Dying HP 3580A spectrum analyzer

[RPuccio]

Hi everyone !

I have an old HP 3580A Spectrum Analyzer for audio work. It used to work okay since I got it - yes, in the bin - but is now showing concerning behaviour.
The problem is the same as described here : https://www.eevblog.com/forum/repair/hp-3580a-spectrum-analyzer-display-faint-how-to-testreplace-a11s-cr-5/

When powering the device on, I need to wait for a few minutes until I see something on the CRT. I need to wait at least 10 min to get comfortable brightness, and turning on my desk lamp is enough to make it unreadable. The display is not linear either, and gets compressed on the left and right of the CRT. Obivously, everything makes me think of a power supply issue in the HV circuit.

I'm attaching the service manual below. HV power supply is at page 200 (also a variant at page 197, I haven't checked which one my scope is). Here are some measurements :
- The +158V power supply only starts at about 80V, increasing to 94V or so after 10min. This voltage is also what drives the vertical and horizontal deflection amps, explaining the squishing issue.
- Voltage at the feedback node is a steady -9.3V and seems unaffected by the HV adjust trimmer. It also seems quite high in value, I would've thought about something in the -20V-ish range
- Other readings are fine (+10V, -10V). The -11.5V is only -11.2V but it should still be good.
- One node is marked with a '9', with its corresponding waveform at page 199. Frequency is good, between 55kHz and 60kHz, but amplitude is way too low : only about 1.5Vpp instead of 7.5Vpp.
From what I understand, the transformer and HV oscillator are acting together to induce oscillation in the circuit, and it generates +158V as well as the CRT heater voltages and the -3kV voltage.
I'm torn between either hacking the scope by adding a DC-DC converter to output at least the +158V supply to get a better look at it, trying to identify the faulty component (best idea) or threwing the CRT and HV boards alltogether and hack the X/Y signals to display it with an Arduino and an LCD screen.

I'd still love to make it work again though, any tips on what I can check ? I'm unfortunately not equipped with a HV multimeter so I can't have a direct reading of the -3kV power supply.

Cheers and thanks for your help ! 
Rodolphe


Link for the 3580A service manual : https://xdevs.com/doc/HP_Agilent_Keysight/3580A.pdf

[RPuccio]

Upping tge thread, just in case someone has an idea ! 

[cncjerry]

sounds like bad caps in the 158v supply.  They are leaking and taking a long time to come up to the correct, or even close to the right voltage.  there is probably quite a bit of ripple on that supply as it is coming up to voltage.  The analyzers I've repaired have had blown or leaky caps in the power supply.  Both my 8566b and my R&S FSB have had bad caps.  8568a has one now.  Measure the ESR and it should be obvious.

[RPuccio]

The thing is the power supply caps are not electrolytic capacitors but either film or ceramic types. Could it be possible that one of those is dead even though they measure fine on my cheap chinese tester ?

[RPuccio]

Schematic of the power supply. Here are a few other tests performed :
- Switched the A8 board with a working unit. My board is working in the other unit, the other unit's board is not working in mine. My conclusion is that the A8 board is working as intended, and something is off somewhere else.
- Compared the winding's resistance of the HV transformer with the working unit's one. I get the same reading.
- Measured all the diodes and resistors of both my unit and the other one. They are all pretty much spot on. I see that my HV diode is a different model from the other unit, both can't get measured in diode mode (Overload) but when using the ohmmeter, I measure about 12MOhm on my diode and 17MOhm on the other one, so I guess it should be pretty much right.
Next test will be to disconnect one tap of the CRT heater and be really, REALLY careful about it, because I am suspecting the CRT to load the power supply too hard.

[factory]

Be aware the CRT heater is referenced to the high voltage supply.
Measuring the resistance of the transformer will not tell you if it has shorted turns (would probably overheat though), likewise if a component is breaking down under HV a low voltage meter probably won't help.
Don't they have any old analogue meters (with 3kV or above ranges) for sale where you are? Or you could build a HV probe.

David

[RPuccio]

Yeah I need to build myself a HV probe I think ! Or at least order one, it can be handy anyways.

Disconnected the CRT heater and the power supply still isn't working. That's actually a good sign at least, I think the CRT might still be good !
So it narrows down to the HV generator section, either A11A1 or A11A2. It could also be the HV transformer though.

