EPA'GOO/:? fit)/Oil
t'chrLiary
An Automated Hanitorinj Systca for fish fhpsioloiy and Toilco]oi?
!>/
Ricln.irH ff. CiHsti^ Cl:ii3 Crejrurv J. |r i en
ir.fi. £:;v i r^nm^Tta I Pr^tflot i «:i A. g^ncy
tr.vL ror.me:ital Research Laba ratorj-I1!! I ut h
fc?Cl Oong^r>n 0-ju>rard
DuL utTi. HV E S-fj.a-1
and
Rruct h- llzlmei.
ftir-cricar Scientific .nt-srr.itmnal , Inc.
Uuluth, MS 55301
Environmental Ekaearch laboratory
Office of Eeaearch and De»etopnent
U.S. EnTironnentaJ Protection Agcnej
Duluth, Iff SEA04
-------
TECHNICAL REPORT DATA
„ tfteott rtai Ihi true Horn on the re tent btfort completing!
1. REPORT NO. 2.
EPA/600/3-89/011
3. RECIPIENT'S ACCESSION NO.
W9 15X212M
4. TITLE ANO SUBTITLE
An Automated Monitoring System for Fish Physiology and
Toxicology.
5. REPORT DATE
February 1989
6, PERFORMING ORGANIZATION CODE
7. AUTHORIS)
Richard W. Carlson, Gregory J. Lien and Bruce A.
Holmen
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING OROAN12ATION NAME AND AOORESS
U. S. Environmental Protection Agency
Environmental Research Laboratory-Duluth
6201 Congdon Boulevard
Duluth, MX 55806
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT MO.
12, SPONSORING AGENCY name and address
Office of Research and Development
Environmental Research Laboratory-Duluth
U.S. Environmental Protection Agency
Duluth, MN 55804
13. TYPE OF REPORT AND PERIOD COVERED
1«. SPONSORING AGENCY CODE
EPA-600/03
IS. SUPPLEMENTARY NOTES
16. ABSTRACT
This report describes a data acquisition and control (DAC) system that was
constructed to manage selected physiological measurements and sample control for
aquatic physiology and toxicology. Automated DAC was accomplished with a
microcomputer running menu-driven software developed with an extended BASIC. An
interface module was bui1t that connected standard sensors and controls to the
computer. Digital I/O signals for sample device control and analog signals from
sensors were multiplexed through the interface module. Tine intervals for automated
DAC were user defined, and test data were displayed on a monitor, printed, stored on
disk, and transferred to a minicomputer for analysts. Automated measurements were
made of temperature, ventilation volume, oxygen content of exposure (inspired) and
expired water, and pH of both waters from f our in vi vo rainbow trout (Sal mo
?* i rdneri) preparations. Oxygen uptake efficiency and oxvren consumption we re
calculated. Urine and expired water samples were also collected from all fish.
Non-automated sampling included ventilation frequency, cough frequency, the
electrocardiogram, and aortic blood from an implanted c&nula. Sampled blood was
analyzed for oxygen, carbon dioxide, pH, hematocrit, and hemoglobin. The
respiratory-cardiovascular data gathered with this system were used to define fish
acute toxicity syndromes (FATS) specific to known modes of toxic action.
17. . KIY WORDS AND DOCUMENT ANALYSIS
1. DESCRIPTORS
b.lDENTll ERS/OPEN (NDID TERMS
C. COSATI Field/Group
11. DISTRIBUTION STATEMENT
Release to public
21. NO. OF PAGES
85
20 SECURITY CLASS (Tha pas*!
Unclassified
22, PRICE
SPA F—m 2230.1 IR»». «-77) *««wiou« coition i» oaiouCTt
-------
DISCLAIMER
The i nfora»ti on in this document has been funded wholly by the United States
Envi ronnentai Protection Agency, It has been subjected to the Agency's peer
and administrative review, and it has been approved for publication as an EPA
document. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
-------
A data acquisition and control (DAC) system was constructed that managed
selected physiological measurements and sample control for aquatic physiology
and toxicology. Automated DAC was accomplished with a microcomputer running
menu-driven software developed with an extended BASIC. An interface module
was built that connected standard sensors and controls to the computer.
Digital I/O signals for sample device control and analog signals from sensors
were multiplexed through the interface module, fine intervals for automated
DAC were user defined, and test data were displayed on a monitor, printed,
stored on disk, and transferred to a minicomputer for analysis. Automated
measurements were made of temperature, ventilation volume, oxygen content of
exposure (inspired) and expired water, and pll of both waters from four in
vj_vj» rainbow trout (Salno gai rdneri 1 preparations. Oxygen uptake efficiency
and oxygen consumption were calculated. Urine and expired water samples were
also collected from all fish.
Non-automated sampling included ventilation frequency, cough frequency, the
electrocardiogram, and aortic blood from an implanted canula. Sampled blood
was analyzed for oxygen, carbon dioxide, pH, hematocrit, and hemoglobin. The
respiratory-cardiovascular data gathered with this system were used to define
fish acute toxicity syndromes (FATS) specific to known modes of toxic
act i on.
Key -words: respi rometer, rainbow trout, fish acute toxicity syndromes,
automated system, electrode chamber, ventilation, computer
monitoring, cardiovascular, respi rate ry
-------
CONTENTS
Abstract , . i i i (
Figures vi
Tables vi j i
Acknowledgments i x
1. Introduction I
2. Conclusions ....... 3
3. Recommendat ions 4
4. Materials and Methods 5
Exposure system 5
Automated data acquisition and control 9
Computer system ....... 9
Interface . . . . . 13
Software 29
Main program and tasks 29
(
Subrout i nes 34
Sampling and measurement 42
Water flow . , ',2
Dissolved oxygen ..... 45
pH 48
Terape rature , . . 47
Water 49
Urine ... 55
Data Management 58
Fish acute toxicity syndrome testing GO
Physiological monitoring 61
Test procedures ..... 61
c
i v
-------
CONTENTS {continued t
(
5. Results and Discussion 84
System evaluation 64
FATS testing 6b
References 72
(
(
V
-------
FSCORES
Mather Page
1 Schematic diagram of respi rome ter-metabo i i sm chamber ... 7
2 Vented enclosure containing exposure apparatus and
monitoring components B
3 Block diagram of automated system ... 10
4 Interface module 15
5 Block diagram of analog multiplexing operation 18
6 Schematic of analog multiplexing card 19
7 Detail of one analog multiplexer 20
0 Block diagram of digital multiplexing card 22
0 Detail of the digital multiplexers used to supply bo 11.1 e
selection code for water sampling 24
10 Detail of the digital multiplexer used to send start
pulse for water sampling and contro1 of flow measurement
valves 25
11 Schematic of control circuits for two-way valves
used in flow measurement 2?
12 Schematic of control circuits for toxicant pumps and
urine fraction collector 28
13 Fiow chart of program TEST 30
14 Water flow measurement device 43
15 Block diagram of pre-tcst sensor calibration for
DO, pH, temperature. and water flow 44
(
v i
-------
FIGURES (continued)
f
Number Page.
16 Schematic of temperature amplifier ............ 48
17 Water sampling device .................. 50
10 Schematic of water sampler control board ......... 52
19 Schematic detail of water sampler bottle selection
multiplexer 53
20 Schematic detail of start sample and water level
detect circuit 54
21 Schematic detail of water sampler osci11ator ....... 56
22 Modified urine fraction collector ............ 57
(
(
vii
-------
TABLES
Number tiLCS.
1 Analog and digital I/O channels used for physiological
monitoring and controlling system functions ....... 12
2 Phys i oIo»ica1 variables monitored in rainbow trout to
define the toxic responses associated with ffsh acute
toxicity syndromes ... 62
3 Average deviation from actual value for computer
monitored water flow rate ....... ... 65
4 Average deviation from actual value for computer
monitored dissolved oxygen concentration . ... 66
5 Average deviation from actual value for corrected
computer monitored dissolved oxygen concentration .... 69
v j i i
(
-------
ACKNOWLEDGMENTS
We thank Dan Burrows for his initial design effort on the electronics for
this system. Mike Ostraski developed much of the software and constructed
working model of the interface module; his contributions were considerable
We also thank James McKim for his assistance and numerous discussions duri
the planning stage, and She 11ey He i ntz for typing the manuscript.
-------
SECTION I
INTRODUCTION
Quantitative structure activity relationship (QSAR) models make predictions
about toxicity of chemicals based upon their phys i cochemi ca1 properties, and
it is assumed that all chemica1s that kilt by the same mode of toxic action
can be node led by a single QSAR (Ve i th et al., 1963; 1987 ). Because
alternative QSARs are required for different and specific modes of action,
optimizing the QSAR approach to predictive toxicology requires an additional
systematic effort to define and predict node of action so that the
appropriate QSAR is invoked to make the toxicity prediction for a particular
chemical. Detailed mechanistic studies to understand mode of action at the
molecular !eve I are simply not possible for all chemicals, but one approach
to understanding causal relationships between chemicals and their effects
developed recently and is termed fish acute toxicity syndromes (FATS). These
are collections of direct and indirect measures of effect, or c1inica1 sign-,
manifested in the animal upon exposure to chemicals that are unique and
specific to a common mode of action (McKi¦ et al,, 1987a). Based on a grouj>
of measurable toxic signs involving the respiratory-cardiovascular system in
rainbow trout (Sa1 mo ?airdneri). FATS have been defined for narcotics,
oxidative phosphorylation uncouplers, acetylcholinesterase (AChE) inhibitors,
respiratory membrane irritants, and the pyrethroi d insecticide f envaIe rare
(McKim et al., 1987b,c; Bradbury et al., 1907).
PATS testing requ i red data acqui s i t i on on 11 respiratory-cardiovascular
variables and the capability to monitor more if necessary. Monitoring was
performed manual 1y during all previous FATS tests and consumed the full
attention of at least three people along with the part-time help of several
others during both a seven hour control period and for up to 48-h during the
1
-------
i
acutely lethal exposure period. Only two fish could be prepared and tested
at one tine, and two tests were therefore required to gather sufficient
information on four fish to reliably define a FATS. Although measurements
were made often enough so that statistical relevance was established, a
higher sampling frequency was desirable for greater confidence. Clearly, a
more efficient system had to be developed before extensive F&TS testing could
be accompli shed,
The objective here was to develop an automated system to efficiently quantify
some of the physiological Functions of a whole fish preparation exposed to
acutely lethal chemical concentrations. It was desired that the system
provide for data gathering on at least four fish, and to do so at
predetermined intervals throughout the test including periods of unattended
operation. Another requirement was that it remain flexible enough so that
sampling and measurement regimes could be changed, singly or collectively,
during a test. With these in mind, a system was designed that could be
constructed from commercially ava iI&b1e sensors, control valves, a personal
computer, a specially constructed interface, and menu-driven software to
coordinate system activities. Although this report provides some details on
how the system was constructed, it is not intended to serve as a construct i !:¦,
manual, It should, however. provide enough guidance so that those with
access to some technical resources could design arid build their own custom
system.
