Text preview for : HP-Bench-Briefs-1981-06-08.pdf part of HP HP-Bench-Briefs-1981-06-08 HP Publikacje HP-Bench-Briefs-1981-06-08.pdf
Back to : HP-Bench-Briefs-1981-06-0 | Home
m
.
I
-. 1
SERVl CE INFO R MATI0N FROM HEWLETT-PACKARD
JUNE-AUGUST 1981
Logical Troubleshooting
IN THIS ISSUE
Logical Troubleshooting
Transistor/Diode
Cross Reference
Service Notes
New HP 1304A
Display Power Supply
spent. Availing yourself to the ex- The Information Funnel that the detector is really working)
periences of others will make you and then to make measurements at
more productive (and more Now we come to another considers- B,C, and D until the defective stage
valuable!) tion in the information-seeking pro- is finally isolated. While this %erial
cess. Once a decision has been method" works, the fastest approach
If no one else has seen the problem, reached as to where a check or is to use "halfsplitting" technique.
you must solve it yourself. At this measurement is to be made, there
stage, apply the Rules of Logical still remains the related question of The main principle of half-splitting
Troubleshooting. This consists of what type of check to make. You is that a check is made at the mid-
"milking" the instrument for all must decide, for example, whether to point of the remaining part of the
symptoms available. U e your eyes,
s flip a switch, take a measurement or circuit that has not yet been
ears and nose. Are lights lit? Read- replace a part. checked.
outs active? Are there any signs of
heat or broken components? Are The information funnel means mak- Looking back at Figure 1 we know
there any abnormal sounds? Pops, ing the type of check that is most that there is no audio output (point
hissing, hum? Is there a smell asso- appropriate to the size of the trouble E) and we suspect that a signal
ciated with a particular area? Es- brackets at a given time. The d e is: exists at point A. (We know for cer-
sence of carbon camp? The symptoms As you progress in locating a trou- tain that one output of the detector
are next sorted by functions -try to ble, vary the type of memurement is ok because of the reading on he
localize the fault to a particular you make in a specific order. Start signal strength meter.) Now what??
function. Many times this can be with front-panel checks, then make
done by using switches on the front adjustments, take waveforms, We should measure at point C be-
panel. With practice, the technician measure voltages, measure resist- cause this is halfway into the ab-
can become quite skilled at isolating ances, and finally, replace parts. normal path. If the signal at point C
the fault to a function and some- is ok,the trouble is isolated to either
times to a particular block or circuit P e d o d n g the checks in this funnel the audio power amplifier or the
within the functional circuitry. With sequence assures you wide coverage loudspeaker. Measuring at point D
the symptom determined, the ab- of trouble possibilities initially, but would indicate the defective area.
normal path of circuitry is with low precision. Then, gradually, Additional measurements may be
identified. as you proceed to localize the trou- used to determine the defective
ble, the checks become more precise. component.
Half-Split vs. Serial
Half-Splitting
Bracketing Comparing the half-split method
In our example, we know the prob- with the serial (or "daisy chain")will
The problem can now be isolated lem exists somewhere between the emphasize the difference. Assume
with a technique called "bracket- detector and loudspeaker. Where that we are troubleshooting a linear
ing", which is simply a means of es- should the first measurement be path with eight stages as shown in
tablishing the broad limits around made? One technique used all too Figure 2. Assume also that there is a
the areas to be tested. For example, often is to measure point A (to verify signal at the input (point A) but no
if you connect a signal to the an-
tenna terminals of a radio receiver
(see Figure 1)idhees neeutput in
i
the speaker, you have bracketed the
problem to somewhere between the
antenna circuits and the speaker
cone (Le., the entire reseiver). If,
however, the signal strength meter
shows a strong indication, the cir-
c i s from the antenna terminals to
ut
the detector are probably good (since
the signal strength meter is usually
driven from one of the detector's
output). Therefore, the problem is
now isolated to the audio section. We
also know that at least some of the
power supplies are working, and this
is valuable information. Figure 1. Simple radio raceiiver.
