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

[Johnny10]

I opened the 3457A and took some measurements.

And also put the scope on U123 Pins 11 and 12
Pin 11 simply went high to 5 volts and stayed until the test ended.
Pin 12 is in the scope pic.
Very stable waveform.
I will keep it open if you need more comparisons.

I also remember that after the repair it took 3 or 4 days of steady power on for Error 64 to clear and pass the "Self Test OK" screen.

[Satbeginner]

Hi Jerry, Johnny,

Yes, the offset adjustment (diagnostic4) was already working, I just wanted to see if this part of the circuit was working normally, to eliminate the usual suspects. (great movie BTW)

I checked about all components in the current-source chain for both test 5 & 6, that´s how I ended up at the bottom part of the test6 circuit. (Page 8-43)

After I changed both the LT1012´s (U211, U212) most of the err 32´s are gone, and resistance measurement is looking normal too.
(I still have the drift, but for now I just assume this is caused by the resistors under test. (metal film, but still Chinese. Higher values seem to drift more than lower values)

It seems I now have a working meter, but with a broken test-circuit? This would be the world upside down.

Like I said, if I do an autozero first, the err32 is really gone, but only until power down.
It might be that this behaviour has something to do with the cause of the err64.
If, what I think, some electronic switches are stuck in a certain position, this may influence the results of test6.

Thanks for checking U123 for me, my U123pin12 is steady at a slightly lower level than a normal ´high´, also lower than all the other outputs of U123. (I find this suspicious)
Pin11 is just high at what appears to be a normal ´TTL High´.
Both of them go high at PowerOn of the 3457A, and stay there.

Johnny, I understand your Pin11 goes up at the begin of the test, but comes down at the end , right?

I hope it is just the MC14094, because I could not quickly find a source for the switches (U402, U403)....

To be continued, many thanks for your input,

Satbeginner, (Leo)

[Johnny10]

Yes, Pin 11 drops after test ends. Jumps to 5 v during Test then back to zero.

[Satbeginner]

Hi Johnny,

thanks, this really confirms there is a hard difference between my meter and a working one!
MC14094´s are shipped, should be here in 1-3 days...

Will keep you posted.

Un saludo from the 1st snow in 34 years,

Satbeginner

[Satbeginner]

Hmm, not looking too good:

in preparation for the arrival of the U123 replacements (MC14094, probably Wednesday), I removed it from the board.
After that I powered-up the meter to check the two inputs of the switches U402 & U403. (U123, pin 11, pin 12)

The input of switch U403 (coming from U123 pin11) now is at a logic ´low´, this looks like normal behaviour, however,
the input of switch U402 (coming from U123 pin12) now is at -2V, this would explain the lower than normal level when U123 was still in place, but the -2V is very wrong.

This -2V comes from the input of the switch, and further checking learned that all pin connected to this part of the switch are at wrong levels. They all are around -2V, this very strongly indicates a broken switch U402...... (I´m thinking somebody applied a serious overvoltage to the input of the meter....).

So I will start looking for this part too.
U 402 is an analog switch (HP partnumber 1826-1205)  made by Intersil, type HI3-390-5 (sometimes indicated as HI3-0390-5)

They still exist, but a minimal order quantity could be a problem.

Update: I think I found suitable replacements at Mouser Europe, they should be here in 2-3 days.

To be continued,

Satbeginner

[Satbeginner]

Again an update:

Today I removed U402 in preparation of the new switch part being underway.

So, at the moment both the broken U123 and U402 are removed, this enabled me to check for other weird voltages in the meter around these two chips.
The good news: All the inputs of U123 are now at normal (TTL) levels, and there are no strange voltages now at any of the pins of U402.

I choose to put sockets in both locations, so I can easily replace and test the functionality of both parts when they arrive.
, hopefully later this week
Since U123 only uses TTL-level signals, so (IMHO) using a socket there is no problem.

With U402 the situation it a bit different:
Most of the pins are power-supply (-15V, gnd, +15V) and (TTL level) control signals.
The analog signals at both switches are used during the selftest, so for now, I assume having a socket there is not critical either and will be of minor influence to the meter when used for taking measurements.