2
-------
i
SECTION 2
«
CONCLUSIONS
1. Automated monitoring of respiratory-cardiovascular variables from
fish resulted in a considerable savings of time and effort when compared
to manual data gathering methods.
2. Automated monitoring provided continual data collection during periods
t
of unattended operation, thus ensuring that data were collected during
times when critical changes may have occurred.
3. The real-time sampling and calculation of vital signs permitted
judgments on the course of an experiment,
4. Automated monitoring allowed rapid data collection at shorter time
intervals than manual If possible. A greater number of samples provided
for greater statistical reliability,
5. Data were easily manipulated and transferred between computers because
they were immediately stored in computer files. (
6. Less manual sampling reduced human exposure to potentially hazardous
chemicals.
3
(
-------
I
SECTION 3
RECOMMEND A.T! ONS
1. The automated systen as described here functioned we 11 and proved
suitable for fish acute toxicity syndrome (FATS) testing. Other
applications such as pharmacokinetics would also benefit from automated
monitori ng.
2. More physiological variables should be automated, including ventilation
frequency, cough frequency, and heart rate. This would result in more
detailed information about the physiological state of an animal in real
t ime.
3. Commercially available components should be used wherever possible when
constructing an automated system. This reduces the in-house research dir
development effort necessary to get such a system into operation.
4. Construction costs for this automated system are estimated at 150,000,
4
-------
I
SECTION 4 (
METHODS AND MATERIALS
This system was designed to perform physiological monitoring on rainbow trout
(Sal mo ga i rdneri) that weighed between 0.6 and 1,0 kg and were exposed to a
lethal concentration of an organic chemical. This required the integration
of several subsystems that included; (l) exposure apparatus that provided
water and toxicant delivery; (2) automated sampling and measurement circuits
and devices that provided automatic data collection and sampling of
physiological functions: (3) non-automated circuits and devices that provided
monitoring of those physiological functions that defied automation at this
time; (4) a micrecomputer system that controlled all aspects of automated
monitoring; and (5) an interface that provided all necessary i nterconnect i ouv
and switching between the computer and external devices.
4.1 EXPOSURE SYSTEM f
The exposure system consisted of two stainless steel headboxes that fed Lake
Superior water at 11 ± 1" into s flow-splitting mix cell, four
respi rometer-nsetabol : sm chambers to hold the fish, and metering pumps that
delivered a concentrated aqueous solution of chemical to the mix cell, The
mi * cell was a 18 cm x 24 cm x 2? cm high glass container that had four 18 mm
diameter bored holes in the bottom. Each hole held a neoprenc stopper that
contained a 10 em section cut from the tip of a 2 ml disposable pipette. By
grinding back the tip so that the bore was about 2 mm in diameter, a 800
ml/min water flow was obtained when the water column height within the mix
cell was adjusted to 20 cm. Constant head height was maintained by a float
valve that was supported by a glass cover placed over the mix cell. Exposure
water was delivered at equal and constant rates through teflon-lined tubes ro
the A compartment of each fish chamber, Chemical solution was delivered to
(
5
-------
I
the mix celt by 12 volt DC Model RP-BG75 metering pu»ps (Fluid Metering.
Inc., Oyster Bay, NY).
Except for modifications noted below, design and construction of the fish
chambers used in this system was that of McKim and Goeden. 1902 (Figure !I.
Standpipc position was changed to a bottom rather than side exit in each of
the three compartments of the plexiglass fish chamber. Water overflow from
the A compartment of fish chamber one and from all B compartment standpipes
was directed into the flow measuring devices; other A compartment and alt C
compartment overflow went directly to drain. From ports located on the sides
of the chambers, water was directed-to the different sensing electrodes
without aeration. Other ports were installed for manual water sampling, an-!
one port was used as the exit point in the C compartment for the urine
catheter. Quick-disconnect couplings threaded into the chambers provided
convenient connection of sampling tubes.
The exposure apparatus was contained in a specially constructed vented
enclosure (Figure 2) to minimize human exposure to potentially hazardous
chemicals and to shield the fish from human activity. The framework of the
183 cm long x 91 cm wide x 238 cm high cabinet was constructed of 1.9 cm
thick plywood coated with chemically-resistant epoxy paint. Laboratory
exhaust connected to the top of the enclosure provided a negative air
pressure relative to room air. Sliding glass doors allowed observation and
access to the apparatus, yet maintained sufficient negative pressure within
the enclosure.
6
-------
Water
T«m§s Contra!
Water
Water
Sample
Fraction
Collector
NOTES:
1. Chamber design & surgical preparation (McKIm &
Gosden, 1M2)
2. Transected 800 -1100 gram Ratabow front (Selmo
3. A Clumber (expoeure wafer) monitored on only 1ol4flali
H r%«MhAtr immwJrmd ttfatofl IVa1 VflflUlflllOCI VOluSM
-------
1.
flaproduced from
bait avKilnbt* copy.
8
-------
4.2 AUTOMATED DATA ACQUISITION AND CONTROL
figure 3 shows the overall layout of the data acquisition and control (DAC)
system that performed sampling, measurement, and calculation of selected
physiological functions. Individual components arid their operation are
described below. The DAC system was designed to monitor pH, dissolved oxygen
(DO), temperature, and flow rate of both the incoming and expired water in
which the fish resided during a test. This was accomplished by monitoring
the expired water (B compartment) of up to four fish chambers and the
incoming water (A compartment) of fish chamber one (Figure 1). The A
compartment was designated chamber five, both in illustrations and within the
computer program that controlled system operation, but was referred to as
chamber 1A on the monitor display. Additionally, samples of both waters and
urine fractions from each fish were collected automatically and held for
chemical analysis. A single water sample was taken from the A compartment
whenever any or all of the 0 compartments were so scheduled. Also, whenever
any fish chamber was monitored for pH, DO or temperature, the A compartment
was sampled immediately afterward so that the samples of inspired and expired
water were as close in time as possible. This was necessary because the
calculations for oxygen uptake efficiency (Ug) and oxygen consumption
(VQ2) involved the difference in DO content of both waters at that moment.
4.2.1 Conputer System
The computer was an IBM PC/XT version specially built for Analog devices,
running at 4.77 MHz, and containing an INTEL 8088 microprocessor, 8087 math
coprocessor, 258 KB RAM, and four expansion slots. One of the slots held an
additional 384 KB RAM, boosting total memory to 640 KB. The basic system
-------
SYSTEM CONFIGURATION
pH
Analog
Mum
mmmmm
DO
Temperature A
cEiSSS3
B
CD I ^11 ^ IOO ET
Flow Pressure Sensors
ti
md
mmmmci
Blfital
Mux
k/D
Macsym
DSO
Card
120
Card
ibm rr
Gmpulsr
Printer
VAX
How Solenoid*
mmmmm
Water Sampler Multiplexer
Urin®
Collector
Toxicant
Pumps
2
later Sample Vahrea
71
On
(*>
-------
a
came sypplled with an 83 key sealed membrane keyboard, 130 W power supply,
(
one 5 1/4" floppy diskette drive, a 10 MB fixed disk drive, color/graphics
display adapter, and a combination adapter card, The latter provided a
serial asynchronous communications port, a parallel printer port. battery
backed-up realtime clock, and a thermal warning system to monitor internal
system temperature, A 33 era diagonal, 16 color RGB industrial display with a
protective screen served as the system monitor. The entire computer was
manufactured to withstand the higher temperatures, vibration, and particulate
contaminants found in the industrial environment.
Automated DAC was provided by an MI0120 multifunction board and a DIG120
digital input/output (1/0) board (Analog Devices). The M10120 contained an
input analog section consisting of an analog multiplexer with capabilities
for 32 channeIs of si n*le or pseudo-differential inputs, or 16 channels of
full differential input. The automated system described here used ten (
channels configured as single ended, bipolar analog inputs in the ± 10 vo1t
range. Six of these channels were reserved for individual analog-to-digital
(A/D) conversion on voltage signals coming from the flow sensors, and one
channel each for A/D on pH, D0» compartment A temperature, and compartment C
temperature signal inputs (Table !). Analog signals from the DO, pH, and
temperature meters were in the 0.8 to 1.2 volt range, and in the i to 8 volt
range from the pressure transducers used in flow measurement. These signals
were fed into the A/D converter which had an A/D conversion resolution of 12
bits (4096 counts). A/D on input signals was accomplished with the MACBASIC
command (AIN) used to poll any channel for its current readings; a high speed
burst mode was also aval I able for sampling but was not used.
i
11
-------
Table I. Ana log and digital I/O channels used for physiological
monitoring and controlling system functions in the automated
test system.
Computer
C&£d
Channel
Type
Funct i on
MI 0120
Analog Input
A Temperature
C Temperature
Dissolved oxygen
PH
Chamber I pressure
transduce r
Chamber 2 pressure
transducer
Chamber 3 pressure
transducer
Chamber 4 pressure
transduce r
Chamber 5 pressure
transducer
M10120
MI 0120
D! 0120
8
14
15
21
Digital Input
Digital Output
Digital Output
Fraction collector micros»ireh
Master A channel select 1i ue
Master B channel select 1 i tit?
Master C channel select 1 i n-
Water sample
Water sample
Water sample
Water sample
Water sample
- k set
- A reset
- B set
- B reset
3-way valve
Flow rate 2-way vaive
Start toxicant pump 1
Start toxicant pump 2
Advance fraction col 1ectm
i i
-------
The MI0120 also contained TTL compatible eight bit digital input and output
ports. Each bit represented one channel of digital 1/0; hence, eight
channels of digital input and eight for output were available. Digital
output simply means that, on command, the computer will place close to zero
volts (low) or approximately + 5 volts (high) on a digital output channel.
These voltages were used to control system functions, such as operating
solenoids or motors, by activating or deactivating transistorized circuits
that controlled those devices. Digital input means that the computer senses
whether a particular input channel has no voltage or about + 5 volts
present. This information may come from switches or other devices and was
used to determine what course of action was required to perform some system
function. All eight digital output channels and one input were used on the
MI0120 (Table 1).
The D10120 was a TTL compatible 24 channel (24 bit} digital 1/0 card.
Although the channels we re divided i nto three eight bit groups (ports) that
could be configured for either digital input (readback) or output, each
channel (bit) was independently addressable through software control. The
system only required four channels for digital output (Table I).
