.4023 i August 1964

Automatic Feed-processing Control

With
Electronic

Weighing

TEXAS A8<M UNIVERSITY

TEXAS AGRICULTURAL EXPERIMENT STATION
R. E. PATTERSON, DIRECTOR, COLLEGE STATION, TEXAS

Contents

\1\103>->r-l>>4>¢-PQ9U909I\'J

Summary
Introduction
Procedure
Weight Control
Sequence Control
Event Control
Prototype Control Unit
Testing the System
Results
References

Appendix

(5 u_ a‘;
 at "

Summary

 

Development of an automatic feed-processin

  
 
  

control system using automatic weight proportio“;
ing is the result of research conducted by the Di
partment of Agricultural Engineering. '

A prototype unit was designed and construct:
incorporating the following control sections A
actions: (1) weight control, (2) sequence control an.
(3) event control. Reliability tests made with thé
unit show that commercial electronic componen F
can be adapted and assembled into an electron”
weighing system which will enable an operator t
program and process a predetermined ration with

out continuous supervision.

  
  
  

A schematic diagram of the control system V_
shown in this report. The actual operation of th_
control sequence is explained in detail. u

 
 

E

Acomplished in the spring of 1961.

Automatic Feed-processing Control
With Electronic Weighing

Ralph J. McGinty and Price Hobgood*

T1112 NEED FOR MECHANIZING feed-processing o-pera-
tions has been well established. The main ob-
ject is t0 free labor from time-consuming tasks. Less
obvious, perhaps, is the possibility of operating small
capacity equipment around the clock and saving
on dollars invested and power costs (1, 2,  The
continued decrease in the number of agricultural
workers and the increase in the size of farms demand
a change to more efficient labor-saving methods of
production.

Work has been done to automate specific pieces
of equipment, such as an auto-matic load regulator for
a feed grinder (4). Also, there have been several
tests and reports on volumetric feed meters for small
automatic mills (5, 6, 7).

As feed-processing operations become more com-
plex it becomes necessary to develop control sys-
tems in which the various components are controlled
as a unit. In a well-designed system each compon-
ent must perfonn a given job with all parts work-
ing together at a specific signal. Most automatic
farm-size feed-processing systems now in existence
use volumetric proportioning which is limited to
free-flowing materials. Many of the larger commer-
cial-type installations use weight proportioning, but
it is usually done manually rather than automatically.
Since weight proportioning is preferable to volume
proportioning, there is a need for an automatic:
feed-processing system using weight proportioning.

A complete search of the literature, as well as
correspondence with leading manufacturers, was ac-
The conclusion
was that no known commercial automatic weight
controller was suitable for producer use, though
the PTL load system came close to the desired spec-
ifications. Therefore, work was initiated to develop
an automatic feed-processing system using automatic
weight proportioning. To keep the system practical,
thePunit was to be designed and built from commer-
cially available components. The unit also was to
have a process control program which was flexible
rather than designed for a specific job.

Procedure
A feed-processing operation usually follows a
denifite sequence of events. However, occasionally
an optional event may be included or excluded.

*Respectively, former instructor and professor and head,
Department of Agricultural Engineering. '

Using weight proportioning with small-capacity equip-
ment requires repeating the sequence of events many
times. Any given event can be controlled by multi-
ple factors which determine if that event should be
included or excluded for that cycle. The stated
conditions determine that the control action must
consist of three parts: (l) weight control, (2) se-
quence control and  event control. The three
parts or control actions were developed separately
and then incorporated in a prototype unit.

Weight Control

The weight control appeared to be the most
critical part and was developed first. Since the Bald-

win-Lima-Hamilton PTL load system came closeg

to the required specifications for the weight con-
troller, a unit was purchased for study. The unit
consisted of a voltage-regulated amplifier and con-
trol relay. In operation a constant voltage was
supplied to a strain gage bridge circuit. The un-
balanced voltage, proportional to the load, was
fed to a two-stage voltage amplifier. The output
voltage from the first amplifier was fed to a second
two-stage amplifier with a plate circuit relay in the
final stage. Two potentiometers in the cathode cir-
cuit of the first stage determined the opening and
closing points for the relay. One of these potentio-
meters was used to balance the unit at no load and
the other to control the weight at which the relay
tripped.

