Couple 1040STf issues
Posted: Sat Oct 17, 2020 7:41 pm
Hey all! Had a couple recent issues with my 1040STf that I wanted to see if anyone's encountered or got some thoughts on. The reason I bring this up now while it all works pretty-well is that I've got a TF536 here and the last bits I need to get it up and going are on the way, and I'd like to make sure the system itself is in as good of a state as possible.
Current state is that I've got a Exxos PSU, Exxos 4MB upgrade, TOS 1.04, and Blitter added to the system. All electrolytic caps are brand new as well. I've got an internal HxC floppy emulator with current firmware, and an external UltraSatan that I assume is also current since the last firmware seems to be from 2014. At this point, I've also gone through the various fixes at https://www.exxosforum.co.uk/forum/viewtopic.php?t=1585 - a couple of which had some interesting results. I'm also using the latest (paid) HDDRIVER.
It's worth mentioning that I've also got a full electronics bench at home including a 100+ channel logic analyzer, so if there are any specific signals that would be useful to capture or anything, I'm fully capable of that - I haven't gone down that rabbit hole on my own yet.
First thing - 1MB mode doesn't seem to work?
I can run YAART all day long on the 4MB module and it's rock solid. When I disconnect the jumper to switch to 1MB mode, I get a black screen and the system will not boot. I thought this was interesting - maybe a symptom of something else - but it hasn't seemed to affect anything otherwise.
Second thing - Occasional weirdness with UltraSatan and HxC
A weird glitch I encounter sometimes that led me down a path of trying things was that sometimes when trying to copy *folders* from the HxC to the UltraSatan, the copy will begin, the GEM window will show that it tries to copy a file named "Aa" or "AA" (which doesn't exist), and then I'll end up with an empty folder on the UltraSatan but TOS acts like the process has completed. What's fun to note here is that this is only when I try to drag a folder over. If I create a folder on the UltraSatan, then go into the folder on the HxC and select all of the files and drag them over to the folder I created, this issue never happens.
Based on other people's file copy issues, I guessed it was something to do with the weak DMA bus and added the various pull-ups and swapped bus resistors as described in the list at https://www.exxosforum.co.uk/forum/viewtopic.php?t=1585 - it didn't seem to cause any problems, but it also didn't seem to change anything.
Third thing - 68k HC processors in an STf - bad idea?
So, going down this path, I finally went ahead and swapped out the stock M68K for a newer HC variant. It didn't seem to help the previous file copy issue, but I did see two interesting things. The first is that they really didn't seem to like the lower-value bus resistors, and I encountered all sorts of stability issues until I swapped back to the stock values. Other pull-ups didn't seem to affect anything one way or the other. This seemed to stabilize everything except for one case - the system would freeze about 50% of the time during VDI text operations in GEMBench. No bombs or anything. I tried a second HC variant and it showed the same behavior, so I ended up just swaping the stock M68K back in and reinstalling the new value bus resistors and haven't seen any significant issues. I just thought the behavior here was interesting and could be a symptom of something else.
Anyone have any thoughts? Any documentation I should've read more thoroughly? Cheers!
Current state is that I've got a Exxos PSU, Exxos 4MB upgrade, TOS 1.04, and Blitter added to the system. All electrolytic caps are brand new as well. I've got an internal HxC floppy emulator with current firmware, and an external UltraSatan that I assume is also current since the last firmware seems to be from 2014. At this point, I've also gone through the various fixes at https://www.exxosforum.co.uk/forum/viewtopic.php?t=1585 - a couple of which had some interesting results. I'm also using the latest (paid) HDDRIVER.
It's worth mentioning that I've also got a full electronics bench at home including a 100+ channel logic analyzer, so if there are any specific signals that would be useful to capture or anything, I'm fully capable of that - I haven't gone down that rabbit hole on my own yet.
First thing - 1MB mode doesn't seem to work?
I can run YAART all day long on the 4MB module and it's rock solid. When I disconnect the jumper to switch to 1MB mode, I get a black screen and the system will not boot. I thought this was interesting - maybe a symptom of something else - but it hasn't seemed to affect anything otherwise.
Second thing - Occasional weirdness with UltraSatan and HxC
A weird glitch I encounter sometimes that led me down a path of trying things was that sometimes when trying to copy *folders* from the HxC to the UltraSatan, the copy will begin, the GEM window will show that it tries to copy a file named "Aa" or "AA" (which doesn't exist), and then I'll end up with an empty folder on the UltraSatan but TOS acts like the process has completed. What's fun to note here is that this is only when I try to drag a folder over. If I create a folder on the UltraSatan, then go into the folder on the HxC and select all of the files and drag them over to the folder I created, this issue never happens.
Based on other people's file copy issues, I guessed it was something to do with the weak DMA bus and added the various pull-ups and swapped bus resistors as described in the list at https://www.exxosforum.co.uk/forum/viewtopic.php?t=1585 - it didn't seem to cause any problems, but it also didn't seem to change anything.
Third thing - 68k HC processors in an STf - bad idea?
So, going down this path, I finally went ahead and swapped out the stock M68K for a newer HC variant. It didn't seem to help the previous file copy issue, but I did see two interesting things. The first is that they really didn't seem to like the lower-value bus resistors, and I encountered all sorts of stability issues until I swapped back to the stock values. Other pull-ups didn't seem to affect anything one way or the other. This seemed to stabilize everything except for one case - the system would freeze about 50% of the time during VDI text operations in GEMBench. No bombs or anything. I tried a second HC variant and it showed the same behavior, so I ended up just swaping the stock M68K back in and reinstalling the new value bus resistors and haven't seen any significant issues. I just thought the behavior here was interesting and could be a symptom of something else.
Anyone have any thoughts? Any documentation I should've read more thoroughly? Cheers!