4.2.2 Interface
The interface fulfilled three important needs. First, informati on in the
form of analog signals, or varyi ng voltages. from the meters measuring
dissolved oxygen concentration, pH. temperature, and from the pressure
transducers for water flow rate must be read into the computer. The
interface provided the switching and interconnections between those devices
and the correct connection on the A/D card in the computer where the analog
13
-------
signals were converted, or digitized, into the sort of numerical information
that the computer could manipulate to calculate values for those
measurements. Secondly, the computer was required to control the operation
of solenoids and motors. The interface supplied electronic circuitry and
connections so that the solenoids in the two way valves used for flow rate
measurements were activated and deactivated at the proper times, as were the
solenoids for actuating the three-way valves used in taking water samples.
The five-way valve used in taking water samples was a 1 so rotated to the
correct port at the right time, water sampling initiated, and sampling ha 1 ted
when the bottles filled. Additionally, the toxicant pumps were contro11ed
and the urine fraction col Iector was advanced through the interface inter-
connections, Lastly, the interface provi ded power to operate or control the
operation of the different sensors and devices attached to it. An internal
power supply provided + 12, + 15, and -15 voIts DC to the system, while
connections to power supplies externa I to the interface supplied + 5, -5, and
+ 24 volts DC.
The interface was constructed within a standard instrument rack that measured
53 cm wide by 39 cm deep x 45 cm high (Figure 4). Two rack mounting frame
kits (Vector CA52-HP119) for holding plug in boards, or cards, were installed
in the box and 72 pin plugboard receptacles (Vector R636-2) were mounted
behi nd frame kits. Most of the circuitry was built on plug in boards with 72
contacts (Vector 3719-4) that mated with the receptacles on the card frame,
and circuit construction techniques used were wire wrap and solder tack. !C
wire wrap sockets were mounted to the cards with hot me 11 glue and wrap ID
we re used to aid construction. Where necessary, test poi nts were made
accessible to the front of the cards.
14
-------
Ff$ H
(
{
13
-------
The upper card frame he 1d ten receptacles and the lower card frame eight; not
all receptacles held cards but were installed to allow future expansion.
Cards in the upper frame included an analog multiplexing board and six boards
that contained two temperature meter circuits each. The temperature meters
monitored the k compartment of chamber one, C compartment waters of each fish
chamber, the incoming lake water headboz. and a chilled water bath for the
fraction collector, or seven temperatures altogether. One board was a
spare. The lower frame he 1d a digital multiplexer, a valve driver board that
contained the circuitry used to ope rate the solenoids in the two-way flow
measurement valves and to start the toxicant pumps, and six boards that
contained the circuits used to control water sampling from each of the B
compartments and the one k compartment; a spare board contained the same
circuitry. Three ribbon cables. two 50 conductor and one 34 conductor,
connected the interface to the computer. Two of these connected the digital
multiplexing board, one going to the D10120 computer card and the other
terminating on the digital I/O section of the M10120 card. The third cable
connected the analog multiplexer to the A/D section of the M10120.
In effect, the analog multiplexer served as a four pole, five position
switch. The computer was required to read the pH, DO, and two temperatures
(four poles) from five di fferent locations (five positions).
The usual sequence was to begin with chamber one and read 00 and pH from the
expired water B compartment, the headbox and fraction collector temperatures.
and pH, DO, and temperature from the A compartment on chamber one. The
1atter was referred to as e i the r chamber five or chamber 1A. This sequence
was repeated for the remaining chambers two through four, skipping any
deactivated chambers.
18
-------
Each of these analog signals could have been fed directly to individual
channels on the A/D card but that would have tied up 17 of the 32 available
channels; seven for temperatures and five each for DO and pH. Combined with
the five A/D channels reserved for reading the pressure transducers used in
flow measurement, few channels remained for future expansion. Hence. it was
decided to multiplex the above signals into four A/D channels (Table l).
Figure 5 shows the multiplexing scheme for one of the four required
measurements. The CD4051, an eight channel analog multiplexer integrated
circuit, was the workhorse af thi s circuit, and there was one CD4051 for DO,
one for pH, and one for each of the two temperature circuits {Figure 6).
Although eight positions were available on each CD4051. this application
requi red only six: one for each chamber and a spare. LM348 quad operational
amplifiers were used to buffer, or isolate, the CD405I inputs and outputs
from other circuit components. The CD4051 switched from one chamber to ^
another whenever the computer program transmitted the binary code
representing the new chamber over the chamber select lines (Figure 5). These
lines were designated Master A, B, and C and originated from the digital I/O
section of the MI012D computer card (Table 1), but also passed through
inverters on the digital multiplexing board. Because the chamber select
lines were common to all the CD4051's, a 11 of the sensors from one chamber
became available simultaneously to the computer whenever the CD405I's
swi tched. The multiplexer circuit for one CD4051 and associated LM348"s is
detailed in Figure 7. The inputs to the circuit were for connecting the
meter output of the sensor for each chamber and a spare.
The digital multiplexing board perforated several critical functions
i nc1udi ng:
(
17
-------
CHRMBER i
SENSOR
- A
SENSOR
RHPL IFIER
I
LM348
CHRHBER 2
(SENSOR
A
SENSOR
AMPLIFIER
I
LM348
SELECT
LINES
C B A
S&LECTEB
0 0 0
1
0 0 1
2
0 1 0
3
0 1 1
4
1 0 0
5
CHRHBER 3
SENSOR
SENSOR
AMPLIFIER
1
LH348
—
CD4051 t
A
LH348
A —¦
R/D CRRO
CHRMBER 4
(SENSOR
A
SENSOR
AMPLIFIER
I
CHAMBER 5
IsensorI
SENSOR
AMPLIFIER
LH348
X
LM348
CHAMBER
SELECT
L INES
OIO CARD
COMPUTER
-------
R-f
—w I Nv 1
, 1 J> •-
/fT^
8-19 >
fi-29 >-
S-3I>-
n-94>-
11-44 >—
fl-i 1 >
B-21 >
8-91 >
H-4I >
8-99 >
*-4l>
R-4 >
fl-9 >
It-1 >
"^t>
US
rp
> i
P|CD.«ggt
i (M-7
OXYGEN
£04031
p H
ftll-9
B-8 >-
ft- SI V
"^t>
n-a»>-
8-96 >-
«-ss y-
~^t>-
8~»i >
»-!?>
8-27 >
8-97 >
n-42 y
(ItB
rm
E
nn-a
fi TEHP
C04MI
Ull
C04II1
hr-6
B TEMP
71
q-'
-------
ft-11 >
n-21 >
fi-3 1 >
fl-4 1 >
ft - 36 >
R - 4 6 >
fl-4 ^
fl-6 >
fl - 8 >
CHAH9IR 2
CD 4 0 51
iOOK
CHIMCR 5
LM34#
1€KJ#€
CHWCT ft
> Rfl-9
MULTIPLEXER OUTPUT
TO fl/D CRRD (H I 0120)
CHAMBER SELECT BUS
-------
Switched to and connected the correct water sampler control board for
a chamber into the circuit when the computer called for water
sampling from that chamber.
Supplied the binary code to the water sampler board to select and
fill correct sample bottle (1-4).
Supplied the start pulse to begin water sampling.
Switched to and activated the proper solenoid valve used for making
a flow measurement.
Operated one or both toxicant pumps when required.
Advanced the fraction collector when required.
The latter two were not mul tiplexi rig operations, hut required digital output
to activate the motor control circuits for those devices.
Whenever the computer program called for a water sample, the inverted binary
code representing the desired chamber was passed to the digital multiplexer
over the chamber select, or Master A, i, and C, lines. These entered the
multiplexer through CD4049 chips (111) used to invert the signals and isolate
the circuit (Figure 0}. The control lines were common to three C34097
di fferential. eight channel rcul t i plexer/demultiplexer integrated circuit
chips designated U5, U8, and U7; the lines were also common to the analog
multiplexer. U5» U6, and U7 all had grounded inhibit lines so they were
always enabled. The master control lines were connected to the A, B, and C
inputs of each CD4097, and the correct binary code switched them to the
desired channel, or chamber. The first two CP409?'s, U5 and US, had flip-
flops built from CD4001 quad, two-input NOR gates connected to the output of
each channel, and each combination of one CD4097 and five flip-flops were
designated Series k and Series B, Once the chamber was selected on the
21
-------
semes n
semes e
CD4C97
CC
9
11
ua
CC
Ul
>—¦O*
>—*£>*-
y——*
CD4C4S
W5
U1I
0
—^ 3
U12
¦4 J
C04I97
UI9
CC
1*
JJ2
> »{>*¦
IS
>11 C04OI
UI 4
U15
—^ IS
• to
US
0
I—» 4
U13
U19
—> 12
f|)—-{>-—
r ¦¦> t
"'"JlBi!
==-~21
U14
UIS
i®
>a«
U7
STMT
>3MHPLC
ruute
now
> HESSUfiEKSMT
1 W»T VRLVCS
-> •
It
-> 14
TO RNftLOS
MUL
:t BUS
~F1
Oq
-------
CD4097's, an inverted set or reset signal was placed on pin l or pin 17,
respectively (Figure 9). This set or reset information was fed through U5
and U8 and controlled the state of the flip-flops connected to the selected
channel outputs. The combination of the k output and B output from the
selected channel, paired flip-flops formed the binary code (00-11, decimal
0-3) required to select the correct sample bottle (1-4) for filling. This
information was sent to the correct water sampler control board.
For instance, if the computer called for a water sample to be deposited into
bottle two on chamber one, the program would firs: set the chamber select
lines so that binary 000 was received by U5, U6, and U7, switching them to
channel one. A reset signal was then applied to pin 17 of the Series A
CD40S7 (U5) resulting in an output of logical zero on pin four (connector
three) of the Series A flip-flop III I. Simultaneously, a set signal applied
to pin one of 116 resulted in an output of logical one on pin 11 {connector
four) of the Series 8 flip-flop Oil. This combination of outputs, AB =
binary 01, represented sample bottle number two, and was sent to the water
sampler board for chamber one.
The A and B Series of flip-Hops and independent sets of AB lines from thera
to each water sampler control board were required to Iatch the code for
bottle selection, Cncc latched, the bottle filling procedure could begin and
continue uninterrupted while the computer went on to different functions,
which may have included activating another water sampler. Once the chamber
was selected and the bottle selection latched for water sampling, an inverted
start sample pulse was applied to pin one of the third CD4097 (U7) at the
moment sampling was to start {Figure 10). Because 117 was also switched to
-------
cixeai
CMBOt
cc
3 >
5 >
7 >
CD4097
C04041
0
3
SERIES P
CD 4 097
CC
23 >
JO«
15
CHAMBER SELECT BUS
SERIES B
• 0
-> 6
-> 10
-> 14
TO RNRLOG
MULTIPLEXER
-------
CD 4097
> 13 >SAMPLE
>17 i PULSE
17 >
BB v
31 >
U 7
> 28
„ . i I FLOW
>29 > MEASUREMENT
> 30 2 WRY
> 31
CC
3
TO ANALOG
MULTIPLEXER
> 10
5
7
CHAMBER SELECT BUS
-------
the same channel as U5 and U6, the start sample pulse was fed to the
corresponding water sampler board where it set a latch to begin the bottle
f i11i ng procedure.