In the unit purchased, two problems developed.
First, the weight-control potentiometer proved to
be the limiting factor in accuracy because of its small
range of movement from no load to full load.
Second, there was no provision for more than one
controlled weight, thus the unit had to be changed
manually for each change in weight. Separate am-
plifiers could have been used for each ingredient,
but the cost would be prohibitive.

To eliminate these two problems the weight-
control potentiometer was removed from the cir-
cuit and a bank of ten-turn precision potentiometers
was used in its place. A separate potentiometer was
used for each ingredient. Each potentiometer in
turn was switched into the circuit in place of the
original weight control. The additional range of
movement from no load to full load greatly in-
creased the accuracy with which the desired weight
could be set. By using a separate potentiometer for
each ingredient, it could be automatically switched

3

Figure 1. Prototype control unit. Front row left to right:
Baldwin strain gage amplifier, modified controller with
four set-point dials and manual select switch, power supply.
Back row left to right: event control relays, time delay
relays, automatic electric stepping switch.

into the circuit when needed; thus, eliminating
manual adjustment for each cycle. Relatively high
wattage po-tentiometers were chosen t0 reduce warmup
drift due t0 switching cold potentiometers into the
circuit. Tests showed that the resistance differences
among the potentiometers were negligible and any
one of the potentiometers could be used to
calibrate the system. Only four weight-control
potentiometers were used in the prototype unit;
however more could be added if needed. The
only limitatio-n to the number of potentiometers
would be the size o-f the switch needed to switch
them into and out of the circuit.

With these modifications, it was possible to set
accurately the weight-control potentiometer to 0.1
percent of the full-scale value or to 1 pound in 1,000.
The modified Baldwin-Lima-Hamilton PTL system
appeared to meet the necessary specifications for
accuracy and reliability, provided a precision linear
load cell was used.

Sequence Control

An automatic electric type 45, five-bank, 50-
position stepping switch was selected for sequence
control. The switch was hermetically sealed, cap-
able of stepping at the rate of 75 steps per second,
with a rated life of 10 million operations. One
switch bank was used to provide a path fo-r step-
ping pulses, two banks were needed to switch the
set-point potentiometers, one bank was for the oper-
ation sequence control and the fifth bank was a
spare. With 50 positions, up to 50 sequenced events
could be controlled. Switches with more or fewer
positions could be purchased if needed. Unused
positions can be bypassed at approximately 75 per

4

second. The extra positions add flexibility to
unit.

   
  
 
  
  
 
 
 
  
  
 
 
 
 
  
 
  
 
  
  
 
  
 
  
  
 
 
 
  
 
  
  
 
  
 
   
  
 
 
  
  
 
 
  
 
 
  

Event Control

Each event controller must be individually I
signed as each event is unique. One event mil
be the weighing of an ingredient, another the 
ing or conveying of grain and ‘still another the 
ing of ingredients. In the simplest case the evf
controller could consist of a relay controlled by 
step-switch. The relay in turn could control a mot
In another case, a time delay might be required
starting or stopping a motor, a time interval
quired for mixing or a pause o-f one auger uni
another auger catches up. Satisfactory circuits f.
the mentioned cases and for many others have lo
been in existence. Thus, standard relays, timi
time delays, pressure switches and photocells 
control each type of event can be selected accordi
to the specific need. a

Prototype Control Unit

A prototype unit was designed and construct
incorporating the three control sections, Figure
The stepping switch provided sequencing contr‘ ~
The weight co-ntroller was activated by the 
switch whenever an event consisted of weighing  A
ingredient. The step-switch also selected the oil-l 
weight-control potentiometer for each ingredie
Each event control circuit was activated in turn  I
the step-switch. I

At this stage of development of the automa
process control system, no attempt was made ~
optimize the actual design or size of the circui
The three control sections and their common 
supply were b-uilt on separate chassis and connec A.
by cables. Extra compo-nents were included 
insure flexibility. All components were deliberat,
oversized to increase reliability. Small pilot rel
were used between the step-switch and power p;
cuits to insure long switch life. A low-voltage 
lated power supply was used for safety. Both 
volts AC and 24 volts DC were provided in r
to experiment with various available AC and I’
relays. Suitable switches were included so that 
system could be operated manually or automatical
Connections to the step-switch were made from t 
inal strips which provided simplicity in changil
the sequence program. Sequence order could p
changed, events added or dropped in a matterl
minutes simply by changing some connections. Ti
feature greatly increased the flexibility of the systif '

