Author Topic: sick Agilent E3634A  (Read 2463 times)

0 Members and 1 Guest are viewing this topic.

Offline djacobowTopic starter

  • Super Contributor
  • ***
  • Posts: 1151
  • Country: us
  • takin' it apart since the 70's
sick Agilent E3634A
« on: April 02, 2019, 01:33:54 am »

I fished an Agilent E3634A out of the trash bin at my place of work. Lucky for me, the unit was never tagged, so it went into regular trash rather than the magic trash from which items cannot be pulled.

Anyway, it doesn't work. Sometimes when I power it up, everything seems to work OK for a few seconds, then I get an Overcurrent Protection message. However, most of the time I turn it on I get two beeps, all the VFD lit, and then it goes out and nothing. There are a few other strange modes where it worked for a minute and locked up, etc.

I downloaded the service manual and measured all the voltages I could. Everything seems to be in order: 5V on the main board, -5V, -15V, 15V, plus the -17 and +17 that go to the front panel. There might be intermittent problems on one or more of these, but at least when I measure, everything seems solid. Service manual really doesn't have that much in the way of debugging -- though it does have the schematics.

Any ideas where to go next? I am suspecting the display board because of the two beeps at startup, which the manual says means that the processor failed to get a response. I tried reseating the connector on both ends. to no avail. But maybe that is just a downstream symptom of a worse sickness?

Anybody familiar with these units? I can't do all that much testing on the display board because most of the components are mounted under the VFD (why????).

 

Offline feedback.loop

  • Frequent Contributor
  • **
  • Posts: 255
  • Country: us
Re: sick Agilent E3634A
« Reply #1 on: April 02, 2019, 03:17:25 am »
Are you saying that sometimes you manage to set some output voltage, and display goes out? Does the output voltage change? If the display board is faulty, then losing display should not affect the output.
Could you try setting things using GPIB or serial interface?
 

Offline djacobowTopic starter

  • Super Contributor
  • ***
  • Posts: 1151
  • Country: us
  • takin' it apart since the 70's
Re: sick Agilent E3634A
« Reply #2 on: April 02, 2019, 04:05:53 am »
Are you saying that sometimes you manage to set some output voltage, and display goes out? Does the output voltage change? If the display board is faulty, then losing display should not affect the output.
Could you try setting things using GPIB or serial interface?

Yeah, I should have been clearer. This thing never outputs anything, at least nothing related to the value that was set in the display. For the moments when the display and buttons work, if I enable the output, I might get 0.7 volts or something, but I don't get what the oltage was set to. But I notice that the current is constantly set to 0.000A so perhaps that's why the actual output voltage is low/zero. I have tried to turn up the current setting, but even though I can hear "ticks" from turning the knob, the numbers don't change.

But 95% of the time the display goes blank and/or the keys lock up, so I can't even get to play with the output.

I have gone through and measured all the bias voltages again and they are all fine. +/- 12, +/- 15, +/- 5. I looked at them on a scope, and they are smooth, too.

Honestly, I was hoping the issue would be something in the "powertrain" of this unit, but it looks more complicated than that. :-(

I don't have GPIB controller. It has an RS232  port. I will try to dig up a USB/Serial adapter. I have the TTL level ones, but I need a real one.
« Last Edit: April 02, 2019, 04:16:16 am by djacobow »
 

Offline wictor

  • Regular Contributor
  • *
  • Posts: 123
  • Country: fi
Re: sick Agilent E3634A
« Reply #3 on: April 02, 2019, 04:34:35 am »
If display is working so that you can see the voltage and current numbers, then it means that communication is ok between main board and front panel. If there would be failure in components under VFD, you would not see any sensible data on display. There are fee tantalums in front panel that could be bad and make the unit shutdown or reset.

Front panel is sending button codes to main board and main board makes decisions to change values and beep when encoder is rotated.

I would concentrate to main board based on symptoms. Check that main cpu reset signal is stable. I think there has been also some RAM failures that could explain that kind of behavior.

Wictor
 

Offline wictor

  • Regular Contributor
  • *
  • Posts: 123
  • Country: fi
Re: sick Agilent E3634A
« Reply #4 on: April 02, 2019, 04:38:26 am »
The current does not change in normal mode if you don't have any load. When you press limit button, you should see max voltage and current setting. There current value should change, when current digit is blinking and you rotate encoder.

