HP 8594E Spectrum Analyzer repair (I hope)

[TerraHertz]

This one:
https://www.eevblog.com/forum/testgear/hp-8594e-spectrum-analyzer-at-last-i-own-a-decent-spec-an/

ebay 122514312297
HP/ Agilent 8594E Spectrum Analyzer + Options: 041, 105, 140 + User Manuals
Programmer's, User's, & Calibration Guide/ Manual
Powers Up, Not Tested, 30-Day Warranty

The options are:
  041, HP-IB & parallel printer interfaces
  105, Time gated spec analysis
  140  Narrow bandwidth and precision freq ref

I got it for US$600 plus shipping (and the Oz gov 10% GST on used goods AND shipping), so it's come to well over a thousand.

Unfortunately there are two issues. One is that the seller left out one of the three manuals and despite initial positive response has now failed to respond since the 8th . Maaaybe they are still snowed under?  They are in Burnsville, MN, and the weather there has been pretty bad.  Of course thanks to Murphy it was the User's Guide missing, which I need because...

The machine has a hardware problem. It powers up, the display seems sensible, but the front panel controls are strange. At first they seemed completely dead except for the Preset button. Then I found that after pressing buttons, a looong time later the machine sort of responds. Some screen updates, something between a few seconds to 15 seconds later.
Also the freq plot on screen, updates about every 10 seconds. Shouldn't it be continuously updating, in this default mode? It says "sweep 58 mSec"

Mystifying. My wild guess would be that somehow the CPU clock is running extremely slow?

Anyway, this machine is a bit daunting to me for several reasons. It's RF magic. It's very densely packed. I've never used a spectrum analyzer before, so have zero confidence in operation and knowing what to expect.

I have all the manuals and schematics (that I know of, including the CLIP set from Artek) but the ones I need for repair are all PDFs. Which slows down comprehension a lot due to the crappy access times (for me anyway.) The paper manuals I have are the Programmer's Guide and Calibration Guide. I've ordered a paper copy of the Quick Setup guide, but that won't be here for a while. As for the User's Guide, and whether I'll ever get the one I paid the seller for, or have to buy another, who knows.

At the moment I'm just reading the service guide PDF, with the machine case removed.

Number one objective is to find out where the battery is, and see if it's flat or not. I had assumed it is flat due to the 'FREQ UNCAL' message on screen, but now it seems it would more likely be saying 'Using defaults' or something like that if the battery was dead. Also the date shown on screen is correct (for the USA.) Wouldn't the RTC use the same battery?
Might the 'freq uncal' be related to the weird CPU lag effect?

Anyway, I haven't yet sighted the battery. It's buried somewhere. Till I figure out where it actually is, I dare not start disconnecting anything. Oh and of course the advice 'first thing, record the cal settings' is useless, due to dead front panel controls.  Manual reading...

One question that someone familiar with this model could answer: When I press the green PRESET button, the screen goes blank for a moment, then for about a second is filled with 'CPU nuts, wrong sync, complete garbage' before a normal display returns. See last photo. Is this normal behavior? Or maybe another 'CPU running very slowly' effect?

Edit to add: A link to the unboxing photos and video. Related to one manual being missing.
  http://everist.org/pics/HP8594e/

[TerraHertz]

Naturally, the battery is on the CPU card. What I meant by 'couldn't find the battery' was 'couldn't find the processor card.'  It didn't take much manual reading to find it. This one:



Somewhat buried under multiple other modules. The battery is about where the oval is, and can only just be seen with a torch. The board has to come out. All good, there's a clear description of the removal procedure in the manual.

Incidentally, by this time I'd noticed something that may be related to the fault. When I'd cleaned the outsides, there was a small amount of a brown residue on the left side. A splash of something. it came off easily. Now I find some had got in through the vent holes, and dripped down both the case inside, and the mainframe. Very little, but maybe some got onto the CPU card. I can't tell yet. At a guess I'd say soft drink. So probably conductive. We'll see.





