[solved] Agilent infiniium 54855A calibration problem

[Aiy]

Hello everyone,

I am having problems calibrating my Agilent oscilloscope, the failure I get is
Trigger for all four channels.

The output of the calibration routine is
Chan 1 Trigger errors
Interp.Gain
AnalogPath Skew
TrigPath Skew
Pred Delay
LTrig Delay

Does anybody know where to start troubleshooting this?

I am using the correct calibration cable and GHz BNC-to-SMA connectors

Best regards

[Aiy]

I noticed today that if I connect the Aux Out to Channel 1 and put Aux Out into Probe Comp mode the signal captured is -1V to 0V. This does not make sense to me, I would have thought it would be a 1kHz 1V peak-to-peak signal for probe compensation.

Can anybody confirm?

I believe that the Aux Out port is connected to a mux with GND, DC-component, 1kHz/1V/p2p and 456MHz-isch demo clock. Maybe one of the supplies for the Probe Comp is bad?

Does anybody know where on the acquisition board the calibration signals are located?

Best regards

[Aiy]

I checked another Infiniium scope, it appears that the probe cal signal is a square wave -0.5V to 0V with ~718Hz frequency, so my gut feeling was wrong this time.
Gonna probe the clocks on the acquisition board to see if there's anything fishy

[Aiy]

I picked up this project again

The scope triggers funny with external and low frequency signals.
It seems the oscilloscope can sample its own internal 1GHz clock just fine without showing any jitter, I put my probe on the VCO output and it works.
This points to a VCO jitter problem. I have traced the Acquisition board VCO to the synthesizer, an old national semiconductor LMX chip.
The synthesizer is programmed when the scope-app is launched.

I looked in the datasheet for the synthesizer and the setup is Fpd = 1MHz, Fref = 10MHz => R=10, Fin = 1GHz => N=1000.

The synthesizer is programmed so that MUXOUT/LD pin puts N counter on it. I have not yet found where it goes.

Best regards

[Aiy]

Hello Everyone,

I tried a disassembler searching for strings in the agscope.exe file (My scope runs version 5.0 after factory reset) and found out that the .exe accepts some interesting flags from the commandline:

Code: [Select]

agscope.exe /scope /internal /hpib /service 

gives access to an extra menu under the "Utilities" menu with LOTS of hidden stuff and extra info in the self-test utility.

Flags I found interesting:
factory
service
servicecal
defaultsetup
internal
testflash
hwdebug
debugknobs

I found some information on the Agilent forum over at groups.io, and apparently the old hp5484x .exe file has similar flags.

I also tried a CW source as input to the scope. It seems that the trigger works fine with frequencies over ~500MHz. I feel that this is a big clue to what is wrong with the scope. Perhaps the trigger IC output drive is somehow defective, but the point is that with high enough triggering frequency it works. Which would point to the circuitry after the trigger circuit (U21 Is the trigger IC I believe).

Gotta try to trace the trigger circuit output to the ADC/ASICs next.

Best regards

[Aiy]

After a lengthy period of troubleshooting I've finally solved the Self-calibration problem.

The trigger MUX sends the trig-out signal to a bunch of ECL circuits at the edge of the board.
The Trigger signal is passed through a 1:2 differential clock splitter as well as first input to a NAND gate.
One of the outputs of the clock splitter is sent to a D-flop and I noticed that one of the diff-pair was behaving strangely. It turns out it was missing termination att the end.
I traced the signals through to the D-flop and a couple of other components, among them are differential analog delay lines (2.0ns and 4.5ns) marked MS200D and MS450D. These are obsolete Susumu Group or TFT (Thin Film Technology) components that are no longer manufactured nor in stock anywhere.
It is an interesting component, a Susumu GL2L5MS450D-C which is a dual 4.5ns delay line in a SOIC-16 package.
One of the delay lines was broken, thus the lack of termination.
I managed to patch a 51Ohm termination to Vee=-5V at the broken input of the delay line (i.e. for the D-flop to have correct termination for the ECL pair), and swapped a 61.9Ohm resistor at the second input to the NAND gate mentioned earlier (to shift the resting voltage to the middle of the differential pair level)

After these patches my instrument now passes Self-Calibration.

