A broken HP 33120A

[vaualbus]

I've recently bought a broken HP/Agilent 33120A with the stable reference in function generator for around 200 bucks.
Yesterday it arrive home, I turn on and for like 5 minutes it turn on and fail self test.
Than the fuse blowed.
I open the instruemnt and I've found a shorted diode CR704 a broken diode CR803 and four tatalium caps burned (one have also demage the PCB unfortunally).
I post some photo of the main board.
What I think it happens is that tese caps burn and take down the two diodes. (Strangly one of the two diode is cracked but still test good meanwhile the other is a short circuit).

Image 1: https://www.dropbox.com/s/gqvwzb979m0lru9/1.jpg

Image 2: https://www.dropbox.com/s/gzpujw8xk2i8vby/3.jpg

Image 3: https://www.dropbox.com/s/l8j6m6ntb0uubah/6.jpg

Best regards, Alberto.

[vaualbus]

I have some problem with dropbox either on my machine or on the server. For now only the first image is  visible.

[HighVoltage]

Can you tell us what the error message was during startup or selftest
I also have one 33120A that is dead but the problem is not clearly to be identified

[vaualbus]

I can't now the instrmwnt doesn't turn on. I need to replace the csps and diodes.

[pgross]

Is the fried tantalums in your 33120A - 20V rated types?

I know that a couple of 20V rated tantalums in the 33120A are mounted directly across the +/- 15V rails.

It almost looks like the same issue with the 34401A multimeter, where some have discovered 20V types
tanatalums have fried and damaged the pcb.

If the “normal practice” of min. 50% voltage derating should be applied, these tantalums should be
retrofitted to 35V types.

Hope that you’re able to clean / repair the pcb, and get it functional again.   

Best Regards

Peter

[vaualbus]

Ya they are 2.2uF 20V caps. I wish that the cap that have burned the pcb havem't demega the internal layers.

[TiN]

I have donor board

[vaualbus]

Next week I'll go and buy the component (on of the two diode, on rs components it will be avaible only in december so I need to buy that on ebay)
and I will update for the results.
Best regards, Alberto.

[vaualbus]

>Any body kno why tantalium capicitor burn when they get very old?

[nctnico]

Tantalum caps should be used in places where the current is limited. Tantalum caps basically are made of thermite. Too much current= iginite. Reverse voltage = ignite. If the circuit doesn't depend on the ESR of tantalum caps then it is better to replace them with low ESR electrolytic or ceramic capacitors (X5R or X7R dielectric).

[vaualbus]

so you suggest to use normal eletroliti caps in the 33120A? All this caps are bypass capicitor on the -18V and 18v rail?
If it is possible it even better for so I haven't to buy the caps.

[vaualbus]

I've some update for you guys:
Fist I replaced all the caps and diode and now the display not work (The fronpanel fall from my desk so the vacum display could be broken but I don't see any crack), but the instrument still not produce any output.
The frontpanel a part from the display it work, I can move the encoder an hear the click, use buttons and ecc.
Probably I find some problem. The negative +15V read -20 meanwhile the positive +15V read only 9V.
I don't kno why I read this voltages?
PS Somebody have some frontpanel or vacum display to sell to me? 

[vaualbus]

I was wrong aboat the volatages I mesearud them with the wrong GND. Both 15 an -15 are right.

[vaualbus]

Also how I can meseuare the output of the dac?
It seem a stupid question but it isn't the dac have a differential current output so I can't meseare that with a scope.
Any idea?

[TiN]

Check getter on VFD. If there is white residue around getter, it means oxygen got in and vacuum is gone now.
If its metallic shiny, your display likely be okay.

[vaualbus]

ya is white as I suspect because when I remove the vacum displat to check voltage on the front panel I saw two cracks.
Now I need to find a new display vacum tube. It seem hard to find.

[vaualbus]

IU have discovered something very very strange.
So If I measeaure the +/- 15V of the output amplifier using the agnd as ground I meseaure -20V an +9
but If I meseard them using the gnd as ground I meseare +15 and -15.
The strange think is that all the bypass cap are grounded using the agnd.
What is going on?
What I should use as ground agnd, or gnd?
(The worst think could happen is that near the caps that burn and that have the demage the pcb, the two ground plane have shorted).
Anybody can meseuare the resistence between GND and AGND?

[vaualbus]

So i find somethingh strange; apparently the two ground are not isolated, there about 800ohm of resistence between.
If I meseaure the voltage across the grounds it reads, a magic number, -5V. A little bit of math with the previous measaruments explain the solution.
-15V + -5V = -20V, 15V + -5V = 9V. So the volatages I read was right the only problem is that the ground is not a zero volt.
I think this is not normal, so looking for a explnation the only component in the entire instruments that are betweeen the grounds, is the semi custom DAC.
(Fortunally it seem to be avaible in ebay).
I'm right? It seem a good explanation to me.
Best regards, ALberto

[vaualbus]

10 years later and I still am perplexed on this! I still own the instrument and never repair it. I burned like 3 or 4 DACs. I still thinking that the grounds should be the same but again have no way to check it! (I now own way more interesting stuff anyway  )
Inherently enough in schematics I see that the grounds are connected at JM1006 but that part seems no where to be found in the PCB layout. What I think is that the part is the big middle mounting hole where the plastic insert is mounted.

[HighVoltage]

Wow, it has been 10 years, time flies fast, crazy!

What if you connect the grounds to bring them all to the same zero level?

[vaualbus]

Indeed and after I have fixed all the broken pads with some "surgery", I still have the output amplifier not working 
I would say, nothing really changed 

Basically I see that I have a signal driving Q709 but no signal on the base of Q710 
I am worried that the DAC is generate a too low voltage,  the SM said it should generate a 800 to 1Vpp signal but when I probe DACOUT+,DACOUT- I see only a 300mVpp signal. Now the DAC is generating an output current so it can be that I am not measuring the voltage using the correct ground. 

[vaualbus]

Well. well, after 12 years I can say I've have finally fixed it  just took a minute or so to did it 
Surely the "fake" RF output power transistor from eBay won't give me linearity or good ft , indeed square waves look sinusoidal above 10Mhz (on a 500Mhz scope and 50ohm matched) but for now I don't care on that part.
(looking at forum it is a normal behavior)

At the end the last strugle was a broken internal PCB trace that prevent the front panel to receive the proper reference voltage used for detecting the digital high low.
(I had to change again the VFD after I had connected backwards the main power plug and send some voltage to the VFD and make it a Christmas tree )
SO yeah good adventure I almost give up until tonight I decide to said let's check for a final time and figured out I was missing a simple 2.5V  and so the entire display was blank and randomly when the floating voltage was "correct" it showed some of the normal menus and stuff.

PS:
By the way indeed the grounds are connected at that point, after I restore the connection with solder I n more destroy DAC (at least for the time being)

PS2
The output amplifier decide to burn again and the original half way waveform retuned (I should haven't left the instrument turn on for so long    ).
At the I decide to replace it with his big brother the 33250A that I repaired. I will see If I want to fix the output again or throw the instrument in the bin