HP 8643A / 8644 operation questions

[fenugrec]

Hi all,
I'm just "finishing" (hopefully) the repair of a 8643A signal generator, and have a few questions for anyone with a known good unit :

- does the unit warn you when you're exceeding the normal output range ? (+13dBm on the 8643A) On mine, I can punch in +19dbm and it won't complain, but looking at the scope I see significant clipping of the top of the waveform, and a spectrum analyzer also shows much higher harmonic content.
Not sure if this normal behaviour, or if mine has other issues that are not detected by the self-calibration and builtin ILD.

- how do you adjust the internal 10Mhz reference ? The instructions I found apply to the the ovenized option which I don't have...

- has anyone ever seen a paper version of the CLIP / schematics for these units ? seems awfully rare. Artek sells a copy; I haven't decided if I'll purchase that.
Actually, if someone has those and wouldn't mind looking up a few things, there's a few areas I'd like to look at :

1- charging circuit for the RAM backup battery (just to confirm what charging current they set; I replaced the battery on mine with slightly different specs)
2- STD MOD/NSM OSC module : I replaced one shorted tantalum cap in there; I'd look at how the others are used and if they're all power rail decoupling, I'll recap with alu-polymer electrolytics instead of tants. Also there was a burnt inductor that I eyeballed with a replacement but curious to see the original part specs.

Thanks !

[wn1fju]

Buy the CLIP from Artek.  It's cheap insurance should you need to repair the 8643A further.  Be advised, however, that the CLIP does not contain the schematics for the two switching power supplies (which, of course, were the defect in my unit!). 

My unit behaves similarly to yours:  at RF levels >15 dBm, I see distortion on the scope and increased harmonics on the spectrum analyzer.  The 8643A doesn't complain however.

The battery is listed as a 45 ma-hr 3.6v NiCd.  Picture of the charging schematic attached.  Actually, what I did with mine was to put three AAA batteries in a plastic holder with a series diode (to prevent charging) and used that.  It has been working just fine.

I have the ovenized oscillator, so I'm not sure about yours.  There is no metal can with a screw covering a hole for an adjustment???

I can't help you with the STD MOD/NSM OSC module since the schematic is five pages long and it is not at all clear from the parts lists which are the tantalums.  Get the CLIP and enjoy.

[fenugrec]

the CLIP does not contain the schematics for the two switching power supplies

That's fine, they're fairly simple off-the-shelf units .

My unit behaves similarly to yours:  at RF levels >15 dBm, I see distortion

Aha, thanks for validating this.

Thanks also for that screenshot ; so they have it rigged for ~ 5mA, that should be reasonable. I have an 80mAh Varta 3.6V NiMH, designed for this kind of application. I was able to fix it onto the DCU and everything fits in normally.
Interesting to note : Vbat goes through the motherboard to power the RAM banks, so if you remove the DCU card from the chassis it clears the memory. Probably to ensure a re-calibration on the next power cycle.

There is no metal can with a screw covering a hole for an adjustment???

I was just not looking properly (upside down with a mirror) and indeed, there is an access hole for the adj screw near the rear BNC ports.
But I did a quick test with the CHU time broadcast on 3.330 MHz and as far as I can tell I'm within about 0.1Hz of the carrier -- not going to even touch the adjustment.

[fenugrec]

Finally got around to uploading a video of the repair process, if anyone's interested.

https://youtu.be/5btPjheYEjs

[fenugrec]

Also, I'm not sure anymore if I have an OCXO or not !

My options sticker says no ("H02 K01" ):
" H02    Improved Performance from 252 kHz to 1030 MHZ"
Couldn't find anything about what K01 means.

But this certainly looks like one : insulated can, with a PCB mounted on the back, 4-5 pins going into the can...

It's labeled "OSCF   OSC92-388" which is of course a custom number, no help there.

[wn1fju]

The 8643A CLIP is even more confusing!  The schematic shows two possible oscillator cans, part numbers 0950-0411 and 10811-60111.  The latter is the usual OCXO found in dozens of HP pieces and has provisions for electronic frequency control.  But let's assume you have something closer to the 0950-0411 version.  The schematic marks pins 1, 2, 3, 4 as GND, +21VDC, GND and +15VDC.  The nominal 21V should be the oven voltage and comes from the smaller (~24V) switching power supply.

My guess is that you do have an OCXO.  If it were me, I would plug the 8643A into AC and let it sit for 30 minutes (presumably the 24V "standby" power is on in this case).  Then unplug it and stick your fingers onto the oscillator can.  I would think you would feel some warmth.