Fluke 5440A repair

[picburner]

According to the manual, this is the area where to look for the fault:

[MegaVolt]

According to the manual, this is the area where to look for the fault:

Thank you )

I turned on the output and even with these errors I see voltages from 20V and below )))) If you select a voltage higher, a guarded power error is triggered.  It looks like these are capacitors.

[picburner]

If the voltages greater than 20V do not work I would check the components highlighted in section A13.

[MegaVolt]

If the voltages greater than 20V do not work I would check the components highlighted in section A13.

Yes. High voltage has problems. I heard breakdown sounds during high voltage self-test.

[MegaVolt]

DAC DIGITAL FAULT
SECOND SWITCH

This error has been fixed. Replacing the optocoupler U18 (HCPL2601) on the A8 board.

The following error may be the reason for running the tests on a cold device )

[MegaVolt]

Warmed up the device. Started the calibration. On high voltage gain calibration, the unit reset and gave an error.

I understand the issue may be that the resistors went a little more than necessary?

[Badwolf]

I also recently repaired a 5440. The diagnostic mode with output on RS232 provides a lot of information and allows you to find your way on the diagram to identify problems. In my case, most of the errors were due to the relay. I didn't replace them ($ 100 each), I just used Dexoit + isopropyl alcohol to remove residue and prevent leakage.

[MegaVolt]

I just used Dexoit + isopropyl alcohol to remove residue and prevent leakage.

Looks like my repair is going to be a little more complicated

I also found that the A9 board has a date of 88.  It looks like it was replaced later.

This change was not an accident. I looked at the reference voltage and was very upset. It looks like the board was already in the hands of a technician and one of the SZA263 is marked with a black marker. This board was put as a known bad one

I also observe oven oscillations that never end

I also checked the output voltages when they were stable. The ratios between 10V and 1000V are very good. Better than I can accurately measure. So there is nothing wrong with the divider resistors.

[Kleinstein]

The oven oscillations are unusual. This is normally a question on how good the initial design is not so much about failing part. There may be a electrolytic cap used in the regulator compensation that may have failed. The other point is too much delay from the heat to the sensor, e.g. from thermal past missing or loose screws.  With a schematics available this should normally be fixable.

For the black dot, this may be from inital selection of the parts. One finds dot's on slected chips quite a lot.

[MegaVolt]

Here are pictures of the inside of the oven.

[MegaVolt]

continue

[MegaVolt]

With a schematics available this should normally be fixable.

What should I change to remove the oscillation?

For the black dot, this may be from inital selection of the parts. One finds dot's on slected chips quite a lot.

It's not like the usual dots. It looks more like it was painted over.

What else could explain this behavior of Vref?

[Dr. Frank]

I also have a (newer) 5442A and investigated on the interesting stacked 13V reference.

This stacking circuit is partly on the small reference board with both SZA263, and partly outside the oven assembly, if I remember correctly.
Maybe you print the different parts of the schematics to understand how this circuit works.

I don't know, where exactly you probe these 13V. W/o extender card it's also not that easy.

I suggest that you connect two cables directly on the reference board choosing ground and  a connection directly at the upper SZA263 output, and then analyze the signal in situ. Pay attention not to create any shorts, especially by proper selection of isolated cables!

I doubt that the SZA263 itself is damaged. Maybe one of the LM308 is breaking down thermally, i.e. one of its input transistors, as the whole assembly is at about 55°C permanently.

Maybe you might also dis - connect the reference board and rebuild the stacking circuit on a test board at room temperature, so you can decide where the failing OpAmp, or other component is sitting. 

Another idea: As the oven might also have problems, check the stability of the supply voltages.

Frank

[MegaVolt]

I also have a (newer) 5442A and investigated on the interesting stacked 13V reference.

They are identical in my opinion.

I don't know, where exactly you probe these 13V. W/o extender card it's also not that easy.

I suggest that you connect two cables directly on the reference board choosing ground and  a connection directly at the upper SZA263 output, and then analyze the signal in situ. Pay attention not to create any shorts, especially by proper selecting of isolated cables!

Here are the connection points.  The only extra element is the relay. Otherwise I have a direct connection to the zener.

