HP 54503A scope attenuators

[dzseki]

Got this scope the other day off e-bay. It was sold as deffective, and eventually showed up as DOA. Fortunately it was only a blown fuse and some corrosion in the internal PSU connetor, so the scope came alive.
I started to check the rise time performance of the scope with my pulse generator. This generator has vague rise time specifications (only says <1ns), but at work we measured with a LeCroy 4GHz scope to be 425ps.
So with this generator I got nearly identical rise time results on CH3 and CH4 around 600ps (which is quite nice from a 500MHz scope... )
CH2 on the other hand has some sort of under damped behavior, so the resulting rise time is around 1.2ns, and is basically independent of V/div setting as I saw 
There is a sticker on the front panel stating CH1 50 Ohm termination is open, and it is indeed open, however using a Tee adapter with an external 50 Ohm terminator it shows similar rise time as CH3 and CH4.
Is there a way to adjust the compensation of the attenuator on CH2?
See picture below, from top to bottom CH4-3-2-1

[dzseki]

So I got around to take a look into the CH1 attenuator: To my biggest surprise, the problem actually is not the 50Ohm terminator being open, but the relay does not make any contact instead. The relay itself is moving by applying voltage on the coil, but the contacts read open all the time. I did not have the gut to try to remove it from the substrate...
As for the CH2 I think there should be some grounding issue with that as at some point the rise time got identical to the other channels.

[shakalnokturn]

Or, just flaky relay contacts on the 2 most used channels...

[dzseki]

Or, just flaky relay contacts on the 2 most used channels...

Indeed -I was away from this topic for a while- and now as I revisited the CH2 problem at lower frequencies it became aparent that the range relays on CH2 are flaky, all of them. It is matter of luck whether it is accurate in any range or not. At higher frequencies this is less of a problem, then this only manifests in the poorer rise time. The 50 Ohm relay seems to be working though on that channel.

Too bad that all my attenuators are electrically OK, only the relays give some headache. Perhaps I should look for a defective attenuator where the relays work at least and investigate the possibility to change them.
Also I was wondering  -since there is not much to loose- what if I drill a small hole on the relay house, apply some contact cleaner (eg. Contact 60) then flush out (Tuner 600 or Contact WL) and cover back the hole...

I know there there are attenuators available around the net, but each of them are more expensive than what I've paid for the whole scope, and in fact from that much money I could buy an other donor scope as well...

[MarkL]

It's actually not that hard to clean the contacts on these attenuators.  I wouldn't drill any holes in the them.  You might create more of a problem with debris.

Each coil comes off by unsoldering only two pins.  The thru-holes in the ceramic are quite wide compared to the pin which allows a solder sucker to be very effective in removing all the solder on the first go.

Inside you'll find gold-plated posts soldered onto the ceramic, and a free floating strip or two of gold-plated spring material.  The strips can be removed and cleaned separately from the posts, but be gentle with them since they are thin and delicate.

I had several attenuators I was able to restore.  I tried several different contact cleaners, but in the end what worked was silver polish.  The polish is slightly abrasive and removed whatever layer of junk had deposited on the contacts over time.

Photos below.  The coil pins are the two, large outermost ones.  Don't unsolder the posts.  I unfortunately didn't take a photo of a coil removed.  This attenuator is from a 16534A logic analyzer oscilloscope card, but it is of the same design.

[dzseki]

So I got a backup attenuatur from here and therefore got brave in the repair, thanks MarkL for the tips.

I was able to fix the not working 50 Ohm input termination on CH1 this way.

Then I went on with the CH2 instability, I managed to lose one of the little springs three times in a row (those were some hard times), but in the end I managed to find it and was able to reassemble.

I used Kontakt 60 applied to a tooth pick (wood) and cleaned the springs and posts with that, then washed down with Kontakt WL.

But strangely CH2 did not get any better, in fact it only became worse. Now CH2 is totally whacko. At low mV/div settings there is only a few LSB of signal appearing (while input was 2Vpp) at 5V/div  this looked more like 15Vpp, in middle ranges (eg. 200mV/div) I got nothing on screen.
I pulled out the attenuator again and checked each relay contact, in both states, and it works fine! No other sign of failure.

