HP8568B calibrator noise / spurious signals

[G0HZU_JMR]

I'm going to be working over in Wales for the next couple of days so I probably won't be able to help much in this timeframe.

If the glitches do appear way back at TP10 then this is the DAC output and this would suggest there may be an error in the way the DAC is being driven by the controller/data bus.

This could be as simple as a poor connection or it could be a memory issue or a duff DAC or a logic chip somewhere.

I suppose you could try very carefully removing the A15 controller card and clean the edge connector in the recommended cleaner. I think this procedure is covered in the HP8566 service manual and you may also finf it in the HP8568 manual. I just use isopropyl alcohol and a soft lint free cloth for stuff like this. Don't use anything abrasive on the PCB edge pads as they are gold plated and the plating is microscopically thin.

But sometimes faults like this can be cured simply by removing and reseating a card in the motherboard. The glitches on your scope plots look to produce similar pulse widths and also similar pulse depths each time they appear. So this could be due to a timing issue or missed bit somewhere.

But then again the glitches could be caused by something else and maybe they aren't related to the fault anyway. But they do appear to sometimes hit the tuning ramp and this would kind of fit the symptoms we see on the display.

If you do end up peeping in at the A15 controller card then it's worth looking at the onboard backup battery to see if it is showing any signs of leaking. if it leaks it can damage the surrounding PCB and also the motherboard. My battery is fine and is probably the original battery.

Note!

There should be a warning label on the A15 controller cover that tells you not to run the analyser without the cover in place. I think it uses the cover as a means to channel air over the CPU etc so it needs the cover to prevent the card from overheating.

[Chris Wilson]

Originally Posted by G0HZU_JMR

Also, there is a test point TP4 on the A6 module. Can you transfer channel 2 of the scope to TP4 on A6? Does this have the glitches on it?

See image below for the location of the A6 module and TP4.

TP10 in your previous post shows spikes, and they are worst when the display shows a lot of noise spikes. I just used one probe and one channel initially.

I moved the probe to A6 TP4 and again spikes corresponded with periods of display noise. Changing from under a 1MHz span to over a 1MHZ span changed the waveform from ramped to square wave, but even on the the > 1MHz square waves the spikes were seen. i have photos of the scope screen showing them if needed. Sadly my digi scope is proving hard to use to see these, I am relying on old Tektronix technology


Thanks again!

[Chris Wilson]

Originally Posted by G0HZU_JMR

I'm going to be working over in Wales for the next couple of days so I probably won't be able to help much in this timeframe.

If the glitches do appear way back at TP10 then this is the DAC output and this would suggest there may be an error in the way the DAC is being driven by the controller/data bus.

<SNIP VALUABLE GUIDANCE>

Thanks Jeremy, I had already pulled that card and cleaned the fingers with IPA. I also checked the battery for leakage and it was fine. Talk of A to D stuff is very worrying, it's a TOTAL mystery to me, the maths seems insurmountable, and the manual baffling.... Oh dear, this may be curtains for my very limited talents, does it sound fixable to you?

Enjoy Wales, I hope the weather holds.

[Radio Wrangler]

Everyone is much more talented than they give themselves credit for.

If you can add, subtract, multiply and divide (calculators are allowed, text books may be open) then you can handle the maths. It's a complex instrument, but that only means that it's a lot of simple circuits flying in formation. No-one can handle the whole lot at once. You need to understand simple bits of circuitry one at a time, and you need to understand the 'big picture' but only in general terms.

A Digital to analogue converter of the sort in that analyser is fairly simple. There is a voltage reference, something like a more precise cousin of a zener diode. It biases an array of resistors and these resistors create currents to ground of say 1uA, 2uA, 4uA, 8uA, 16uA, 32uA, 64uA, 128uA, 256uA and 512uA. The DAC has switches on each of these current paths which can either route the current straight to ground, or route it into the input of an amplifier set up as a current to voltage converter (an opamp with a single feedback resistor) This amplifier porduces a voltage proportional to the sum of all the currents switched into its input. If its feedback resistor is 1k Ohm, then 1uA makes an output of 1mV. If we turn all of the current feeds on, it gets 512+256+128+ etc uA which is 1023uA and so the amplifier gives out 1023mV or 1.023 volts however you like to write it.