The thi rd CD4097 (U7) also provided the switching necessary to select the
proper solenoid v&lve used in flow measurement. After channel selection by
the chamber select lines, an inverted signal placed on pin one was fed
through U? to the corresponding chamber circuit on the valve driver board
(Figure 8). Here the + 5 volt signal was passed to the NPN transistors
2N4123 and TIP29 connected as a Dar1ington pair (Figure 11). This caused the
TIP29 to conduct and closed the normally open solenoid valve. This circuit
had to renain active throughout a flow measurement.
The valve driver board also contained the circuits necessary to control
toxicant pump operation and the fraction col lector. Control lines for these
circuits originated on the DI0120 card (Table 1). An on/off switch 1ocated
on the front pane 1 of the interface (Figure 4) provided manual controIs for
each toxicant pump or placed them under computer control. In either case
they could be operated singly, concurrently, or alternately. A request from
the computer program to activate either pump brought the input line BB19 or
B817 to a logic zero (Figure 12). This signal was inverted by U3, a 4049
chip, and passed to a Dar1i ngton pai r where the hi gh output from U3 allowed
the TIP29 transistor to conduct and the toxicant pump to run.
Although part of the interface, circuit descriptions for the temperature
meters. urine fraction collector, and water sampler control boards are
deferred to the sampling and measurement section where they are explained
along with their associated hardware.
26
-------
M WtTS 0»
P
5 f
UM14S
smiis
r
->72 >
nwm J_
¦> 6 >
* 24 VOLT* Q>-
W41«
m*aa$
->72 >
7 f
¦) 8 )-
~ 34 VOLTS ®»
Wi#
->72 >
9 f
TtT-W"
««• I3_
-> ib>
* 24 VOLTS 0 » ) 72 )-
imimlk
11
^12>
4.7 K
UP»
~ M VOLTS 9~
smtm
mmm
->72 >
13 14 >
t UMV SOLXNOSD
% M»Y MLSHOSB
a met mjmm
s mr sm,smm
3 MY MUSMt
27
-------
V
A
CD
u
28
-------
4.2,3 Software
The MACSYM 120 included Concurrent CP/M-86 as its operating system, a '
multi-tasking system that managed the I/O of all devices attached to the
computer, provided file management, and loaded and ran the operational
program for DAC. This program was coded in measurement and contro1 BASIC
(MACBASIC), an expanded version of BASIC that had been optimized for realtime
measurement and control. MACBASIC features included spec i a I key words to
designate common measurement and control functions, a line-by-line compiler
that directly translated program statements into code as they were entered,
and multi-tasking. The latter permitted several program operations, or
"tasks," to be performed concurrently and independently of each other.
The operational DAC program was written in-house and named "TEST." TEST
contained 1083 Sines of source code and required 55 KB of memory for the
undocumented source code or 48 KB for compiled object code. However, TEST ^
was not optimized to save space or to run faster; a more judicious use of
code would shorten the program considerably. TEST consisted of a short main
program to begin and di rect program execution, two tasks running
concurrently, and 25 subroutines that performed a 11 the functions required by
the main and task portions of the program. The flowchart (Figure 13} shows
the main program and task interrelationship, but the reader is referred to
the subroutines Ii sted below for a detailed explanation of their operation as
they are called by the main and tasks.
4.2.3.1 Main Program and Tasks
After the fish had been prepared and placed in their chambers, the source
code for TEST was entered and updated with the current coefficients used in
the pH, DO, temperature. and flow algorithms; more wilt be said about these
I
29
-------
main program
START-UP
FIRST
SAMPLE
TASKS
o /START
\tiue
TIMING k CONTROL
^ /wOgMOtK
N™* /
tocksxJE
CLEAR
NO
SIT
FLAGS
FLAG
INTERRUPT
(K^D
DISPLAY
SUSPENO
TASK 2
PROGRAM
CHANGE
MENU
IASK 2
(W)
-------
later. Once TEST was started, the main program made immediate calls to
STARTUP and SETSTIM, then waited until the start of monitoring; normally,
this was 0500 h the next morning. When monitoring began, calls were made to
the routines PHDOTMP and FLOW to perform the first sampling and to PRNTSTOR
to save the results. The display monitor was set up to show sampling times,
intervals, and results by a call to MOM TOR, after which tasks one and two
were activated and the main program was exited.
Task two is described first since it assumed control of program execution
once it became active. It contained the timing sequence that kept track of
what and when to sample and whether or not stock bottles became empty when
the toxicant pumps were running. The timing sequence was simply a loop in
which, for fish chambers one through four, a flag was set whenever it was
time to sample pH, DO, or temperature, make a flow measurement, advance the
fraction col lector, or take a water sample. The sequence within the loop was
as follows; First, a timer variable was updated to the current total time in
seconds that had elapsed since the start of the main program. A toxicant
pump flag was checked to see if pumps were running; if so, the total liters
remaining in the stock bott1e(s) were determined by calculating how much had
been used based on pumping rate and elapsed time of pump operation. If total
liters remaining was < one, a call was made to TOXPMP to either switch pumps
along with stock bottles or to stop the pump.
Upon return to the timing loop the interval variables for the fraction
collector, pH, 00, temperature, flow, and water sampling were checked to see
if it was time to sample. This was done by subtracting the last sampling
time for each variable from the current total time. If the difference
31
-------
was > to the initialized sampling interval, a flag was set and the timing
loop was completed for that particular fish chamber and then the loop was
cheeked again for all fish chambers to determine whether or not any other
sampling flags should be set for that time period. Also, when it was
determined that a flag should be set to signal sampling, the current elapsed
total time became the new last sampling time for that particular variable.
Because the program was occasionally interrupted for some reason and sampling
was not executed at a specified time, a factor was calculated and applied to
the last sample time update so that upon return to program execution,
sampling would still occur at specified times. For example, we preferred to
sample DO, pH. and temperature every 15 r»i n on the quarter hour, advance the
fraction collector every one or two hours on the hour, and take water samples
every four hours on the hour. If the program was interrupted at, for
instance, 10 min before the hour and restarted 20 min after the hour pH, DO,
and temperature sampling would occur immediately upon restart because the
elapsed time since last sampling exceeded the i ni tiali zed 15 rain interval
and, without applying the factor, sampling would reoccur at 15 min intervals
from that point. However, the readj usted last sample time using the
calculated factor for missed sample times would insure that sampling would
occur on the quarter hour as desired.
The timing loop for chambers one through four was executed continuously until
a flag was set to signal sampling. Once a 11 flags for a particular sampling
time was set, a jump was made from the timing loop into a sampling loop.
Here, for fish chambers one through four, calls were made to the proper
subroutines to perform sampling, print, store, and display the results.
Separate loops for timing and sampling were necessary if it was desired to do
-------
these operations in sequential order. If calls were made directly from the
timing loop to sampling subroutines when it was time to sample a particular
chamber, a larger number of temporary variables and increased code would be
necessary to rearrange the data into order since it could not be predicted
which fish chamber would be sampled first. Because time was tracked in
Tractions of a second, it sometimes happened that the timing loop would sense
a sampling time when the loop was somewhere other than on chamber one, and,
without separate loops, results were often printed and stored in some order
other than 1-4.
When all operations were completed for that sampling period, the necessary
variables for data collection and handling were re-initialized to their
before sampling states, and a jamp was made out of the sampling loop and back
into the timing loop. Task two continued program execution in this manner
until interrupted and TEST was stopped.
Task one was a program interrupt that waited until it sensed a lower case "i"
input at the keyboard. At that time it suspended execution of task two and
displayed the interrupt menu below:
1.
Set or change sample times
2.
Advance fraction collector
3,
find of file entry
4.
Deactivate chamber
5,
Start or stop toxicant pump
8.
Continue test
7.
End test
8.
Activate printer
9.
Toxicant pump operation
33
-------
The task would wait until a selection was made from the menu, performed that
choice by a jump to the proper subroutine, and redisplayed the menu until
continue test or end test was selected. On continue test, MONITOR was called
to update the display and program control was re 1i nqui shed by reactivating
task two. End test halted TEST and exited the program.
4.3.1.2 Subroutines
Listed alphabetically and not in the order in which they were called or their
order of occurrence within program TEST.
1. ACPRNT - set a flag that signaled the printer was on-line.
2. ADVANCFC - advanced fraction collector one sampling position.
ADVANCFC called STEPFC three times to accomplish this.
3. DEACT - queried user for fish chamber number and deactivated all
sampling for that chamber by resetting the sample time intervals
for pH, DO, temperature, flow, and water samples to 600,000 s.
This time period was longer than any foreseeable test duration.
4. DIG10UT - set digital output Ii nes to select the proper fish
chamber for sampling. Chamber number was passed to this formal
subroutine as a parameter; each chamber number was specified by
binary code and the code was sent out through the Master A. B,
and C chamber select lines to the digital and analog
multip1exers.
5. DSKERR - MACBASIC had error trapping routines that diverted
program execution to another part of the program or a subroutine
when certain run time errors were detected. One of these was
"channel timeout." Program execution would hang for a specified
time when an attempt to perform some operation over an internal
34
-------
a
I/O channel could not be completed because of a hardware or ^
software problem, and the program crashed when the default amount
of time (2 m) had passed. DSKERR was a subroutine that was
called when an attempt to write to disk failed for any reason.
and would display a message on the monitor, reset & flag to
signal the disk was off-line, and returned control to the next
1i ne in the program.
6. ENDTST - was called only from menus. Queried user whether or not
test was to be ended, or returned control to program or task that
was running. If TEST was to end. ENDTST called EXIT.
7. EOF - some information about the fish and test was entered for
printout and storage at the end of the test file. Fish number,
sex, stage of maturity, and termination times were entered after
prompts. Dose, depuration, redose, and total exposure times were
calculated by calls to T1MCALC, and the times where 25 and 75 (
percent exposure occurred we re calculated f or each fish. All
information was stored on disk and a formatted output was printed
for each fish. This subroutine was normally called from the
interrupt task menu and only when a test was terminated.
0. EXIT - closed a 11 files and I/O channe1s. Ended program and
exited to MACBASIC.
9. PILEHDR - called from menu displayed by subroutine STARTUP.
Information entered for the file header included test number.
chemical used in test, toxicant code, date (month, day) that a I I
fish piaced in their chambers, fish number, the time (hour,
minute) that each fish was placed in chamber, and fish weight.
(
35
-------
Information was stored and printed. A1so printed was a header
containing column labels.