But first...
In all my electronics career I've never bothered with 'anti-static stations', wrist straps, conductive mats and so on. I prefer the alternative - just learn simple movement habits that control and dissipate static charges via paths other than through electronics. Always touching metal equipment frames first, or ground points on cards. Keeping some skin touching metal frames at all times when working on gear. If picking up or putting down some static sensitive part, first palm-flat on the surface, to equalize potential (desk surfaces are always slightly static conductive, just from normal dirt, sweat, etc contamination.)

So far as I know, this always worked - I never static blew a component. (Fumbling a charged CRT EHT lead against a deflection PCB in a Tek 7104 scope - that's a different kind of screwup.)

But with this spectrum analyzer, I want to be super cautious. Today I bought an anti-static mat and wrist strap from Jaycar. The only type they had. If I'd bought a real one from the net, I'd have had to wait for it to arrive. This:



As expected, the wrist strap lead end to end is 1 meg ohm. The black ground lead is direct wired from alligator clip to wrist strap sockets, and 1 meg to the mat pop-clip. So that's fine. But what about the mat? It looks like just a sheet of vinyl, but with a pop-clip in one corner. Hmmm... This is made in China, why not be suspicious?

First check, that I didn't expect to work: Press a metal pad on it about 1cm from the pop-clip. Put 25V DC between the pad and clip. See if a multimeter (Fluke 77) can see any differential voltage between either pads, and points on the sheet between.
A: not a sausage. No detectable conductivity.

I guess it's expected to be very high impedance, to only move electrons around a little. Expecting a garden variety multimeter to see anything is unreasonable.
OK, so how to give it a fair test?
I have this:




It's a hand held tesla coil thingy. Applies really really high voltage, high frequency excitation to whatever is near the tip. Typically used to find leaks in vacuum glassware, because electrons will find even the tiniest passage. You look for where the glow goes.
Point being, to this thing even many 'non-conductive' objects will pull an arc from the tip. Dry concrete, wood, paper, etc. You can judge how conductive so-called insulators are, by how much the corona discharge is drawn to them. Here it is with concrete.




Let's try this on the 'anti-static mat.'  First quick try:



The test platform is a sheet of perspex, spaced up off the ground on cardboard boxes. An adjacent concrete step serves as 'ground'. (No really, for this it's good enough.)   It's late dusk, 2 second exposure, camera on tripod.
The mat other end is sitting on the concrete. HV tester tip is about 1cm from the closer mat corner.
Hmm... From my experience with the 'leak tester', that mat is no more conductive than any ordinary plastic.

For a comparison, a bit of wire with its other end point-touching the concrete. Competing in arc length with the 'anti-static mat.' Note that almost anything will draw the arc a little, even small objects in free space, since the arc is HF AC, and most things have at least a little charge mobility. Even plain vinyl plastic.




All right, I need something that could be considered a fair competition. Let's see how the 'anti-static mat' compares to... bubble wrap. The setup:


And running:


Good heavens. There's no discernible difference. This made in China 'anti-static mat' is no more conductive than bubble wrap. It probably is just a bit of plain vinyl sheet.
That's really funny.

Meanwhile, no anti-static mat.

Great.

[Bicurico]

Hi,

Did you have a look at Keysight: https://www.keysight.com/en/pd-1000002155%3Aepsg%3Apro-pn-8594E/portable-spectrum-analyzer-9-khz-to-29-ghz?pm=PL&nid=-32440.536881805&cc=ZM&lc=eng

You can download all relevant manuals here.

There are some posts/threads in this forum regarding battery replacement.

I will be following this thread in a near future...

Regards,
Vitor

[MarkL]

The unit is definitely sick.  The display should not do that on PRESET or any selection.  It should also be updating the sweep a lot faster.

The fact that it responds but is very, very, slow makes me suspect the processor is being pummeled by interrupts, maybe from some failing sub-system.  The screen garbage on PRESET may be a consequence of the slowness.  The CPU could be resetting the video controller and then taking a really long time set up the registers again for a proper display.