Wictor
« Last Edit: April 02, 2019, 04:40:04 am by wictor »
 

Offline djacobowTopic starter

  • Super Contributor
  • ***
  • Posts: 1151
  • Country: us
  • takin' it apart since the 70's
Re: sick Agilent E3634A
« Reply #5 on: April 02, 2019, 05:15:59 am »

I would concentrate to main board based on symptoms. Check that main cpu reset signal is stable. I think there has been also some RAM failures that could explain that kind of behavior.


So the reset signal PON_RESET is not stable. It is sort of hanging around 2V and sometimes jumps up to 3 or 4V (it's fast) and when it does so, at least some of the time the unit beeps and goes a bit crazy.

PON_RESET is connected to the main uC's (it's an Intel 80C196KB) /RESET through a little RC network. It also disables some thyristors that are used to switch which windings the PSU uses for the power path.

PON_RESET is driven by an LM2925. The datasheet for that part claims the reset output should be either 5V or <0.6V, so what I'm measuring points to that part of the circuit.

Too tired to proceed further tonight, but I'll be thrilled if I just need to replace that part... assuming I can get it!

« Last Edit: April 02, 2019, 05:53:33 am by djacobow »
 

Offline wictor

  • Regular Contributor
  • *
  • Posts: 123
  • Country: fi
Re: sick Agilent E3634A
« Reply #6 on: April 02, 2019, 06:10:24 am »
Don't worry, there should not be anything fancy with that reset circuit.
 

Offline cbutlera

  • Regular Contributor
  • *
  • Posts: 104
  • Country: gb
Re: sick Agilent E3634A
« Reply #7 on: April 02, 2019, 03:45:58 pm »
I think there has been also some RAM failures that could explain that kind of behavior.

Yes, I agree that it would be a good idea to check the RAM.  Try cooling or heating the RAM while monitoring the behaviour of the display.  A drop of IPA on the RAM will probably do if you have no freezer spray.  I have an E3642A that would reset itself randomly after it had been switched on for an hour or so, which turned out to be a faulty RAM.  In my case the RAM worked fine when cold but failed when it was heated.
 

Offline djacobowTopic starter

  • Super Contributor
  • ***
  • Posts: 1151
  • Country: us
  • takin' it apart since the 70's
Re: sick Agilent E3634A
« Reply #8 on: April 02, 2019, 05:39:32 pm »
I think there has been also some RAM failures that could explain that kind of behavior.

Yes, I agree that it would be a good idea to check the RAM.  Try cooling or heating the RAM while monitoring the behaviour of the display.  A drop of IPA on the RAM will probably do if you have no freezer spray.  I have an E3642A that would reset itself randomly after it had been switched on for an hour or so, which turned out to be a faulty RAM.  In my case the RAM worked fine when cold but failed when it was heated.

Some progress. I removed the resistor that connects between the reset output of the LM2925 and the RESET/ input of the 80196. I turned the thing on and ... it started working. I guess the in-built reset circuit of the 196 was able to do a reasonable job of resetting the processor after power on. On the 2925 side of things, the RESET output is a flatline at gound, so even though the main output of the regulator is working just fine, the reset output is  dead. So I suspect replacing the part and putting the resistor back will do OK.

And it sounds like the RAM is also frequently a problem, so I probably will replace that, too, while I'm in there.

Cautiously optimistic! I've been playing with the unit in CC and CV mode for the past ten minutes and nary a blip.
« Last Edit: April 02, 2019, 06:02:42 pm by djacobow »
 

Offline wictor

  • Regular Contributor
  • *
  • Posts: 123
  • Country: fi
Re: sick Agilent E3634A
« Reply #9 on: April 02, 2019, 05:59:52 pm »
Nice progress!

And if it turns out to be too boring, you can always play with LM2925 equivalent circuit found from here: http://www.buchty.net/ensoniq/replace.html.

Wictor
 
The following users thanked this post: djacobow

Offline djacobowTopic starter

  • Super Contributor
  • ***
  • Posts: 1151
  • Country: us
  • takin' it apart since the 70's
Re: sick Agilent E3634A
« Reply #10 on: April 02, 2019, 06:05:24 pm »
Nice progress!

And if it turns out to be too boring, you can always play with LM2925 equivalent circuit found from here: http://www.buchty.net/ensoniq/replace.html.

Wictor

Cool. I was just thinking about whether I felt like just putting in a 7805 in there along with a few other components to make a delayed reset signal.

I'm having all kinds of deja vu. First of all, the 80C196Kx series is a processor line I worked on back at the beginning of my career. I was a designer on a follow-on product at Intel, the 80C196SA. I also once owned an Ensoniq SQ-80.

Ah, the early 90's. Good times.
 