Since the ON-resistance of a U402 switch (according to the datasheet) is supposed to be smaller than 35 Ohm, having a socket in the circuit should not be a problem.

If reality on the long run proves different, I will remove the sockets.

P.S. the funny thing is, that the meter -even with both U123 and U402 removed) still is capable of measuring voltage and resistance.
The only setting now giving OVLD is DCI (DC Amps), this because the ground circuit of the low-range Amp-measurement is now open, caused by U402 not being there.

To be continued,

Satbeginner (Leo)

[Satbeginner]

Again...... progress!!

Today the replacement parts for switch U402 arrived, so I placed it on the board and performed a self-test......
No go, a big series of errors were thrown in my direction! (32, 64, 256)

I also tried to measure a 100kOhm resistor, that worked, but measuring a 10V reference voltage failed bigtime.
The reference voltage dropped to about 1.8V!?!, even with the meter switched off!!

I quickly switched the meter off, and went to have a look at the schematics (again).
After some testing, it turned out that BOTH relays K301 and K302 had their single make contact fixed at the ´ON´ position, so this is really bad.
Somebody must have done something very, very wrong while using this meter.    

Although these red relays are a special, metal case, sealed type of relays, I replaced them with a more basic version of a 5V relay.

Wow, Wow, Wow, the selftest now works, and gives a "Self Test OK" message, also the measurement of the 10V DC reference voltage now works.

However, happy with this result, I now tried a AutoCalibration (ACAL).
This does give an error, the message is "1 Cal. Failed".
ACAL by means of the ACAL button performs all 3 routines, 2 AC tests and 1 Ohm test.

If I perform the ACAL functions individually from the menu (Blue, A, Arrow Right, Arrow Down, select Ohm, Enter) it turns out this is the calibration routine that gives the error.
Performing the AC tests this way does not give any errors.

So, there is definitely progress, but the Ohm circuitry is still under suspicion.
Or maybe it just needs calibration because of the exchanged amplifiers.

Un saludo,

Satbeginner (Leo)

[Satbeginner]

Almost there......

After the repairs so far, I found the following weird behaviour in the Ohms measurement, and I think this is related to the ACAL error in the Ohms routine:

If I (on autorange) measure a metal film 100 kOhm resistor it measures around 100,470 kOhm.
However, when I switch to one range higher it measures 0,10080 MOhm. (at least similar)

Here is the weird thing:
Again, one range higher (in the 300 MOhm range), the measurement changes to 00,1048 MOhm (suddenly 5% higher??)
And finally in the 3 GOhm range it measures 000,105 MOhm (Again 5% higher??)

This suggests it maybe is related to
Option 1: the "Extended Ohms Range Circuitry" , or,
Option 2: it could be related to the Hybrid U101, because pin1 in the Ohms setting is only used with the 300M and 3G range, or,
Option 3: maybe there is still a problem in the switching of the Ohm current-source.

To maybe rule out option 2: a DC voltage of 1V does measure OK when switching ranges.

But so far I could not find an obvious failing component.

Any suggestions??

Un saludo,

Satbeginner (Leo)

[Kleinstein]

The slightly different readings could be still a calibration (e.g. ACAL part) problem, since ACAL seemed to have not fully completed. It is still rather hard to tell were the error might be.

One thing would be to go back to the basics and use second meter to measure the current in the ohms ranges. As an extra test one could add some series resistance to see if the the current is constant.

A second point to test could be the input bias current in the volts range. Just to make sure there is no extra leakage path.

It might also be a good idea to check the low current ranges. Some of the shunts seem to be used in the ACAL procedure for the current source too.

[Satbeginner]

Hello Kleinstein,

thanks for the suggestions, will check them tomorrow.

Although I am not sure what you mean by checking "the input bias current in the Volts range"?
Do you mean if there is current flowing into the meter when measuring voltages?

The ACAL function does end, but with an arrow above the ERR indicator, the error is "1 Cal. Failed".
If I run the ACAL AC functions, it ends error-free.
The Ohms test ends with an error.

Will let you know tomorrow.

Un saludo,

Satbeginner

[Kleinstein]

Form a damage due to over-voltage, there is small change that the voltage measurement will not be that low current any more. So than there will be extra current of more than the usual 10-50 pA. Also contamination of the board could case such a problem. Obviously some extra current could upset the measurement of high resistors.