On your RTC question, the battery does run the RTC.  Hint: When/if you remove the small card that the battery is mounted on to replace it, the RTC will reset.  This is because the RTC is on the main board and receives power from the small card.  So don't panic when that happens.


This sounds like a HW failure to me, and with any failing equipment, I'd start with checking the power supply outputs.  Then follow Chapter 4 in the Assembly Level Repair manual to see if any of their troubleshooting hints apply and could provide more clues.

One of the many things in the manual is to do a DISPOSE USER MEM via the CONFIG menu to clear the memory of any misbehaving DLP's (programs).  I don't give this a high likelihood of making any difference with these symptoms, but it's simple to try.

[SaabFAN]

That almost looks like someone spilled coffee on the unit.
Fortunately, the affected side is far away from the sensitive RF-Components.

I'd suggest replacing the battery. With a battery approaching 23 years of age (unless the date on the back of the unit is wrong) i wouldn't trust the calibration-data to be correct.

Btw. I went through the service manual a few days ago as I wanted to buy a similar device and it seemed to me that only the flatness-correction needs special equipment (a signal generator with a known and stable signal level from 9 kHz to 3 GHz). The remaining entries, for the YIG for example, are generated by the unit itself when running the cal-routines. For that the CAL-Output needs to be working.

If you don't have a signal generator capable of these specifications, the ADF4351 Module made by SV1AFN have a pretty flat amplitude over their whole range and can be controlled by an Arduino with ease.
Also have a look at my Sig-Gen project in the Open Source Hardware section here.

[TerraHertz]

Hi,

Did you have a look at Keysight: https://www.keysight.com/en/pd-1000002155%3Aepsg%3Apro-pn-8594E/portable-spectrum-analyzer-9-khz-to-29-ghz?pm=PL&nid=-32440.536881805&cc=ZM&lc=eng

You can download all relevant manuals here.
There are some posts/threads in this forum regarding battery replacement.
I will be following this thread in a near future...

Thanks for the link! All downloaded. That is an overlapping set but with some differences to another Agilent site URL. Here's my current set of doc resources for this machine:

http://artekmanuals.com/manuals/hp-manuals/    search for 8594E  finds:
 * Service     8590-90316   $10.00   ASSEMBLY LEVEL SERVICE ( NO SCHEMATICS)
 * CLIP     5963-2951   $25.00   FULL SCHEMATICS SCANNED AT 1200 DPI (fax mode, not quite high-res enough. Includes the above.)

https://www.keysight.com/en/pc-1000000308%3Aepsg%3Apgr/discontinued-859x-series-portable-spectrum-analyzers?pm=LB&nid=-32440.0&c=204590.i.1&to=79830.g.0&cc=AU&lc=eng
   Discontinued 859x Series Portable Spectrum Analyzers  (list of docs, 4 pages)

https://www.keysight.com/en/pd-1000002155%3Aepsg%3Apro-pn-8594E/portable-spectrum-analyzer-9-khz-to-29-ghz?pm=PL&nid=-32440.536881805&cc=ZM&lc=eng
   8594E Portable Spectrum Analyzer, 9 kHz to 2.9 GHz    (list of docs, 2 pages)



@ MarkL
Thanks for the confirmation re the screen. I had felt pretty confident it was not normal, since any product that ever did that would be a constant source of irritated user queries for the manufacturer. I figured it had to be related to CPU slowness in running through CRTC setup.
Interrupts deluge - ah, yes, that's another strong possibility. Should have thought of it myself, thanks. With 'drink on PCBs' anything is possible. Suspect is the CPU card on that side. For starters, a visual check when I get the card extracted.


Because I have to pull the CPU card anyway for that visual exam for surface contamination, I'm going to do the battery replacement with the whole card out. Which means I can supply an alternate backup floating voltage during the swap. (Card floating during soldering. Or maybe there's a battery rider board?)
This means I don't _have_ to solve the CPU speed problem (and keyboard inoperability) first, to do a backup of the config settings.
Also given the probable battery age, replacing it asap is the 1st priority.

I'll check the PS outputs first, but I note that all rail status LEDs on the PS are lit. And I'll read Chapter 4, thanks for the hint.