The analyser uses a number of these sorts of DACs to make tuning control voltages, sweep waveforms and so on. You can think of a DAC as a tiny computer-controlled power supply that makes a voltage which a program has decided on.

The analyser adds various tuning voltages together and turns the result into a fairly hefty current which is fed to the tuning coills in the YIG oscillator. YIGs are tuned by applying magnetic fields.

There's quite a few of us on this forum who are comfortable with these technologies and will provide support.

An 8568 is a superb instrument and well worth fixing.

David

[Chris Wilson]

Nice post David, that is both very informative without being overwhelmingly detailed, and quite comforting to the reader as to a possible prognosis, you'd make a good doctor

My maths is appallingly bad, but I do try and follow the divide and conquer routine when faced with complex technology. I am far better at understanding something when hands on, I can see a design fault in a transmission pretty quickly if I have to strip and rebuild it, but might well miss it looking at a dimensioned drawing

My other fear is I will get to the sateg where thing needs recalibrating due to some cal memory being lost. I am not sure what that would entail, physically and financially.

To rub salt in the wound a friend who bought a Signal Hound to replace his stuttering 5868 decried it after a month or two as having suspect software. Apparently, and much to many people's surprise, who thought the project dead in the water, they have recently released updated software that works superbly. There are times recently where a little modern box that fits in your pocket and just "works" is quite appealing, but I am a plodder and won't give up just yet!

Thanks for the morale boost!

[Andrew2]

Hello Chris! I see the saga of the noisy 8568b continues. I think I mentioned in our last e-mail exchange that my old 8568b had periods of microphony and noise on the trace which showed up worst at the top end of the coverage and at narrow spans. A quick tap would often clear it.
I'm the man with the Signal Hound and I can report that I am more than happy with it now. It looks better, the ergo's are hugely improved and the bugs have been banished. I've not had a single crash. It's so good now that I rarely use the Rigol, which has phase noise about 30 dB worse. I may flog it.....
Sorry for jumping in and deflecting the thread slightly. Good luck.

[Chris Wilson]

Hi Andy, good to know it's now running well. I will persevere a while longer, it keeps the brain active, and the electric bills coming But if the worst comes to the worst a SH may be on the cards sometime. And thanks for the good luck, I will need it!

[Radio Wrangler]

If you do decide to pack it in with the 8568, then I expect that it'll still feature on these pages until it is fixed, though maybe with someone else doing the physical stuff. They are serious machines and still have a lot of advantages over pocketable analysers.

Even a dead 8568 is still worth good money, especially if the CRT is good. The top box is the same as for the 8566 and top boxes alone sell for more than bottom boxes alone because of the demand for CRTs.

Anyway, it's worth fixing and it will out perform any new affordable analyser. I can say nice things about the 8566/8568 because though I was a senior design engineer with Signal Analysis Division (the spectrum analyser part of HP) I wasn't involved in them so I can claim a degree of impartiality

David

[Chris Wilson]

My wife is now telling me there's a life beyond fixing "that junk upstairs". As the life involves tiling and decorating the jury is out as to whether she's right or not <LOL>. These repair jobs can get a bit too time consuming, as many here will probably vouch, but drinking is becoming too damned expensive, and I am sick of being ridiculed pretending I am still the womaniser I once may have been, so I need a good hobby like fixing stuff up!! That's my story and I'm sticking to it.

[G0HZU_JMR]

Originally Posted by Chris Wilson

TP10 in your previous post shows spikes, and they are worst when the display shows a lot of noise spikes. I just used one probe and one channel initially.
I moved the probe to A6 TP4 and again spikes corresponded with periods of display noise.