FLOW - this subroutine, as did all other sampling subroutines,
first called DIGIOUT to set the digital 1/0 (Master A, B, and CI
lines for the chamber being sampled, and noted the time from the
system clock (hour, minute) for use in monitor display update
after sampling was completed. FLOW closed the solenoid on the
flow measuring device and waited one second to allow head
pressure to begin building on pressure transducer. Before the
flow rate was calculated. A/D conversion of the voltage signal on
the pressure transducer output was required. This voltage
increased as the water level within the flow device increased.
A/D was performed in a loop that cycled from one to up to 60
times. During each passage through the loop, A/D was done on 50
samples from the pressure transducer in less than one second;
these were averaged and converted to a digital whole number.
When this number exceeded 8.0 volts the water level in the flow
device had reached its maximum allowable head height and a jump
was made out of the A/D sampling 1oop. This occurred whenever
the flow rate was greater than approximately 200 ml/min.
Otherwise, FLOW waited unti1 one second had elapsed, including
A/D sampling, and then recycled. If the entire loop was cycled
60 times, then the flow rate was less than 200 ml/min and A/D
sampling had lasted 60 seconds. Thi s ensured that sampling was
not prolonged during periods of low flow rate. Once A/D
conversion was completed, the f1ow solenoid was opened and flow
diverted to drain. FLOW then calculated the volts/min that
36
-------
the pressure transducer output changed, inserted this value into
the appropriate linear equation for the chamber being sampled,
and calculated flow rate. A print Flag was set and the
subrouti ne exited.
11. HBFCTMP - called from PHDOTMP every time that subroutine was
called. Temperature of the incoming lake water within the main
headbox and of the cooling water flowing through the fraction
collector bath were each obtained by performing 50 A/D
conversions on their meter outputs, averaging the mv outputs, and
using these values in linear equations.
12. INITFC - called from the STARTUP menu to move the fraction
collector into its starting pos i t i on. Advances were made one
step at a time by calling STEPFC. Also used before and after a
test to load and unload the fraction col lector.
13. INTVLCALC - called from MONITOR to calculate when the next
sample times would occur based on last sampling time and present
time interval for any particular variable being sampled. Set up
temporary variables for times and called INTVLTIM to make the
caIculati ons.
14. INTVLTIM - a formal subroutine that had the hour and minute of
the last sample along with the sampling interval (in minutes)
passed to it, calculated the hour and minute for the next sample,
and passed that information back to INTVLCALC.
15. MONITOR - the monitor display was updated after every sampling
period to show the results of that sampling, time of sampling,
the sample interval, and the time of next sampling for pH, DO,
temperature. flow, water sampling on all four B compartments and
the A compartment of chamber one. Also, fraction col lector
advance times were displayed. MONITOR first called
-------
INTVLCALC to calculate next sample times. Formatting of the
display was done and all information was sent to the monitor.
16. PHDOTMP - called DIGIOUT to set digital 1/0 lines for chamber
being monitored. Obtained current time (hour, minute) from
system clock. Performed A/D on signals from the pH, DO, and
temperature meter outputs by taking 1000 samples in rapid
succession for each measurement. The average voltage value was
inserted into the appropriate Ii near equation and pH, DO, or
temperature calculated, Except when monitoring chamber 1A,
PHDOTMP also called HBFCTMP to determine temperatures in the main
headbox and fraction collector bath. Before returning, PHDOTMP
set the print flag.
17. PRNTMAL - this subroutine used the same error trapping routine
as DSKERR to display an error message on the monitor. set a flag
to signal that the printer was off line, and continued program
execution in the event that the printer malfunctioned and caused
an I/O channel timeout.
18. PRNTSTOR - obtained system date and time. Called TIMCALC to
calculate the elapsed time between start of program and the last
sample time. Kept track of the total number of data records
stored on the hard disk. Data were written to disk if the disk
write flag was set and printed if the pri nte r flag had been set.
PRNTSTOR calculated VOg and Uj before printing since only raw
data values we re stored on disk.
19. SAMlNTVL - provided a monitor display of the current settings for
PH. DO, temperature, flow, fraction collector, and water sampling
time intervals. A change menu was also displayed and any or ail
time interva1s for sampling on any fish chamber or compartment A
38
-------
i
could be changed. Although all times were displayed or changed
as minutes, the program converted all sample intervals to seconds
for timing considerations. Called from either STARTUP or
interrupt menus.
20. SETST1M - set the starting time for program execution. If start
hours = 0, returned to main program immediately. Otherwise, the
routine sat in a loop checking the clock until starting time
arrived. Simultaneously, an interrupt task was activated that
served as an interrupt to the waiting period. Upon sensing an
Interrupt by any input on keyboard, a new start time was entered
and the waiting task react i vated. Returned to main program as
soon as starting time arrived.
21. STARTUP - this was the main subroutine for starting program
execution and was the first call made from the main program. It
initialized all variables including dimension statements for all ^
arrays, flags, default sample times, logic levels on the digital
I/O lines, and format statements for monitor display of sampled
data. After Initialization, user was queried whether output
should go to printer and hard disk, file names to be used, and
whether or not any chambers should be deactivated. Lastly, a
menu was displayed that contained options for other pre-test
operations and included;
1.
Enter file header information
2,
Enter EOF
3.
Initialize fraction collector
4.
Set sample intervals
5.
Set toxicant pumps
6.
Start test
7.
Exit program
38
-------
After choices were performed by calls to the proper subroutines,
STARTUP was exited by choosing to start test.
22. STEPFC - when called, advanced fraction collector one step at a
time. This subroutine was used to position the fraction
collector to its starting place or for loading and unloading
centrifuge tubes that collected the urine samples. Advancing the
fraction collector required monitoring mieroswitch action to
determine whether or not the advance was completed. A safety
feature was included so that if the f ract i on col lector continued
to advance for more than one second, longer than required to move
one step, an error message was printed and displayed, a fail flag
set, and the subroutine reset. As soon as the fraction collector
was successfully advanced, or after error handling, a return was
made to origin of call.
23. TIMCALC - formal subroutine that accepted two different times
(month, day, hour, minute) and calculated the difference in
hours.
24. TOXPMP - consisted of five distinct sections. The first part.
called from STARTUP, initialized the pumps by prompting the user
for pump flow rate (ml/min) and stock bottle capacity (liters)
and calculated the time required to empty the stock. Also,
options inc1uded operating two toxicant pumps, e i ther
simultaneously or alternating; default condition was one pump
operating, Second, a monitor display of current pump settings
and operational mode was called from task one, the interrupt
task. If changes were required these could be specified. Third,
when a call to start the toxicant pump(s) was made from the
40
-------
interrupt task menu, the timer clock was polled for elapsed time
and this became the starting point for calculating the time
required to empty the stock bottle(s). The system clock provided
the dose start time (month, day, hour. minute). Fourth, when &
cal1 was made from the interrupt task menu to stop the toxicant
pump{s), the dose stop time was obtained from the system clock
(month, day, hour, minute). This was used later along with dose
start time to calculate total dose times. Fifth, a call from the
timing loop in task two occurred when it was time to switch from
an empty stock bottle to a full one. The program continued to
switch from one stock bottle to an alternate when empty time
arri ved so long as the toxicant pumps were initial I zed to the
alternating mode. 1n this way stock solutions of toxicant were
continuously available to the exposure system.
WATSAMP - after calling DIG1OUT to select the chamber and
obtaining sample time, WATSAMP set the I/O lines on the digital
multiplexer to move the five-way valve to correct port, closed
the three-way solenoid valve on the drain line so that chamber
water was diverted into the sample bottle, and updated the
variable keeping track of sample port. Water sampling was done
in sequence 1-4 and a notation was made on the printer output
when a sample had been taken on a particular port so that sample
times were accurately tracked. Once a water sample was started.
the program could go on to other sampling without having to wait
until the sample bottle had filled; an independent ci rcuit was
responsible for unlatching the water sample circuit. The printer
flag was set before returning to call origin.
41
-------
4,2.4 Sampling and Hensurencnt
4.2,4,1 Water Flow
Ventilation volume (Vq) was the amount of water a fish pumped over its
gills in a unit of time and was measured directly from the expired water B
compartment overflow drain. The expired water drained into a flow
measurement device constructed frotn clear, rigid, series R-4000 PVC pipe
(Exceion), a model 180PC pressure transducer (Micro Switch 0-10" HgO), and
a two-way, normally open solenoid (Va1 cor 16P8408-5). The device was
configured similar to a trap found in any drain system (Figure 14). Expired
water flowed down a 2.54 cm diameter pipe that was reduced to 1.27 cm
diameter at the top of the trap. Tygon tubing connected the pressure
transducer to a nipple mounted in the pipe and water left the other side of
the trap at a point just above the level of the transducer orifice thereby
maintaining about two cm head pressure on the sensor.
During a flow measurement the computer selected and closed the normally open
solenoid valve and the pressure exerted on the transducer rose along with the
water level as the pipe filled. The computer digitized the increasing
voltage output from the transducer as explained in the FLOW subroutine and
calculated flow rate based on a volts/nun change in output. The regression
equations used in the FLOW subroutine were established for each chamber
before each test by regressing four or five measured water flows against the
rate of change in transducer output voltage (Figure 15), The resulting
coefficients for slope (m) and intercept {b) were inserted into the FLOW
algorithm before a test, and ventilation volume was calculated as ¥g
(ml/kin) = mx + b where x = volt/mitt. A flow device was also placed on
chamber one, compartment k overflow; this monitored only the difference in
flow rate between incoming water and expired water for chamber one.
42
-------
Reproduced from
b»ti available copy
'mmS
tt>.££*V*7Ji : i> *. -*•¦ .v;
43
-------
SENSOR CALIBRATION
m
A/D
Voltage
1
UL
Kaotra
Values
Flow
Sensor
—
A/D
Min
VolU/Uin
Notes:
1. A separate software program is used for calibration.
2. Linear response is assumed.
3. Known values include: a. 0 & saturated for D.O.
b. pH 7.0 k 10.0 buffers for pH
c. measured flows between 0-600 for flow
-------
4.2.4.2 Dissolved Oxygen
Determination of oxygen uptake efficiency (Ug) and oxygen consumption
(VOg) by the fish required the measurement of the DO content of both the
incoming water and the expired water. Sample ports in the A and B
compartments of each fish chamber were connected to needle valves (Clippard.
Model MFL-2) that regulated water flow into flow-through electrode cells to
approximately 10 ml/min. Sampled water was reconnected to the flow
measurement device to maintain accurate Vg readings. The electrode cells
( Beck/nan 572-934) he 1 d standard poIarographi c oxygen sensors (Model 39557,
Beckman Instruments, Inc., Fullerton, CA). The electrodes were connected to
Becknan Model 0260 oxygen analyzers.