So I put in my newly acquired spare attenuator and with this it works good, only that it reads some 10% lower voltage than it should, I guess I have to recalibrate that channel? If so how to do that?

For reference I've taken a picture of the relay springs.

[jaimebcn]

Hi everybody,

Nice to meet you.

Finally I find some info about this issue so I have the same in my HP 54610B (Two channel).

When I try to compensate my probes and after 5 minutes approx the square signal goes down and down like I show below  (it happens only in the Channel-1) :



Then goes to this:


And finally


- It happens when I go up to 200mV/div.

- If I change the scale to the next upper step, 500mV / div and over ... I've got the signal that you can see in the 3rd image.

I tried to follow the Self Calibration and I've got a Fail in the Gain Channel 1:



I swapped the modules in between themselves (Ch-1 on Ch-2 and Ch-2 on Ch-1) and... Now the calibration fails on Ch-2 and I got the same problem on Ch-2 when I try to go from 200mV/Div to 500mV/Div.


(I took the picture from above as an example. Mine is exactly the same but it has 2 PCBs. One for each channel.)

The most interesting thing is that it happens when I left the time pass. If I try to turn off the scope for a long time (like an hour) and I turn on it again the Ch-1 works fine again ... until 5 minutes pass approx. Then fails again in the same way.

Thank you!
Cheers,

[precaud]

I encountered similar issues in the past with HP 54100-series attenuator modules. In most cases, it was caused by dirt or dust finding its way into the relay mechanisms, accumulating over time and and hindering the actuator. Most of them were fixed by soaking in IPA (using a syringe to spray it in), actuating it while wet, rinsing with IPA, and repeating if necessary. Use a vacuum to suck out the excess IPA and let dry completely before testing. I'd guess I had 75% success rate doing this.

[jaimebcn]

I encountered similar issues in the past with HP 54100-series attenuator modules. In most cases, it was caused by dirt or dust finding its way into the relay mechanisms, accumulating over time and and hindering the actuator. Most of them were fixed by soaking in IPA (using a syringe to spray it in), actuating it while wet, rinsing with IPA, and repeating if necessary. Use a vacuum to suck out the excess IPA and let dry completely before testing. I'd guess I had 75% success rate doing this.

Thank you precaud!

I am so sorry but what do you mean with "IPA" ? IsoPropyl Alcohol?

So as I guess, the way to clean the relays could be unsoldering it from the PCB and then clean it with IPA? That's correct?

Cheers,

[precaud]

I am so sorry but what do you mean with "IPA" ? IsoPropyl Alcohol?

Yes. Its an acronym I learned on eevblog 

So as I guess, the way to clean the relays could be unsoldering it from the PCB and then clean it with IPA? That's correct?
[/quote]

Only unsolder if necessary. Disassemble it the minimum amount to gain an access to the moving parts of the relay with a syringe needle. The 54100's required no desoldering and minimal disassembly.

[MarkL]

...
I used Kontakt 60 applied to a tooth pick (wood) and cleaned the springs and posts with that, then washed down with Kontakt WL.

But strangely CH2 did not get any better, in fact it only became worse. Now CH2 is totally whacko. At low mV/div settings there is only a few LSB of signal appearing (while input was 2Vpp) at 5V/div  this looked more like 15Vpp, in middle ranges (eg. 200mV/div) I got nothing on screen.
I pulled out the attenuator again and checked each relay contact, in both states, and it works fine! No other sign of failure.

Was this a visual check?  I would check the ohms with a DMM across the contacts, in both positions, if you haven't already done so.

On many of my relays there was a thin layer of something that no contact cleaner could remove, which is why I had to resort to liquid silver polish.  I used a toothpick to rub it on at the contact areas on both the posts and the springs, then cleaned well with isopropyl.

So I put in my newly acquired spare attenuator and with this it works good, only that it reads some 10% lower voltage than it should, I guess I have to recalibrate that channel? If so how to do that?

There is a self-calibration procedure described in the manual (pg 12-9):

  http://bama.edebris.com/download/hp/54503a/54503A%20User.pdf

It doesn't say what the procedure is, but rather just to follow the displayed instructions.  During the procedure it will probably want you to connect a BNC jumper to the back DC calibrator output.

