Long story short, I have an abusive relationship with this machine. It was one of my first PS/2 systems and from day one I've had to fight with it to try and pull it out of its eternal slumber. From a dead diskette drive, to a blown fuse for the FDC on
the system board to a failed memory module (and many other fun things in between), its really been reluctant to come back to the land of the living.

Just yesterday I thought I finally had it, but it turns out it had one more funny trick up it's sleeve; After running configuration and restarting, I was greeted to the display "scrolling up" at a seizure-inducing rate. I swapped monitors thinking
something on the other had went out, but that was a no go. I then stuck an XGA adapter in Slot 3 and what do you know, I had normal, stable video once again.

So, now it looks like I need to figure out what else on this system board has died and try and fix (if that's even possible) the on-board video. I'm not really sure where to start here, so I'm hoping someone has an idea what's died and some helpful tips
on what to check.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 26.9.2021 22:40, 404time...@gmail.com wrote:

Long story short, I have an abusive relationship with this machine. It was one of my first PS/2 systems and from day one I've had to fight with it to try and pull it out of its eternal slumber. From a dead diskette drive, to a blown fuse for the FDC on

the system board to a failed memory module (and many other fun things in between), its really been reluctant to come back to the land of the living.

Just yesterday I thought I finally had it, but it turns out it had one more funny trick up it's sleeve; After running configuration and restarting, I was greeted to the display "scrolling up" at a seizure-inducing rate. I swapped monitors thinking

something on the other had went out, but that was a no go. I then stuck an XGA adapter in Slot 3 and what do you know, I had normal, stable video once again.

So, now it looks like I need to figure out what else on this system board has died and try and fix (if that's even possible) the on-board video. I'm not really sure where to start here, so I'm hoping someone has an idea what's died and some helpful

tips on what to check.


Vertical roll? Sounds like a missing or out-of-spec VSYNC signal.

VSYNC is pin 14 on the HDD15 video connector. Make sure the pin isn't
bent, pushed back into the connector or corroded. Then check the path
back to the VGA chip - looking for bad solder joints. There will
probably be some passive components right before the connector - check
for any opened resistors or shorted caps.

Then there's the external sync source from the AVE MCA extensions and
the associated switching logic. In the older VGA design (type 1 & 2
planars) this is realized by a discrete logic - U14(F) 74ALS04 and
U25(B) 74LS125. See here (page 3 physical):

https://ardent-tool.com/docs/pdf/schematics/m80_sch.pdf#page=3

(note that there is an error in the schematic, the VSYNC line should be connected to pin 14 of the video connector, not to the grounded pin 6)

I'm not sure if this also applies to the Type 3 (Axx) planar or whether
they moved the switching logic to the VGA GA/ASIC. But it shouldn't be
too difficult to figure out where the signals go from the AVE
connectors. AVE pinout:

https://ardent-tool.com/tech/MCA_Pinouts.html#AVE

We are interested mainly in the ESYNC signal on pin BV10:

https://ardent-tool.com/tech/MCA_Pinouts.html#sig_AVE

This signal should be at logic high normally (~2.7 - 5 V). If it's low
(<0.8 V), and you have no AVE adapters installed in the system, then
there must be a short somewhere along the way or the pull-up resistor is
open. But ESYNC stuck low would probably cause the picture to disappear completely, since the other signals would also be sourced from the AVE
lines instead of the VGA chip. Fault in the switching logic could lead
to the described behavior - more specifically the buffer marked as
U25(B) in the schematic above (it will be marked differently on the Axx
board, if it even exists as a discrete part). You can bridge the two
pins corresponding to the particular gate with a - let's say - 100 ohm
resistor and see if you get vertical lock on the monitor.

It may also be worth checking output frequency of the two crystal
oscillators - but again, if these were off, there would be more serious
image problems most likely.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Lot's of good information in here, but alternatively, the onboard video for these generations of PS/2's is in terms of speed and colors at higher resolutions is horrible. Your model 80 will happily boot an XGA-2 card and you'll appreciate it. It will
go in any slot that is not the AVE slot. You'll need to upgrade the reference diskette by booting the Option Disk and using the menu to update your reference disk, then run the reference disk.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

https://ardent-tool.com/video/XGA2.html

https://ardent-tool.com/video/xga_files.html

XGAOPT.exe (zipped image)

Installs the updated (as of '92) SC.EXE, XGA/2, and SCSI, SCSI w/cache
ADFs, plus updated *.dgs. For 5565, 5657, 70486, 7080, 7677, 90, 95,
P70, P75. For any system with refdisks after 92/93, these files are
probably redundant.

