Dingy Agilent E8358A VNA repair

[dzseki]

We have this „boat anchor” at work. Our company got this instrument through an acquisition of an other company many years back. As good as it was, we did not have a use for this instrument because we either work with frequencies above 10GHz, or bellow 10MHz. It was rarely used, and was generaly uncared-for. Now we have an upcoming project where we can (to better put: could) use it finally. Unfortunately as it appears this VNA didn’t age very well.

There are various problems with this puppy:
-   CMOS battery dead (easy fix).
-   Built in screen mirrors in horizontal plane from time to time (external display is OK).
-   Front panel buttons not always working (then VNA software complains about some error with ’fpservices.exe’).
-   Front panel USB over molested, does not work.
-   Self test throws an error for Port 1 receiver attenuator, for all steps. (this appears to be the most serious issue).

The VNA operates with the 266MHz CPU board and Windows 2000, with its original spinning HDD. I have a feeling that we should make an image of this drive while we can…

I have gathered basically all related documentation that is available from Keysight site: Service notes, Service manual, etc. I am optimistic with the USB/display/front panel thing, but a little bit unsure about the attenuator. To be honest for our target application 1 port would be enough (to check reflection vs. frequency), so at the moment it is not a must to have both ports completely flawless, and disassembling may do more harm alone than the projected yield…

Does anyone have any in depth experience with this line of VNAs?

[joeqsmith]

There is a thread in the test equipment section for it.   

CMOS battery is a bit odd.  You can find the originals still but you may want to consider using something more common.  We covered it in the other thread.   

Attenuators may just be the bumpers (small O-rings).  There is a video on-line about replacing them.  DO NOT FOLLOW THEIR INSTRUCTIONS!!!  It was brutal to watch, similar to the old Samsonite luggage commercials with the gorilla in a cage. 

I would make an image before anything, but that's just me.  Had someone write me about messing with the drive and loosing everything.

[joeqsmith]

The power of the search engine:

https://www.eevblog.com/forum/testgear/agilent-e835xa-pna-vna-notes/

https://www.eevblog.com/forum/testgear/agilent-e8357a/

[Bud]

Does anyone have any in depth experience with this line of VNAs?

You could try sending a PM to Dr.Joel (of Keysight) with a pointer to this thread and asking to help with troubleshooting the attenuator. He normally hangs out in RF and Microwave board of EEVBlog.

[dzseki]

Does anyone have any in depth experience with this line of VNAs?

You could try sending a PM to Dr.Joel (of Keysight) with a pointer to this thread and asking to help with troubleshooting the attenuator. He normally hangs out in RF and Microwave board of EEVBlog.

Sorry, but what is his exact screen name here? I've tried to look him up in the members list but to no avail.

[Bud]

Joel_Dunsmore

Edit: Here is a similar thread on HP VNA coupler repair:
https://www.eevblog.com/forum/testgear/agilent-8753es-a21-test-port-coupler-repair/msg3992420/#msg3992420

[dzseki]

In the meantime I've installed a CR2032 holder instead of the old one, it was basically a drop in replacement. Disassembled the front panel stuff as well. Basically found nothing interesting on it, BTW this one has the E8356-60003 FPI, the USB connector IS in good shape. It is a litle bit odd that they in fact populated a complete USB hub on that board. Considering that I had problems with all three functions served by the front panel it must be a general problem with the board, most trivial would be an intermittent contact on the bus connector.
Sure enough reassembling everything on the front panel made the front USB work again (for now at least), monitor is also good. I could not test the front buttons, because HDD is not installed yet. I don't want to put it back until I can do the image backup.

[dzseki]

- Saving the HDD image was successfull, as I've tested the image on a different disk, and worked as it should.
- I've spent a lot of time with finding out what could be the problem with the front panel. I came to the following conclusion: If there is no external USB keyboard attached to the system the front panel works (no complain about error in running fpservice.exe), if I use an USB HUB at the back of the system with a keyboard and mouse plugged in, I always get error with fpservice.exe, and thus the front panel doesn't work. If only an USB keyboard is plugged in at the back of the system, then fpservice.exe may work or may not, adding the USB keyboard only after the vna software strats, then it is working without a problem.