This is a bit confusing. I chose these two test points for various reasons but I was hoping to see spikes on just one of them.

These two signals appear to (eventually) meet and sum together in a summing opamp U19 on A22. Opamp U19 then goes on to drive the FM coil of the YTO via a buffer amp made from Q24 and Q25 and this is where your original TP3 test point for FM Sweep is located. In theory the sum port/pin of the opamp should be a virtual ground so there should remain high isolation between these two signals. So little or no crosstalk?

TP10 on A22 comes from a DAC and TP4 on the A6 module is from a circuit comprising a phase detector + sample/hold loop filter + buffer amp.

So they are very different circuits and they are on two different modules within the analyser. Have you looked at the +15V PSU rail and the -15V PSU rail on the scope for these spikes? These rails are common to the output of both of these circuits.

[Chris Wilson]

Oooh! Now, I am trying not to get too excited here Jeremy, but tonight, on a cold start up the display was fine, but after 10 minutes I cam back to it and the noise was pretty chronic. I am not good with my TEK 7854, it's a complicated beast, and my probe has a fairly long ground lead. I may also be probing in a less than ideal place (A24 regulator board, pins TP1 for ground, and pins TP2, TP3, TP4, TP5 and TP7 for +15V, +20V, -15V, +5V and -5V respectively. That aside, there was a marked difference on the -15V pin. Pin TP4 on A24. As soon as I see display noise there are very noticeable spikes on the -15V. These are NOT apparent on any of the other pins. Please critique my scope set up though. I may be on another red herring....But surely plus 45mV is a big spike?. Some are negative too. I will zip up some screen shots. The spiked ones are from the -15V pin, the quiet ones from other pins, these are all much the same. Probe is set X1.

Is this a possible cause?


I also tried a 1000uF 50V electrolytic on Pin TP4, the spikes on the scope disappeared pretty much completely, and the display cleaned up to at least 50% less noisy spiking. Hmm.

Give me a while to do a couple of little chores, and I'll upload some scope shots as http://www.gatesgarth.com/SA/scope-shots.zip

[G0HZU_JMR]

Looks promising

I'm still busy with work related stuff but I would advise patience and caution at this stage.

If it is related to the -15V rail then I think it's worth looking at this with some care because it may be possible to cause carnage throughout the analyser if a diagnostic mistake is made that corrupts one of the PSU rails.

I have no experience with the PSU sections in this analyser so I would need some time to look at the PSU circuits etc. Maybe others have more experience with HP PSUs. I'm at my best with RF related stuff

Note: Your link above doesn't work for me. So I can't download the zip file.

[Chris Wilson]

Uploaded now, I see the spikes on my digi scope too, again only on the -15V pin. The last two shots are the +15V pin.

[G0HZU_JMR]

I've looked at the PSU reg circuit for the -15V rail and the spikes on the output look to be 'faster' than the response time of the PSU (based on my guesstimate of the PSU operation)

So this makes me think that the -15V rail may have a problem with its output smoothing caps.

Can you check C49 on the motherboard with your ESR meter? I think it should be 6.8uF

This PSU regulator only has a small amount of (local) smoothing capacitance and I think it needs a healthy C49 on the motherboard near its output.

The regulator has current limit and overvoltage protection so it should be quite forgiving of our attempts to fix it

But don't go overboard with adding extra (as in big) smoothing caps on this rail. I don't think it's the right thing to do.

I still need more time to study the PSU circuit in detail but I think every smoothing cap on the -15V rail needs to be healthy. Especially C49 on the motherboard. Hope it's easy to get to! Obviously, you will have to unplug the 230V AC mains connections to the analyser if you are going to be fishing deep inside the unit to get to the motherboard. I don't know where the exposed 230V hazard areas are.

Also, have you measured the -15V rail to see if it is accurate?

