HP 3457A multimeter repair? Advice needed... [Solved!]

[Satbeginner]

Hi all,

recently I bought a nice, OK looking 3457A from the USA on eBay.
It was advertised "as is", so I knew it would be my WRP. (Winter Repair Project)

After I cleaned it -mostly outside, inside was 200% immaculate, I started testing.
( I love the build quality of machines from this era!)

And now the funny thing: everything seems to work: DCV, ACV, DCI, ACI, Ohm, etc. and all values seem to be spot-on!
However: the extra tests, initiated by pressing ´Shift´ followed by ´Test´ does result in a "Test Failed" message.
The error-code is 96, so 32 and 64.

After studying the schematics I came to the following conclusion:

"The meter is not broken, but the relay K201 has a dirty contact, because this relay is hardly ever used....."

Based on the schematics, I think the relay K201 is only there to be able to do these tests, and since the tests are performed maybe once a year, the contact now (after 30 year!!)  is dirty, and causing the failure of the tests.
The relays in the measuring circuit are of a different, higher quality, type then the ´test´relays like K201.

- Does anyone has experienced anything similar with an 3457A??
- Is there more information about the function of relay K201??


Un saludo,

Satbeginner

[Kleinstein]

Form the schematics from the service manual, K201 is used to separate the ohms current source. So a slightly dirty contact should not be a problem. So it is not that critical and thus likely not a really special relay. Neither low thermal EMF nor low on-resistance are needed. Only low leakage / high voltage rating could be important.

[Johnny10]

This is the same issue solved by cncjerry and I no that long ago.
My unit is now 100% after repair.
Passes Self Test and measures all ranges.

I started by checking voltages at several points in the Test 5 circuit.

Look for that post and you will see the voltages to look for on the opamps and U101.

(HP 3457A Multimeter No Ohms Measurements)
Error 96
After rereading my posts I see I never finished the description of repair results.

I will get on that.

[Satbeginner]

Thanks for the tip to the thread, I am reading now!

Update after reading:

Your thread made me look again, and yes, I too have a problem in the Ohm ranges.
It does measure the proper values, like 1k, 10k, 100k, 1M, and 10M, but in all ranges he is drifting slowly upwards in the 3th, 4th and 5th digit behind the comma.

So now I will check -guided by your thread- the voltages around the hybrid and the U211 and U212 chips.

Thanks!!

[cncjerry]

Let me know if you need some help.  J10 is right. First thing to check are the DC voltages around the input hybrid.  If any of the bias voltages are off then solve them first.  You'll note that there are simple voltage dividers inside U101 at pins 8 and 5.  Start there and measure pin 8, 21, 17 and 18.  Most likely you have a problem with the Ohms current source but it's best to check the simple things first.  Follow J10's advice and problem determination in his note and then report back on the results.

There is another person with a 96 error around, maybe that was J10?  Can't remember, thought I saw another one.

You're lucky.  I had a nasty bug, very subtle, in the ohms pre-charge circuitry.  I had to debug a problem where the pre-charge takes out the very slight offset caused by the input capacitance charge in that area.  When I read thru the design of that meter and that circuitry in particular, I gave up all hope of ever designing a meter.  These guys are smart!  The input capacitance charging causes a measurable error in the reading?  Go figure, who would have thought of that?  It makes sense of course but someone had to think about everything when designing a meter that's good for 7.5 digits. 

Now that you have a 3457a, you should download the HP Journal articles on it.  Pretty amazing design process.  I know that meter gets a lot of shrugs because it doesn't have certain features common in meters today, but I like my two.  I paid $150 for one and maybe $140 for the second and that is quite a deal.

Jerry

[Satbeginner]

Hi Jerry,

thanks for your offer to help, I really want to fix this meter.
I found this post with voltages in it, so I will check things this afternoon.
https://www.eevblog.com/forum/repair/hp-3457a-no-ohms-measurements/msg1047822/#msg1047822

First I have to go to the postoffice here to collect another parcel, this may contain a Fluke 45.
(different story, already have one, but I love the dual display on these things. Both DC and AC at the same time. Check a voltage AND the ripple in one reading)

Un saludo, groeten,

Satbeginner

P.S. something came up today, so maybe tomorrow

[Satbeginner]

Hi Jerry,

today I was able to take some readings around the hybrid U101.

