STE V1 32Mhz booster - hoping to get some ideas on this..

[oraclejedi]

If I am reading the marking right, they are stamped A222J, so yes that would be 2.2K right?

[rubber_jonnie]

If I am reading the marking right, they are stamped A222J, so yes that would be 2.2K right?

That's correct as I understand, but if you want to be sure you can measure between pin 1 and any other pin with a multimeter.

[Steve]

If I am reading the marking right, they are stamped A222J, so yes that would be 2.2K right?

I just checked through the instructions for this version of the booster and I think you're supposed to leave the motherboard bus resistors as-is.

exxos page: https://www.exxosforum.co.uk/atari/last/V1STE/index.htm
user guides:
https://www.exxosforum.co.uk/atari/last/ ... /index.htm
https://www.exxosforum.co.uk/atari/last/ ... /index.htm

There is no mention of changing/removing bus resistors in the instructions here.

Have you checked your u405 chip to see if it is 74S257 or 74F257.

[Steve]

I also had the 'increasing wobble with time' issue on a USA STe a while ago. I did fix it by playing with the routing of the u405 cable. Please see the image below for the routing that worked for me. Note: The cable is lifted about 1cm from the motherboard surface.

Make sure the wire is single-core.

32mhz line.jpg

[oraclejedi]

Thanks for the update Steve.

It was those original two install guides I used when I first installed this, probably in 2017 or 2018. And yes, they did not call for removing the SIL resistors from the bus.

I was following the mandatory upgrades linked by @rubber_jonnie where it ultimately links to https://www.exxosforum.co.uk/forum/viewt ... 514#p50402.

If I understanding that right, we are now expected to remove them.

I did remove them. P101 thru P105. I left P100 in place as shown in the pic. I am not clear if that was supposed to come out or not.

Anyway, after repairing one trace that tore as one of the SILs was removed, the system still boots.

I get about 10 minutes of stable operation and then the wobbling starts. Slowly at first and then more intrusively until about 20 mins later the vertical sync basically goes and the picture starts rolling. This however is overall an improvement on where things stood when I opened the thread, so there's progress at least.

It is a little bit challenging to measure progress and sometimes a modification can be installed and tested before the unit cools down. If I see an improvement, is it due to the change made, or the fact that maybe the unit was off for an hour while I made it?

Its interesting you had the same (or similar) problem. I have routed the 32Mhz line to the left of the shifter chip and then to the accelerator, but felt I got better results going to the right of the shift and along side the RAM making the line shorter. I also experimented with routing across the foil side of the PCB, but that was even worse.

In all cases however I have been running the line directly over the motherboard surface. It seems you have managed to suspend the line using hot glue on top of some of the ICs. I may have to try this approach and see.

Anyway thanks to you for the insights and for all the ideas suggested in this thread, they are greatly appreciated.

[Steve]

Yes and don't forget to check you have the correct type of u405 chip and using single core cable.

[rubber_jonnie]

@oraclejedi @Steve

Just to be clear, the advice to remove the SILs if the booster was fitted with 2.2K SILs came out of a conversation I had offline with @exxos, so please don't just consider it my advice, but based on a discussion held with the designer of the booster.

If I'm honest. this seems like a thermal issue since it takes time for the problem to show itself. If it were due to the routing of the 32Mhz clock line, I'm confident it would be immediately apparent.

Have you checked the video out socket to see if there are any bad solder joints?

Have you checked the ICs for temperature once the wobble starts?

Also, is it possible to determine if the desktop is still present when the wobbling is at its worst? If so then that may indicate a problem with the video circuit that may have been brought about by handling the board during the upgrade, they can be quite fragile and need some care.

BTW, if you remove the 32Mhz wire from U405 does the problem stop?

[oraclejedi]

The thermal suggestion makes sense, and I have done a finger-test of the major IC's which hasn't revealed anything conclusive yet. I will make a visual inspection of the solder joints when I have the unit in pieces next.

At one point the accelerator was socketed. And if I removed the accelerator and put back the original CPU, all the issues went away. Of course that was probably 3 years ago so I need to be careful of being over confident that that observation rules anything in or out.

Whatever software is running - GEM, Frontier etc, appears to continue to be running as the issues gets worse, so the core OS appears to still be running.

Interesting idea to remove the 32Mhz wire. I figured that would break everything. I will have to try that idea and observe the result.

I will try to spend some time this weekend. I will visually inspect the solder joints of the video connector, and reroute the 32Mhz wire again.

Thanks again for the help guys.

[rubber_jonnie]

The thermal suggestion makes sense, and I have done a finger-test of the major IC's which hasn't revealed anything conclusive yet. I will make a visual inspection of the solder joints when I have the unit in pieces next.

IIRC the shifter is in a socket, so it may have, ahem, shifted in the socket, so a reseat/clean of the IC legs and socket plus a tweak of the socket fingers might help here.

At one point the accelerator was socketed. And if I removed the accelerator and put back the original CPU, all the issues went away. Of course that was probably 3 years ago so I need to be careful of being over confident that that observation rules anything in or out.

Have you tried removing the 68HC000 and dropping your old CPU into the booster, should still work, just only at 8Mhz.

Whatever software is running - GEM, Frontier etc, appears to continue to be running as the issues gets worse, so the core OS appears to still be running.

So suggestive of a problem in the video circuit if the OS continues to run fine. Have you recapped the video circuit?

Interesting idea to remove the 32Mhz wire. I figured that would break everything. I will have to try that idea and observe the result.

Well, it's a nice quick test. As long as you ensure the booster is not in 32Mhz mode it should be just fine.

I will try to spend some time this weekend. I will visually inspect the solder joints of the video connector, and reroute the 32Mhz wire again.

If I'm honest I think the Shifter or something in the video circuit is the problem.

Thanks again for the help guys.

No problem.