On 9/26/2021 23:42, lharr...@gmail.com wrote:

Lot's of good information in here, but alternatively, the onboard video for these generations of PS/2's is in terms of speed and colors at higher resolutions is horrible. Your model 80 will happily boot an XGA-2 card and you'll appreciate it. It

will go in any slot that is not the AVE slot. You'll need to upgrade the reference diskette by booting the Option Disk and using the menu to update your reference disk, then run the reference disk.


--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Unfortunately, the P/N could be changed for tracking manufacturing,
cost, or moon phase.

So, "it depends"

Good luck.

On 9/28/2021 10:29, 404time...@gmail.com wrote:

This is some fantastic information here. Unfortunately, I don't believe the attached schematic will be of much help to me as the P/N for the VGA chipset differs from what the Axx planar has (P/N 90X8941), so I'm assuming that this solution differs from

that of the Type I and II planars

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Vertical roll? Sounds like a missing or out-of-spec VSYNC signal.

VSYNC is pin 14 on the HDD15 video connector. Make sure the pin isn't
bent, pushed back into the connector or corroded. Then check the path
back to the VGA chip - looking for bad solder joints. There will
probably be some passive components right before the connector - check
for any opened resistors or shorted caps.

Then there's the external sync source from the AVE MCA extensions and
the associated switching logic. In the older VGA design (type 1 & 2
planars) this is realized by a discrete logic - U14(F) 74ALS04 and
U25(B) 74LS125. See here (page 3 physical):

https://ardent-tool.com/docs/pdf/schematics/m80_sch.pdf#page=3

(note that there is an error in the schematic, the VSYNC line should be connected to pin 14 of the video connector, not to the grounded pin 6)

I'm not sure if this also applies to the Type 3 (Axx) planar or whether
they moved the switching logic to the VGA GA/ASIC. But it shouldn't be
too difficult to figure out where the signals go from the AVE
connectors. AVE pinout:

https://ardent-tool.com/tech/MCA_Pinouts.html#AVE

We are interested mainly in the ESYNC signal on pin BV10:

https://ardent-tool.com/tech/MCA_Pinouts.html#sig_AVE

This signal should be at logic high normally (~2.7 - 5 V). If it's low
(<0.8 V), and you have no AVE adapters installed in the system, then
there must be a short somewhere along the way or the pull-up resistor is open. But ESYNC stuck low would probably cause the picture to disappear completely, since the other signals would also be sourced from the AVE
lines instead of the VGA chip. Fault in the switching logic could lead
to the described behavior - more specifically the buffer marked as
U25(B) in the schematic above (it will be marked differently on the Axx board, if it even exists as a discrete part). You can bridge the two
pins corresponding to the particular gate with a - let's say - 100 ohm resistor and see if you get vertical lock on the monitor.

It may also be worth checking output frequency of the two crystal oscillators - but again, if these were off, there would be more serious image problems most likely.

This is some fantastic information here. Unfortunately, I don't believe the attached schematic will be of much help to me as the P/N for the VGA chipset differs from what the Axx planar has (P/N 90X8941), so I'm assuming that this solution differs from
that of the Type I and II planars.

I did test pin 14 on the VGA connector and received 0.115v. I also tested pin B-V10 on both AVE (with no adapters installed) slots and received 0.002v, so it's definitely low across the board. I also did try and back-trace the traces, but I didn't have
much luck with that, either. There is also very little in the way of passives on this board (at least as I know them). There are a handful of electrolytics around the MCA slots, but that's it. Everything else just seems to be surface-mount ICs for the
most part. I'm really at the limit of my component-level troubleshooting skills, though.

