HP 33120A, It's time to fix it

[vaualbus]

So after almost 10 I decide finally to have this instrument fixed. I bought this back 2014 from ebay and found some burned components, some zener have burned the board.
Than the problems really started:
  - Had to source a front panel boards as I broke the VFD during the initial repair apparently
  - Than I was able to turn it on for the first time, I quickly realized that something was wrong in the output amplifier, as I got some clipping on the waveform.
  - Than one day the output was gone I traced it back to the main DAC been bad.

Than phase 2 of the repair started:
I started bought components from  China to replace the DAC and some the transistor I though were bad, replace them and burn again the DAC.
When I had to change it again I destroy some of the pads on the mainboard  so the first stop moment started.
After a lot of years I decide to give another go at this, so I bough a new DAC and hack it on the remain pads with a lot of bodge wires, still it work for some time than the DAC was bad again.
So basically I give up.

Years passed and now I wanna get this working again finally!
So I decide to try to repair the broken pads with some PCB epoxy and the SMD pads kits you now find on ebay,
than I will resolder yet another DAC and give another go. This will start next month.

Now that you know the full story, I can begin the real post I wanna make about something that always bothered me:

Apparently the AGND you found in the schematics is not connected to the GND, the supply voltage of chips in the analog output section are referenced to AGND.
Those are supplied by a +15V volts that is generated from the +18V via a zener diode. The 18V is generated by U1002 that is referenced to GND.
The strange part is that the bypass caps on the 15VA is referenced to AGND!

In the schematics you can see that  apparently GND/AGND/AGN2D/PGND are connected together to JM1006 but I cannot find any such component on the PCB (I suspect it is the big middle hole where the plastic support is inserted by the way).
When I measure the voltage between AGND and GND, I see around 6.5V!, If I connect the two ground together using a multimeter in current mode is see around 0.152A of current flowing between the grounds!
Logically I think those ground should be the same but I cannot really confirm it in any way. If the ground should be the same I see why I am destroys DACs as probably the output are biased above the maximum allow value due to this voltage difference!

One of the burned caps  that I had to replaced (C707) did burn the board and I had to dig in it to remove the burned FR4, so I could have destroy an important trace, that with my luck was the one that connected the grounds together?

Anyone have any idea about this ground situation nation ?

Or have a 33120A that can open to test if there is any continuity between AGND and GND? AGND can be found at pin 2 of U603 as indicate din the subscreen, GND is the ground of J401 SMB connector  .

Despite all I still expect to have a failed output amplifier part as I was never able to really debug it due to the main DAC always broke before I can do any serious measure, so yeah this is really a never ending story 

PS
In this 10 years I got a lot of interesting instruments but all results in a easy repair as changing a cap or reconnect a cable (or not possible repairs due to hybrids gone bad  ), so really this could be said that is the first "real" repair that I am doing ahah
Also I need to investigate this before installing yet another DAC because I get bored of broken DACs )

[Swake]

Found JM1006 on the board in de image on page 142, Component Locator Diagram.
About 1/3rd from the top, and 4/5th from the left.

They might have wanted those grounds to be separated physically to avoid ground loops between those circuits, and then connect them in single central reference point to 'earth'.

GND = 'earth'. That is certain as it is mentioned as 'Earth referenced circuit' and in the schematic some safety caps are connected to it.
PGND and GND are certainly connected. Else U1001 and U1002 (page 140 top) cannot function normally. The ground 'after' the regulator received another name.
The same happens to the GND after CR703 & CR704 (page 137 top/left) to make the + and - 15V from the same 18V. After those zeners it is named AGND.

There is also EGND, this one is not to be connected with the other. It is the ground for the GPIB bus and that bus is isolated from the rest by opto-coupler U901.

JM1006 is not in the component list, but I think that is just an omission as JM1001 to JM1004 & JM1051 is in there (page 118 top) with quantity '7'. So there might also have been a JM1005 at some time.

[vaualbus]

Thanks! Indeed I was able to found it too this morning, and be hold there was a clear broken trace nearby, I guess I either broke it while probing or, as it is near a cap could be due to leaking caps?
And of course was the AGND trace so yeah I was 100% right that all the non earth referenced grounds need to be connected together! I feel kinda stupid to have not notice this over the years  ,
but I always was perplexed by this ground "problem" so I am kind happy to have figured it out finally

So now I will fix my pads situation and than solder a new DAC, I really hope I won't burn it on again!
It makes sense, again, If I had 6V across the grounds to demage the DAC as it is one of the few components that were across both grounds!

So I will update this post once I have the required stuff for repair the PADS!

[vaualbus]

So I can provide some updates finally!
I decide to buy the required stuff for my surgical operation yesterday, so I perform the pad reconstruction and soldered finally a new DAC (and I do not burn it in a day  ), so finally it is time to really fix the generator.
As I suspect I still have not output, but I have sync signal so yeah the DAC is working! I quickly trace down that there were some bad soldering in one of the chips used for pre-attenuation (U601) so I quickly resoldered it correctly, and now I have no huge offset in the AMP_IN+, AMP_IN- signals (a small offset is normal but before resoldering the analog MUX I had +V offset in the AMP_IN+ signal!), now why I have no output at all?
I know by other posts about this output amplifier, this is current feedback amplifier, and it is tricky to repair.
I begin to trace the signal and I quickly realize that something is wrong, I have a signal at the base of Q709, but no signal is present at the opposite PNP transistor Q710.
It seems that I have no signal at the base of that transistor at all. I replace both Q702 and Q710 but not difference at all.
I guess the bias of Q702 is bad, so the transistor is always in saturation hence no amplification and I only see a DC offset at the emitter of Q710.
I have no idea why Q702 is in saturation though, I am thinking of remove the feedback form the amplifier (removing R702/R740) and see if when it is in open gain it work?

If I read the service manual correctly, I am seen that DAC output should be between 800 mVpp and 1Vpp, but in my case I see only 344mVpp when measured at the +DACOUT referenced to ground. I am not sure if this is an issue or is just that we should not measure it referenced to ground as the DAC has current output.