Electronics > Repair
Series defect on agilent 167xx boards?
(1/50) > >>
DocBen:
Hi there,

I recently acquired some used 16750a boards for my 16902a logic analyzer. Curiously some of them seem to have similar problems in the selftests:
Zoom Acquisition Chip Select Test Failed!
Chip 0: Master Clock from Chip 0 Test Failed!
Comparator Test Failed! (only one Pod connected to one Chip fails, not always the same one)

When inspecting the boards for damage I couldnt find anything odd except some small green nodules forming next to the plastic parts on the bottom. When I took them off I noticed that the adhesive material was hard as rock (should be soft I think) and some of the copper traces underneath are erroded. (really erroded not ripped off  ;)

I think the failing selftest are related to the damage to the traces and more specifically to the clock traces (differential pair on the bottom left).
Did anybody repair a problem like this?

When looking at working cards they seem to develop the same problem. Right around the edges there are small green nodules. But the traces underneath have not been damaged (yet).

After reading
https://community.keysight.com/message/57746
is it just me or is this a common problem?
simmconn:
The way I see it, it is a common problem, and you've found the cause. Either some chemicals in the adhesive/gasket is corrosive, or it absorbs moisture and causes corrosion. I don't see a good way to fix it as in many cases the residue of the adhesive/gasket is very difficult to remove, and repairing the broken trace/vias in a multi-layer board is no easy task. I may be pessimistic, but I think all 16500 series and 16700 series plugins with the plastic stiffener will eventually die because of this. It's just a matter of time.

Working cards may have the same corrosion, only that it has not eaten away enough copper to cause problem. I usually clean the corrosion with IPA but they some back after a while.

The 16900 series plugins use different stiffener (shorter and thinner) and adhesive (transparent instead of foam/felt like). Maybe Agilent knew something and took on a new design.
MarkL:
simmconn is right.  This is a known problem with the adhesive on the plastic runners.  I have 20 or so of various 165xx and 167xx cards and they all have this corrosion to some degree.  It seems to be particularly pervasive on the 1675x cards.

The only fix is to remove the runners and, I also agree, clean up the board with IPA.  Do an end-to-end continuity check on all traces running under or near the runners.  Don't trust if they look "ok".  I've had some where the traces look fine and later found out they were corroded and severed *under* the solder mask.  I've seen some with corrosion down the vias, which could render the board unrepairable.

I've also seen boards where people have not been careful about inserting it into a chassis and there are cracked/broken components and PCB traces on the underside.  Do a thorough survey under a microscope.  One of my boards had an invisible fracture in a ferrite bead which cut the power to a chip, which in turn killed the acquisition clock.

I have a 16702B chassis which runs HPUX.  On it there's a command line utility, /usr/sprockets/bin/pv, which is the equivalent of the self-test GUI tool.  It can be used to turn on more verbose output from the self-test routines (set debug, mode, and/or result levels to something other than 0).  Some of these values can also be set from the GUI under the "Options" menu on the self-test.  I have no experience with the 16902A, but I would poke around for something similar.

Regrettably, there's no documentation for any of the verbose self-test levels or the meaning of the output.  You have to just play with it and guess as to the meaning.  However, I found that some of the output, at least under HPUX, will actually point you to specific chips or signal lines that aren't behaving as expected.  It's better than nothing.

That all being said, it's not guaranteed you can fix one of these boards.  I still have a bunch of dead ones even after fixing all the severed traces I could find.  If you bought it as "Used", I would consider returning it since it's supposed to be functional.  It can turn into a huge time sink.

EDIT: For the curious: Added photos of a 16755A with a severed trace (fixed now), and a picture of the corrosion.
DocBen:
Great!

Well, not great, but I thought I'd been cheated on those boards.

I wonder if Agilent did that on purpose to limit the lifetime of their boards?

@MarkL: how did you repair them? Did you just bridge the missing part of the trace or reroute the connection entirely with wire?

I've attached a selftest log for posterity.
And I agree without actually knowing what part is tested where its hard to understand.
That and the fact that most technical information is in the Logic Analyzer Software Help File and not in the cards datasheet ;)

All I see as pattern is clock. Except for the last 16752 which probably also has a defective memory chip I guess.
we'll see when I have time to try to repair them
MarkL:

--- Quote from: DocBen on November 20, 2017, 08:54:06 am ---I wonder if Agilent did that on purpose to limit the lifetime of their boards?

--- End quote ---
I doubt it.  It took many years to manifest itself before they even knew they had a problem.


--- Quote ---@MarkL: how did you repair them? Did you just bridge the missing part of the trace or reroute the connection entirely with wire?

--- End quote ---
The damage was too long to be bridged with solder.

I scraped the soldermask off both sides of the break until I found good solid copper instead of crumbly powder.  I then bridged the path using a single strand extracted from one conductor in a ribbon cable (around 0.12mm == 36 AWG).

Hint: Tin the tip of wire, and then use extra flux on the bare trace and the wire before tacking it in place.

Below is a photo where I repaired 15 paths in one area.  The white outline is where the runner sat, and it totally destroyed 11 or 12 of the traces directly under it.  I patched them all to be sure.  There were several other areas with breaks, but unfortunately repairing them all did not bring this particular card back to life.
Navigation
Message Index
Next page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod