Unit Key Data

[20190930 and 20191001] Initial Power-On and Inspection

  • Checked mains power settings, connected both parts of the unit (with a cable duo picked up later at Peter’s lab) and switched the unit on.
  • Pretty much everything appeared to work (control panel, noise levels on display).
  • However, when feeding a signal, it became clear that the RF attenuator unit in the hp8566b was broken.
  • By now I already had ordered and downloaded all available ArtekMedia documents for the hp8566b.
  • I opened the unit, removed the attenuator module (HP 85660-60122), and verified independently that it was indeed broken.
  • Made an attempt to repair the attenuator, but this failed.
  • Partially reassembled the analyzer and put it in the repair queue.
  • Decided (a few weeks later) to purchase two hp85660-60122 on ebay; one from alltest_instruments and one from patentix_ltd. (XXX Need to add to Events.)

[20191114] Replacement of the Attenuator Module

  • Upon arrival of the hp85660-60122 from alltest_instruments (XXX link), I made another attempt at fixing the hp8566b.
  • About half the control wires (rainbow flatcable) were cut through; so much for ‘tested and working’:
  • I re-soldered the wires and used heat-shrink tube to isolate the solder connections:
  • Then I wrapped it all together using isolation tape, in an attempt to add some strength to the construction:
  • After re-installing the attenuator unit, I reassembled the entire unit, this time using the official HP documentation :-)…
    • There was no need to remove the RF unit from the frame, as I did.
    • The PLL unit (?) that has to be removed in order to reach the DIL connector for the attenuator control requires some strength to remove.
  • After reassembly, I checked the complete hp8566b unit with various signals and compared readings with my hp70000 system. The unit proved to function just fine; the readings matched very well with those on the hp70000! More often than not, readouts were less than 0.1 dB apart!
  • By playing with the reference-level and (manual) attenuator settings, I verified the proper operation of the “new” hp85660-60122! → SUCCESSFUL FIX!
  • However:
    • At some point it went out-of-lock completely after full-sweep for almost an hour. A power cycle solved the problem, at least for now.
    • Pretty much all traces show signs (at > 60 dB below REF LEVEL) of (internal) reflections. This analysis still needs to be verified. It does not render the unit unusable, though, far from it.
    • The frequency offset measured using the pa3gyf-gpsdocxo-v1 was quite high in view of the fact that the hp8566b has an oven oscillator. The offset was, IIRC, around 11 Hz on 10 MHz, so worse than 1 ppm. More details are likely to follow later, but perhaps the OCXO needs some fine-tuning.
  • Decision [20191114]: Put the unit into operation and make performance assessments.

XXX TODO Transfer (remainder of) the .odt blog to this site!

[20210819] Low-Frequency Accuracy

Today, I measured the low-frequency accuracy of the unit, connecting it to the output of the HP-3325B generating a -40 dBm sine wave of 1 kHz. The SYNC out of the HP-3325B was used to assure that the output frequency was (well) within 1 Hz (10^{-3}), using the HP-5316A and the GPSDOCXO. Final conclusions were derived after more than 1 hour after startup.

The settings on the unit were: fc = 1 kHz; span = 200 Hz; RL = -40 dBm; ResBW = 10 Hz; VidBW = 30 Hz; SwTime = 6 s; Attenuation = 10 dB (all ‘coupled’; none manual, where applicable).

First, the power readout was completely spot on (using PEAK SEARCH): -40 dBm +/- 0.5 dBm (conservative estimate).

Second, still, to my eyes, weird side-band noise readings. This has been described earlier, so I won’t go into details, but it’s still there….

The frequency readout started at 1.00 kHz initially (i.e., once the oven was heated and the PLLs were in lock, but then gradually drifted. After roughly 1 1/2 hours, the readout had gone down to 976 Hz…

Hence, the center frequency in this case is of by more than 2%…

From this reading, I set the frequency on the HP-3325B to 1100 and 900 Hz, respectively, in order to check the horizontal scaling (SPAN).

  • 1000 -> 1100 Hz: Readout from 977 -> 1076 Hz
  • 1000 -> 950 Hz: Readout from 977 -> 927 Hz

We conclude that for this case we have a frequency offset problem; the frequency span appears to be OK.

Is this within spec or not? Also, I wonder what would happen if we lock the HP-8566B against our GPSDOCXO

Time to get something to eat…

[20211117] Frequency Response around 6 GHz

Today, I performed a small experiment checking the frequency response of the HP-8566B in the 5-7 GHz range, actually in an attempt to get some feeling on the frequency range of the coaxial cable used. The full set of pictures is here.

I entered a signal from my HP-8350B, sweeping very slowly (20s sweep time) between 5 and 7 GHz into the HP-8566B, and used its two traces to captures maximum (top trace) and real-time detector values, and here’s the result after several minutes:

Clearly, the response is nowhere near flat… Something serious is happening at around 5.8 GHz. Clearly, there is a problem in frequency response when the HP-8566B switches to mixing against a different LO harmonic at around 5.8 GHz (or thereabout)…

When I connected the signal to my Wiltron-560A (using its Channel A, see the photo set for the detector probe used), I obtained:

Which is certainly a lot better,,,

Here’s the detector used for the Wiltron-560A [i.e., it it the one with the red marking on its “LF Cable”]:

Well, not sure what to think of this yet, but the HP-8566B looks like in need of some serious calibration…