[Andrew2]

The beer and crisps are out here, I'm thoroughly enjoying this.

[Chris Wilson]

I renewed all four tants on the motherboard, C49, C50, C51 and C52 ( a typo on the board layout shows two C50's....). A Russian owner says these are the most common cause of issues with noise on the rails. I actually found new old stock to match the originals precisely. The old ones measured fine though, as did the new ones.

The big electrolytics have me wondering. In an earlier post (#44 I think) I mentioned that my Peak ESR meter wouldn't handle one of them as the capacitance was beyond it range. I would have to check just which one this physically was, but the voltage rating may have been too low for the -15V supply, which I think is about 25V pre regulator. However, the values of these caps do not match those shown in the parts list. wonder if they have been renewed at some time? Would an iffy pre regulator cap cause issues? the other three big electrolytics tested fine.

I am hoping I am finally getting somewhere, as I am sure others are, as i am consuming a lot of bandwidth with this! Thanks.

EDIT: The big cap I couldn't measure was C4. It seems it's for the unregulated 10V supply, that regulates down for the 5V rails, so it probably isn't that. It is far bigger capacitance than the parts list cites though, and is physically bigger than the others. C3 looks to be the one on the -25V unregulated -15V supply. I will check it again.

[Radio Wrangler]

Just take it slowly Chris.

Finding spies on the -15v rail is very very good.

It's going to be a good clue leading towards the cause.

BUT

Are the spikes on the -15v rail the cause of the LO frequency jumps which make the display look noisy OR are they an effect of whatever is causing the jumps?

Whichever they are, you have found a hot trail. If you trace it to find what is putting pules on the -15v rail you either find what is making the pulses which disrupt the LO, and so fixng it fixes the problem, or else you track it back to jumps in the YIG coil current loading the supply and trace that back to whatever is making the tuning currents jump, and that's your root cause.

Regulators can be very fussy about what sort of capacitor is used on rails before and after them. They can sulk and oscillate.

David

[G0HZU_JMR]

I've looked again at the various scope images from Chris and I think I'd like to see a scope image of the spikes at TP10 on A22. Are they fairly big?

Even if there are spikes on the -15V rail, there ought to be enough supply rejection in U10 on A22 to prevent much in the way of spikes leaking to TP10.

So I'm back to wondering there is an issue way back at the DAC that is causing sharp glitches in current from the YIG coil driver that the -15V rail can't cope with.

Is it possible to also probe pin 1 of the DAC (U14) on A22? You will probably have to solder a short wire to it to allow it to poke out enough to clip a scope probe to it.

If you do this then take a scope image of it during the time when the display is noisy.

[Chris Wilson]

Good day! I grabbed a few more scope shots, I hope the file names are self explanatory. All taken at 20MHz centre frequency, a tad over 1MHz span. Probe set X1. I looked at the schematic and thought some probing at TP17 may be helpful, as it seems it's before the buffer U17a. There's noise there as well. I VERY carefully soldered a probe wire to PIN1 of U14 and did some shots from there, but note in some I changed the the volts /div scaling substantially. I am not sure what I should be seeing there...

Maybe coincidentally I had a period of really bad jitter after soldering the wire to PIN1 U14. I thought I had done something `orrible, but it settled down.

I tried to grab a video, but I chose the wrong whatsit and it's come out at 1.6GB, so I won't be posting that!

Shots should be zipped at http://www.gatesgarth.com/SA/more-scope-shots.zip

Thanks.

[Chris Wilson]

What effect would a noisy R39 trimmer resistor (multi turn) in the YTO Main Coil Fixed Driver circuit have? I believe I disturbed it with vibrations when removing A22. You only have to look at R39 now and it jitters. I tweaked it half a turn each way and it settled, but occasionally it goes wild again. Just the lightest tap or scratch on R39 without even touching the adjuster screw starts the jitter.

http://www.gatesgarth.com/SA/R39.jpg