HP 8566b low signal dBm

[bristpi176]

Hi,

I'm working on a new to me HP 8566B and I'm trying to narrow down a problem.  The analyzer reads on frequency but low on dBm.  When I hook up a signal, whether it be the internal cal signal or one from my generator, it reads about 26 dBm lower than actual.  It won't do anything on the calibration process (Recall .  No sweep that I can see.  On (Recall 9) it begins the process but stops when it doesn't get the expected signal and says to calibrate the amplitude. 

The amplitude readout responds to the calibration adjustment on the front of the machine, but can't bring it into spec.  Has anyone else seen this fault and can point me in the right direction?

Sorry for the less than technical description of the problem.  This is what a business degree gets you.

[0culus]

My first assumption here would be a bad frontend attenuator. If you take control of it manually (via the appropriate "coupled function" key on the left side of the RF (lower) unit keyboard, can you exercise it through all settings (note that 0 dB attenuation can only be keyed in manually with data entry for safety)? Put a known good 0 dBm stable CW signal on the input. The attenuator should go from 10 dB-70 dB setting, with corresponding movement of the noise floor displayed. If the displayed CW amplitude changes on one setting of the attenuator vs. another, you have a smoking gun.

The bad attenuator can basically be two things. Either (1) someone smoked it by keying a transmitter directly into the front end (DO NOT do this) or (2) mechanical failure inside the attenuator itself. The latter (2) situation can be the disintegration of tiny rubber rings that help cushion the force of the solenoids moving. This can be rebuilt, but it's like doing surgery on a Swiss watch. You'll need a good stereo microscope. The former (1) situation can be a very expensive problem. If the attenuator took the brunt of the damage you may be ok with replacing it, but this overpower excursion can also damage or blow up the first mixer (which is very bad).

[bristpi176]

Thanks. I’ll check that out tomorrow.

[0culus]

Another thought...particularly if this thing has been sitting for a long time before you got it...the act of exercising the attenuator might alleviate the problem if it's just something stuck. I wouldn't bet on it, but suspecting mechanical issues in old HP stuff is often a good place to start troubleshooting.

[bristpi176]

I set up the test and got a change in dBm at the -40 dBm mark.  It went from -26 dBm to -6 dBm on a 0 dBm signal.  So it looks like the attenuator is toast.  I guess I'll start looking for one on eBay that comes with a return policy if it doesn't work.

[0culus]

You may yet have another problem (getting a known good attenuator will show if further problems exist), but I think that's a pretty good indicator that something is wrong with the attenuator.

If you remove the attenuator (it's kind of buried in the RF section, rather than being attached to the front panel like on the 8568B), you can try taking it apart and visually examining the contacts and resistive pads with a loupe or microscope. They are fragile, so don't touch. It will be immediately obvious if one or more pads are burned and charred from excessive power input. If they look fine, the attenuator is possibly fixable with a rebuild. If there is evidence of burning and charring, the first mixer could be damaged as well.

[bristpi176]

Good news and bad news.

I put in a replacement attenuator and that problem was gone.  It attenuates properly and works through "recall" 8 and 9.  Its reading signals at the right frequency and amplitute.  Next problem is that it won't complete the calibration process.  It appears to get through 1/3rd to 1/2 of the process and stops saying "cal amptd". 

I changed out the 1st converter and no change.  The internal cal signal looked very noisy on the screen, so I swapped the phase lock assembly.  Not any cleaner and no change on the calibration process.  I then took a clean signal from my generator and plugged that in and still didn't get through the calibration process. 

Two questions, how important is the internal calibration process and do you have any ideas on where to go next?  Thanks for your help.  I'm getting closer and it'll be great if I can bring this al the way back to life.

[bristpi176]

Here's a pic of the display from the 100 mhz cal output on the 0-2.4 ghz scale

[0culus]

That looks normal for the cal signal. You didn't say you had, so make sure you adjusted the amplitude calibration and freq zero calibration before you run the cal routine. After the instrument has warmed up for a while (I'd give it half an hour), connect the calibration signal to the RF input, using a decent quality cable. Then, key in RECALL and then 8 from the data entry pad. This calls up a build-in routine. Using a small flat screwdriver or tuning tool, twiddle the AMPLD CAL pot (just left of the RF input) for a marker amplitude of -10 dBm. Then, key in RECALL and then 9 and using your turning tool to adjust FREQ ZERO (the bottom of the three adjustment pots under the CRT intensity knob on the IF/display unit) to maximize the amplitude of the response shown on the CRT display. Then, trigger the self cal routine with SHIFT and W. Once finished, you can enable or disable cal constants with SHIFT X and SHIFT Y, respectively. You can display the constants on the screen with SHIFT w (small w). All the constants should be small, ideally a mix of positive and negative but won't necessarily be. As long as none are anomalous (e.g. a constant shown -5 dB is indicative of something wrong) everything is fine.

The self cal constants help correct for small errors in amplitude and frequency. Whether they are critical to use or not, really depends on what you're doing.

Also, when the instrument is going to be on but not in use, SAVE YOUR CRT by using SHIFT + g to turn off the CRT beam and SHIFT + h to turn it on again. As the trace is digitized internally before display, this won't affect the ability to program the analyzer over GPIB at all.

[bristpi176]

I did successfully complete the procedure for amplitude cal (recall and frequency zero (recall 9) before trying the calibration process.  It still won't pass through the process.  I have attached a pic of the constants but I don't see anything out of the ordinary in them.  I'm not really doing anything with my collection at this time except for learning how they work and playing around with the older stuff.  I'm trying to learn something useful along the way.  I guess I'll button this one up and call it good.  I'll see how it performs along the way.  Thanks for your help.

I've got more pieces to work on that I scavenged up.  All HP.  An 8662A, 16702b, and a 54111d so I'll be back as I pick through them.  It's been a learning experience.

[0culus]

You're welcome!

Those constants look quite normal. Those constants were stored from the last time someone ran the self cal routine. Should be able to use them no problem with SHIFT-X and SHIFT-Y. They are stored in battery-backed memory will last as long as the memory backup battery located on the CPU board does.

If the thing appears to be working otherwise, use it and learn how to use it. You can learn a lot from troubleshooting HP RF stuff. It's well documented for the most part and they had a large cadre of VERY smart RF engineers working for them. Recommend checking out the HP Journal (still hosted by today's HP here: https://www.hpl.hp.com/hpjournal/journal.html, but also available elsewhere). They tended to dedicate issues to new products that were released, with multiple technical articles written by the engineers that designed it. The August 1979 issue covers the then-revolutionary 8566A.