I did take these photos the the VGA chipset and surrounding components. The only things I can readily identify are the chipset itself and the VRAM modules.

https://imgur.com/a/XgWQQoA

I will say I'm not really optimistic about this being something that's repairable, but I've been proven wrong before. I would like to get the on-board graphics working again as I had some AVE cards I was planning to use in this machine.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 28.9.2021 17:29, 404time...@gmail.com wrote:

This is some fantastic information here. Unfortunately, I don't believe the attached schematic will be of much help to me as the P/N for the VGA chipset differs from what the Axx planar has (P/N 90X8941), so I'm assuming that this solution differs from

that of the Type I and II planars.

It's a different P/N but the two parts are probably very similar
internally, maybe even identical. The package is different than on the
Type 1 and 2 boards, so you will have to figure out what is what by
checking continuity to the nearby components (if it comes to that...).

Thanks to David we have some hi-res photos of the board:

https://ardent-tool.com/8580/Type3_Planar_Photo_Front.jpg

I can see one 74F125 right next to the VGA chip, but without probing the
board I can't tell whether it's related to the AVE switching or not.

I did test pin 14 on the VGA connector and received 0.115v.

Either the sync signal has negative polarity (very much possible, 0.1 -
0.2 V sounds about right for that scenario) or there's a problem. You
would need an oscilloscope to see what exactly is going on there, the DC measurements gives you a very limited view...

Or we could measure the DC voltage on a known-good unit and compare the results. Be aware that the polarity is video mode-dependent, so I'd
suggest to measure it during the POST process as the mode used there
should be the same across all units.

I also tested pin B-V10 on both AVE (with no adapters installed) slots and received 0.002v, so it's definitely low across the board.

Now, that's not right. Are you sure you have made a good contact with
the pin (it requires probes with relatively sharp tips) and that it was
the correct one?

If it really is low, then there's your problem!

I also did try and back-trace the traces, but I didn't have much luck with that, either.

Yeah it can be tough to figure out what goes where on a multi-layer PCB
like this one. But we know what we are looking for, so I'd suggest
sticking one probe to the AVE connector pin BV10 and then sweep the
board with the other probe - especially the VGA area and any 74x04 parts
you can find.

There is also very little in the way of passives on this board (at least as I know them). There are a handful of electrolytics around the MCA slots, but that's it. Everything else just seems to be surface-mount ICs for the most part. I'm really at the

limit of my component-level troubleshooting skills, though.

The passives are on the solder side, together with some glue logic:

https://ardent-tool.com/8580/Type3_Planar_Photo_Back.jpg

I did take these photos the the VGA chipset and surrounding components. The only things I can readily identify are the chipset itself and the VRAM modules.

https://imgur.com/a/XgWQQoA

I will say I'm not really optimistic about this being something that's repairable, but I've been proven wrong before. I would like to get the on-board graphics working again as I had some AVE cards I was planning to use in this machine.

If it really is the AVE switching logic then it should be very much
repairable - unless they moved that part of the circuit to the ASIC/GA
and the failure is internal to the chip...

I'd suggest the following:

1) Look for any obviously missing or damaged parts on the solder side,
and perhaps measure the ones in the VGA area.

2) Verify the BV10 measurement and if it indeed is low, then switch the
meter to the resistance or continuity mode and check whether the line is shorted to ground (with the unit turned off). If it's not shorted then
try pulling it high with a 1-10K resistor stuck between BV10 and the 5V
rail. If this bodge makes the picture stable then you are probably
missing the ESYNC pull-up (either electrically or physically).

The sync source circuit may be a little bit more complex on the Axx
board, because IBM Japan added the second AVE slot. Though I'd expect it
to work the same as on the more usual systems with just one AVE slot -
pull-up on the planar and either of the two potential AVE adapters can
ground the line to indicate its intentions (of course not at the same
time, but that part would be covered by the software side).

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 28.9.2021 17:52, Louis Ohland wrote:

Unfortunately, the P/N could be changed for tracking manufacturing,
cost, or moon phase.

Yeah, wouldn't be the first time... But in this case there's a good
reason for the P/N difference, because at the very least the used
package is different (silver-cap PGA vs QFP).

The package change is not surprising since the Axx is a later, Japanese, design. And IBM Japan rarely used the silver-cap package, they usually manufactured all chips locally using more modern packaging technology
(except for cases when they reused the original board as-is).

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

404time> This is some fantastic information here. Unfortunately,
404time> I don't believe the attached schematic will be of much help
404time> to me as the P/N for the VGA chipset differs from what
404time> the Axx planarhas (P/N 90X8941), so I'm assuming that
404time> this solution differs from that of the Type I and II planars.