[dzseki]

Hi MarkL,

Of course cecked the contacts with DMM, all read below 1 . Although, I am not worried about this too much, providing I can succesfully self cal the "new" attenuator.
Thanks for the link for the manual!

Also, I am wondering if anyone has ever tried to check the attenuators on-the-fly, tied via extender wire?

[dzseki]

CH2 Self cal went through without any problems. In fact this attenuator seems to be quite nice since the measured rise time on this channel stays below 500ps, the other channels are ranging between 510-710ps, while the generator rise time is around 425ps... 

[jaimebcn]

I encountered similar issues in the past with HP 54100-series attenuator modules. In most cases, it was caused by dirt or dust finding its way into the relay mechanisms, accumulating over time and and hindering the actuator. Most of them were fixed by soaking in IPA (using a syringe to spray it in), actuating it while wet, rinsing with IPA, and repeating if necessary. Use a vacuum to suck out the excess IPA and let dry completely before testing. I'd guess I had 75% success rate doing this.

Hi precaud,

I've got a spray can of IPA.
As  you said you fixed this issue by soaking in IPA the relays and then actuating it while wet.

How did you activated the relays while they were wet? Did you do this with the attenuator module out of the oscilloscope (I guess) ?

Thank you very much.

[precaud]

As  you said you fixed this issue by soaking in IPA the relays and then actuating it while wet.

How did you activated the relays while they were wet? Did you do this with the attenuator module out of the oscilloscope (I guess) ?

Thank you very much.

Yes, with the 54100's you can remove the input module fairly easily. And the relays can be actuated manually too, all of the actuating levers are accessible with a small screwdriver.

[jaimebcn]

Yes, with the 54100's you can remove the input module fairly easily. And the relays can be actuated manually too, all of the actuating levers are accessible with a small screwdriver.

Hi precaud,

My attenuator module is just like this:



The only way that I could clean the 3 relays is by unsoldering the relays from the PCB. And then spraying and actuating the relays as you told me. I guess.

I will try this weekend or as soon as possible. I will let you know.

Thank you!

[jaimebcn]

Hello,

I still have the same problem.
Sometimes when I change the impedance between 50ohm and 1Mohm I get an overvoltage error.
Is it possible that I left the relay magnets on the wrong position when I reassembled it?

Thank you very much.

Regards,
Jaime.

[precaud]

I'm sorry, but I'm out of suggestions, Jaime. Maybe someone else has some experience with them?

[jaimebcn]

I'm sorry, but I'm out of suggestions, Jaime. Maybe someone else has some experience with them?

Thank you very much precaud. 
Anyway I probably will buy a replacement later (I have to think about this , due this is an expensive part). It looks it will be the solution.

Cheers!

[dzseki]

Putting the magnet in the wrong place should be no problem, as once you step through all the ranges they should find their place. In my case I found that some relays were harder to move than others, thosey may stuck "inbetween" places causing unusual errors.

[jaimebcn]

Putting the magnet in the wrong place should be no problem, as once you step through all the ranges they should find their place. In my case I found that some relays were harder to move than others, thosey may stuck "inbetween" places causing unusual errors.

I know that it will sounds strange or maybe savage, but if the issue could be about the "moving magnets" and his movement ... If I try to put a really small quantity of lube/grease non conductive inside? Or I am saying something really stupid? 

Cheers,

[dzseki]

I'd think this is related to mechanical wear in the rail or the "car", ie. the magnet-car easier gets stuck in the rail, especially if the spring forces are not balanced, no lubrication can help on that.

[rdebets]

The magnets are polarized and if you put them in the wrong way the relais will switch the wrong way as well. I had to re-unsolder some of them when I noticed that. The relais are driven directly from the header pins so you can easily apply 12V in both polarities and check that they move in the right direction. I cleaned by taking the contact springs out and used Contact 60. It visibly took the dirt off. Contact resistance on some of the bad switches was up to 300mohms and went down to 30 mohms after cleaning (not absolute numbers, it was what my 34401A with 2 wire resistance measurement & nulled would tell me).