Offline feedback.loop

  • Frequent Contributor
  • **
  • Posts: 255
  • Country: us
Re: sick Agilent E3634A
« Reply #11 on: April 02, 2019, 06:39:49 pm »
I see quite a few LM2925T on eBay.
 

Offline djacobowTopic starter

  • Super Contributor
  • ***
  • Posts: 1151
  • Country: us
  • takin' it apart since the 70's
Re: sick Agilent E3634A
« Reply #12 on: April 02, 2019, 06:43:17 pm »
I see quite a few LM2925T on eBay.

Yup. I ordered two last night. :-)

$4.50 each, though, plus $3 S&H.

This seller claims they are NOS and not pulls. Not that I have any way of knowing.....
 

Offline SoundTech-LG

  • Frequent Contributor
  • **
  • Posts: 788
  • Country: us
Re: sick Agilent E3634A
« Reply #13 on: April 02, 2019, 07:11:43 pm »
Got a bag of those regs, just in case my E3631A needs one...
Thanks for the repair  details. Good to know the 2925 failure symptoms include your type of failure.
 

Offline SoundTech-LG

  • Frequent Contributor
  • **
  • Posts: 788
  • Country: us
Re: sick Agilent E3634A
« Reply #14 on: April 02, 2019, 07:16:06 pm »
Sounds like you may be near if you worked at Intel. I'm just over the hill from there. Quicker than ebay for the 2925...
 

Offline djacobowTopic starter

  • Super Contributor
  • ***
  • Posts: 1151
  • Country: us
  • takin' it apart since the 70's
Re: sick Agilent E3634A
« Reply #15 on: April 02, 2019, 11:46:27 pm »
Sounds like you may be near if you worked at Intel. I'm just over the hill from there. Quicker than ebay for the 2925...

Santa Cruz?

I actually did work in Santa Clara for awhile, though I did my work on the 16b controllers in Chandler, AZ.

I'm in the East Bay now. In any case, the part is ordered, so I might as well use that one.

Funny thing is that it works just fine without a reset signal at all. The reset node is connected between the uC and an ASIC, so I suspect one or the other has some kind of weak pullup combined with open collector pulldown, and a sequencer that once reset is triggered, holds it down for some period and then release it, and which by luck or design, works even without being driven at all.

While I'm waiting for the part I've been exercising the unit all day against my electronic load and there's been nary a blip.
 

Offline wictor

  • Regular Contributor
  • *
  • Posts: 123
  • Country: fi
Re: sick Agilent E3634A
« Reply #16 on: April 03, 2019, 03:39:54 am »
It may work without reset, but when that cpu voltage starts to drop or has not reached proper operating voltage and reset is released, you may get undefined behavior, corrupt some memory etc...

Wictor
« Last Edit: April 03, 2019, 08:01:14 am by wictor »
 

Offline djacobowTopic starter

  • Super Contributor
  • ***
  • Posts: 1151
  • Country: us
  • takin' it apart since the 70's
Re: sick Agilent E3634A
« Reply #17 on: April 12, 2019, 03:28:59 am »
Just to close out this thread. The unit is buttoned up and works just fine. Huzzah! I wish the VFD was brighter, but beggars cannot be choosers.

I ordered the replacement LM2925 parts on fleabay. I got two in case one was bad, which turned out to be a good idea because of my ham-fistedness.

These parts were the ones that came with the pins in a row, rather than staggered like they are on the PCB. I thought, no problem, I'll just bend them myself. I've done this a zillion times before with no problems, like when I "convert" a TO-200 into TO-263. Anyway, this time it was not a good idea. The first pin broke right off at the moment I tried bending it. I don't think I've seen metal fatigue on the first bend before. Suspecting that maybe these devices are not the real deal.

In any case, I just sort of feathered out the pins on the second one to get it in the footprint and soldered it in and it worked fine. I might get a few more from a different Ebay seller just in case.

I was surprised to see that this model PSU is still available from Keysight today!
 

Offline TheDefpom

  • Frequent Contributor
  • **
  • Posts: 702
  • Country: nz
  • YouTuber Nerd - I Fix Stuff
    • The Defpom's Channel
Re: sick Agilent E3634A
« Reply #18 on: April 12, 2019, 03:38:10 am »
This MIGHT be useful to you, I had the same part fail in an E3646A, I actually found that the 5V regulator output wasn't even used, it was only used for the reset circuit !

I did a couple of videos on it, I even made a little substitution board with a different supervisor IC on it until I received some "NOS" (actually they were used) parts.



Cheers Scott

Check out my Electronics Repair, Mailbag, or Review Videos at https://www.youtube.com/TheDefpom
 
The following users thanked this post: djacobow


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf