Tek 7603 with 7L13, lots of little problems

[okto]

I recently picked up a 7603 with the 7L13 spectrum analyzer at a hamfest. It powers up and works okay, but it has some issues:

1. There is a constant but irregular vertical jitter/noise in the signal trace. I would describe it as “jumpy”, but there is also clearly a harmonic element to it, IE, I’m seeing what looks like an additional signal of randomly varying frequency heterodyning with the base level of noise. It reminds me of having a dirty potentiometer in which surface oxidation has formed a crude rectifier.  I don’t have the technical knowledge to describe this succinctly, so I’m attaching a video.

2. The readout cannot achieve full scan height with the scan trimmer turned to the stop. The service manual (and all photos I’ve seen) indicate that the readout should be positioned at the top and bottom graticule division; mine will only get within about a division and a half of that point, IE, above the second division and below the seventh.
The signal trace seems to have full range of vertical deflection, from just below the first graticule division to off the top edge of the screen.

3. Signal amplitude/offset changes (and more baseline noise becomes visible) when 300kHz resolution is selected in 10dB vertical deflection. 30, 300, 30k, and 3M Hz all draw the baseline at the same height, and 300kHz will draw the baseline at this same height when 2dB or LIN are selected.

4. The UNCAL indicator lights frequently, and seems to turn off in an unpredictavle fashion. It doesn’t like increased sweep speed or certain combinations of frequency resolution and span. Usually I cannot go past 10ms with the UNCAL light remaining dark, and never last 5ms. Some of this may be operator error; this is the first analog spectrum analyzer I’ve worked with.

5. The service manual indicates that with the calibrator connected to the RF input, with frequency span set to MAX and resolution set to 3M that I should be able to see the comb pattern of the harmonics of the calibration signal. At these settings,  I cannot see any vertical trace at all unless I reduce resolution to 300kHz or less.

6. The signal peaks for the 50MHz calibration signal and its harmonics jitter at the same inconsistent frequency as the noise baseline does at 300kHz resolution. This noise seems to affect the signal trace at all settings, and 20MHz span/3MHz res I can see that the baseline trace has a sawtooth ripple that’s out of sync with the horizontal scan rate (the ripple appears to be slowly marching from right to left).

So far, I have done the following according to the service manuals for the 7603 and the 7L13:
1. Check low voltage supplies. All are right within spec and show zero ripple on my 465B.
2. Adjust vertical amp bias and gain.
3. Adjust readout height, scan, row, column, and vertical span. Responds as expected to all adjustments except scan, which as previously mentioned will not occupy the full screen height indicated in the manual.
4. Calibrate 7L13 frequency dial, sweep, log, amplification and contrast per manual procedure, although I am not 100% that the instrument is in-spec enough for this calibration to be accurate. The frequency dial, for example, will not stay aligned with the 50MHz signal peak, either while the instrument is running or between power cycles.

My test equipment is limited to a Tek 465B which is well-behaved, a 2MHz function generator, and a true-RMS DMM.

I appreciate any pointers y’all can provide before I load the parts cannon.

300kHz noise and amplitude problem:
https://youtu.be/bygs0DUs_EU

UNCAL light coming on:
https://youtu.be/Zt8jx3kTHUU

[0culus]

I'd be willing to bet that the readout position/scam issue is a bad capacitor somewhere, probably in the mainframe itself.

[kirill_ka]

It doesn’t like increased sweep speed or certain combinations of frequency resolution and span.

I'm pretty sure this is what it is supposed to do. It warns you that the sweep is too fast for the BW/VBW. RTFM?

[okto]

It is doing its job in the most basic sense, but I don’t believe it’s supposed to be unable to use the whole top half and lower third of the sweep knob.

[0culus]

I don't have a manual handy, and I've never used a 7L13, but that could be normal. Sweep-tuned spectrum analyzers will always have optimal combinations of sweep time, RBW, and video BW with your normal frequency span. If you deviate far from optimal settings, you will eventually set things to a point were stuff doesn't work right anymore, which is what the uncal light means. Some spectrum analyzers (such as the HP 8566 and 8568 systems) provide coupled function mode, which you can manually override if desired, which automatically sets everything optimally no matter what you set the main functions (center freq, span, reference level) to. Even more primitive HP spectrum analyzers that have mechanical controls typically have a way to couple or uncouple RBW and span by pulling on the knob that controls it.