At last HPIB is working on the 8568B! Woohoo and all that.
It turned out to be the A12 RF section interface card. According to the manual this card 'contains digital functions that allow the front-panel controls to interface with the A15 Controller assembly' (CPU).
Well, it does a little more than that.
Having drawn a blank by replacing the CPU card I went back and retraced ground I'd already been over. And finally, I noticed that the reason the display froze was that sweep stopped. Since it's a digital display it continues to show the last update on screen even if sweep is stopped. (My poor excuse for taking this long to notice is that I was looking at the screen on the left, not at the Sweep indicator LED on the right.) I'll return to this shortly.
Having exhausted every possiblity I could think of I decided to check out why the REM and Addr'd LEDs didn't light up. First possibility was dead LEDs. Not likely when every other LED on the front panel was working and doubly unlikely when the two non working LED's were status indicators for the very fault I was trying to fix!
So I can see an LREM signal coming out of the CPU card, low for remote, high for not. And sure enough, when the instrument is addressed that line goes low. Transfer the probe to the front panel traces and the line stays high whether the instrument should be in remote or not. So I traced back from the LED and discover it goes through the A12 assembly.
And there, on page 2 of the A12 assembly description was the answer. They talk about the Service Request block and it contains the following passage.
The priority encoder, U1, collects the outpus of all interface functions, such as keyboard, RPG, sweep, HP-IB, phase lock errors, and so on. When any input is active (low), the priority encoder:
1: Stops the Sweep
2: Alerts the main processor...
etc etc.
So now we know why the display seems to freeze but we're not a lot closer, yet.
Now the keyboard was working, ditto the RPG and sweep. I've never seen a phase lock error in this machine so I can't talk to that. Only HPIB wasn't working.
So let's pull the card and see what can be seen. According to the manual it should have been either an 85680-60168 or an 85680-60181 card. What I had was an 85680-60121 which looks like a similar layout to the 60168 but not exactly the same. As I don't have an extender card to lift it out whilst still running I decided it was time to stop and see if that extremely generous EEVblog member had one he was willing to let me have. He did. An 85680-60168 card. It arrived in the mail this week and the machine now works correctly on HPIB.
I used to repair these machines for Hewlett-Packard in Melbourne aeons ago, when they were HP8568A's and stopped shortly after the B's went into production. I remembered that HP sold an upgrade kit and it's documented on the Keysight website. They don't have a description of the 8568 kit, only the 8566 kit but the description and parts list shows that they changed the CPU card AND the RF Interface card. If they changed it for one model methinks it likely it was also changed for the 8568.
So I'm beginning to suspect that my 8568B is a 'bitzer' (bits of this, bits of that). I've already noted that the firmware wasn't the right version for my serial number and it seems the A12 assembly also wasn't right for my serial number. I can only speculate that someone had a roomful of faulty 8568's (A and B's) and was charged with getting a few up and running so they could be flogged on Ebay. I doubt that, having determined that it works manually, they bothered checking if HPIB worked as well. And I'm not complaining - I paid $200 including shipping for mine (it was in the same city, Phoenix, hence the low shipping charge).
Lessons learned. The first is that the circuitry is somewhat more complex than appears on the surface. The CPU assembly has the CPU, firmware etc but it also has the HPIB interface chips (SN75160/SN75162) and a TMS9914A HPIB interface chip. So looking at the schematic for the card it looks like ALL of the functionality is confined to that card. It's not at all obvious that things go off card and return.
I should have noticed that sweep stopped! That would have led me to investigate that part of the instrument and, I have no doubt, would have brought me eventually back to the A12 assembly and I would have seen the description quoted above.
I wish I could say that replacing such and such a chip repaired the card but, frankly, I'm just glad it's all working and feel no inclination to dig further. I suspect it's the wrong card for the machine anyway (though I'll be glad to be corrected on that point). As for why only HPIB failed, I have no idea.
Incidentally, the firmware still isn't the right version for my serial number - the CPU cards I received had the 7.4.1987 release, which seems to work just fine.
And I'd like to record my gratitude to the EEVblog member who so generously parted out an 8568B and made the path to this resolution so painless.