But now I face a more serious problem 
While poking with fpservice thing (plugging in and out the USB devices, and several system restarts) the VNA part refuses to work at all... 
The software starts, but is very slow, the UI is basically frozen, no trace is shown, and time to time this error message pops up: 'Unknown hardware error'. I also miss the relay click sound that I usualy heard when the program stood up.

I have reseated basically all boards, including the CPU board, without success.
If I disconnect all boards (except for the CPU and FPI), then VNA software throws an error message not finding any related hardware.
If I put back the SPAM card, then VNA software can start, but the problem is the same: Unknown hardware error...

I would think about checking the supply rails first, but I did not find any testpoints for those either.
Any ideas?

[dzseki]

Not having better idea I've tried the other HDD on which I've checked wheter my backup image was working or not. Then the picture have changed, the VNA software started, but it complained about not having a serial number for the tool (previously unheard problem), and 'source unlevelled' error, nevertheless the software started and drew some trace.
These problems were eventually solved by reseating again the SPAM board, so with the backup HDD the system worked.
I put back the original HDD, but the 'Unknown Hardware error' message came back, with the frozen VNA software. I don't know what happened, but wiped the HDD and restored to my working image (thankfully I had that!), so the VNA is functional again, well, except for the receiver attenuator.

This tool has the configurable test set option, I see in the SM that VNAs without this option doesn't even have this attenuator in the signal path at all, why is that?

[Miek]

This tool has the configurable test set option, I see in the SM that VNAs without this option doesn't even have this attenuator in the signal path at all, why is that?

That option is intended to allow measurement of higher power devices, so it adds those two step attenuators on the test receiver inputs to be able to handle the higher power (they don't exist at all in the standard configuration).

It's a bit odd that you're getting an error for all steps, if it were stuck or if some pads were damaged I'd expect some steps to still pass. Are all the front panel jumpers present & undamaged?

[dzseki]

I also wondered about how all ranges could be wrong...
Anyway when I examined the front panel board I had to remove the jumpers, these are now installed again, but I doubt that I reinstalled them in the same order, yet the error remained the same. The "cables" look nice, except for one, which have a small dent, but that is now installed at Port 2 anyway.

I have disassembled the attenuator to the point until it was easy. Now I can at least operate the cams manually, one of the steps is indeed very flaky, the second is so-so, the third appers to be rather good. with all steps off the feed through is good, all tests at DC obviously...

[dzseki]

Attenuators may just be the bumpers (small O-rings).  There is a video on-line about replacing them.  DO NOT FOLLOW THEIR INSTRUCTIONS!!!  It was brutal to watch, similar to the old Samsonite luggage commercials with the gorilla in a cage. 

It was rather convenient to slide off the cams with a needle (under microscope).
For sure the O-rings are sick, one is outright cracked/broken, the rest are simply flattened, now I need the O-rings

[dzseki]

While the O-rings on their way, I've just found the HP Journal artice where these attenuators are described:
HP Journal 1991 April
from page 47.

[nexus]

Interesting thread. I just about wrapped up my restoration of a once dead E8358A. It was an early unit with opts 015 and 010. Clean and low hours from the looks. 15V standby rail went dead in PSU (and replaced all the RIFAs), so that was a quick fix and it came back to life.

Mine had the 266 MHz CPU running Windows 2000 and very early software on a 10gb HDD. I first made a forensic image of the disk using macrium reflect, so I thought i was safe. Well, I goofed the PNA software while messing around and in a panic I reimaged the original HDD. Guess what, my disk image was not bootable, and no matter what I did I couldn't get win2k to boot. NEVER REIMAGE A BOOTABLE ORIGINAL DRIVE. Wish someone had told me that sooner.