2 BENCH BRIEFS JUNE-AUGUST 1981 WWW.HPARCHIVE.COM
there, we would next measure point
the half-splitting technique will lead occur in Stage 1or 2, but many tests C and find no signal there. Next
you to the problem with the least are required to find measure point B and find no signal
number of measurements (and pre- the end of the chain there. Therefore, three meas-
sumably the least amount of time, so urements are needed to find the fail-
you are more effective in your job). ure. Assume Stage 2 has failed.
urements required to find a failure Measure points E and C. Finding no
Serial Metho in this eight-stage circuit is 43/8. f
O signal at either point, measure point
course, you cannot make a fraction B and find a signal present. There-
any measurements will be of a measurement; that is juet an has failed. Three meas-
, on the average, to find a average. Another way to look at it is e needed.
using; the serial method? 4375 measurements would be
This technique ( is NOT rec- Similarly, any other stage can be
ommended) will be to measure each %solated with three measurements, a
successive point until the signal goes sharp contrast to the "serial" method
away. which required an average of 4-3/8
Now let's use the half-splitting - an increase of 45%! Or to repair
Figure 3 lists the number of meas- method to find the problem. Assume 1000 instruments, 3000 meas-
urements required to find a circuit that Stage 1 has failed. Split the urements would have to be made
failhe in the various stages. Note chain into two sections by measur- using the halfsplit method vs. 4375
that very few measurements are ing at point E. Finding no signal for the serial method. Is anyone in-
terested in making 1375 unneces-
sary measurements? The difference
in effdtiveness of these two tech-
niques becomes even more dramatic
is increased.
High Failure History
In the calculations above, we as-
sumed that each stage has an equal
chance of failing (i.e., similar relia-
bility). This is probably a good as-
Figure 3 Serial method (not recommended).
. sumption if we don't have any prior
experience or knowledge of the
product being repaired.
Many times, though, we do have ad-
ditional information that is useful. If
previous experience indicates that
one stage is particularly prone to
failure, this should be considered in
deciding where to make the
measurements.
Looking back on Figure 2, assume
that Stage 8 has poor reliability.
Flgure 4. H.lf-,pmmothod (recommendad). This is often the case because of
WWW. HPARC HIVE.COM JUNE-AUGUST 1981 BENCH BRIEFS 8 ;-
higher power than the previous Other Considerations output signals in light of the particu-
stages, plus, if it'$an output stage, lar circuits. Good documentation
possibly being subjected to excessive The half-splitting technique also as- showing waveforms or other data for
loads, &ort circuits, etc. sumes that all test points are key t s points is needed. Prior ex-
et
equally accessible. Many times perience with the product can also be
h g n i z i n g that the failure very shielding, gaskets, cover plates and used to advantage in lieu of thor-
likely exists towards the end of the other mechanical restrictions make ough documentation.
chain would suggest that we shift hst points inaccessible, especially in
our initial measurement point to- high frequency products. This will In conclusion, half-splitting will lead
ward the troublegome area. There- similarly alter how you split the ab- you to the problem in the least time.
fore, a good rule of thumb about normal path. If you know nothing about the relia-
-hooting is to half-split at the bility of the circuits under test,
paint where there is approximately Taking time to decide where to make make the test measurement midway
an equal chance of finding the fail- a measurement will usually save between the point of a known good
ure on either side of the test point. time, but only if the decision takes signal and the point of a known bad
The odds are good that you will only less time than the measurement. signal. If one end of the signal path
make one or two measurements. Pondering two minutes about a has a bad failure history, shift the
measurement that could be made in initial measurement closer to the
In our example, specific failure rates one minute is not an effective tech- troublesome area.
could be used in a calculation to nique either.
prove the validity of this method. Using a logical troubleshooting ap-
However, exact failure rates are not It is also worthwhile to point out proach will make you more effective
needed nor should you perform any that when making a measurement, in your job. Let's try it on the follow-
calculations. Just recognize that a you must have information avail- ing problem.
high failure rate in one end of the able that tells you whether or not
string should cause you to s i t the
hf the signal is correct. The main rea- A Troubleshooting
initial point measured toward the sons for the popularity of the highly Example
troublesome circuit. A good rule of inefficient "serial" method are lack
thumb is 50%. Unless a particular of good documentation and lack of This problem concerns a typical mal-
circuit has at least !50% of the fail- experience. It is sometimes difficult function in the electrical system of a
ures, do not modify the initial to measure in the middle of a long car. Referring to Figure 5, let's make
method. In fact, when a failure rate chain and determine if the meas- a brief rundown of the way the sys-
is high enough, it may make sense urement just observed is correct. It tem is supposed to operate. At the
not to even make a measurement - may appear easier to move along one left of the diagram is the generator.
simply change the suspected part. stage at a time comparing input and The shaft or armature, is driven by
Figure 5. SlraPHfird -n
r dan autcmoMkelefMealsysfm.
4 BENCH BRIEFS JUNE-AWUST 1981 WWW.HPARCHIVE.COM
The result of grounding the field small positive charge rate on the either resistor. This is an application
terminal on the generator is this. ammeter. You eliminated the fan of the half-split rule and information
The ammeter goes to maximum in belt, battery and accessory divergent funnel rule.
the positive charge direction even path from the problem which left the
though the engine is turning over at generatorhegulator. When you The result of measuring the resist-
a fairly slow rate, so it appears that shorted the F terminal of the ances is 0.1 ohms kr each one - a
the generator is functioning and ca- generator, you found that the charge normal indication.
pable of producing a large enough rate on the ammeter increased to a
current. You can move the left very high value. What sort of infor-
bracket a step to the right now and mation was this? Well, it moved the Now what? Remember when we
eliminate the generator, since it is left bracket in the past generator to were discussing what the effect
outside the bracketed area. And the left of the two series control re- would be if the voltage regulator coil
since the lights are still on (aren't sistors. The two resistors were sus- were open? It wouldn't operate and
they?), you can add a bracket across pect as well as the contacts of the the resistor RV wouM be perma-
that divergent path. What's left two regulators. The current reg- nently shorted, causing high field
then? The regulator. ulator coil was eliminated since you current and high charging current.
were able to reach a high charge rate But our problem is the opposite -
The next check is to shut off the en- by shorting the F terminal (i.e., you low charging current. So what would
gine (and lights) and remove the knew that you had to have a good the symptom be if the volt-
cover of the generator regulator. Re- series path from the generator age regulator (or current regulator)
start the engine and visually inspect through the ammeter). contacts were open (dirty), which
the voltage and current regulators would cause resistors Rv and & to
for operation as you let the engine However, there is one other concept always be in the circuit? Low charg-
idle. Then increase motor speed to a under suspicion. Let's puzzle out ing current!
point where the regulators should what would happen if the voltage-
operate. This is another check that is regulator coil were open. W w would
o The best check here is to visually
similar to front-panel "milking". this affect the charging amperes? If inspect the voltage-regulator con-
the voltage-regulator coil were open, tacts, or clean them with ordinary
The result of watching the voltage then it would never operate, and the paper or a burnishing tool. The
and current regulators while you resistor & would be permanently easiest way to do this is to take a
vary engine speed is that neither shorted. And if the resistor were piece of coarse paper (not
relay operates. Now turn the lights permanently shorted, then the-field sandpaper), insert it between the
on and see what happens. The result resistance would be low, you'd have contacts, and apply light pressure
is that the cutout relay opens up. Is lots of field current, high magnetic with your finger on the armature as
this normal? Yes. When the field, and you should have a high you pull and push the paper back
generator does not supply enough charging rate limited only by the and forth. As a result of cleaning the
current and voltage to operate the other series resistor, which would contacts, you find that the generator
accessories (for whatever reason), become unshorted when the current is back in operation.
the relay should open up to prevent became too high. But that doesn't fit
the battery from discharging the facts in this case. The field cur- Let's review how, we tracked this
through the generator. rent is low, keeping the charging problem through to conclusion. We
current low. started with front-panel milking and
Up to this point, you have-made all noted that we had zero, or a ve*
your checks and tests using only the The next cheek at this point is to go small, charging rate as shown on the
car's ammeter (assuming it has ahead and measure the resistance of ammeter. When we shorted the field
one). Now you need to apply the in- the voltage-regulator coil. The result terminal of the generator, the am-
formation funnel rule and get out of measuring the resistance of the meter jumped to a very high rate,
the VOM for some serious voltage-regulator coil after you have showing that the main series path of
measurements. disconnected it is this: Your ohmme- the car's electric system was alright.
ter reads 80 ohms to ground verify- The trouble appeared to be in the
But before we take any meas- ing what you had assumed - that feedback path. Then we inspected
urements, let's review the brackets the coil was ok. the regulators and found them not
and circuit and determine what re- operating. When we found the series
sults we expect t o get from the The next check is to disconnect the F control resistors okay, we promptly
measurements. wire to the field of the generator and suspected the regulator contacts. We
measure the resistance of each of the found that the series resistor Rv was
Originally you found that the bat- control resistors RV and &, while always in the circuit, so the
tery was not being charged, as you hold the regulator armature generator voltage was always low
shown by a zero or insignificantly down to remove the short across and the current consequently never
6 BENCH BRIEFS JUNE-AUGUST 1981 WWW.HPARCHIVE.COM
got very high. When we cleaned the and using the information funnel. How many tests or questions do you
contacts the generator ran the As a test for yourself to prove the have to ask t o determine exactly
accessories and charged the battery. power of halfsplitting, take a stand- what card you have chosen? Look for
ard deck of 52 playing cards and the algorithm in the next issue of
This example shows the power of es- remove one without looking at it. Bench Briefs.