Testing the System

Initial laboratory tests on weighing water fl ‘
ing into a barrel indicated the weight-control '
was satisfactory. To test the entire system it 
necessary to weigh batches of grain repeatedly w

 

testing for (1) absolute accuracy, (2) repeatability
and  reliability. The first two items depend
mainly 0n the weight-control unit while the third
depends upon all components in the system. The
three items can be tested by virtually any type of
processing program since the main requirements are
weighing and repetition. Many factors which would
be important in an actual process would have no
bearing on the performance of the control system
itself. For example, the thoroughness of mixing
would depend upon the design of the mixer used and
not upon the relay which turned the mixer on and
off. Since only the control system was being tested,
many simplifications could be made for convenience.

Since many repetitions were to be run, it was
desirable to use the same grain many times; this
ruled out grinding and mixing operations. Unground
sorghum grain was used to simplify conveying. Be-
cause of limited space only two ingredients, both un-
ground sorghum grain, were used. For this test,
it was immaterial what actually took place during
the interval between weighing batches, so a hypo-
thetical sequence was simulated, Table 1.

In the simulated system a weigh tank, consist-
ing of a lfi-cubic-foot ho-pper-bottom barrel, was
suspended from a steel pipe frame, Figures 2 and 3.
A Baldwin-Lima-Hamilton type TXX precision
linear loa(l cell with a capacity of 1,000 pounds
formed a link in the suspension system. The load
cell was electrically connected to the amplifier. Two
storage bins, similar in design to the weigh tank,
were mounted a few yards from the weigh tank.
Each of the storage bins had an auger running from
their hopper bottom to a point just above the weigh
tank. One of the storage tanks was arbitrarily labeled
A and the other B. Grain from A and B was augered
into the weigh tank. Power relays mounted on the
weigh tank frame controlled motors operating augers

TABLE l. PROGRAMMED SEQUENCE OF EVENTS‘

Event number Event

I Weigh X pounds of ingredient A

2 Weigh Y pounds of ingredient B

3 Fill mixer

4,5,6 and 7 Mix and while mixing refill weigh tank
with X pounds of ingredient A and Y
pounds of ingredient B; then wait until
mixing is complete

8' Empty mixer

3i Fill mixer

4,5,6 and 7 (Same as 4, 5, 6 and 7 above)

n-2 Stop gsignal

n-l Finish cycle and mix last batch

n Empty mixer, stop

"The reader should keep in mind that the actual sequence
Ifollowed was entirely arbitrary. In practice any 50 events
could be arranged in any necessary order.

DUMMY MIXER

3-5" AUGER
\4" AU can

suwwormus 3m A
/ \ / FRAME
|o' ----- -- .

______ .1

4"AUGER

‘ mGRAVlTY FLOW

4‘ AUGER 

WEIGH TANK BIN f}

CONTROL TABLE

Figure 2. Plan view of test system.

from tanks A and B. A short, horizontal auger if

mounted in the bottom of the weigh tank emptied
the weigh tank into a hopper. A second auger simul-
taneously elevated grain from the hopper into the
dummy mixer. The dummy mixer consisted of a
fourth bin identical to the storage bins. Lights on
top of the dummy mixer were used to signify when
“mixing” was taking place.

Pressure switches were located near the bottom
of both the weigh tank and mixer. These switches
activated time-delay relays when the bins were nearly
empty. The time-delay relays allowed the bins
and augers to empty completely before a new event
was started. An interval timer controlled the mix-
ing time. After mixing was completed, the mixer
was emptied. An auger mounted in the hopper
bottom of the mixer elevated the grain to the top
of the two storage bins, filling A then B. A sche-
matic diagram of the control system is shown in
Figure 4. The actual operation of the control se-
quence is explained in detail in the Appendix.

In order to test the accuracy of the weight con-
trol, a set of crane-type scales were mounted in the
suspension linkage above the load cell. The scales
were calibrated over their entire range from 0 to
500 pounds at 20-pound intervals and found to be
accurate within iV2 pound. The scales were marked
at l-pound intervals and could be estimated to 1/2-
pound. To avoid implying ultraprecision the scales
were read to the nearest pound.

