Help Identifying 1 Ghz clock on board

[Generalquark]

Hi,
 I am working on an Agilent 8647A Signal Generator that has a 613 Reference out-of-lock at 1 Ghz error. I have removed the A4 board from the unit and the 10 Mhz clock seems to work fine. I am having trouble finding the 1 Ghz VCO that is phase locked to the 10 Mhz signal. Do you think they used a normal VCO or circuitry I am not familiar with to generate the clock? I would assume it is in the shielded area, but I am not seeing anything that looks familiar. I figured I would ask on this board. I am attaching a picture so perhaps someone could identify it. I have attached the block diagram where I think the 1 Ghz out from the A4 card is labeled. Thanks for the help!

[DaJMasta]

I..... don't know if it's generated on this board.

I don't know the instrument or schematic, but I agree I would expect to see a VCO or similar in an instrument of this age.  The thing that makes me question it is the lack of any high speed connector... there are some controlled impedance traces going to the connectors, so maybe, but those things aren't going to keep a 1GHz signal contained and clean, I'd expect that to go through coax - if it were going through a backplane at all, it would be through an integrated coax connector.  This is a bit of HP test equipment, after all, not some consumer gear, so I would expect every high speed signal  or signal they want to keep interference out of (no bonus mixing products) would be through coax.


That said, have you tried looking up the main chip part numbers?  If there is a 1GHz synth with no discrete VCO, it's gotta be an integrated synthesizer, which is only gonna be one of the larger chips.  If neither of the PLCCs are, I'd say it's unlikely.

[Generalquark]

Hmm,
 I will try to look up the datasheets on all the large parts. I have attached the service manual notes that seem to suggest that it is indeed generated on the A4 board, unless I am misunderstanding it. Thank you for the input.

[DaJMasta]

Well that certainly sounds like proof it's on here somewhere.  Could the the 10MHz reference just be multiplied a bunch?

I'd look at the smaller PLCC as being an important part... though the marking seems to be coded.  The SO-8 in the middle can could be important, as with the larger RF package in the top can if it's got a number.


Those pairs of vias with silkscreen circles around them may be your test points for various stages of the circuit, looks like one is always the controlled impedance trace and one is always through to an internal/lower layer - especially if it's multiplying, the first few stages from the 10MHz source would be visible on a regular scope.

[ArthurDent]

If you look at the pinout in the manual that has the schematic of the A4 board you have, you will see what the connections are. If you have the 10Mhz out, the first part of the circuit is working, and if the divide by 100 has an output of 10Mhz feeding the phase comparator, you should have 1Ghz on pin 3 of the connector as shown.  The signal lines should be coax as well. The sections of the phase-lock loop multiplier are in the shielded enclosures and should be easy to trace back from the connector pins.

Also check that there is a 200Khz output as well. All the signals with the exception of the 1Ghz should be easy to check with any counter or scope. If these signals are all there it probably is another board causing the problem. If the frequencies are out of spec, the oscillator may have drifted out of range or failed and using an external 10Mhz reference may allow it to work.

[SeanB]

Q6 U21 looks a likely suspect, with the diode there being the tuning for the oscillator. the single loop of gold plated copper is the inductor, the rest being all bias, input and output. U15 looks like the prescaler and control, feeding the output down to U16 to provide the output, with a tap and U26 driving the divide by 50 U18.

Pretty much all in a line down the board, and it looks like J2 selects internal or external clock.

[Generalquark]

Ok,
 I have some new updates to this one. I was able to get it to boot without the error once, but then the 1 ghz signal shifted over after being powered on for a bit. Since then it has been shifted over. I have tried heat and cold to the entire board. I am still unable to find where the 1 ghz is generated from for sure. The 10 mhz clock output seems good as Ext Clock reads right at 10Mhz. I tried looping that back into the Reference input, but still had the same issue. The 200Khz output is good also. I think it is safe to say the 1 Ghz is generated on this board, but I can't determine where for sure. I'm not used to Troubleshooting RF without a schematic and a known good board. Even then it can be tricky. Does anyone have any ideas? I have uploaded a picture of the board with locations where I took some spec an measurements. It seems like the signal is shifting quite a bit. It is right around 1 Ghz at U18.  I have a spec an to check signals with if anyone has an idea of where to look. Thanks!

[TheSteve]

That board is the same or extremely similar to the reference board in the 8920A/B series of service monitor. The schematics for it are available!
I doubt it is Q6, I have also had the same/similar error before and it was one of the MMIC's failing.

edit: here is a link to my original repair thread:
https://www.eevblog.com/forum/repair/hp8921a-service-monitor-high-stability-reference-noisy/

Note that since my post in 2015 the schematics have been released.

[Generalquark]

TheSteve Is this the schematic Page you were mentioning, or is there something more detailed?

[TheSteve]

That's a block diagram, not a schematic.

Try this:
http://www.petergottlieb.com/docs/hp/8920/HP%208920%20CLIPS.pdf

[Generalquark]

I knew that  . This is the first time I have interacted on this board and wasn't sure if the same terminology is used here as in Manufacturing. Thanks for finding that schematic for me. I have taken more measurements and am looking for input. I am wondering if CR11 might be causing a problem? C37 read good and the Cap (C32) and Inductor (L11) that measured a bit off might just be the capability of my equipment. The 10V instead of 11V on the rails could be a result of me running through a db9 extension cable from the main board instead of plugging right into it. Does anyone know what the Tuneline_VCO line is for? Is this used for phase locking? Thanks for all the help. I appreciate it.

[TheSteve]

Just how are you probing these points you're testing? A regular probe won't cut it @ 1 GHz.

[Generalquark]

Well.... I am slightly embarrassed to tell you. I took a bnc end, put a piece of home electrical solid copper wire into the socket and soldered it in place. I then sharpened it to a point with an exacto and put an alligator clip ground on it. I believe it is measuring pretty close to correct because the one time the unit booted up without the 1 ghz out error, I was able to see a 1 ghz output signal with the probe. The signal was not shifted like it currently is when I get the error. I will now gladly accept my lashing for not using the correct probe for the job.

[Generalquark]

I found that the 10_PLL_10MHZ signal coming into U10.11 is not there on the board. I was able to confirm that it is there on a good board. I cannot seem to tell if this is an input or output. It looks like an input, however I cannot find the source of where it would be coming from on the schematic. Does anyone see where this signal is generated? Thanks!

[shakalnokturn]

PLL divide by 50 section. U27, MC12013L, Pin 7.

[Generalquark]

Ended up finding that U14 was intermittent with the home made probe. It actually reads quite well in the frequency range I was using it. Replaced it and the unit is working fine. Thank You all for all the help in troubleshooting this board!