[George Edmonds]

Hi

At this point, although desirable, I don’t think that you really need a high voltage probe.  What you need to do is to determine is , if I have understood your posting correctly, just what is loading down the high voltage oscillator.  Now you say that the A8 board has been replaced and did not fix the problem.  You have also disconnected the CRT heater supply and the fault remains.

Logically it must therefore be on the A11A1 board or the A11A2 board and possibly a faulty HV transformer.  As a rule when HV transformers fail the driving oscillator stops and the driving transistor is destroyed, which yours has not done so as it is partially still working.  Any suspicion about the CRT can be removed by pulling the base connector off and monitoring the 158VDC supply

The A11A1 board contains two HV DC supplies derived from windings on the HV transformer.  What you need to do is to disconnect the feed to pin1 whilst monitoring the 158VDC supply with a bench meter (NOT A HAND HELD ONE), if this allows the 158VDC supply to raise the fault is possibly with the circuitry between pin 1 and pin 3 of the A11A1 board.  Due to the feedback circuitry involved do not do the same check by disconnecting pin 2 of the A11A1 board.

My best guess is that one of the high voltage capacitors C1 to C6 has failed and is loading down the high voltage oscillator.  The other possibility is that there is a fault on the A11A2 board, again disconnect pin 3 on the A11A2 board and check the 158VDC supply.

Be very careful to isolate the incoming wires and do the checks as quickly as possible.  I may be worth first removing the A11A1 board and doing some resistance checks but keep in mind the low test voltages of a DMM may not show the problem, please keep in mind that CR1 to CR5 are not simple diodes and will have a greater forward voltage drop that you might expect as a result they may not show as diodes on a DMM.

Hope that this helps, PLEASE KEEP IN MIND THE HIGH VOLTAGES AROUND THE CRT AND NEVER USE A HAND HELD DMM.  IF YOU ARE NOT HAPPY WORKING WITH HIGH VOLTAGES DO NOT DO SO.

G Edmonds

[RPuccio]

Thanks a lot for your help !

Disconnected pin1 of A11A1, no change. Disconnected pin 3 of A11A2, no change. I'm starting to think the HV transformer might be shot at that point.

The weird thing is that -9.3V at the feedback node means the HV supply is around -2530V, which should really be enough, right ?
I tried disconnecting the tap on the transformer that's supposed to generate the +158V supply (pin 5) and tying it directly to a +158V-ish supply but the display is still really, really dim.

Could it be that the HV transformer is partially dead both on the +158V secondary AND the heater winding ? I've never had a dim CRT before, but could it be that the heaters have too low of a voltage accross them ?

[RPuccio]

Last test I did is definitely explaining a whole lot the problem : I disconnected the heater wiring and connected it to an isolated DC power supply. Fortunately, some guy on Ebay is selling a replacement CRT with the heater info written on it : 4.65V @ 235mA. Turned the power supply slowly to the right voltage, monitoring the current, and I powered on the analyzer. I immediately got a trace on it, with the same distorsion problem as before obviously because the +158V is still too low (90V-ish).

I guess it is time to find another way of generating those voltage.

[g4tuz]

So, what is the voltage across C15? the one after the 3 diode string?

[RPuccio]

Voltage at C15 is 104.5V, when measuring about 90V at the output of Q19.

[g4tuz]

OK, do you have approximately equal voltages across each of the 3 diodes?
If you lift the emitter of Q19 and one leg of R68 what is the voltage then?

[factory]

Q19 is on the A8 board, which I thought had been verified as working in the other unit. 

David

[g4tuz]

Reading this on my phone so not picking up every detail. So forget about the diode question, still wondering about loading of the 153V line off board.

[RPuccio]

Yep, it could still be loading of the 158V PSU.

I also measured the +11.5V to be about +9.7V, which is significantly lower, so maybe the low voltage regulators are not happy with something either.
I want to try to patch a 1:10 transformer with a MOSFET to basically replace the current HV transformer. I'll see how things work and will keep you all in touch !

Anyways, if that thing gets back to life and dies later on, I'll certainly end up selling it for parts on Ebay or anything. I mean, the analog front end and the analyser part still work !
Or maybe I should get an old X-Y recorder and be really patient 

[George Edmonds]

Hi

I have very carefully re read this entire posting which contains conflicting information, it is now time to do a reset and act in a logical manner.  The large Elephant in the room is the lack of high voltage testing facilities.