At any rate, I tried so hard to get the old cpu with win2k working, an anonymous benefactor took pity on me and donated a 500 MHz CPU with a bootable win XP drive. Back in business. (Saved the new spinning drive and copied the image onto an IDE SSD directly from the VNA using an old version of reflect. Boots from SSD now, nice speed boost.)

After that, I rebuilt the attenuators. It's an option 015 unit so it had both source and receiver attenuators on both ports (4 total). Quite a few hours and frayed nerves, even though I've probably done about 50 HP style step attenuators by now. I do a full breakdown which involves desoldering the control PCB from the solenoids, carefully disconnecting the push rods from the contacts with two sets of tweezers (one holding, one guiding/pushing), and separating the solenoid block from the main body. Only then can you remove the pushrods and replace the o-rings properly. I use a small drop of clock oil on the o-rings and the solenoid piston. Works a treat and is safe with the Buna rubber. I've done this many times on units that went to Keysight for cal, never had an attenuator issue on the measurement report.

Got it cal'd and she is working great! I would recommending running through most of the adjustments (10 MHz, 3.8GHz PMYO, source and reciever power leveling). The only one I didn't run at the cal lab was the LO power adjust, since you have to pull the cover and start moving things around. Op check passes just fine, so I think LO power to the reciever mixers is at an acceptable level.

[dzseki]

I took me a while to get the O-rings replaced.
Interestingly I still some got error with the attenuator.
With DVM all positions make a reading, but software complains about the 5dB steps, By manual switchin (I mean from software) I can't even hear the relay switching, for the 5dB, only for the major steps.
First I tried this attenuator at the other position -> same problem.
Then I disassembled to access the coils and came to the conclustion that the 5dB relay is strange, while the other relays try to remain in their position if moved by hand, but this 5dB relay is not like that, it actually likes to flip to the other position, rather than staying where it was. Could it be that I accidentally moved the magnet the other way around?

[dzseki]

It was the magnet. But Yesterday with the 5dB steps I only got error wit hthe 5dB step related point, the rest was OK. Now as all steps are working I again get error on all steps.
By manually switching the attenuator and comparing it to the other's (outputs open) I don't see much difference.
So I am a little bit puzzled with this one...

[Miek]

Could you post a screenshot of the Receiver Display (System -> Service -> Utilities -> Receiver Display) with nothing connected to port 1 or 2?

[dzseki]

Could you post a screenshot of the Receiver Display (System -> Service -> Utilities -> Receiver Display) with nothing connected to port 1 or 2?

Here it is.

[dzseki]

In the meantime I think I've just figured out what's going on.

The OpCheck always giving some 60dB Trace-pp value.

When I reassembled the attenuator I noticed that at the 5dB step one of the edgeline center conductors (as they call it) flips a little later than the other, as if it would stick to the attenuator card.
When switching the attenuator from the softwre there is a transient that can be seen on the trace, for a moment because of this "timing issue" the attenuator breaks up (see attachment, which was caught by 'Print Screen'). The other attenuator does not have this transient.
The operator's check is exercising the attenuator with each step. So the 5dB is either swithed on or off at each step. On the other hand when the 5dB step did not work because of the flipped magnet, the 10 and 20dB related steps did not give any errors, only for the 5dB steps (no transients). Probably if there would be more dead time between switching and measurement, the OpCheck would also pass.

So afterall the attenuator works as it is, but during switching it gives a transient, I don't see this at the moment to be a serious problem. We'll have to live with this...

[Miek]

Yeah, that 60dB pk-pk result was what had me wondering about the receivers, but those traces looked fine.

Nice find! I agree, it doesn't seem like a serious problem - in real use you don't want to be switching that attenuator constantly anyway. I guess the only worry is that it might become stuck open at some point, but it'd be easy to spot if it did happen.