During sampling, the computer selected the chamber through the analog
multiplexer and digitized the oxygen analyzer output. The subroutine PHDOTMP
calculated DO content by using the correct regression equation. Linear
equations were established for each DO sensor and meter by regressing
corresponding voltages against two known values, zero and saturated. The
zero value for DO was obtained by shorting out the electrode output, and the
saturated value of incoming water was determined by the mod i f i ed Winkler
titration method. The slope (m) and intercept (b) of this 1i ne were used as
coefficients in the PHDOTMP subroutine algorithm and DO was calculated as DO
s mx + b where x = volts output by the DO meter. The coefficients were
determined and inserted into the subroutine before each test,
45
-------
The difference in DO content between incoming and expired water was the
for that fish at the time of sampling and was expressed as & percentage of
the incoming DO. The PHDOTMP subroutine calculated oxygen consumption
(V02) as:
vo2 = no x uE x vG
wt
Where:
VOg = mg/g/min
DO = oxygen content of incoming water (rag/L)
Ug = uptake efficiency (percentage)
Vq - ventilation volume (ml/min)
Wt = fish weight (grams)
Values for VOg were converted and expressed as mg/kg/h when used in data
analysis.
4.2.4.3 pH
These measurements were made similarly to those for DO. Incoming and expired
waters were directed through need 1e valves to combination pH electrodes
(Sensorex 450C) mounted in flow-through cells (Sensorex G12255}. Rater
passing through the cells was redi rected to the flow measurement device. The
pH electrodes were connected to Beckman 3550 pH meters. During sampling, the
computer selected the chamber through the analog multiplexer and digitized
the pH meter output. The subroutine PHDOTMP calculated pH by using the
correct linear equation for that sensor and meter. The coefficients for
these equations were established by regressing corresponding voltages against
known pH values. These values we re obtained by using buffer solutions of j>B
7.0 and 10.0 as recommended by the manufacturer to bracket predicted values
for the test water; pH of Lake Superior water was close to 8.0. The slope
and intercept from this regression equat i on were used as coefficients in the
46
-------
PHDOTMP subroutine and were inserted just before a test. pH was calculated
as pH = mx + b where x = vo1ts output by the pH meter.
4.2.4.4 Temperature
Seven system temperatures were monitored; all C compartments, the A
compartment of chamber one, the incoming lake water headbox, and the fraction
collector cooling bath. YSI thermistor, type 403, stainless steel
temperature sensors {YeI low Springs instrument Co.. Inc., Yellow Springs.
Ohio) were inserted directly into the compartments being monitored, and
custom built ci rcuits amplified the signals f rom each sensor. There were two
temperature circuits on each of six boards in the interface, and each board
contained one LM324 quad operational amplifier to support both circuits.
Each circuit used a 50 k ohm resistor in series with the YSI sensor to form a
simple voltage divider (Figure 18). Because the YSI thermistor had a
negative temperature coefficient, resistance decreased as temperature
increased. With the YSI sensor connected to the bottom of the voltage
divider, the op amp input sensed a voltage decrease whenever temperature
increased. The first stage of the op amp provided an approximate gain of 16
while the second stage was set for unity gain and acted as a buffer stage.
All of the capacitors were added for filtering. The 10 k ohm potentiometer
set the offset voltage during calibration, and the 47 k ohm resistors
connected to the potentiometer provided a more precise adjustment. A test
point was supplied to measure the temperature ci rcui t output during
calibration. Card level calibration was accomplished by inserting a 7500 ohm
resistor in place of the YSI sensor and adjusting circuit output to an
arbitrary + 500 nv.
47
-------
p i£.. IL*
r~
(
v
(
48
-------
Temperature sampling and calculation were the same as for DO and pH.
Pre-test coefficients were obtained by regress i ng voltage outputs against
four or five known temperatures proximal to the expected test temperature.
Coefficients were determined for each temperature meter and inserted into the
PHDOTMP subroutine. Temperature was then calculated as TMP = mx + b where
x = voltage output of the temperature meter.
4.2.4.5 Water
Chemical analysis of incoming exposure water was required for most testing
done with this system. Normally, water sampling was done manually from
chamber sampling ports. However, an automated sampling device was necessary
if samples were needed during periods of unattended operation, and water
samplers were built and attached to the drain line coming from each flow
measurement device to meet this contingency. A 24 volt DC normally open,
three-way valve (Peter Paul Electronics, Model 76 Z 00310GM) was installed in
the line so that, ordinarily, water flowed to drain but when switched by the
computer drain water was diverted into a sample bottle.
The water sampler was constructed with a five-way valve (Hoke Inc., Creski11.
NJ, Model 7841GGY) attached to a 24 volt DC actuator (Hoke, Model 0172L2P)
and four one-liter glass bottles (Figure 17). Each bottle was fitted with a
neoprene stopper that had two stainless steel wire electrodes inserted
through it, a vent hole, and a stainless steel fill tube. The electrodes
extended down into the bottIe and were part of a water 1 eve 1 detect circuit
that stopped water f1ow into the bottle when the desired amount was
col 1ected.
49
-------
/
f
r?
50
-------
The interface contained a separate water sampler control board for each
sampler unit that contro11ed sample bottle selection, start of sampling, and
sample ha 11 when the bottle filled. Each board was connected to the digital
multiplexer by a separate set of control lines that were. by convention,
labeled A. B and SET or connectors 10, 14, and 6, respectively {Figure 18).
Independent control lines were necessary so that the computer would not have
to wait until one sample was complete before starting another. When the
computer program selected a particular water sampler to activate, the digital
multiplexer placed the binary code for bottle selection on the AB tines as
explained previously. These lines were connected to three CD4052 analog
multiplexer/demultiplexer integrated circuits labeled U2» U4» and U6, and the
AB binary code switched then to the correct channel for that sample bott1e
(Figure 18). The output of U6 was fed into a Darlington pai r causing the
TIP29 transistor to conduct and drive the corresponding coil in the five-way
valve actuator (Figure 19). This rotated the five-way valve to the correct
port and connected the drain line to the sample bottle.
Once the five-way valve was in position, the water sampler control board
received a start sample pulse from the digital multiplexer over the set line
(connector six). Thi s pulse was accepted by U5, a bi stable multivibrator
constructed from CD4001 quad, two-input NOR fates and designed to serve as
both a water level detect flip-flop and a latch to start and hoid the bottle
filling procedure (Figure 20). The start sample pulse set the f1ip-flop and
the output from pin four on U5 was applied to another Dar1ington pair,
causing the three-way drain valve to switch from drain to the collect
position. This diverted dra1n water into the sample bott1e. The other half
of the flip-flop was connected to the inhibit pins of U2 and U4 and with the
f1ip-f1 op set, a logic low was placed on the inhibit pins, thus enabling both
51
-------
n, ri ri n n
*»** to » MA m * ran
«Xl^r-6
1
3 *_ » *> ¦* *>
U _ H |* * Hi £
¦ '¦l^gpll I
il^ LI, l_i Li
&
r r j~ r
IpuMMbmi Asg^uElkHari^
SiT-Jl S
fn
:^r
JHs
L_$ ®
t 3
1-1
52
-------
CD 405 2
TER SAMPLE BOTTLE
SELECT LINES
10 >
1 4 >— —
J.H4140 I
ak
72
<36 4
2M4X2S n TXPM
*1
z—r
BOTTLE ONE
5 WRY
+ 24
tM41» 1 Tint
M41U 1 TlfM
72
<40 4
"h
z r
BOTTLE TWO
5 WRY
72
3
UMt«a
<44 4
+ 24
»
/* t o Z'
/ Z \L
BOTTLE THREE
5 WRY
m
vs^-
< 48
jBOTTLE FOUR
15 WRY
SM41ZB 1 tirz*
-------
OSCILUtrOH
ml
X xo
XI.
¥ XS
MS
HO
Yl
V*
•*»
im
10 >
14 >
MM K
« VOLTS O o ~ * VOLTS
a MC«
m k
tp a
Z . X
h!
[
J
WATER LEVEL
DETECT CIRCUIT
SAMPLE BOTTLE
+ 24
BflR. 6 >
•OTTLK SKUCCT LINKS
xo
X
XI
xa
V
x«
*o
VI
3
XMH
ft
B
4.T K
«T K
<-» o
/ z
32
ttPWt
DRAIN
LINE
3 WAY
VALVE
Tl
fS
o
-------
U2 and U4. At this point the f1i p-fI op was latched, the water samp)e
started, and the computer free to go on to other functions.
When the AB lines switched U2 and U4, the water 1 eve 1 detect lines coming
from the electrodes for the correct sample bottle were connected into the
circuit (Figure 18). When 112 and U4 were enabled by the start sample pulse,
a free running oscillator (Figure 21) fed its signal through U2 and. so long
as the bottle was not filled with water, an AC signal was applied to only one
electrode. When the water level reached both electrodes the low level AC
signal was passed to the detector circuit where the signal was ampli f i ed by
U3, an LM324 op anp. The amplified signal passed through U4 to pin six of 1'5
and cleared the water level detect flip-flop allowing the inhibit lines to
become active. This inhibited the operation of U2 and U4, shutting down the
AC signal going to the selected bottle electrodes, and removed the detected
signal f rom the f1i p-f i op. It was necessary to stop the signal to the ^
flip-flop so that the water sample control circuit could become enabled on
the next request for a sample.
4.2.4.6 Urine
The fish urinary catheter was connected to a hypodermic needle which In turn
was connected to a port in the C compartment. On the outside another
hypodermic needle connected the port to capillary tubing which drained the
urine into centrifuge tubes placed in a fabricated plexiglass carrousel
mounted on a modified fraction col lector (Instrumentation Specialties Co..
Lincoln, NE). The fraction col Iector {Figure 22) was originally designed to
operate with 45 positions for small vials, but only 15 of the larger
centrifuge tubes per fish could be placed in the custom holder. This meant
that the fraction collector had to be advanced three times for each sample
(
55
-------
* • VM.1S
UiB
22 N
JU*
rl>M>
juic
CP*
10 >
14 >
-JJl
IQ K
lOO tC
to K
3H412I
1 K
m.
X XQ
X
V XX
X9
p)
VI I I -
I 1
VI 9
I MM I .1
M k*|—
e 1—*—
B VOLT*
tMfER MHPkC 0OTTL« SELKCT I.XMES
TO |W»|.f MTIUI
MATSR UCVfiO. ELECTR6CMES
FROM mm MATER
LEVSt. DSTKCTeS
rt.ip n.«f
31
-------
n*produc«d from
bW tvjjtobfr eepy.
Ft# . 2 Z.
(
S7
-------
position. The computer accomplished this by monitoring the internal,
cam-operated mi c roswi tch that closed each time the collector moved into one
of the 45 original positions. This was monitored over the digital input
channel zero of the MI0120 card (Table 1) and three microswitch closures were
counted for each new sample pos ition. The 110 volt &C motor that rotated the
collector was Isolated from the computer by a solid state relay (Figure 12).