@ SaabFAN
Yeah, coffee or coke. Not very much of it, fortunately. Actually it looks like it was a spray, hitting the side of the machine in a small area. Horizontally in through the case vents, then dribbled downwards on all surfaces. There are only two PCBs to check for contamination - the CPU, and the PCB in that flat metal case at the very bottom.
There's so little it might be unrelated to the fault.

Calibration gear - that's a future issue. I'm ignoring it for now. I may have something.

[MarkL]

The SRAM has a supercap which will keep it powered while you replace the battery.  You have about 8 hours.

This is explained in the battery replacement procedure in the Assembly Repair manual.  Referenced in your first thread.

[TerraHertz]

Assuming the supercap is still functional after 20+ years, and was charged to normal 5V or full battery level, rather than 'battery on the verge of going below working voltage' level.

I think I'll just provide external power to be safe.

Today I returned that non-anti-static mat to Jaycar, and ordered a real one from Element14.
Still reading manuals.

Oh, on the 15th  the ebay seller finally responded again, but the issue of the missing User's Manual is still not resolved. They claim 'can't find it', yet have that exact manual listed for sale, saying "Qty available: 4". I'll give them till early next week then if they haven't stopped with the stalling, will give up on them, leave a negative (with link to full documentation), and buy one elsewhere.

Edit to add on 16th: Ah ha! The seller hadn't checked their own listing photos. Because the machine is an 8594E, they were searching for a manual called "8594E User's Guide". Despite their own listing photo showing "8590 E-series and L-series Spectrum Analyzers User's Guide". So they ignored their pile of 4 of those exact manuals (one of their other current listings, with photo looking exactly the same.) Looks like *that* problem is resolved.

[TerraHertz]

The anti-static mat jokes just write themselves. The one I ordered from Element14 arrived. An anti-static mat, packaged in an anti-static plastic bag.

But it is a real one. 2 mm thick black rubber with a non-slip texture, definitely conductive; about 1 K between any two corner pop studs. Actually you could use it for a heater pad, assuming some way of making contact to it that didn't develop stinky hot rubber contact points.

Only one tiny complaint. The direction they rolled it, when placed flat with the pop-studs facing up, the sheet ends want to curl UP.

[Floopy]

I just (somewhat finished) repairing a HP 3561A. Hp service manuals are amazing! It's hard to track down a fault when you don't know the history of the machine. You don't know when it stopped working and what might have happened to it. In my case the unit had been sitting in storage for over a decade and something went out of tolerance. (and the power supply decided that life wasn't good enough)
Good luck, I want to see how this evolves.

[CJay]

The anti-static mat jokes just write themselves. The one I ordered from Element14 arrived. An anti-static mat, packaged in an anti-static plastic bag.

But it is a real one. 2 mm thick black rubber with a non-slip texture, definitely conductive; about 1 K between any two corner pop studs. Actually you could use it for a heater pad, assuming some way of making contact to it that didn't develop stinky hot rubber contact points.

Only one tiny complaint. The direction they rolled it, when placed flat with the pop-studs facing up, the sheet ends want to curl UP.

Yeah, I've got some of those mats, I ghave up trying to get them flat and moved on to some blue ones from Farnell instead, they are heavier and lie flat, more 'rubbery' than plastic.

Try not to forget that they're conductive like one of my colleagues did, he spent a good half an hour wondering where the burning smell was coming from and why the bare board PSU he'd just repaired had gone bang.

[TerraHertz]

Yeah, I've got some of those mats, I ghave up trying to get them flat and moved on to some blue ones from Farnell instead, they are heavier and lie flat, more 'rubbery' than plastic.

Since I have to find some matching pairs of male & female pop studs anyway, and the pop-studs on this mat are badly pressed, I'm going to remove the existing ones, replace just one on the appropriate corner but the other way up. Then it will lie flat.

Try not to forget that they're conductive like one of my colleagues did, he spent a good half an hour wondering where the burning smell was coming from and why the bare board PSU he'd just repaired had gone bang.