First may I say that I doubt that the problem is with the High Voltage transformer, normally when these fail it is a total failure as a result of shorted turns or the breakdown of the insulation, this normally kills the driving transistor Q21 and or blows F1 and F2.  The accepted method of testing these High Voltage transformers is to first carry out what is known as a “Ringing Test” with the High Voltage transformer out of circuit, if the High Voltage transformer passes the “Ringing Test” it is then submitted to a high voltage insulation test.  If both of these test are passed you can be about 99% certain that the High Voltage transformer is OK, but not a 100%.

I see that the latest posting is querying the +11.5VDC supply, this begs the question is the -11.5VDC OK as the High Voltage transformer is driven across both supplies, both of which are fused by F1 and F2.  The simple first test of the +11.5VDC supply is to remove J1 (the feedback connector) is the +11.5VDC supply then correct?

Now rule one when fault finding CRT supplies is to suspect the highest voltage one FIRST (3KV supply) as these components are normally the most stressed, in the case of the 3580A this is complicated by the 3KV High Voltage supply also being used as the feedback path.

This is a simple matter for me as I am used to working with high voltages and have high voltage test facilities.  Prime suspect must be CR5, BUT this is not a standard diode and cannot be tested by a DMM.  The forward breakdown voltage will be in the range of 20 to 30VDC and the reverse isolation voltage some 5KV, such diodes are commonly used in Microwave Ovens.

There must also be a question regarding the high voltage (leakage) performance of  C4, C5 and C6 on the A11A1 board, again a simple matter for me to test as I can do leakage tests on capacitors up to 5KV.  It may just be worth pulling the A11A1 board and doing a DMM test for leakage of the High Voltage capacitors.

G Edmonds

[George Edmonds]

Hi again

I have now found a good copy of the HP3580A service manual.

At first let us concentrate on the +11.5VDC supply, this supply is on the A6 board and is a conventional series pass transistor type using Q1, Q2 and Q3 with NO current limiting provision.

This means that if it is working correctly the output voltage must only be either at +11.5VDC or zero with F1 blown, F1 is a 375ma fuse whose only role is the protection of the +11.5VDC supply to the A8 board and the high voltage transformer.  The +11.5VDC supply is not used anywhere else in the instrument according to the service manual schematic.

Start by checking that you have 18.5VDC on the emitter of A6 Q3, if not look at the on/off switch on A10 first.  If this is so you should find 11.5VDC on both sides of F1.  Do you have +10VDC on the emitter of A6 Q1, this is the reference voltage for the +11.5VDC supply

I have never seen a HP3580A, but I have seen a number of HP instruments where they have used a socket for TO66 transistors like Q3, the Base and Emitter contacts provided by these TO66 sockets are frequently poor/intermittent.

Q3 is a PNP device type 2N3740

Until both the +11.4VDC and -11.5VDC supplies on the A6 board are correct do not carryout any further work on the high voltage section.

Let me know how you get on.

G Edmonds

[RPuccio]

Sorry for the long delay and thank you all for your help.

George, you got it right ! I was indeed thinking 'hey, 9.3V instead of 11.5V is way off, the oscillator might not run properly because of this".
Checked the +11.5V fuse and yup, here it was : high resistance ! Apparently it was still passing some signal though, but I replaced it with a wire to check things and the power supply now runs smoothly, and I can adjust it between +148V and +154V which is the right range. Now I'm gonna try to remove my CRT heater hacking and put it back in its original position.

EDIT : yes, confirmed, it now works ! Time for a calibration though, but I'm so glad you helped me all. Thanks once again 

[SilverSolder]

Got to love how often it turns out to be a "mechanical" problem!  switches, pots, connectors, fuses....

[RPuccio]

Not really mechanical though ! The fuse was an inline fuse with lead, so basically a through hole component. Still, I don't understand how it was measuring "open" on my ohmmeter and still passing 9V. Anyway, it is now fully calibrated and reads wonderfully. Bright trace on the CRT, not the thinnest I've seen though, but performs right. One more tool saved from my school's recycling bin !

[SilverSolder]

If the fuse sits in a fuse holder, it qualifies as "mechanical" from my jaundiced perspective! 

[RPuccio]

Oh yeah, for sure it does ! One simple fix : use a nail instead   

Anyway, thanks once again for all your help. This forum is an incredible community and being able to repair gear with some pro's help is a true blessing.

[factory]

You'll just turn the PCB tracks into fuses instead, if you do that and something like a capacitor decides to short. 

David

[SilverSolder]

You'll just turn the PCB tracks into fuses instead, if you do that and something like a capacitor decides to short. 

David

He did say "simple" fix....   not "good" fix!