AC measurement is largely separate so no surprise that at least AC ACAL would work Ok. The more interesting part would be the (DC) current modes as they are more linked to the Ohms.

[Johnny10]

I am sorry to hear you are still having problems with your unit.
I picked up a 100k 1% resistor and checked on my 3457 today.

       5 digit readout.              6 digit readout

Auto.            099.681 K.                     099.6813 K
Up Arrow.         0.09968 M.                    0.099682 M
Up Arrow.       00.0997 M.                    00.09971 M
Up Arrow.     000.100 M.                    000.0997 M

If you need any other measurements just ask.

[Satbeginner]

Hi all, thanks for all the help, tips and feedback: I am happy to say my 3457A is back in business!!

My latest actions:

I thoroughly cleaned the A2 board and the U101 Hybrid with IPA, no change.

I checked the input resistance of the U101 Hybrid (10 MOhm), spot on, so all OK.

I checked the Amp- and DC Voltage ranges with my HP 3468 and a Keithley 2000, all OK.
 
I checked the Ohms-currentsource again, this time by using my Keithley175A as Amp-meter.

This one has a 200,00 uA full scale setting and with this I could confirm all currents -even the very low ones- were there.
In the lowest Ohm range (30 Ohm) the current at the HI-LO front connectors was 1,0156 mA, stepping down in factor 10 steps to 0,10 uA in the 30 MOhm and 300MOhm ranges, so this was good. 

Still, the display would be about 5% off (high) when using the 2 highest ranges......
My conclusion: it had to be the gain-calibration of these ranges!

As far as I could find, this meter has all separate calibration values stored for each individual setting and range, so I decided to "calibrate" the 2 highest 2-Wire Ohm-ranges, this leaving the rest of the meter´s calibration untouched.

I first did an Offset-calibration on all the 2-Wire Ohm ranges, since this is easy and only needs a copper wire bridge.
After that I made myself a 30 MOhm "standard" resistor by combining several metal-film resistors to a 30,0 MOhm resistor, I used my Keithley 175A as "the reference" for this.

Well, it turned out there is only one High range to calibrate, that would be the 30 MOhm range. It looks like the 300 MOhm range is derived from the 30 MOhm range calibration.

So I manually selected the 30 MOhm range, connected my "standard resistor" and went for "Blue", followed by "C, Offset Comp", and the "CAL" appeared in the display.
Now I entered 30000000,3457 followed by <ENT>.    (P.S.: I had to reset the security code by changing the jumper on the A1 board)
It said "Calibrating" and after that it read 30,0120 MOhm.

I tried a calibration with other values as well, but it looks like calibration is only possible when using Full-Scale values, plus- or minus a small deviation.
So, when I tried to set calibration in the 30M range with a 10M resistor, it would display "Calibrating", but nothing would change.
Also, when I used my 30M resistor and entered 30800000,3457 <ENT> it would not change anything.
Entering 30200000,3457 does work.

Anyway, after the calibration with my 30M resistor using 30000000,3457 I performed an "ACAL" and hurray, it finished without any error-message or indicator.
Also the simple test with my 100 kOhm resistor is good now.

So, here is my resume of this (successful) repair:

- 2x OpAmp in the currentsource U211, U212
- Q212 in the Ohms Overvoltage protection
- U402 Analog Switch, replaced by a MAXIM DG390 (Mouser)
- U123 MC14094, serial to parallel shift register
- 2x reed relay K301, K302, temporarily replaced by simple, small enclosed 5V relays, will be replaced by 1A05 DIP4 reed relay 5V reed relays. (Don´t think the metal magnetic shielding is really necessary here)

All these parts are in the input-circuitry, so this definitely suggest that the input was exposed to a huge overvoltage.

Again, many, many thanks for all the help!! 

Un saludo,

Satbeginner (Leo)

[Johnny10]

 

[Kleinstein]

The metal shield with reed relays is not only for shielding. It also helps to make the relay more sensitive, so it can work with less current and thus lower power.
The relays K301 and K302 do not look like they are really critical for small DC voltages. So normal reed relays could be OK.