Ha ha. That's why I measured it. Wanted to know if it was safe to place a CPU card with dying battery on it. It's not.
The first mat (red one from China/Jaycar) was not conductive at all. This one is way TOO conductive. But I can live with it (and am extra poor atm, so have to.)

[Bicurico]

OK. Today I received two broken HP/Agilent 8594E units.

Both were sold to me as broken.

The Agilent one switched on and booted but after 1-2 minutes of operation the PSU died.

The HP one is surprisingly working without a fault so far:

I swapped the good PSU into the Agilent unit and it boots, but the symptoms are exactly as described by OP: the unit does not refresh the spectrum and respons very slow to any key.

The reason is simple: someone removed some parts between the RF input connector and the next stage, so the input connector is not connected to anything. This makes it impossible to do the CAL.

But when I try to do the CAL, after the error message, the device starts responding OK: fast response and spectrum refresh rate.

I think the problem is AT LEAST due to the missing input and lack of CAL.

Worst case scenario is a broken YIG.

Now I am stuck with the broken PSU: I disassembled it, inspected it, cleaned it and inspected it again. There are no broken electrolytics or other burned components.

Any suggestion what would be the most common faults on this PSU?

This Agilent unit is newer than the HP one and has more options: 041, 105, 140, 151 and 163.

Does anyone know if I can transplant the modules into the working unit? Or would I need to activate them by personality cards, too? (I don't have any)

Regards,
Vitor

[Bicurico]

Managed to connect RF input port to cable - it was just disconnected.

However, no signal appears and thus the device is responding very slowly to any command and has slow refresh rate. After failed CAL responsiveness gets to normal, but spectrum only shows noise - no signal at all.

The CAL OUT does produce the 300MHz signal at -20dBm - I measured it with another spectrum analyser.

So now I have on my TODO list:

1) Get the PSU running again - this will be a hard one, as I am not experienced with PSU repair. I guess I will have to start replacing components, which is a rather lame approach. Any suggestions/help is welcome.
2) Disassemble the device and figure out why there is no signal at all. I fear it is the most expensive and hardest to obtain component: the YIG
3) Replace the battery on both devices

Question: I ordered a memeory card (32KB). Can this be used to store the calibration data, in case the battery fails?

Regards,
Vitor

[CJay]

PSU repair, check high value resistors and low value caps on the primary, check *all* diodes on the secondary, a single rail shorted could cause the whole thing to be dead.

[Bicurico]

Found a dead resistor on the PSU's main PCB. Should be 215 Ohm but was dead. Replaced it with a 220 Ohm one I had in stock, but unfortunately that did not fix the issue.

Was kind of hoping for something simple as described here: https://community.keysight.com/thread/5394

But the resistors pointed out in above thread all measure correct values.

I'm unsure what to do next, as all three boards look perfectly OK and there is no evidence of any failed component.

I tried to measure diodes and capacitors onboard and all the equal ones measure the same. I know I should measure them off-board but that is a hell of a job.

Grateful for any further tips.

@Terraherz: I hope you are not mad at me for "highjacking" your thread - I just thought to tune in as the device has apparently a similar fault as yours (apart from the broken PSU).

Regards,
Vitor

[Bicurico]

Arghhh...

Now my second unit died. It was the PSU, too.

Now I have two broken PSU's! Shit!

The dying was like this:

The device was running.
Then it suddenly turned off.
But just by touching the pressed in power button, it turned on again.
Then it turned on/off/on/off by just touching the device.
Then it turned off.

I disconnected the power cable and smelled if there was the typical burned smell. There was none.

After 1-2 minutes I turned it on again and it switched on.
The device run for another 5 minutes and then it switch off.

Definitely.

I tried the PSU on the other device, to make sure that it was the PSU that failed and indeed - no sign of life on the second unit.

Because the PSU's both died in the same way, I suspect this: can it be that the temperature sensor on the heatsink or the associated logic is broken? Shutting down the PSU?

Uff... going to bed now - frustrated.

Was beginning to write a new version of the "VMA Simple Spectrum Analyser" software for the HP 859X series...