The computer placed a low on channel 21 of the DI0130 which was connected to
BBS of the valve driver board (Figure 1Z). Stage C of the 4049 inverter (U3)
inverted the low voltage and in turn activated a 2N4123 transistor. This
closed the solid state re I ay and the f ract i on collector motor ran. Norma!ly,
the fraction col 1ector was advanced once every one or two hours. The urine
samples were maintained about four degrees C by cooling water flowing through
a water bath within the carrousel, and were collected daily for storage in a
freezer until analysis.
4 .2. ft Data Management
After each measurement or sampling operation the computer monitor screen was
updated to show the results for that time interval, a continuous hard copy on
a Decwri ter 3 printer {Digital Equipment Corp., Maynard, Mass . ) was appended.
and all data were appended to a file residing on the hard disk.
Some of the detailed information about these operations was given above in
the software section on subroutines. The printed output consisted of a file
header containing information about the fish and test, a header with 18
column labels foil owed by the data output, and an end-of-f i1e area reserved
5fl
-------
for information about dose times for each fish. The co 1 um,n labels and the
data they represented i nc1uded:
1. DATE - the month and day that the sample was taken.
2. FISH - fish number.
3. 5NUM - each fish had their own set of staple numbers.
Every time a sample or measurement was performed, SNjM was
increased by one,
4. ETIMi - the elapsed tine in hours that the fish had been in the
chamber.
5. ApH - pH of incoming exposure water.
8. BpH - pH of expired water,
7. ADO - dissolved oxygen content in tag/L of the incoming exposure
water,
8. BDG - dissolved oxygen content in rng/L of expired water.
9. 02UP% - percent oxygen extracted by the fish from the inspired
water. Also called uptake efficiency (Ug).
10. VOg - oxygen consumed by the fish expressed as mg/g/min.
11. AFLOW - flow rate (ml/min) of water leaving the A compartment
drain of fish chamber number one.
12. BF10W - flow rate (ml/»in) of expired water leaving B
compartment,
13. ATEMP - temperature of incoming water.
14. CTEMP - temperature of C compartment water.
15. WATS ~ first column; no sample = 0, water sample = i. Second
column: water sampler port number I 1-4).
16. URNS - number of fraction collector advances.
59
-------
Only raw data we re stored on the hard disk file; hence, the disk file
contained the DO results shown for ADO and BDO but not VOg or 02UP%.
However, the disk file also contained a record of the headbox and fraction
collector temperatures; these were not shown on the printer output due to
lack of space.
Information in the hard disk file was stored under a CCPM format, but the
MACSYM 120 operating system had the capability to transform CCPM files into
the DOS format. The DOS file was written to floppy disk, and transferred via
modem to a VAX minicomputer (Digital Equipment Corp. ) where appropriate
information was extracted and merged with other files in the FATS database.
4.2.6 Fish Acute Toxicity Syndroms (FATS) Testing
Thi s system was constructed primarily to perform the required sampling and
variable measurement to fulfi11 FATS testing, and the description of FATS
testing included here was only to demonstrate how the system was used in our
research program and how it behaved during testing. Methods and procedures
were as described in Bradbury et a I. (1988); a brief description is provided
here but will not include toxicant preparation and analysis, water
characteristics, or data analysis. However, some non-automated sampling and
measurement procedures were included since these were an integral part of the
overall system, and some of these may be automated in the future. Also.
Bradbury et a 1. ( 1988) defined a new FATS for a group of chemicals termed
po1ar narcotics; the five chemicals tested represented about a third of the
test runs completed with the system to date. Although some system functions
were not used for those tests, the following description includes all that
the system was designed to perform.
-------
4.2.6.1 Physiological Monitoring
During each FATS experiment measurements were made on the physiological
variables shown In table two. Vg, YQ«>, and Ug were moni to red
automaticaliy while the remainder were done manually. Ventilatory frequency
(fv) and cough frequency (fc) were determined from portions of
strip-chart recordings made of the trout ventilatory patterns. These were
monitored from non-contact stainless steel wire electrodes placed in the B
and C compartments of each fish chamber. Each electrode pair was connected
via patch cords to a j unct ion box mounted above the fish chamber and from
there shielded two-conductor cable led to high gain capacity-coupled
preamplifiers set to a time constant of 0.3. Frequencies above 30 Hz were
filtered from the signal as it passed through channel amplifiers and the
ventilatory signals were recorded on a physiograph recti1inear strip-chart
recorder (Narco Biosystems. Houston, TX). EKG electrodes were connected
through to the recorders similarly to record heart beat (f^), but EKC
recording also required a third electrode to he Ip reduce electrical noise;
this was inserted into the dorsal muscle mass of the trout and connected to
ground potential on the junction box.
4.2.6.2 Test Procedures
Spi na1ly-transected rainbow trout were surgically prepared as described by
McKim et a I. {1987). Each fish was f i tted with a latex rubbe r membrane that
separated expired water from incoming water, a dorsal aortic cannula for
blood sampling, copper wire electrodes for monitoring the EKG, and a urinary
catheter. Four trout weighing between 0.6 and 1.0 kg were exposed to each
chemical, or were used as controls. After surgery the fish were placed in
individual respi rometer chambers, the electrode connect i ons made, and the
urinary catheter was connected to the C compartment port.
61
-------
Table 2. Physiological variables monitored in rainbow trout to define
the toxic responses associated with fish acute toxicity
~syndromes (FATS).
Vari able Units
Ventilation Volume
no./mi n
Cough Frequency
(fc)
no./min
Heart Frequency
(
no./rain
Total Blood Oxygen (arterial)
{TaOg)
g/100 mL
Total Blood Carbon Dioxide (arterial)
(TaCOg)
mmoI/L
Blood pH (arteria! )
( pHa)
pH units
Hematocrit
I Hct)
%
Heraog1 obi n
(Hb}
g/100 mL
62
-------
Program TEST was entered and the starting time for automated monitoring was
set. Physiological monitoring was started between 18-20 hours after the fish
had been placed in their chambers, usually 0500 h the next morning. Predose
values for aft physiological variables we re obtained during the control
period between 0500 and 1200 h. The computer measured and calculated Vq,
VOg, and Ug every 15 min while fy, f q, and f^ were determined from
recordings made every 30-60 min. One set of blood chemistry variables was
collected during the predose period, The fraction collector was advanced
hourly and, if needed, automatic water sampling was activated.
Toxicant delivery was started at 1200 h by interrupting TEST and selecting
the proper routine. The fraction collector interval was set to two hours at
this time but the sampling schedule for other variables remained the same fcr
the test duration. When needed, the water samplers were activated and set
for 4 h intervals. During periods of unattended operation, the physi ograph
recorders ran continuously. Blood chemistry variables were measured in
arter i al blood samples taken from each aortic canula two to five times during
an exposure. Periodically, water samples were taken from each fish chamber B
compartments and the A compartment of chamber one and analyzed for DO content
by the Winkler titration method to provide a check on automated measurements.
Temperatures were also checked. Whenever a fish died, that chamber was
deactivated for monitoring, and the test was terminated whenever all fish
died or a 48 h exposure was completed.
63
-------
SECTION 5
RESULTS AND DISCUSSION
5.1 SYSTEM EVALUATION
To date 17 tests involving 68 fish have been completed using the system.
Testing included three freshwater control runs, two control tests on a
carrier solvent used to aid dissolution of some test chemicals, and 13 tests
with organic chemicals used in describing fish acute toxicity syndromes
(FATS). In a sense. al1 of these tests constituted evaluation and validation
of the stated objectives for the system, although only the first three runs,
two controls and a previously tested chemical, were used to certify the
system as ready. Collectively, the results from these tests showed that the
system performed as designed despite some sporadic electronic malfunctions
and problems with sensor calibration during some tests.
11 seldom occurred that the computer monitored values were exactly the
same as the actual value measured for DO, pH, temperature, or ventilation
volume (Vq). Tables three and four show the average percentage deviation
from actual value at those times where simultaneous sampling and measurement
were done manually. Computer monitored values taken from the flow
measurement pressure transducers for Vq were consistently close to actual
value, generally within five percent (Table 3}. The only exception to this
was flow senior 4B during tests four and five in which case the deviation was
17 percent high and 11 percent low, respectively. For 17 tests overall, the
deviation for absolute values showed that sensor 2B performed the best, with
a mean deviation of 2.3 percent. despite the fact that during test five it
was off by 11 percent.
The average deviation for DO measurements was considerably more erratic,
ranging from a -54 percent for sensor IB to + 61 percent for 3B (Table 4).
One recognizable problem that occurred during DO monitoring was air bubbles
64
-------
Table 3. Average deviation (percentage} from actual value for computer
monitored water flow rate.
Flow Sensor
Test Ik LB M M 11
1
a
-9
a
a
a
2
a
a
a
a
a
3
a
a
4
4
9
4
a
0
4
3
1?
5
1
-1
11
-7
-11
6
4
0
0
3
3
?
S
3
-3
-6
4
8
B
6
0
4
a
9
0
3
1
2
-2
10
0
6
1
-3
4
11
3
3
0
2
4
12
-4
-9
-1
-3
-1
13
2
0
0
3
1
14
4
0
-1
2
-1
15
-1
2
-2
6
3
16
5
3
-2
1
1
1?
1
-3
-4
4
2
8 - Aetna 1 values not recorded.
65
-------
Table 4. Average deviation {percentage) frosi actual value for computer
monitored dissolved oxygen (DO) concentration.
DO Senior
Test
1A
IB
28
3B
48
i
-3
-7
-1
-5
-7
2
1
-6
-7
-10
-7
3
-1
-6
-3
-9
1
4
3
-2
-B
3
-4
5
4
2
2
-9
6
6
1
•7
6
3
-t
?
a
-10
2
-5
— s
a
5
-8
14
-5
4
9
0
-11
-6
-4
-5
10
2
-3
0
-11
-10
ii
2
-2 ,
11
61
¦>
ft-
12
-24
-14
17
-5
27
13
a
-54
18
-37
-8
14
a
0
10
-24
b
15
-10
-4
11
15
-20
16
-11
2
5
16
-37
1?
-14
-5
9
-12
-14
a - Electrode malfunction: DO monitored manually
^ - Chamber not used for this test.
66
-------
getting trapped in the electrode cell holders when saturated water degassed
as it flowed past the electrodes. Thi s was an ongoing problem with chamber
1A since the electrode there always measured DO concentration of the incoming
water; it was less of a problem in the other electrode cells where fish were
extracting oxygen from the test water before it flowed into the electrode
cell. Lightly tapping the electrode cell or forcing water through them with
a syringe bulb dislodged the air bubbles and alleviated the problem. Also,
the electrode cell hoider frow chamber 1A was replaced with a redesigned
electrode hoider constructed so that the water flowed Into the bottom of the
cell and f1 owed vertically upwards past the DO electrode rather than
horizontally through the cell.
