HP8566B wobbly trace

[w2kv]

Hello to the forum,
My HP8566B has developed a problem where the trace when looking at a signal (either externally applied or from the 100 MHz CAL port) is noticeably wobbly at spans below 100 MHz.  At a span of 5 MHz or below the trace is a broad mess.  It looks like one of the LOs has come unlocked.  Any ideas on which one?  How to check?  I have used 8566s at work and home for many years but never had to repair one, so any thoughts would be appreciated.
73 Dave W2KV

[KE5FX]

Treat it like you would any other radio.  Start by getting the service manual.  It's a big 'un.  Then, check the power supplies for voltage and ripple.  The specs will be in the manual. 

If you don't own an ESR meter, well, now you have an excuse, because probably 75% of the things that might go wrong with an 8566 are due to bad axial electrolytic capacitors.  The shotgun approach is very reasonable here.  Pull all the boards in the RF section, check all those caps.

[cncjerry]

I had all kinds of weird problems with my 8566B recently.  Turned out it was the -10V supply cap.  I would check the DC voltage and then put a scope on them to make sure the ripple isn't high.

[w2kv]

Good advice.  I have downloaded the service manual, 800 pages!!  An ESR meter is one gadget I don't have so I guess I need to get one.  I can see where it would be a useful instrument.  The 8566 returned to life after running a couple of hours and still looks OK after being off overnight but these sort of problems generally don't get better on their own.  Now if I don't get a hernia moving the thing...
73 Dave W2KV

[Johnny10]

I remember a post about a coffee maker on the circuit causing wonky display.
Not that this is your problem.
But something to think about.

[tggzzz]

An ESR meter is one gadget I don't have so I guess I need to get one.  I can see where it would be a useful instrument.

You don't often need to measure a capacitor's ESR. You do need to know whether or not it is "good enough", and frequently a sufficient test is to measure the ripple on the PSU rail. There are exceptions, e.g. where a cap's minimum ESR is used to ensure sufficient stability in a switching PSU's feedback loop.

When curious, I've crudely measured the ESR by applying a square wave to a resistor, R,  in series with the suspect capacitor, and used a scope to observe the voltage across the capacitor. Choose the frequency so that the scope's input isn't saturating. When the input changes there will be an initial short-duration step before the expected RC exponential waveform is seen. The amplitude of that voltage step is determined by the ESR and input current change (V=I*ESR, of course). The input current change is I=V/R where I is the square wave amplitude.

The 8566 returned to life after running a couple of hours and still looks OK after being off overnight but these sort of problems generally don't get better on their own. 

Indeed.

[KE5FX]

You don't often need to measure a capacitor's ESR.

You do if you're working on an 8566. 

The PSU bulk capacitors are easy enough to service without an ESR meter, but there are hundreds of small axial tantalum electrolytics that are not amenable to testing any other way.  Those are the ones you 'shotgun' with an ESR meter as a first troubleshooting step when, for instance, one of the PLLs has trouble staying locked.

[tggzzz]

You don't often need to measure a capacitor's ESR.

You do if you're working on an 8566. 

The PSU bulk capacitors are easy enough to service without an ESR meter, but there are hundreds of small axial tantalum electrolytics that are not amenable to testing any other way.  Those are the ones you 'shotgun' with an ESR meter as a first troubleshooting step when, for instance, one of the PLLs has trouble staying locked.

I presume they are an integral part of some analogue loop; are they the rather expensive "milspec" range? If so, not replacing them unless proven faulty seems prudent.

Is the ESR specified somewhere? If not, what can you presume from a measured value of X ohms?

[KE5FX]

I presume they are an integral part of some analogue loop; are they the rather expensive "milspec" range? If so, not replacing them unless proven faulty seems prudent.

In that particular instrument, HP used them for almost every application that called for capacitance between 0.47 and 100 uF.  Among other places, they were used for point-of-load bypassing, LF coupling, emitter degeneration, noise suppression, and general-purpose feedback/control loops.  PLL loop filters were more likely to use film capacitors that don't cause much trouble.

The axial electrolytics weren't inherently bad parts like the SMT electrolytics that you see in Tektronix gear, so they don't need to be replaced unless actually defective.  It's very obvious when they are.  The bad caps will either appear open on the ESR meter, or their ESR will be several times higher than other capacitors on the same board with similar ratings.   

In fact, the bad caps often show some visible corrosion.  Those should obviously be replaced regardless of what the meter says.

Is the ESR specified somewhere? If not, what can you presume from a measured value of X ohms?

There's often a chart on the meter itself that shows what ESR to expect for a good capacitor with given voltage and capacitance ratings.  Usually less than a couple of ohms.  Most capacitors will pass that criterion easily, even after 30+ years.  Opinions vary, but I like to replace anything that's over 2x the value in the chart, or that's more than 2x worse than identical parts elsewhere on the same board. 

[cncjerry]

I mentioned on the other thread where you posted that you should check all the voltages for ripple  I had a weird one where the 12V supply was dropping out and it turned out to be the cap on the -10V supply.  As John posted, these things have caps everywhere but start at the large main supply caps in the RF unit.  Also, I don't suggest you wholesale replace the caps as you will more than likely introduce another problem.