[Satbeginner]

Thanks for the extra info.

I ordered the reed relays, because I do not want to stress the U101 too much in the long run.

The original relays have 330 Ohm coil resistance, the normal relays I use now have only 150 Ohm, so they are pulling 30 mA from the U101.
The new ones to come have a 500 Ohm resistance, so less current from U101.

[Satbeginner]

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?

Hi, yes, I checked all voltages around this array (U216 in the Pre-Charge circuit), and they all seemed normal.
With the meter powered off, I also checked the forward- and reverse voltages of the transistors, that looked normal too. (although I did the tests with the array still on the board)

Not sure what you mean by VOS?

un saludo

[Kleinstein]

To check the precharge offset, one could test the input bias current / input impedance for the 300 mV or 3 V range. A wrong precharge would cause excessive input bias, especially for short integration time with AZ active.

It might be a good idea to check this after a repair near the input section anyway, as the bias might also be influenced by contamination as well.

[HUGOMAN]

Hi Satbeginner

really you made a good work on your 3457A, my CONGRATULATIONS.

By this opportunity i hope that you can help me to solve almost the same problem as yours with my 3458A giving error 205 out of range 95, it fails during autocalibration on 1/10µA it goes out, then no measurement on ohm ranges.

i hope you could analyse the schematic to find out the faulty components.i will ready to make measurements.

Thanks
HUGOMAN

[Satbeginner]

Hi Satbeginner

really you made a good work on your 3457A, my CONGRATULATIONS.

By this opportunity i hope that you can help me to solve almost the same problem as yours with my 3458A giving error 205 out of range 95, it fails during autocalibration on 1/10µA it goes out, then no measurement on ohm ranges.

i hope you could analyse the schematic to find out the faulty components.i will ready to make measurements.

Thanks
HUGOMAN

Hi Hugoman,

I will have a look, will get the schematics first.

I am sure you have seen this site: 
https://www.element14.com/community/thread/46107/l/worklog-repair-of-20-year-old-precision-multimeter-hp-3458a?displayFullThread=true

Here I found something about the err 95, has something todo with the ref.voltage:

"DCI ranges 10nA, 100nA, 1uA, 10uA, 100uA showin zero, no matter what current supplied into meter. So those ranges are broken. That cause self-test fail on value 95 as well."


And this one:
https://xdevs.com/fix/hp3458a/

And I found this in another manual:

Agilent Technologies
3458A Multimeter
Assembly-Level Repair Manual

Page 77, 78

------------------------------------------------------------------------------------
205         Test Value Out of Range                See text

"TEST VALUE OUT OF RANGE" Error Message

This error message indicates that a test did not meet certain pre-defined
limits. The number displayed next to the message indicates the test limit that
was exceeded during the test

The number can be used to determine the assembly that caused the test to
fail. Numbers from 62 through 189 indicate a failure on the DC Circuitry
(A1) assembly.

Numbers from 190 and above indicate a failure on the AC
Converter (A2) assembly. Refer to the Assemblies Removal/Installation
Procedures in Chapter 3 of this manual to replace a defective assembly.

------------------------------------------------------------------------------------

[ManateeMafia]

Hi Satbeginner

really you made a good work on your 3457A, my CONGRATULATIONS.

By this opportunity i hope that you can help me to solve almost the same problem as yours with my 3458A giving error 205 out of range 95, it fails during autocalibration on 1/10µA it goes out, then no measurement on ohm ranges.

i hope you could analyse the schematic to find out the faulty components.i will ready to make measurements.

Thanks
HUGOMAN

Hugoman,

If you start a new thread with the description of your error, you can bet that more people will respond. The link by Satbeginner is also one of your best resources.

[lowimpedance]

Hugoman,

If you start a new thread with the description of your error, you can bet that more people will respond. The link by Satbeginner is also one of your best resources.

He did on Jan. 27th.

[ManateeMafia]

My bad. I missed that one.

[Satbeginner]

Hi all,

Still very happy with my working 3457A.
Today I received the replacement reed relays, so I replaced the temporary ones.

I used these: 5V 1x NO.

Un saludo,

Satbeginner

[wiesl]

Where did you get them?