Agilent DSO6034A issues with one channel

[tymm]

I recently picked up a DSO6034A (with MSO option) at a pretty good deal, though found the second channel isn't working properly and am trying to track down the issue (and hoping it's not hybrid related).

The channel seems to digitize just fine but there is a very large offset (across all ranges and input settings - shows in the 10s of volts in the millivolts ranges, up to something like 375V with 10x probe setting in the higher voltage ranges) and a lot of noise. When I feed in signals on the channel, they clearly are showing up, but due to the noise and offset the channel is unuseable.

I've spent a while digging in and think I've found part of the issue -- the noise and offset seem to be introduced in the channel biasing path -- I can see the noise and signal being fed into the biasing amp (for CH2 this is the second amplifier in U2003)... but I haven't been able to track down where the biasing voltage is being generated. I haven't found any appropriate DACs on the board (though it's quite possible I just missed them or they are hard to see)... at this point I'm assuming it's PWM generated but really don't know where...

I'd love help if anyone here is familiar with the analog side of these scopes.

I'm attaching annotated pictures from my spelunking in case they are useful to other people working on these. Note that for components, I've put the in-circuit measured values, not the values of the parts as they would be measured if removed. Also - apologies for the especially poor handwriting from iphone photo annotations.

First 2 pictures are from a likely unrelated look for the probe offset voltage path - I'm also unclear where this voltage is initially generated but was tracking back from the cable to the front panel channel 2's offset, to a precision op amp.

Third picture was measuring components in channel 2's channel offset voltage path. Here channel 2 component measurements match up with channel 1; the issue looks to be wherever the voltage being fed in is generated. I also see the noise and an offset coming in to the op amp's + input.

Fourth picture was from looking for points where the noise was obvious in the channel's input stage, before I tracked it back to the offset voltage, with some added details. X's are places where the noise does not seem to be present, O's are where it is present. To track things down I probed the locations with a probe hooked to CH1, so I could visually correlate noise on channel 2 with readings from channel 1 (note - the noise here was inverted vs. readings on CH2). Also includes some details on relays, SSR and digital switch. To add -- it appears that the 5 pin parts in the channel frontend are op amps. The SOT23-5 on my board marked "A36" appears to be an OPA336; the smaller 2 (SC70-5?) marked ".H3B" I'm not sure about.

Last picture - my low-voltage tracing probe hack (cheers to Shannon for the ST42 design that keeps excitation voltages low)

Thanks for any help!

[coromonadalix]

have you done a self diagnostic before playing in it ?

from page 101 of the user manual
HP_Agilent_Keysight/HP 6000 DSO Series User.pdf

free registration at Xdev's to get it, or web search ...



Utility & Service & Start Self Test performs a series of
internal procedures to verify that the oscilloscope is operating
properly.

Successfully passing Self Test does not guarantee 100% of the
oscilloscope's functionality. Self Test is designed to provide an
80% confidence level that the oscilloscope is operating properly.

[tymm]

Yup - passed self test multiple times without error

[tymm]

I have been starting to reverse engineer the analog front end, but for a quick test I lifted the output pin of the channel of U2003 that supplies offset voltage for CH2 and fed the output of CH1's offset voltage into CH2 - this worked great, though it means that CH2's offset only follows CH1's. This might be an OKish workaround if I really can't figure out where the non-buffered voltages are being generated but it's definitely not ideal. It does suggest I'm right though and that the problem is with wherever generates the unbuffered offset voltage.