Regards,
Vitor

[TerraHertz]

@Terraherz: I hope you are not mad at me for "highjacking" your thread - I just thought to tune in as the device has apparently a similar fault as yours (apart from the broken PSU).

No not at all. I've been too busy with other urgent chores to begin on disassembly of my unit. Hoping to get to it today actually.

I don't know why you'd think a disconnected input would make the machine run slowly and not sweep. Be aware that there are some gold SMA and SMC(?) connectors visible in the 8594E that don't have anything connected. That's OK, they're for other options. I'm pretty sure mine is not missing anything. Though of course I'd rather I did have the extra option cards and tracking generator.

[Bicurico]

Update:

After sleeping over it, I tried again this morning to turn on the "good" device (HP). And to my surprise it did turn on!

So whatever happend yesterday was due to heat:

1 - The temperature senser might have shut down the PSU: I don't know how this sensor works (it is attached to one of the heatsinks) - can this get broken? How to replace it?
2 - The electrolytics might be dying and due to heat failed (temporarily), resulting in PSU shutdown
3 - Some other component fails when heating up
4 -

Can someone point me in the right direction? What would be the most probable cause?

I imagine that the PSU will evetually fail completely, so I will have to do something about it. Also, I think the other PSU must have faile for a similar reason.

@Terraherz: You are right! Just connecting the input port does not make any difference. I have not looked at it / disassembled the device in order to check if some other component is actually missing (might have been removed) or is some stage is just broken.

Regards,
Vitor

[CJay]

Regardless of how the temperature sensor (if it is one) works, if it's getting hot enough to shut down the PSU then it's got a problem.

Overheating, random shut downs and non starts, caps caps caps...
(I know, sounds like a stuck record but it's always a first port of call for 'odd' PSU problems)
Test and replace accordingly.

While it's running check voltages, if it's possible check current draw and see how clean the rails are.

No idea if these machines use tantalums on the boards but...

[Bicurico]

Hi,

I started a closer look at the PSU.

Here are the first results:

--- Main PCB ---

3x RIFA PME 271M 40/100/56/B 250VAC 100n X2
3x WIMA 250-Y2 MP3 300-X1 2n2
4x KMG 470uF 200V NEGATIVE (M) 105ºC
2x WIMA KP1 - rest not readable without de-soldering

Measured on-board with L/C Meter:

RIFA PME 271M 40/100/56/B 250VAC 100n X2 -->  81nF
RIFA PME 271M 40/100/56/B 250VAC 100n X2 --> 100nF
RIFA PME 271M 40/100/56/B 250VAC 100n X2 -->  86nF
WIMA 250-Y2 MP3 300-X1 2n2 --> 6593pF
WIMA 250-Y2 MP3 300-X1 2n2 --> 6604pF
WIMA 250-Y2 MP3 300-X1 2n2 --> 3134pF
KMG 470uF 200V NEGATIVE (M) 105ºC --> 866uF
KMG 470uF 200V NEGATIVE (M) 105ºC --> 875uF
KMG 470uF 200V NEGATIVE (M) 105ºC --> 875uF
KMG 470uF 200V NEGATIVE (M) 105ºC --> 864uF
WIMA KP1 --> 1641pF
WIMA KP1 --> 6uF ?

--- Output Regulator PCB ---

2x SXE 25V 2200uF (M) 105ºC
1x SXE 10V 6800uF (M) 105ºC
1x SXE 50V 1000uF (M) 105ºC
3x SME 25V 470uF (M) 85ºC

Measured on-board with L/C Meter:

SXE 25V 2200uF (M) 105ºC --> 2.63mF
SXE 25V 2200uF (M) 105ºC --> 2.59mF
SXE 10V 6800uF (M) 105ºC --> 7.00mF
SXE 50V 1000uF (M) 105ºC --> 1.02mF
SME 25V 470uF (M) 85ºC --> 2.63mF
SME 25V 470uF (M) 85ºC --> 2.60mF
SME 25V 470uF (M) 85ºC --> 6.98mF ?