As can be seen from the previously outlined
event schedule, Table 1, some weight readings could
be taken during a pause (while mixing) and thus
read very carefully. However, the weight of in-

5

Figure 3. Test system and controls.

gredient A (if both A and B were weighed) had to
be read quickly while the scale indicator was mov-
ing. Such readings had a possibility o-f human er-
ror and so were labeled “estimated” in the data.
Frequent checks were made on the accuracy of A
by setting B to zero pounds for a few cycles which
then permitted A to be read during a pause.

A record of faults and their causes was kept
along with the weight record. This was for the

R4

4444

 
  
  
  
     
  
  
  
  
    
  

' A77
'-

purpose of determining the reliability of the vari
components. More extensive tests of a final 
will be required to prove or disprove long-term 
liability. *

Results

Reliability tests were startediin the fall of 
but no attempt was made to; check the accur
of weighing. A few mechanical problems appea 
and were eliminated. No trouble was encounte i
in the control circuit itself. A

A precision lo-ad cell was obtained the latter 0 
of November, and weight tests were started. 
system was first calibrated on December 2, ‘t;
A minor adjustment in calibration was made
January 8, 1964. Experimental changes were w;
in the calibration on February l0 and ll. Othert A
these, the only other adjustments consisted of ch-
ing the zero setting at the start of each test. i

To be practical, the system would have to
main accurate over a reasonably long interval i
time and under widely varying conditions. To 
the accuracy under severe conditions, only occasi
adjustments were made on calibration, and the u’
was operated with only a 2-minute warmup in  
peratures ranging from 25° F. to 74° F. (high

  

 

 

  

 

   

      
        

  

 

   

  

   

 -—  e-JI @ @ t.
CONTROLLER POIA 4, Pore vorc Porn "'

I j, ‘_ - ‘__ a 24v Ac j

4 > CONTROL RELAY 
n‘ _y_ . I SPARE I "WAG 
nt-o- 6 § ll ‘F f g 1f
7 "

nz-u- - l a is. . “'7 i
g 23v v2! "—
us-s-zo s; s5" d‘ us-4-zo @ ‘n4 x x 6' 5

' .1 - 24v Ac »

a4vAc ,' - '= @ 1 _ g;

' 510v [I9 H 

ZERO 1

24v Ac m i

24v Ac ;

l-IV Ac -> »<

s-v I20 5 V,

- wz _§

moomco comnottcn cmcuur liv"! ;

z4vAc _

E ' i i I -z4v oc I E i;

‘l! stew swncu \__.. 1

PRESSURE swncn m l? MAGNETIC MOTOR i"

82 mxcn " ‘

=< + — '0‘ s

76 69 H f

24v 0c 0 1 ,
woven mm. 
I lie f
\ a |8  » i?
r~— J. svow smut ,
I545 '4'“ EMPTY MIXER
j__ ' "V" anus!

0.25.1 moron STARTING IS-l 
“V” CONTACTS. _L  '

1, sum smut j

uuv Ac "O ‘z t:

New,“ uc. CONTACT comnottco "VAC <

2‘vAc BY INTERVAL TIMER ( LT.)
-l I~ ‘
u: v Ac F34‘ Mm ..
'4}- ss-z-az “b”. I mo. 
wencn um “q. ‘
PRESSURE sw. us_u_u m __
n20 j W 
NC. I

‘I V.

as-4-_1

6 7O q‘ i. ' PM C POY c fi ' L‘,

ss-s-uo m 95 j 4|- FILL MIXER. I

Q NS-II ,0, u AurouAnc stAnt- st
6 z‘v   P010
a4vAc m v l5 -H-
r-T DIS-IS -||- EMPTY _w:|cu TANK.
z4vAc %
64
as-s-s g az-4-@ E I20
n-a-no 9
m ' n4 4 l- us-v AND us-z ARE
I T "9"" _H_ v45; SPARES AND ARE nor _
z4vAc “v Ac snovm. NOTE :- ABBREVIATIONS :-
88-3-0 63 70 R¢-4-@ | Q l. no SCALE l. MS- MANUAL swncn
"-3-" ‘ " zvnmuc TERMINAL POINTS ARE z. 55- sup swncu
i "3 _“_ uuuacnco :50. 3 RE _ REL"
l l UQ-lo  IRELlY TERMINALS ARE ENCIRCLED.