tablishing brackets, half-splitting,
Safety-Related BLUE TAG REPAIR
Service Notes
Service needed
Service Notes from HP relating to CALIBRATION ONLY 0 REPAIR
personal safety and possible equip- OTHER
ment damage are of vital importance Should one of your HP instruments need repair,
the HP service organization is ready to serve you.
to our customers. To make you more However, you can help us serve you mom effec- Observed symptomrlprohlems
aware of these important notes, they tively. When sending an instrument t o HP for
FAILURE MOOE IS:
repair, please fill out this card and attach it to
are printed on paper with a red bor- the product. Increased repair efficiency and 0 CONSTANT INTERMITTENT
der, and the service note number has reduced turn-around time should result.
SENSITIVE TO:
a "-S" suffix. In order to make you 0 COLD 0 HEAT 0 VIBRATION
immediately aware of any potential COMPANY
FAILURE SYMPTOMS
safety problems, we are highlighting
safety-related service notes here AOORESS
with a brief description of each prob-
TECHNICAL CONTACT PERSON
lem. Also, in order to draw your at-
tention to safety-related service PHONE NO. EXT.
notes on the service note order form
at the back of Bench Briefs each MODEL NO. SERIAL NO.
appropriate number is highlighted
P.O. NO. DATE
by being printed in color.
Accessories returned with unit
If unit is part of automatic system list modcl
0 NONE CABLE(SI
numbers of controller ani! other related
0 POWERCABLE 0 AOAPTEAK) components.
OTHER ~ ~~~
7245A/B over 9320-3896 Printed in U.S.A.
PIotter/Printer
Should one of your HP instruments are filled out by you and attached to
A shock hazard may exist at the need repair, the HP Instrument Re- any instrument sent to HP for re-
power cord socket (or at the male end pair organization is always ready to pair. Increased repair efficiency and
of the power cord) of plottedprinter serve you. Toward this end, we are reduced turnaround time are our
serial prefixes below 2047 when the promoting the use of the "Blue Re- goals. Please help us help you. Ask
power switch is in the OFF position. pair Tag." These tags are available your HP representative for some of
from your HP representative, and these cards today.
If the power switch is OFF the line
filter capacitors retain a charge at
the power socket that can shock an Correction choice since it uses far less memory
operator.
All About CRT's than dot writing. The characters are
formed from addresses and strokes
To eliminate the possibility of this
In the previous issue of Bench Briefs burned into ROMs that are address-
hazard, it is necessary to install a
(March-May '81) page 4 under the
on able using a single memory location.
bleeder resistor across the line ter-
heading VECTOR, I alluded that And Figure 4b is actually an exam-
minals of the power socket. Safety
"dot writing" would be a good choice ple of this technique, not dot writing
Service Note 7245NB-3-S describes
of displaying annotation on the CRT as originally stated.
the procedure for installing the free
safety modification kit HP part along with the main picture. In real-
number 07245-60130. ity stroke writing may be the better Editor
WWW.HPARCHIVE.COM JUNE-AUGUST 1981 BENCH BRIEFS 7
HP P I N JEUEC NO. 111821 --
-- 1902-0761 1N3155 --
-- 1902-07139 11148111 --
-- 0122-0025 1NMl4 --
-- 1901-0786
--
lN821
11823 --
--
1902-1275
1902-0033
IN3156
113208 --
--
1902-0790
1901-0421
1N4936
1114937 --
-
