Agilent E3631A readback issue

[dave the sound guy]

I'm finally having a go at a sick E3631A power supply that I've had on a shelf for a while.  It passes self-test and all three outputs function normally when measured externally (and correctly respond when adjusting the encoder wheel), but the voltage and current readback is very wonky for all three channels. 

I'm proficient at repairing audio equipment, but sophisticated test gear like this power supply is a little out of my comfort zone.  Needless to say I'm up for the challenge and would gratefully welcome any guidance. 

I have the service manual and have been studying it.  The first thing I did was to check bias voltages as described on page 98...everything was in order.  I also checked resistors in the ADC circuit and found none were open.

[dave the sound guy]

I wanted to eliminate calibration error as a possibility, so I attempted a re-cal and it mostly failed.  Oddly it did accept calibration on the 6V voltage but yielded no improvement to the readback.

I'm having a little trouble following the schematics and I'm not sure where to start looking for the fault.  I have a second 3631A that I can use for reference however.

Here's a short video that better illustrates the unit's behavior:

[SoundTech-LG]

I could be wrong, but that sure looks like a rotary encoder issue. When you turn it, it does not seem to raise or lower the voltage correctly, and jumping around. Common issue with these too! I see the DVM looking correct though, so maybet not...

[dave the sound guy]

The encoder definitely works as it should.

I might be on to something.
I opened up my good supply and started to compare measurements with the faulty one.  With both set to the 6V output at 6V I had a look at the FLASH output in the ADC circuit at R29.  The top image is the good supply, the bottom is the defective one.  The signal was also unstable on the bad unit.  Looks like I should start working backwards...

[dave the sound guy]

Output of U5.  Top is from the healthy unit.  Note the amplitude difference.

I measured both sides of R4 between U3 and U5.  U3 side is a steady -10 on both, U5 side is -29mV and steady in the reference unit, and -200mV and slightly fluctuating on the bad one.

Bad U5?

[SoundTech-LG]

U5 likely suspect...

[free_electron]

check the resistors around the analog mux. (4051)
you can't measure that signal. the convertor uses current injection. The voltages tell you nothing as the convertor tries to steer against the error.
Especially the reddish colored resistor or the black ones with yellow lettering. sometimes they are blue too. i have seen them go bad. they are subject to moisture and the electrodes dissolve.

i've seen R4 R5 and R15 R16 go bad. especially R5 and R4.
The AD706 U3 creates a positive and negative reference. R5 and R4 are the current 'injector' resistors for the reference voltages
U7a takes the signal to be measured and R15/R16 create an injector current for that.

The way this converter works is it charges / discharges C19/C18. The comparator U7b lets the controller know if the converter needs to steer up or down. the flash converter takes a reading to know by how much it is off. basically COMP tells you direction and Flash magnitude.
if FLASH overloads then they force a charge/discharge using the positive and negative reference to bring the signal back into range. The controller can adjust the time it injects the charge. Essentially this thing takes a 'guess' , injects a current to remove or add a bunch of charge on the capacitors and then uses the a/d convertor to digitize the remainder. The time used to inject the charge is a direct measure for the voltage on the caps. The end goal is to have zero volts on the capacitors.

Why not use a better a/d : those are pricey especially if you want many digits. you also needs precision resistors and everything needs to be trimmed to get correct readings.
The beauty of this kind of converter is that you do not care about trimming the resistors or any other analog muckery. the whole thing can be calibrated digitally.

[dave the sound guy]

Thanks for the explanation.  That makes a little more sense to me than the theory of operation as written but it's still a lot to wrap my head around, not having a formal electronics education to speak of... 

I re-checked the resistors...they test correctly and agree with my reference unit.  For good measure, I also probed on the devices connected to either side of the resistors to ensure the solder hasn't failed anywhere.

What would be the best way of verifying if U5 was good or bad?

I just had a look at voltage on C18/19 and read 2.4v (stable) on the reference, and around 2v fluctuating on the not-so-good one. 


I think I'll borrow a hot air station and try swapping U5 and see if anything changes. 

[TheDefpom]

I did some repair videos on the E3646A, they share a lot of the same design aspects, one of the issues was a bad read back which I solved by replacing an opto-isolator if I remember rightly, sorry I don't remember which video it was either, it was a few years ago now, but you might want to troll through them anyway in case it helps.

[Fabi1]

Hi,
Thank you for your analysis which helps to solve a readback issue on Agilent E3631A power supply,
R4 & R5 resistors around U3 were open circuit, the power supply now works well after replacement with original 30k value.

note: the first fail was the 625 error, solved by U1 & U2 optocoupler replacement, with HCPL-2211-360E reference.

Fabien.

The attached schematics comes from the full Agilent service guide, available here (too big file for forum attachement)
https://engineering.case.edu/lab/circuitslab/sites/engineering.case.edu.lab.circuitslab/files/docs/Agilent_E3631_Power_Supply_Service_Guide.pdf

[free_electron]

there is something about those 30K resistors that is stinky... i have seen at least 5 power supplies with those things open. it's not an overload , it must be a manufacturing defect, either the resistor itself is bad , or they get damaged. i know certain fluxes , if not removed properly, can eat the electrode interface ( where the metallic electrode touches the carbon material that makes the resistor. ). this is typically because of sulfur remnants (or a high sulfur environment) . That's why we have anti-sulfur resistors now.

https://www.koaspeer.com/applications/anti-sulfuration/?gclid=Cj0KCQjwntCVBhDdARIsAMEwAClvGT0babTzf0MHSws_FCh5rYm5wF5QRzDd8UvpjXrffHSQrXFEBBEaAuR8EALw_wcB