ZOVAC

Figure 4. Schematic diagram of the

control system.

TABLE 2. SUMMARY OF TEST RESULTS

Number Set-point Mean scale Igsiialfium Zialfilttfum
f - ht d. v ion evla ion
salgples d; rlfzuzjlgs’ from set-point, from mean,
pounds pounds
164 50 50 i2 ' i2
236 100 98 __3 +2
230 150 147 _4 +2
223 200 198 -_3 +2
178 300 299 _3 i2

temperatures did not occur during the period of the
study) .

Because of the infrequent adjustment of cali-
bration, the actual weight often varied from the
set-point weight by 3 to 4 pounds. However, this

. could have been easily corrected by either adjust-

ing the calibration or off-setting the set-point by a

i. corresponding amount. Calibration adjustment re-

quired approximately 15 minutes to perform.

Grain was weighed at the rate of 300 pounds

a per minute and fell approximately 5 feet into the

weigh tank. Over a period of a few hours, the

‘ repeatability of the weights was usually 11/2 pound.
~ Over an entire day, including the period of warm-

up, the drift might reach 3 pounds. Allowing 15

,2 to 2O minutes for wannup, instead of the 2 minutes

initially used, cut the daily drift by 5O percent. A
summary of test results are shown in Table 2.

The use of daily calibration checks and a 15-
minute warmup decreased the deviations by ap-
proximately 50 percent. As expected, the percent-
age error decreased with increasing weight. For
very small weights it would be necessary to use a

i smaller load cell.

The results of the reliability tests given in this
report indicate that the prototype unit is practical
and show that commercial electronic components can
be adapted and assembled into an electronic weigh-

ing system which will enable an operator to pro-

gram and process a predetermined ration without
continuous supervision.

References

l. Hienton, Truman E. et al. Electricity in Agricul-
fttiral Engineering. john Wiley 8c Sons, Inc. New
York, 1958.

2. Andrew, F. W. et al. Controls For Farmstead
Automation, Laboratory Manual for Electric Con-
trols Workshopsi Department of Agricultural
Engineering, University of Illinois, 1961.

S. Paine, Myron D., Harold Winterfield and Dennis
L. Moe. Low Voltage Flexible Sequence Auto-

matic Controls. South Dakota Agricultural Ex-

periment Station Bulletin 500.

4. Puckett, H. B. Electronic Controller for an
Automatic Feed Grinder. Illinois Agricultural
Experiment Station Bulletin 615, August 1957.

5. Butt,  L. Results of Performance Tests of a
Small Farm Mixer Grinder. Alabama Agricul-
tural Experiment Station Progress Report Series
No. 57, August 1955.

6. Hobgood, Price and W. E. McCune. Feed Pro-
po-rtioning, Grinding and Mixing on the Farm.
A Summary of Progress Report to the Texas
Farm Electrification Committee. Department of
Agricultural Engineering, Texas AScM University,
June 26, 1959.

7. Puckett, H. B. and Robert M. Peart. Volumetric
Feed Meters, Their Performance for Automatic
Feeding Systems. Illinois Agricultural Experi-
ment Station Bulletin 618, September 1957.

Appendix

Control Sequence

In the actual operation the control sequence
was as follows:

The manual switch designated MS in Figure
4 was turned to Position 5 and the power was turned
on.

The equipment was allowed to warmup for
several minutes and the zero adjustment on the con-
troller was checked and adjusted if necessary.

The manual switch was advanced to Position 6
which grounded Position 52 on Bank 4 of the step-
switch (that is, SS-4-52). This in turn grounded
Relay 4. With Relay 4 grounded, its contacts
closed and grounded the step-switch magnetic motor.
The pulsing action of Relay 2 and the interrupter
springs K-2 and K-3 supplied pulses of current to
the step-switch motor MM-S and caused it to ad-
vance all five banks of the step-switch (SS-1, SS-2,
SS-3, SS-4 and SS-5) simultaneously to Position 6.
This started the automatic operation.

In Position 6 the weight-control potentiometer
for ingredient A, designated Pot. A in Figure 4, was
connected into the controller circuit through SS-1-6
and SS-2-6. SS-3-6 connected power to Relay 5
which activated Relay 15 and caused ingredient A
to be conveyed into the weigh tank. When the set-
point weight for ingredient A was reached, the con-
troller activated Relay 1 grounding Relay 4 through
S-4-6. This supplied power to the step-switch motor
causing it to advance to Position 9. (The step-
switch was advanced from Position 6 to Position 9
to give all relays time for their contacts to disengage.)

In Position 9 the weight-control potentiometer
for ingredient B designated as Pot. B in Figure 4
was switched into the controller circuit by SS-1-9

7

and SS-2-9. SS-3-9 connected power t0 Relay 6
which activated Relay 16. This started the con-
veyor which augered ingredient B into the weigh
tank. When the set-point weight for ingredient B
was reached, Relay 1 was activated and through
SS-4-9 in turn activated Relay 4. The closing of
Relay 4’s contacts along-with the pulsing action of
the interrupter springs K-2 and K-3 and Relay 2
provided power to the step-switch motor causing ‘all
five banks to step to Position 10.

In Position 10, SS-3lconnected power to Relay
7 which activated Relays 9, 10, 17 and 18. Relay
13 was not yet activated because the grain in the
weigh tank held the weigh tank pressure switch
(in series with Relay l3) contacts open. Relay 9
turned on Relay 17 which emptied the weigh tank
and Relay 18 which conveyed the grain to the mixer.
Relay 10 was used to switch the relatively large DC
current for Relay 13 which was a DC variable time-
delay relay. YVhen the weigh tank was nearly empty
the contacts of the weigh tank pressure switch closed
and Relay 13 started its timing interval. After a
short time which allowed the conveyors to com-
pletely empty themselves of grain, the contacts of
Relay 13 closed. This shut off Relays 9, 17 and 18
and grounded Relay 4, activating the step-switch
motor to advance to Position 14.

At Position 14, SS-1 and SS-2 reconnected po-
tentiometer A into the controller circuit. SS-3 sup-
plied power for Relay 5 which started the process
of weighing ingredient A into the weigh tank. SS-5
supplied power to Relay 11 which in turn connected
the interval timer (I.T.) to 115 volts AC. The in-
terval timer contacts supplied power for Relay 19
which started the mixing process. When the de-

i\

510 21

sired weight for ingredient A was reached, the stei
‘switch was stepped to Position 16. An inter -
holding contact in the interval timer kept the‘,
interval timer running although Relay ll was dis-f.
connected by SS-5. In Position 16, weight-contr A
potentiometer B was connected into the controller-

circuit by SS-1 and SS-2. SS-3 supplied power

M .  »

Relay 6 and started the process. of weighing ingredi- g
When the desired weight for ingredient 
was reached, the step-switch advanced to Position! l
17 where it remained until the interval timer con-A it
tacts in series with SS-4-17 closed, grounding Relay 

ent B.

4 and advancing the step-switch to Position 18. 

In Position 18, SS-3 supplied power to Relay 87 f

which activated Relays 12 and 20. Relay 14 was A

kept off by the pressure switch in the mixer. Relay f.

20 controlled the conveyor motor which emptied-j
the mixer. When the mixer was nearly empty, the
pressure switch contacts closed and Relay 14 started ?‘
its delaying operation.

the step-switch to advance. Contacts
step-switch to advance to Position 52. When operat- l
ing on automatic control, Position SS-4-52 was *
grounded through manual switch MS-6-6; this 
caused the cycle to repeat. The grain weighed in
the previous cycle would be empited into the mixer, g
the mixer started and a new batch of grain weighed. 
At the end of the operation, the stop signal would ;
switch MS-5-20 from MS-6-6 to MS-5-5 which would

stop and step-switch when it reached Position 52. ‘
The stop signal would also cause SS-4-16 to be <
grounded thus preventing any further weighing of

grain. The last batch would be mixed and emptied. Q

 
    
     
 
 

At the end of the delay‘
period when the augers had emptied, Relay 14 ii
grounded SS-4-l which grounded Relay 4 and caused 
ss-4-19 ‘
through SS-4-51 were grounded which caused the 5