From Schematic                         My Measurements
Pin 4    10V                                10.15
Pin 8     8V/5.6V                          5.96
Pin 21   12V/8.4V                        8.53
Pin 17    8.4V                               8.12
Pin 18    8.4V                               8.11
Pin 31    +5                                  5.01
Pin 22    -15                             -15.27

They seem to be OK, but yes, I still have the error-code 96 ( 32 + 64 ), although Pin 8 is a bit high.

On U212 pin 7 (should be 15V) I have 14,92V

Readings were taken with meter set for DCV / Auto.

The thing that does worry me a bit is the reading of Pin4, the reference voltage should be 10V, so the difference of 0,15V is weird, seems big? But, on the other hand, the DCV ranges are not drifting nor giving errors, so I will be focussing on the Ohms ranges.

I will check more voltages in the Ohm area U111, U211 and U212.

Any suggestions are very welcome,

Un saludo,

Satbeginner

[Johnny10]

Pin 4 the voltage reference.
On my unit it was 10.19

If you watch Dave's YouTube video teardown of the 3457A you will see him measure the voltage reference and explain why exactly 10Vdc or 10.15 or 10.19 doesn't make much of a difference
it is the temperature coefficient that is important.

EEVBlog #426 - HP 3457A Multimeter Teardown

28:54/ minutes in the vid Dave measures Voltage Ref

[Satbeginner]

Yeah, I read in your thread your ref-voltage of 10,19V , I agree, stability is more important than absolute value.
Value can be corrected by calibration.

If I look at pages 8-41 and 8-42 of the service-manual where the cause of an error-code 32 is described,
and pages 8-42 and 8-43 where error-code 64 is described, besides U101 (hybrid) the common components are U211 and U212, so they must be checked more.

If U111 would be bad, also an error-code 16 would be expected.

So I will check voltages around U211 and U212 better tomorrow, and try to find the same drift as I see when measuring resistances.
If so, my next step will be to replace them.

Are both units of the type LT1012CN8 ?
Can they be replaced / did you replace them by LM11C ?

Un saludo,

Satbeginner

[Johnny10]

I used the LT1012CN8
Couldn't find the service note changing to LM11C.

I would check the voltages on both U211 and U212 as I found the voltage for each opamp follows a different path on the board.
If one measures 15 volts the other must be checked for correct voltage also.

[cncjerry]

J10, you also had some cracked land patterns, correct?  I was wondering if that area of the board should be checked as well.

I would check U211/6 and the voltages at U216.  I had a U216 failure in one of mine but I can't remember the code.

Also, does it impact both 2W and 4W modes?

[Satbeginner]

Really good suggestion, I did not think of testing the drift with 4-wire at all.

I will test this today.

Update: Yes, I see a basically good value when measuring resistance in the 4-wire way (eg. 100kOhm), but it´s having the same drift upwards in the 3rd or 4th digit behind the comma.

[Johnny10]

I would check the voltages at U101 Pin 8 again. This should be 5.6 or 8 volts.
Determined by the resistor divider and switch in U101.
When my unit was not working properly I had that drift problem also, ohms measurement kept moving up or down.

[Satbeginner]

With an 100kOhm resistor connected as ' device under test' it shows 100,345 kOhm and slowly drifting up.

I measured pin8 of U101, and now, depending on de range setting, it is 8,12V or 5,68V.
There is no drifting in this voltage so I checked the output of U211 and U212 to see if there is drift there:

U101 Pin8         U211 pin6          U212 pin6
5,68V                8,52V                5,23V               both no drift, where the display reading does drift up         
8,12V                12,18V              9,21V               both no drift, where the display reading does drift up
   
So, maybe Q211 is causing the drift?   To check this I measured U101 pin21

U101 Pin8          U101 Pin21
5,68V                 8,72V going up and down !?!
8,12V                 12,20 looking stable

So, if the drift depends on the range the meter is in, what does this tell us?

What are your thoughts on this?
Is the type of Q211 FET known?