It's a different P/N but the two parts are probably very similar internally, maybe even identical. The package is different than on the Type 1 and 2 boards, so you will have to figure out what is what by checking continuity to the nearby components (if it comes to that...).

Thanks to David we have some hi-res photos of the board: https://ardent-tool.com/8580/Type3_Planar_Photo_Front.jpg
I can see one 74F125 right next to the VGA chip, but without probing
the board I can't tell whether it's related to the AVE switching or not.

Louis>> Unfortunately, the P/N could be changed for tracking manufacturing, Louis>> cost, or moon phase.

Yeah, wouldn't be the first time... But in this case there's a good reason for the P/N difference, because at the very least the used
package is different (silver-cap PGA vs QFP).

The package change is not surprising since the Axx is a later, Japanese, design. And IBM Japan rarely used the silver-cap package, they usually manufactured all chips locally using more modern packaging technology (except for cases when they reused the original board as-is).

Fortunately, we also have the Model 30 286 (and by shared use, the Model 25 286 - both the 'Type 1' and 'Type 2' planars only differ in a few areas) schematic, with a P/N 15F6864:

https://www.ardent-tool.com/docs/pdf/schematics/m30-286_sch.pdf#page=6

The 15F6864 should be pin-compatible with the 90X8941 and the rarer 90X8999, since I do have sample planars (one of each) using those chips. The Model 70 'Type 1' and 'Type 3' also commonly use the 15F6864, with substitutions of the 90X8999 and ever
rarer 92F0616 (I've got a 'Type 3' sample that is the only instance I have seen or heard of this chip P/N).

I'm currently mapping out (between the two schematics) pinout translations between the silver-cap PGA 72X8287 and the four known QFP packages. That's going to take time - and probably a multimeter later - but I'm the one that has the sample planars (and
8-bit adapter). Each package style has a microchannel (with AVE extension) and 8-bit/ISA versions - I can probably do pictures to aid my work.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 29.9.2021 18:59, IBMMuseum wrote:

Fortunately, we also have the Model 30 286 (and by shared use, the Model 25 286 - both the 'Type 1' and 'Type 2' planars only differ in a few areas) schematic, with a P/N 15F6864:

https://www.ardent-tool.com/docs/pdf/schematics/m30-286_sch.pdf#page=6

The 15F6864 should be pin-compatible with the 90X8941 and the rarer 90X8999, since I do have sample planars (one of each) using those chips. The Model 70 'Type 1' and 'Type 3' also commonly use the 15F6864, with substitutions of the 90X8999 and ever

rarer 92F0616 (I've got a 'Type 3' sample that is the only instance I have seen or heard of this chip P/N).

Nice find, David! If the PCBs in question are exactly the same, then
yes, the parts must be pin compatible, at least partially (some chips
may have extra functionality and I/O...).

I'm currently mapping out (between the two schematics) pinout translations between the silver-cap PGA 72X8287 and the four known QFP packages. That's going to take time - and probably a multimeter later - but I'm the one that has the sample planars (

and 8-bit adapter). Each package style has a microchannel (with AVE extension) and 8-bit/ISA versions - I can probably do pictures to aid my work.

Fantastic! Just be aware that the schematics are riddled with errors,
they are not the original IBM design documents and were most likely
created by reverse-engineering. I'd help you with the verification
process, but I'm already behind the schedule with the other updates, so
I should probably finish that stuff first. But let me know if you get
stuck on something...

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Nice find, David! If the PCBs in question are exactly the same,
then yes, the parts must be pin compatible, at least partially
(some chips may have extra functionality and I/O...).

In the samples I have, it seems to be a one-for-one substitution on the QFP chips - Benchmarking and eventual de-lidding might reveal more to why four designs (two of them are fairly uncommon in my sample size) appear to have identical pinouts. A shift
to plastic RAMDACs from ceramic might also have an angle.

Fantastic! Just be aware that the schematics are riddled with
errors, they are not the original IBM design documents and
were most likely created by reverse-engineering. I'd help you
with the verification process, but I'm already behind the
schedule with the other updates, so I should probably finish
that stuff first. But let me know if you get stuck on something...

I've seen that with the schematic for the 8086-based Model 30 - which also transfers nicely to the Model 25 (minus the RTC) since the components are identified exactly the same. We need to apply corrections. Updates to be supplied as I have them.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Tuesday, 28 September 2021 at 18:28:48 UTC+1, Tomas Slavotinek wrote:

On 28.9.2021 17:29, 404time...@gmail.com wrote:

This is some fantastic information here. Unfortunately, I don't believe the attached schematic will be of much help to me as the P/N for the VGA chipset differs from what the Axx planar has (P/N 90X8941), so I'm assuming that this solution differs

from that of the Type I and II planars.

It's a different P/N but the two parts are probably very similar
internally, maybe even identical. The package is different than on the
Type 1 and 2 boards, so you will have to figure out what is what by
checking continuity to the nearby components (if it comes to that...).

Thanks to David we have some hi-res photos of the board:

https://ardent-tool.com/8580/Type3_Planar_Photo_Front.jpg

I can see one 74F125 right next to the VGA chip, but without probing the board I can't tell whether it's related to the AVE switching or not.

I did test pin 14 on the VGA connector and received 0.115v.

Either the sync signal has negative polarity (very much possible, 0.1 -
0.2 V sounds about right for that scenario) or there's a problem. You
would need an oscilloscope to see what exactly is going on there, the DC measurements gives you a very limited view...

Or we could measure the DC voltage on a known-good unit and compare the results. Be aware that the polarity is video mode-dependent, so I'd
suggest to measure it during the POST process as the mode used there
should be the same across all units.

I also tested pin B-V10 on both AVE (with no adapters installed) slots and received 0.002v, so it's definitely low across the board.

Now, that's not right. Are you sure you have made a good contact with
the pin (it requires probes with relatively sharp tips) and that it was
the correct one?

If it really is low, then there's your problem!

I also did try and back-trace the traces, but I didn't have much luck with that, either.

Yeah it can be tough to figure out what goes where on a multi-layer PCB
like this one. But we know what we are looking for, so I'd suggest
sticking one probe to the AVE connector pin BV10 and then sweep the
board with the other probe - especially the VGA area and any 74x04 parts
you can find.

There is also very little in the way of passives on this board (at least as I know them). There are a handful of electrolytics around the MCA slots, but that's it. Everything else just seems to be surface-mount ICs for the most part. I'm really at

the limit of my component-level troubleshooting skills, though.

The passives are on the solder side, together with some glue logic:

https://ardent-tool.com/8580/Type3_Planar_Photo_Back.jpg

I did take these photos the the VGA chipset and surrounding components. The only things I can readily identify are the chipset itself and the VRAM modules.

https://imgur.com/a/XgWQQoA

I will say I'm not really optimistic about this being something that's repairable, but I've been proven wrong before. I would like to get the on-board graphics working again as I had some AVE cards I was planning to use in this machine.

If it really is the AVE switching logic then it should be very much repairable - unless they moved that part of the circuit to the ASIC/GA
and the failure is internal to the chip...

I'd suggest the following:

1) Look for any obviously missing or damaged parts on the solder side,
and perhaps measure the ones in the VGA area.

2) Verify the BV10 measurement and if it indeed is low, then switch the meter to the resistance or continuity mode and check whether the line is shorted to ground (with the unit turned off). If it's not shorted then
try pulling it high with a 1-10K resistor stuck between BV10 and the 5V rail. If this bodge makes the picture stable then you are probably
missing the ESYNC pull-up (either electrically or physically).

The sync source circuit may be a little bit more complex on the Axx
board, because IBM Japan added the second AVE slot. Though I'd expect it
to work the same as on the more usual systems with just one AVE slot - pull-up on the planar and either of the two potential AVE adapters can ground the line to indicate its intentions (of course not at the same
time, but that part would be covered by the software side).

My rough analysis of +ve/-ve VSYNC and HSYNC generation would be the combinatio of chips U6 (74ALS00 Quad 2-input NAND https://www.ti.com/lit/gpn/sn74als00a) U4 (74F32 Quad 2-input OR https://www.ti.com/lit/pdf/sdfs044) and U5 (SN75150 2-input, 2-output
Line Driver https://www.ti.com/lit/gpn/sn75150).
I expect pins 7 & 8 of U5 should connect to either one or the other of the HSYNC and VSYNC output pins.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Tuesday, 28 September 2021 at 18:28:48 UTC+1, Tomas Slavotinek wrote:

On 28.9.2021 17:29, 404time...@gmail.com wrote:

This is some fantastic information here. Unfortunately, I don't believe the attached schematic will be of much help to me as the P/N for the VGA chipset differs from what the Axx planar has (P/N 90X8941), so I'm assuming that this solution differs

from that of the Type I and II planars.

It's a different P/N but the two parts are probably very similar
internally, maybe even identical. The package is different than on the
Type 1 and 2 boards, so you will have to figure out what is what by
checking continuity to the nearby components (if it comes to that...).

Thanks to David we have some hi-res photos of the board:

https://ardent-tool.com/8580/Type3_Planar_Photo_Front.jpg

I can see one 74F125 right next to the VGA chip, but without probing the board I can't tell whether it's related to the AVE switching or not.

I did test pin 14 on the VGA connector and received 0.115v.

Either the sync signal has negative polarity (very much possible, 0.1 -
0.2 V sounds about right for that scenario) or there's a problem. You
would need an oscilloscope to see what exactly is going on there, the DC measurements gives you a very limited view...

Or we could measure the DC voltage on a known-good unit and compare the results. Be aware that the polarity is video mode-dependent, so I'd
suggest to measure it during the POST process as the mode used there
should be the same across all units.

I also tested pin B-V10 on both AVE (with no adapters installed) slots and received 0.002v, so it's definitely low across the board.

Now, that's not right. Are you sure you have made a good contact with
the pin (it requires probes with relatively sharp tips) and that it was
the correct one?

If it really is low, then there's your problem!

I also did try and back-trace the traces, but I didn't have much luck with that, either.

Yeah it can be tough to figure out what goes where on a multi-layer PCB
like this one. But we know what we are looking for, so I'd suggest
sticking one probe to the AVE connector pin BV10 and then sweep the
board with the other probe - especially the VGA area and any 74x04 parts
you can find.

There is also very little in the way of passives on this board (at least as I know them). There are a handful of electrolytics around the MCA slots, but that's it. Everything else just seems to be surface-mount ICs for the most part. I'm really at

the limit of my component-level troubleshooting skills, though.

The passives are on the solder side, together with some glue logic:

https://ardent-tool.com/8580/Type3_Planar_Photo_Back.jpg

I did take these photos the the VGA chipset and surrounding components. The only things I can readily identify are the chipset itself and the VRAM modules.

https://imgur.com/a/XgWQQoA

I will say I'm not really optimistic about this being something that's repairable, but I've been proven wrong before. I would like to get the on-board graphics working again as I had some AVE cards I was planning to use in this machine.

If it really is the AVE switching logic then it should be very much repairable - unless they moved that part of the circuit to the ASIC/GA
and the failure is internal to the chip...

I'd suggest the following:

1) Look for any obviously missing or damaged parts on the solder side,
and perhaps measure the ones in the VGA area.

2) Verify the BV10 measurement and if it indeed is low, then switch the meter to the resistance or continuity mode and check whether the line is shorted to ground (with the unit turned off). If it's not shorted then
try pulling it high with a 1-10K resistor stuck between BV10 and the 5V rail. If this bodge makes the picture stable then you are probably
missing the ESYNC pull-up (either electrically or physically).

The sync source circuit may be a little bit more complex on the Axx
board, because IBM Japan added the second AVE slot. Though I'd expect it
to work the same as on the more usual systems with just one AVE slot - pull-up on the planar and either of the two potential AVE adapters can ground the line to indicate its intentions (of course not at the same
time, but that part would be covered by the software side).

My rough analysis of +ve/-ve VSYNC and HSYNC generation would be the combinatio of chips U6 (74ALS00 Quad 2-input NAND https://www.ti.com/lit/gpn/sn74als00a) U5 (74F32 Quad 2-input OR https://www.ti.com/lit/pdf/sdfs044) and U4 (SN75150 2-input, 2-output
Inverting Line Driver https://www.ti.com/lit/gpn/sn75150).
I expect pins 7 & 8 of U4 should connect to either one or the other of the HSYNC and VSYNC output pins.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)