HP8566B needs repair?

[eriklidgren]

Hi

I have bought an HP 8566B spectrum analyzer. I'm trying to figure out if it is behaving normal? Or if it needs repair/calibration?
What i'm seeing is that for RBW´s of 100kHz and above there is a shift in frequency(-1.5 MHz with a 100MHz signal). For RBW 30kHz and below the frequency is very close to the actual value.
I have run the error correction routine prior. I haven't analyzed the values too deeply, but all are below 1db.

Any comments/suggestions?

Erik

[MarkL]

What happens if you slow your sweep speed to 300msec when your RBW is 100kHz (span still 100MHz)?

[TSL]

Hi Erik,

Where are you getting the 100Mhz from ?

Is this the marker from CAL out ?

On my 8566B, using CAL out, and the parameters in your first screen grab, my marker is precisely 100Mhz.

You may need to do a calibration routine to bring it back in spec.

I'd at least run through all the checks in the service manual to determine precisely where the problem may lie.

regards

Tim

[eriklidgren]

Hi

Thank you for your replies. Mark, I increased the sweep time to 300ms and it's dead on 100.0MHz. It moves closer and closer as I increase the sweep time. Tim, yes 100MHz comes from the CAL output. Thank you for verifying on your analyzer. I'll do som reading in the service manual and do some more tests.

Regards
Erik

[MarkL]

If you have a signal generator, I would try moving the frequency around to see if the observed offset at fast sweeps changes across  the span, or if it's a constant.  If you don't have a generator, you could change the center frequency of the sweep, but it's not as good a test since it's potentially changing other variables.

This would at least help you understand if you have a sweep ramp gain, offset, start delay, or perhaps some kind of non-linearity issue at higher sweep rates.

Examining behavior at other largely different frequencies and other spans might provide some clues too.

And, as TSL said, all this assuming you've run a successful cal.

Also take a look at the center frequency accuracy spec.  I still think it's odd that it does what you see, but if I'm interpreting this right you're still in spec being off by 1.5MHz @ 100MHz span:

Center frequency readout accuracy
  Spans ? n x 5 MHz ± (2% of frequency span + frequency reference error x center frequency +10 Hz)
  Spans > n x 5 MHz ± (2% of frequency span + n x 100 kHz + frequency reference error x center frequency)
    where n is the harmonic mixing number, depending on center frequency:
    n center frequency
    1 100 Hz to 5.8 GHz
    2 5.8 to 12.5 GHz
    3 12.5 to 18.6 GHz
    4 > 18.6 GHz


from: http://literature.cdn.keysight.com/litweb/pdf/5091-3385E.pdf

I don't have this analyzer, so I certainly yield to those with more direct knowledge of its expected behavior.

[G0HZU]

My 8566B checks out much better than this for accuracy.

I think the sweep system phase locks the LO at the start of the sweep and for accuracy across the rest of the sweep it then relies on the linear tuning of the YIG oscillator and a linear ramp generated with a programmable current source charging a capacitor.

So I would (typically) expect to see the error get worse towards the end of the sweep. But with a 20ms sweep of 100MHz with a 100MHz span with 100kHz RBW mine shows about 200-400kHz error at the far end of the sweep if I feed a comb signal into it with 10MHz pips. To my eye the centre pip at 100MHz still looks to be accurate. There is a slight improvement if I slow the sweep speed down and it then looks spot on across the whole span. So mine looks a lot better than your first plot.

So I think your analyser has a slight issue here.

At first, you could try checking the ESR of all the large tant and electrolytic caps on the A16 card. If you don't have a means to measure ESR then I'd suggest getting an ESR meter that can test 'in circuit'  if you have an HP8566B as the caps do tend to dry out and fail on lots of cards in the RF unit. Even if you fix this fault it's also worth taking the time to check all the large tants and electrolytics on all the cards in the RF unit. Especially the cards associated with the YIG filter and the YIG oscillator.

Also, I think you can access a buffered version of the tuning ramp on a rear panel BNC connector. It might be worth looking at this ramp on a scope  for different sweep rates etc.

Also, be aware that the PCB card edge connector pads on all the plug in cards in the RF module are plated with a microscopically thin layer of gold. So if they look like they need cleaning you must not use anything abrasive or you will remove the gold. There's a note in the service manual on how to clean the PCB edge connectors correctly and safely.

[eriklidgren]

Mark, I'll do some more test this weekend with a signal generator.
I followed the calibration instructions, adjust amplitude cal pot, then frequency zero pot, and run the error correction routine and store the result.
G0HZU, thank you for confirming expected behaviour on your analyzer. I'll do some looking around in the RF unit and measure some capacitors.
Regards
Erik

[eriklidgren]

I did some looking around on the A16 board, but didn't find anything suspicious. So I continued with A19-A21, some electrolytes on A20 and A21 looked dodgy, so I ordered new ones. After I replaced them and tested the SA, the problem is gone. I changed all capacitors at the same time, but I think the C7 on A20 might be the culprit. The other ones on A21 are just decoupling.
I'm attaching some plots. The cal data looks better, the frequency calibration data tapers down to 1 Hz, last time it stopped at around 400Hz.

The next thing I'm going to look at is the 1kHz and 30kHz RBW filters. The noise floor is noticeably higher with those RBWs than with adjacent RBWs, 5-10dB.

Erik