I'm doing this write-up if anybody comes across the same issue with their oscilloscope.

The special menus that become available with the extra flags mentioned above are helpful in order to jump in to the middle of the calibration algorithm when experimenting.

Best Regards
AIY

[super_electron]

amazing work and thanks for the tip to get the service menu. I have a DSO8000 scope which has the exact same issue. Pass all the self tests and vertical cal  but fails trigger during calibration. Based on your tip I was able to capture the calibration debug logs but not sure what to make out of it. the logs are attached.

[Aiy]

Hi there,
The txt file was a mess to read in notepad, but after some \n insertion I found this:

Code: [Select]

Starting ---- IntOscDurMst Chan 1 ----
MST Osc Duration Cal
Failed < 5ns - no trigger found
>>Time 2 Seconds
Finished ---- IntOscDurMst Chan 1 ---- FAILED
Starting ---- ExtOscDurMst Chan 1 ----
MST Osc Duration Cal
Failed < 10ns - no trigger found
>>Time 3 Seconds
Finished ---- ExtOscDurMst Chan 1 ---- FAILED

There might be a hint in there what to look for, "MST" and "Osc" and "duration" ... Could be a missing oscillator signal on your acquisition board.
Do you have high-res photos of your acq board?
A tip: I found that you can run parts of the calibration routine separately using the command-line options (can't remember which right now).. You get access to a special panel where you can run only the test your having problems with while debugging.

[super_electron]

some pictures.

[super_electron]

picture of the back side.

[Aiy]

Hi again,
I have made an initial annotation on your photos, trying to illustrate the different areas of interest. I'm doing it from memory since I do not have my notes at home..

In my case, the delay lines were faulty close to the trigger ICs. Check them with continuity tester (look up datasheet for pinout, they do not match mine from what I can see).
The ECL/LVDS circuits are also part of the trigger / calibration circuitry, as well as signal conditioning for AUX and LINE triggers.
My suggestion for debugging is to take high-res photos of the ECL/LVDS stuff, create a list of all ICs (Uxxx silkscreen marks) and look up their datasheets (modern equivalents by ONSEMI should be possible to find).
Then proceed with a continuity tester to find how they interconnect, unless the traces are visible on the top/bottom layers of course.

If you do not find anything obviously wrong, like open connection on the delay lines you should proceed to power the board and probe around carefully with another oscilloscope to check the MC100 ICs. If any of those are broken, they can be ordered and replaced quite easily.
NOTE: you have to create sufficient airflow over the heatsinks or the oscilloscope will shutdown from overheating!

You did not post a description of how the oscilloscope trigger problem shows itself when you try to use it, the self-cal is only part of the troubleshooting. You should power it up and put a sine signal on one of the input and sweep it through the oscilloscope's frequency range and watch the screen. Post screenshots of the screen if you see symptoms of problems with the trigger.

I see that you also have problems with your motherboard in another thread, but perhaps you have a way of powering and debugging the the acquisition board while you find a motherboard replacement.

Best regards,
AIY

[Aiy]

Acquisition board backside annotated image.

Best regards,
AIY

[super_electron]

Thanks for highlighting the critical areas in the picture. Here is the picture with part number of the delay line . The part number on the delay line is (MS160S). I believe it is a GL1L5MS160S. The ECL parts are fro ONSEMI with part numbers 10h131g an 10h116.

[Minsoon]

Nice to meet you friend
My equipment seems to have a similar problem.
When I took a picture of the resistance value of my delay line, the resistance of one side of the line was very large.
Can I request more details, photos, or drawings of the delay line repairs?
thank you good luck!