I think I found it here: http://www.simonsdialogs.com/hp-crossref/,
so it could be a Fairchild J304, datasheet here: http://pdf.datasheetcatalog.com/datasheet2/9/0pq0dqtwxu9wlx3ogyikcxchzfwy.pdf


Update: I removed Q211 and it tested OK. (cut off voltage 2V, 14mA at Vgs=0V)

So I put it back, thus eliminating a cold solder joint there. Nothing changed.

un saludo,

Satbeginner

[Satbeginner]

Because I had no real clue how to go on from here, I started reading my copy of the Service Manual. (RTFM.....)

In there I found a list of changes implemented over time, and one of them involved the U211 and U212 OpAmp´s.

In my meter they are already the newer type, 1826-1381, these are supposed to be LT1012, to reduce noise and drift.
These were implemented to replace the original part, 1826-1133 (LM11C).

So, if I cannot think of anything else, I think I will order a new set of LT1012´s to eliminate a possible cause of the problem.

Un saludo,

Satbeginner

[Johnny10]

You checked the -15 and +15 on both U211 & U212?

No variation?

I found after replacing U212 that I still had fluctuating Ohms measurement.
I had 13.84V instead on the one chip.
Unsoldered the transistor and checked on curve tracer that was fine.
I eventually followed each power supply trace back till I found an intermittent for me it was a cracked trace.

U101 supplies either 5.6 volts or 8 volts to the first op amp which amplifies the signal 1.5X making voltage 8.4 or 12v.

There are also a couple of capacitors you can check.

I have been busy on my bathroom remodel, keeping me from pulling mine apart again to check voltages with yours.
Something is amiss if you are getting the drifting measurement and the self test is catching the error.

[Satbeginner]

Hi Johnny10,

I checked the power supplies on both U211 and U212, I was a bit confused at first, because I had 0V instead of -15V on U212pin4, but after checking with the schematics, it turned out that this was correct. (At least in my version of the schematic)
Both OpAmps had the same +15V.

What I did find was the -15V being a bit high, -15,29V, where the +15V was 14,92V.
So I replaced the -15V regulator by a new LM7915.
First I looked in my stock for a regulator closest to the -15V, but when I placed this one, the voltage on the OpAmps was about 0,5V low, so -14,52V. Maybe because of the rather thin tracks on the board, or would there be something drawing a high current causing this drop?

Anyway, to compensate for the drop I searched for a -15,5V sample LM7915 and installed that one.
Now both voltages on the OpAmp´s are the same, so +14,92V and -14,93V with only about 1mV ripple.

Long story short, no change, error-codes are still 32 and 64, so 96, and resistor measurement still drifts.

Eg. a 100kOhm resistor starts at 100,15 kOhm, but after drifting up to 100,378 kOhm, but it will drift down to 100,354kOhm and back up again.

When I check the voltage on the outside of the 2-wire connection while measuring Ohm´s with another meter (Keithley2000) it is stable at 1,015V, so this would suggest the 3457A current-source is stable and not the cause of the problem?

If so, that would indicate it is the internal voltage measurement circuit used for resistance-measurements.......

Tomorrow I will look into the schematic to find where this is done.
The weird thing is, that DCV measurements are as solid as a rock, but maybe that will help to identify the source of the problem,

To be continued,

Satbeginner

[cncjerry]

Sorry, been traveling and off again tomorrow.  I'll get the manual out tonight.  I'm having a hard time accepting a good current source, not that I doubt your measurements. 

We'll fix it.

[cncjerry]

The major subcircuit in test 5(32) and 6 (64) is the ohms current source.  When you are seeing this big drift, is it in all ranges?  I still think your current source or range select is the problem.  The current source takes a lot of abuse.  You could have two problems of course but if test 5 fails from current source then 6 fails as well.

What I don't like is U101/21 going up and down.  Is the drive constant?  Look at U212/6 while watching U101/21 with a resistor being measured.  Try different values and ranges and report back.  Just a couple, like 100 ohms, 10K and 100K for example and keep your eye on those two pins.  I don't think Q211 is the problem as you found, it would be the drive coming from U212/6.

And what is U101/17 doing while U101/21 is going up and down? 

There's nothing subtle here.  Look at the circuit on page 8-41 and 8-42 in the manual.  Other than U211 and U212 you have Q111 and Q112 involved. At this point, my experience with these meters is putting money on U211 or U212.  The input hybrid is pretty robust.  It looks like you should have 8.4V on the output of U212.  So as J10 teaches, 5.6 x 1.5 x 1 should be a constant drive to Q211 to U101/21.  If U101/21 is going up and down then I think you have a problem in U211 or U212.  There is a 200pF cap there on the output of U212 and that's probably OK.

I'm looking at your readings.  With U211/3 at 8, you should have U211/6 at 12;  With U211/3 at 5.6, you should have U211/6 at 8.4.  As U212 is a unity gain driver, anything appearing at U212/3 should be at U212/6.  Your U212/6 values seem off:

U101 Pin8         U211 pin6          U212 pin6
5,68V                8,52V                5,23V               <---------- U212 is wrong, should be 8.52   
8,12V                12,18V              9,21V               <---------- U212 is wrong, should be 12.18

This all depends on U212/2 being correct.  Give us a measurement on U212/2, please.

It looks like you are seeing some feed thru onto U212/6 from the FET gate to give these readings.  According to these readings, U212 is probably open (or not functioning because of U212/2 which would mean a problem in U101, unlikely).  I would have to think about this if U212/2 is moving around.

I pulled one of my meters apart quickly tonight but forgot how hard it was to get to U212 so I can't test them until I get back the first week of February.  I put a clip on it but didn't have the right probes for the clip and it spiraled into too much time with my trip coming up tomorrow.
 
I would order U212, replace, clean all the flux from that area and test.

After all these short paragraphs, I am now betting on U212.  I hate putting my reputation on the line like this but after all the thinking, your results, having had problems with U212 and U211 in other meters, what the heck.   If it doesn't fix it I'll send Dave another donation.

Jerry

[Satbeginner]

Hello Jerry,

first of all: Have a save trip, and you will see my feedback soon.
Many thanks for your effort and time, I will go into your requested measurements today.

In parallel: I ordered a set of LT1012´s at Farnell here in Spain, so they should arrive in a few days.

un saludo,

Satbeginner (Leo)

[tombi]

Hi,

With the 100K test, how much does it drift? I tried this on my 3457A with a carbon resistor I had laying around and found it will drift up for easily 30s or 1m before it settles. If I touch the resistor is falls again. I presume this is just tempco at play (it's damn hot here tonight too). It easily drifted 300 ohms before it settled.

I tested the resistor with another meter and they both measure very close to the same value after settling.

Tom

[Satbeginner]

Hi Tombi, Jerry,

After one afternoon of fooling around with this 3457A, Tombi, I think you are right about the drifting, it is temperature related.

Every time when I connect eg a 100kOhm resistor, I do this by holding the resistor, causing it to warm-up, and therefore lowering its value. (Neg.Temp.Coef.)
When I then start the resistance-measurement, the resistor cools down because I do not touch it, so I see the drifting-up.

However, what not has changed is after reaching a ´final´ value (=stabile temperature), the value shown on the display keeps going up- and down.
This behaviour seems the same with different values of resistors, eg. 100Ohm, 1kOhm, 100kOhm.

And when I perform a reset (blue, reset) followed by a selftest (blue, test) it still gave a hardware error, code 96 (=32 and 64).

I measured the following values on U211pin6: 8,1251(6/7)V and 12,203(7/8)V, so I consider this very stable, since only the last digit is fluctuating.

So the output of U211 is stable with the correct values.
About the output of U212, these values should not be the same, because the FET is used as a voltage follower, so the Source acts as the actual output of the OpAmp U212 based 1x amplifier.

The circuit-explanation in the service manual says about this, the FET is added to improve linearity of the OpAmp circuit.
By having the FET behind the output, the output of the OpAmp does not have to work at 12V, that would be too close to the 15V powersupply to maintain optimal linearity.

The good news: at the source of the FET I see the same stable values as at the input of U212pin3.

I changed location of U211 and U212 without any change to the symptoms.
Since the voltage at the output (FET source) is stable, I wanted to check the result of this stable voltage, that should be a stable current through the DUT (the resistor I want to measure, connected to High/red and Low/Black).

Instead of connecting a resistor, I connected a second mA meter to measure the current the 3457A is sending through the external resistor. (I used my Keithley 2000 for this)
Depending on the resistor range, I see the following current change from
01,01542mA
00,10146mA
00,01004mA
00,00091mA
00,00002mA

All these values are stable, only minor fluctuation in the last digit.
If I repeat this measurement but with a 1kOhm resistor in series with the external Ampmeter, the values change a bit, but remain stable.
But, the displayed value on the 3457A keeps going up- and down.