I doubt that the SZA263 itself is damaged. Maybe one of the LM308 is breaking down thermally, i.e. one of its input transistors, as the whole assembly is at about 55°C permanently.

Maybe you might also dis - connect the reference board and rebuild the stacking circuit on a test board at room temperature, so you can decide where the failing OpAmp, or other component is sitting.

Thank you for your words. They are encouraging. An op amp is no big deal I'll keep looking

[Kleinstein]

The oven part looks a little odd. It looks like just a simple proportional regulator. As shown it could be relatively sensitive to ripply and noise coming in and not very stable.  As the resistive heater has a square law the loop gain thus goes up with higher heater power. It also looks like the transistor gain would have an effect.

I had expected more like an PI regulator. The high values for R115 / R116 look odd and may be a left over from an original plan to have a capacitor in series to R116. 
My suggestion to fix / improvementthe regulator would be to reduce the gain at higher current. A possible way would be to replace R114 with something like 2x 10 K in series and than something like 3 K and a LED in series to ground from the center. So at the higher end the gain would be reduces about 3 times. This would still be only a small step not a really good solution. A better way would be more like changing to a PI regulator - though tricky with a electrolytic cap (likely need something like 100 µF) at the warm temperature.

[chuckb]

I noticed that the heater is made up of 8 Carbon Comp Resistors.
If their total value has drifted significantly then the heater gain would also change. A lower total resistance would have a higher "voltage to watts" gain.

Is the thermal oscillation present with all the normal insulation and installation in place?
If the heater block is being testing on a cool bench (or even with covers off) it will need a higher voltage to reach temperature and it will have a higher gain from the square law issue.
If it was installed in a warm calibrator the heater will have a lower voltage and less gain from V^2/R.

Good Luck!

[Kleinstein]

The jumpy reference voltage could be from a cold soder joint, failing bond wire or maybe a bad relay contact.

For the heater part the transistor to control the heater is itself a powerful heater.

[MegaVolt]

To get a better understanding I have redrawn the schematics of the Fluke 5440 voltage reference.

[MegaVolt]

Today is a strange day. I found a problematic part. And I should be happy... But it's a R18 fluke 1.68 kOhm unobtanium resistor and I don't know where to get a replacement (((

When you heat it up, it increases many times. I have seen a figure of 27 kOhm. And it takes a very long time to get back to the initial value. The hysteresis in the picture is 0.4 ohms which is 300ppm. Maybe it's a bad contact inside. Does it make sense to fix it?

Maybe someone has a spare pair of 5.18kOhm and 1.68kOhm resistors? Or should I look for all 4 resistors: R18, R22, R19, R21?

[Kleinstein]

The resistors are not that special - mainly trimmed relatively stable resistors. The 5 K part is even not that kritical. The point may be that the resistors can be trimmed / selcted to match the reference chip they are used with. With this resistors one can trim the TC of the reference to a reasonable low value (e.g. < 5 /K). The final precision than comes from a stable temperature.  So one could still search for a suitable value in the 1.4 to 2 K range and than get a relatively good resistor to replace it. I don't know how critical the resistor is exactly, but expect a attenuation similalar to the LTZ1000, so something like an drft attenuation by a factor like 100 to 500. So 100 ppm drift of the resistor would translate to 0.2 to 1 ppm change in ref. voltage.

At least for the start a relatively normal resistor can do.

[MegaVolt]

While I'm looking for a replacement resistor I tested the rest.

Some show stable hysteresis to one side. For example, the resistor in the picture adds 5ppm after each heating.

As a result after 20 turns on of the device we will have 100 ppm. I think this is a lot.

Is it worth to change such resistors?

[MegaVolt]

I returned

1. Corrected the scheme. C5 fixed to C16.
2. I drew a board with tracks. (this file is for Photoshop layered. Let me know if you need it >50MB)

[leighcorrigall]

...

For the black dot, this may be from inital selection of the parts. One finds dot's on slected chips quite a lot.

It's not like the usual dots. It looks more like it was painted over.

...

The black dot does appear on another reference board that I have. It may help distinguish one SZA263 from the other when selected resistors are being installed.