1901-1065
1901-0831
1116096
31162701 -
--
1901-01335
1902-0986
1902-1 3 07
1902-1 3 08 --
--
11152768
1N47341
111825
1N825 --
--
1902-0509
1902-0777
1N3203
1N3209H --
--
1901-0032
1901-0346
1114960
1115000 -
--
1902-0974
1901-0775
1N6303
211174 --
--
1901-0913
1850-0035
-- --
~
1902-1 312
1902-1 3113
--
1 ~ 9 ~
111298RA
9 111825
~
111825
--
1902-0913
1902-1221
1N3210
lN3212 --
--
19oi-004~
1901-0423
1115002
1115002R -- 1901-0492
-- 0122-0215
1901-0843
211335
2113681 --
--
1854-0026
1851-0024
1902-1 3 20
1902-1 3 23 -
--
114566
lN58578
111827
11829 --
--
1902-0680
1902-0625
lN3260R
1N3252 --
--
1901-OS11
1901-0312
1115003
1151f9 --
-
1901-0494 2N3968
2wsoa1 --
--
1850-0128
1850-0154
1902-1 3130
1902-1 3 32 --
--
*IN750
11158518
1N831
111914 --
--
1900-0019
1901-0710
11132aZR
lN3283 --
--
1901-0315
1901-0307
1115140
1W51401 --
--
0122-0247
0122-0248
211684
211685 --
--
1884-0267
1884-0015
1902-1 3 3 4
1902-1 3 36 --
--
11155338
11160018
11914
e11914 --
--
1901-1029
1901-0830
lN3289R
*1N3290R --
--
1901-0308
1901-0655
1115141
1115142 --
--
0122-0249
0122-0251
21688
2N708, --
--
1884-0016
1854-0085
1902-1 3138
1902-1 3 40 --
--
1114557A
lN5355R
lN935
1N936 --
--
1902-0772
1902-0785
11133160
1N3491 -- 1902-1369
1901-0406
115143
1115144 --
--
0122-0253
0122-0255
*211709
-- 1854-0009
1902-1 3 4 1
-- *lN974R 111936
-- 1902-1241 1113491R--
-- 1901-0343 11151441
-- 0122-0256
217201
21743 -- 1 ~ 5 4 - 0 0 i n
--
1854-0750
1854-0004
1902-1 3 42
1902-1 3~3 --
--
1N47421
1~5330
1N937
--
1902-0786 1N3492
1113494 --
--
1901-0424
1901-0426
1115145
1N5f46 -
--
0122-0257
0122-0259
211834
211835 -- 1854-0P11
1902-1 3145
-- 11153658
11938
lN9381 --
--
1902-0787
1902-1216 1N3563
-- 1901-0742 115147
-- 0122-0261 211910
1902-1 3 46
1902-1 3148 --
--
1N53558
1114746A
111941
1N941 --
--
1902-001R
1902-0569
1113644
1NS662 --
--
1901-0752
1901-0131
11151471
115148 -
-
0122-0262
0122-0263
211916
211918
1902-1 3150
1902-1 31 5 1 --
--
11153648
115350R
lN9411
1N942 --
--
1902-0698
1902-0791
113712
1113714--
--
1912-0009
1312-0007
1151 A
48
1115163 -
--
0122-0264
1901-0913
211950
21941
1902-1 35 3
1902-1 3 54 --
--
1N2985R
1N298SRtl
11943
-- 1902-0793 113716
-- 1912-0002 lN5182
-- 1901-0405 21941
--
1902-1 3 56
-- *lN4753A
111944
1N944 --
--
1902-05111
1902-0793
1113717
i ~ --
m ~
--
1912-0015
1912-0006
1NS2360
1152370 --
--
1902-0630
1902-1305
211964
21976 -- 1850-OQ99
185I-OlS7
1902-1 3 57
1902-1 36 1 --
--
IN53748
lY4756A
1N9658
-- 1902-1304 1113719
-- 1912-0017 11152700
-- 1902-1306 2111131
--
*- 1855-0019
1902-1 3 62
1902-1 3 6 3 -- lN767H
1N749A
1119670
1N 9 n 0 --
--
1902-1362
1902-1341
1113720
1113721--
--
1912-0014
1312-0019
11152760
1115283 --
--
1902-1307
1901-0758
211183
211304 --
--
1830-0064
1851-0017
1902-1 3 67 --
-- 115353
e11979
lN989R --
--
1902-0972
1902-3512
1113881
113889 --
--
1901-0756
1901-0791
1115214
115297 --
--
1901-0846
1901-0869
2N1358
2111482 --
--
1158-0150
1854-onw
1902-1 369
1910-0 0102 --
--
1N33168
lN38fl
11995
--
--
1910-0025 1138898
--
--
1901-0511 lM5305
-- 1902-0526 211484
--
--
185*-0235
1910-0 0 0 3
-- lN551
lNlll6
lNll831
--
1901-0069
1901-0315
elW3889R
--
111389OR
1901-0735
1901-0684
1115312
111533&8 --
--
1903-0627
1902-0996
2111500
211523
--
1850-0051
1850-0194
1910-0 0I14
1910-0 0123 -
-
1N277
1N270
1111183AR
lNll84A --
--
1901-0315
1901-0317
INS891
--
--
1113891%
1901-0717
1901-0774
111S33YA
115336 --
--
1902-0923
1902-1543
2N1595S
2Nl67 1A --