I repeated the test, but now measuring the external voltage across a resistor under test, that voltage should be similar with the 3457A reading.
The external voltage is fluctuating along with the 3457A reading on the display.

I know, these two tests are conflicting..... :-(

So, based on this, I have to assume there is something wrong with the current into the external resistor under test.

Something else: If I perform a selftest after power-on, it fails with a errorcode 96.
If I, after power-on, first perform an "AutoZero", and then directly perform a seltest, the errorcode is 64.....??

To see if there is something with the Offset circuit around Q111/Q112/U11b, I performed Diagnostic-4.
This allows for setting the Vos Adjust.
Although the diagnostic #4 already gave "Passed", it responded well to adjustments of R113, it does show "Failed Lo" or "Failed Hi" when I turned it up and down, so I put it back to "Passed".

I did look for cold solder joints around the circuit, resoldered them, mostly everything around U211, U212, Q111, Q212, Q213 - Q216.

So far for a Tuesday afternoon.

Final thought for tomorrow:
If the voltages going into U101 are OK and stable, it could be the contacts of U101, because the resistors used to determine the test-currents, are inside U101.
I read something somewhere about cleaning the contacts by pulling an IPA drenched cotton wire through the holes.

Un saludo,

Satbeginner

[cncjerry]

Have you done a diode test on U216?  I had a problem getting VOS to pass with a bad U216.  One of the transistors in the array was bad and I was getting a 512 error and now and then a 512 + 32.

I put a 100K resistor on my meter that was at about 55 degrees F.  It warmed to room temp over 3hrs and changed from 99.9986 to 99.9979.  This is a 100K, .01% low tempco.  I've seen other resistors like the Hong Kong brand for instance a 10, start at 9.91427 go to 9.93020 in a few minutes and then go up and down with the air.  All you have to do is blow on them from across the room.

But your meter is failing.  I wish I had more time to do some measurements in that area. I'll take another look when I get back.

Darn, looks like I owe Dave some support...

[Satbeginner]

Progress!!

After again a day of poking around in the guts of this 3457A, there is some progress.
I checked the current source intensively, checked all the diodes and transistors in this area,  I even removed the Ohm-Overvoltage circuit completely (Q212, 213-Q216) and replaced it by a wire bridge. I wanted to exclude any possible interference in the current circuit. But alas....

Later this afternoon I received the 2 LT1012´s I ordered, and now I only intermittently have a error-code 32 during the selftest.
I can eliminate this error 32 completely, by performing an auto-zero when set for Ohms before running the self test after PowerOn.
The drifting of the reading is (mostly) caused by the cheap -quick-and-dirty- (Chinese?) resistor I grabbed to test the meter.
Other resistors perform much better.


And now I have a good idea about the error 64: Imho this error occurs only because the test is not really performed correct, and not because of a failure in the Amp-measuring circuit.

To perform test6 (error64) of the selftest correctly, switching of U402 and U403 is necessary. (Fig. 8-8, pag. 8-43)
During the self test these two switches re-direct the set Ohm current via the Amp shunt, so an Amp test can be performed.
The control signals to do this come from U123 (pin7, pin 11, pin12, etc) (Fig. 8-18 of the schematic)

When I check the control inputs of both switches during a "Blue" "Test" Selftest sequence they do not move.
So I checked all outputs of U123 during the test, they all move during the tests, except for Pin11 and Pin12, they stay always high.....
So, because of these 2 signals being always high, the test-current cannot be seen in R401 (Amp-shunt), leading to the error 64.

I will order a few MC14094BCP (U123) and maybe a HI3-0390-5 (U402) too.
With respect to the error 64, my money is now on U123.

It would be great if there is a member who can scope pin11 and pin12 of U123 during a selftest.
I expect a TTL level with some changes during the selftest. Can somebody confirm this please?
The pins 11 and 12 can be reached without removal of the AC board. The aluminium covers do have to come of though.

I will keep you posted,

Un saludo,

Satbeginner (Leo)

[cncjerry]

Sat, did you check the precharge chip I mentioned?  It is a transistor array, can't remember the number, but it caused weird test problems.  Also, are you able to do the VOS successfully now?