For the stated objectives, a + 5 percent deviation was within acceptable
limits, but when computer monitored values varied by more than about two
percent, a separate computer program was used after a test was over to
readjust monitored values obtained for those particular sensors to correspond
more closely with actual values. For example, DO electrode and meter output
p
was checked periodically by collecting water samples simultaneous to
automated monitoring and DO concentration was determined using the Winkler
titration method. If the monitored results from a particular electrode and
meter showed continual variance compared to the actual DO concentration then
it was assured that the pre-test calibration was incorrect or that DO
electrode performance was faulty and the automated measurements were
accordingly in error. Usually, the actual DO concentration was consistently
higher or lower than the automated result by a certain amount, but sometimes
the variance would gradually increase either high or low as the test
continued i ndi cat i ng that the digitized output was drifting. In either case.
ratios of actual DO concentrations/computer monitored concentrations were
67
-------
regressed against the corresponding times in hours from the beginning of the
test that the actual DO samples were taken. The slope (m) and the intercept
(b) were used as coefficients in the linear equation y = mx + b where x =
elapsed time in hours from the test beginning and y = a correction factor
used to estimate the actual DO concentration for any test time. The
recalculation program read the old data file record by record, recalculated
each DO value as DO = y times (monitored DO), and rewrote all data into a new
data file in exactly the same format as the old. Table 5 shows the average
deviation (percentage) from actual values after DO concentrations were
corrected by the recalculation program.
Errant values for water flow measurements by any one pressure transducer were
handled similarly except that actual flows versus computer monitored flows
were regressed from data sets obtained just before and after a test was
performed; flows were not measured manually during a test. Temperature and
pH values were not readjusted, although a hard copy record was desirable to
ensure that they were close to expected values during periods of unattended
operati on.
5.2 FATS TESTING
Using strictly manual data gathering methods, McKira et al., 1987b,c, defined
FATS associated with narcos i s-i nduc i ng chemicals, oxidative phosphorylation
uncouplers, acetylcholinesterase inhibitors, and respiratory membrane
irritants. The first chemical tested with the automated system,
2,4-dinitrophenol, was tn the original group of uncouplers, and the results
obtai ned using the automated system were cons i stent with those obtained
manually. This verified that the automated system was suitable for FATS
testing as well as showing that the responses used to define a FATS we re
reproducible.
68
-------
Table 5. Average deviation (percentage) from actual value for corrected
computer monitored dissolved oxygen (DO) concentration.
DO Sensor
Te st
Ik
IB
ZB
3B
4B
t
0
0
0
0
0
3
0
0
0
0
2
3
0
0
I
0
6
i
0
0
0
0
0
5
0
0
0
0
0
6
4
5
6
3
1
7
0
0
2
0
a
8
3
7
16
4
i i
9
0
0
0
0
0
ID
0
0
2
0
0
11
1
6
0
0
0
12
0
2
3
5
0
13
a
8
4
3
5
14
3
0
0
0
b
15
0
0
0
1
0
16
1
8
2
4
7
17
1
4
12
11
5
a - Electrode malfunction; DO monitored manually,
k - Chamber not used for this test.
69
-------
The main advantage gained by using the automated system for FATS testing was
that four fish could be monitored simultaneously for the i r ca rd i o-re sp i ratory
responses. Thi s was possible because some of the labor intensive manual
methods were simply replaced by the system, saving considerable time and
effort. When done by the automated system, measurements of Vq and DO were
rendered effortless and it became possible to monitor more than two fish per
test. This at 1 east doubled the number of FATS tests that could be done in
the same time frame.
Also, round-the-clock monitoring was now possible and this greatly increased
the number of measurements done for Vq, Ug, and 07. This ensured data
gathering throughout a test for every fish and increased the confidence that
data were not missed for periods of critical change.
Another advantage to using the automated system was that certain judgements
concerning the course of an experiment could be made while it was in
progress. For instance, it is character Istic of narcosis-inducing chemicals
that their effects on an organism are reversible even at the point of
apparent death, usually def i ned as resp i ratory arrest in aquatic toxicology,
whereas effects induced by chemicals with more specific nodes of toxic action
are irreversible. It was necessary to prove that phenol was a narcotic in
the sense that Its effects were reversible before a new FATS could be defined
for the suspected polar narcotics (Bradbury et a 1.. 1988). By following Vg
and VOg on the computer printout as well as locomotor activity,
vent iI at i on, and the EKG on those recordings, the fish could be revived at
various stages of intoxication with toxicant-free water and recovery
70
-------
monitored. Primary toxic responses by phenol we re found to be reversible,
even when the fish were brought to the point of complete cessation of
venti]ati on.
All of the automated system features were not required for FATS testing.
Because most tests ended within 24 hours, automated water sampling for
chemical analysis was seldom necessary, Also, urine analysi s was not
performed since the only variables used in defining FATS at this time were
associated with cardio-respiratory functions. Hence, the f ract i on collector
was idled during FATS testing. However, both of these features would be
useful for certain pharmacokinetic studies where chemical analysis of
incoming and expired waters is critical to understanding uptake by the gills
and where metabolites present in the urine would shed light on internal
physiological and metabolic processes. Other types of physiological testing
may require additional automated features, blood pressure or the
electroencephalogram for example. The system was built to allow for
expansion should such measurements become necessary.
There are some functions that should be automated to make FATS testing more
efficient. More time and effort could be saved if fy, f^, and f^ were
automated. Manual counts and data input for these would be eliminated, and
it would be useful to observe these in real time. Other physiological
processes simply defied automation at this time such as blood sampling and
ana lysis.
71
-------
REFERENCES
Bradbury, S.P.. J.M. McKim and J , R. Coats. 1987. Physiological Response of
Rainbow Trout f Salmo gai rdneri) to Acute Fenvalerate Intoxication.
Pestic. Biochem. Physiol. 27: 275.
Bradbury. S.P., T. R. Henry, G.J. Nieroi, R.W. Carlson and V.M. Snarski.
1988. Use of Respiratory-Cardiovascular Responses of Rainbow Trout
(Sa1 mo gai rdneri) in Identifying Acute Toxicity Syndromes in Fish: Part
3. Polar Narcotics. M.S. in review.
McKim, J.M. and H.M. Goedcn. 1982. A Direct Measure of the Uptake
Efficiency of a Xenobi otic Chemical Across the Gills of Brook Trout
(Sa1velinus fontinali s) Under Normoxi c and Hypoxic Condi t ions. Coup.
Biochem. Physiol. 72C: 65.
McKim, J.M. , S.P. Bradbury and G.J. Niemi. 1987a, Fish Acute Toxicity
Syndromes and Their Use in the QSAR Approach to Hazard Assessment.
Environ. Health Perspect. 71: 171.
McKim, J.M., P. K. Schmi eder, R.W. Carl son and E.P. Hunt. 1987b. Use of
Respiratory-Cardiovascular Responses of Rainbow Trout fSa1 mo gairdneri)
in Identifying Acute Toxicity Syndromes in Fish: Part 1.
Pentachlorophcnot. 2,4-Dinitrophenol, Tricaine Methanesulfonate and
1-Octanol. Environ. Toxicol. Chem, 6: 295.
McKim, J.M., P.K. Schmi eder, G.J. Ni emi, R.W. Carl son. and T.R. Henry.
1907c. Use of Respiratory-Cardiovascular Responses of Rainbow Trout
(Salito gairdneri) in Identifying Acute Toxicity Syndromes in Fish: Part
2. Maiathion, Carbaryi, Acrolein and Benza!dehyde. Environ. Toxicol.
Chem. 6: 313.
72
-------
Veith, C. D. , D.J. Call and L.T. Brooke. 1983. Structure-Toxi c i ty
Relationships for the Fathead Minnow, PLmeohaIes promelas: Narcotic
Industrial Chemicals. Can. J. Fish. Aquat. Sci. 40: 743.
Veith, G.D. and S.J. Broderius. 1987. Structure-Toxicity Relationships for
Industrial Chemicals Causing Type(II) Narcosis Syndrome. In: K.t.E.
Kaiser ( ed. ), QSAR in Environmental Tox i cology-II. D. Reidel Publishing
Company, Dordrecht. Holland, p. 385.
73
-------
FIGURES
1. Schematic diagram of respi rometer-metabol i sm chamber.
Connections for ventilatory pattern and EKG are not
shown.
2. Vented enclosure containing exposure apparatus and
monitoring components.
3. Block diagram of the automated system,
4. Interface module containing multi plexers, temperature meters,
control circuits, and system interconnections.
5. Block diagram of the analog multiplexing operation for one
type of senior. Binary code for chamber selection is
shown within the inset.
6. Schematic of the analog multiplexing card. Master control lines
for chamber selection are A-4. -8, and -0. Outputs from each
multiplexer, AA-3, -5» -7 and -9, are fed to the analog-to-iigital
converter.
7. Detail of analog multiplexer for one sensor. Multiplexer is a
CD4051 IC chip.
8. Digital multiplexer card. Connectors CC-3, -5, and -7 are the
master contro1 lines for chamber selection.
9. Detail of the digital multiplexers that supply bottle selection code
to water sampler control circuit. The paired output from the A and
B series flip-flops provide the binary number for the correct sample
bottle {i-4) on the selected chamber.
10. Detail of digital multiplexer used to send start sample pulse to
water sampler control circuit and for control of two-way valves used
in water flow measurement.
11. Schematic of contro1 circuits for two-way, normally open solenoid
valves used in flow measurement. Located on interface valve driver
board.
12. Schematic of the control circuits for toxicant pumps and uri ne
fraction col lector.
13. Flow chart of software program TEST used to control all automated
system functions.
74
-------
14. Water flow measurement device. Two-way valve was normally open;
when closed, rising water level In tube at 1 eft increased pressure
on transducer connected to lower part of tube.
15. Block diagram of pre-test sensor calibration used to supply
coefficients (slope and intercept) to linear equations in program
TEST.
16. Schematic of one temperature meter ci rcuit.
17. Water sampler consisting of a five-way val ve with actuator and four
sample bottles. Three-way valve in upper part diverted water from
drain line into one of the sample bottles,
16. Schematic of water sampler control board. There was one control
board for each sampler in the system.
10. Schematic detail of the bottle selection CD4052 multiplexer. Sample
bottle select lines received the KB binary code for bottle number
from the digital multiplexing board.
20. Schematic detail of the start sample and water level detect
circuit. Start sample pulse was fed through 4001 f1ip-flop.
Detector circuit sensed oscillator signal when bottle filled and
shorted electrodes. Detected signal cleared fIip-flop and inhibited
CD4052 multiplexers.
21. Schematic detail of oscillator used to provide low level ac signal
to water 1evel detect electrode in sample bottle.
22. Modified fraction collector for urine collection. Centrifuge tubes
extend down into cooling bath within the fabricated carrousel.
75
(
------- |