[carl_lab]

Measured on-board with L/C Meter:

SXE 25V 2200uF (M) 105ºC --> 2.63mF
SXE 25V 2200uF (M) 105ºC --> 2.59mF
SXE 10V 6800uF (M) 105ºC --> 7.00mF
SXE 50V 1000uF (M) 105ºC --> 1.02mF
SME 25V 470uF (M) 85ºC --> 2.63mF
SME 25V 470uF (M) 85ºC --> 2.60mF
SME 25V 470uF (M) 85ºC --> 6.98mF ?

Did you measure that in circuit?
Maybe the last 3 caps are parallel to the first 3 caps?

[Bicurico]

Yes, I measured in circuit. I know, not ideal, but I did not want to randomly de-solder components just to measure them.

I do think, too, that some values are off due to being in parallel with other caps.

My next step will be to open the other PSU, which today is working again and measure the same caps to see their values.

Anyway, any help/idea/suggestion is welcome, as I venture into PSU repair. Never had too much success with that, to be honest.

Regards,
Vitor

[Bicurico]

New update:

I have the slight feeling that on my HP with the "good" PSU, the PSU shuts down only after a few minutes of using the Agulent 82357B GPIB USB interface.

Really weird: this morning I was coding a bit on my software for this device and of course I used the interface. After like 1-2 minutes the PSU started to shut down.

No I am trying again, without the GPIB interface attached and it has been like 10 minutes without any problem.

Can it be that the PSU does not like the extra power consumption needed for the GPIB? My software is constantly requesting the whole trace, so there is quite some activity going on.

Regards,
Vitor

[Bicurico]

New update:

The device shuts down even without using the GPIB-USB interface. This is kind of a relief, as it means that the problem is contained within the PSU.
On cold boot the device shuts down after aprox. 15-20 minutes. After 5-10 minutes I can turn it on again.

Before testing this, I had opened the PSU to clean it and to measure it like the one that is fully failed.

This is the result (measured on-board with L/C meter):

Code: [Select]


Component BROKEN PSU OKISH PSU
RIFA PME 271M 40/100/56/B 250VAC 100n X2 -->    81nF -->   115mF ???
RIFA PME 271M 40/100/56/B 250VAC 100n X2 -->  100nF -->  54.6mF ???
RIFA PME 271M 40/100/56/B 250VAC 100n X2 -->    86nF -->   116mF ???
WIMA 250-Y2 MP3 300-X1 2n2 2200pF --> 6593pF -->  6175pF
WIMA 250-Y2 MP3 300-X1 2n2 2200pF --> 6604pF -->  6185pF
WIMA 250-Y2 MP3 300-X1 2n2 2200pF --> 3134pF -->  3565pF
KMG 470uF 200V NEGATIVE (M) 105ºC -->  866uF -->   988uF ???
KMG 470uF 200V NEGATIVE (M) 105ºC -->  875uF -->  1004uF ???
KMG 470uF 200V NEGATIVE (M) 105ºC -->  875uF -->  1004uF ???
KMG 470uF 200V NEGATIVE (M) 105ºC -->  864uF -->   988uF ???
WIMA KP1 --> 1641pF -->  1699pF
WIMA KP1 -->   6uF -->  8.72uF
SXE 25V 2200uF (M) 105ºC --> 2.63mF -->  2.82mF
SXE 25V 2200uF (M) 105ºC --> 2.59mF -->  2.78mF
SXE 10V 6800uF (M) 105ºC --> 7.00mF -->  7.35mF
SXE 50V 1000uF (M) 105ºC --> 1.02mF -->  1.03mF
SME 25V 470uF (M) 85ºC --> 2.63mF -->  2.82mF
SME 25V 470uF (M) 85ºC --> 2.60mF -->  2.79mF
SME 25V 470uF (M) 85ºC --> 6.98mF -->  6.xxmF
??? 180uF 40VDC --> -->   253uF

I guess I will order the caps pointed out with "".
Any recommendation for equivalent caps? Especially the RIFA and WIMA ones?

Thanks,
Vitor