exxos's Atari, Amiga & retro forum

Just trying @Steve's board with current13 firmware. Currently using slow slew.

TTram test is powert up test with patched 206. Also TOSCOPY used (as it copies and verifies ROM)

FROM COLD POWER UP - TTram tests fine.

FREEZER SPRAY on SDRAM & PLD - TTram tests fine.

HEATED UP (Gas iron) - TTram tests fine.

@Steve Yours has a bodge wire across the SDRAM clock line now. I was testing to see if the length was a issue or not (seems not so much) .

I think was been puzzling me probably the past three weeks is a thermal fault which I have simply not known about :roll: like when the board has warmed up I could literally leave it running for days and it would not trip up even once. Leave it alone for a couple of days, and it was unstable again.. I would then mess around with the code and then think I have fixed it when in fact I haven't, and the board has simply warmed up and started working again.. I think this is why I have been constantly stuck going around in circles trying to figure out fault which was constantly "moving".

I think the code changes also likely shifted the logic around just that tiny bit that it would shift the thermal profile in the IC adding another layer of problems.

I only caught onto this issue on Friday because it was hot in my room and everything was working fine until I put my big fan on in the window.. The temperature in my room dropped about half a degree, and then the board started being totally unstable even though nothing had changed over those few minutes..

It also seems that slow slew indeed required. Originally I think TF thought it was a issue with the SDRAM clock. I have serious resistors on the current board but it does not seem to have any effect. So it is currently a bit of a mystery (one I don't like) as to why fast slew causes problems.
When I tested slow and fast, there was virtually no difference between them..

I think I might know what is happening with the "PCB hot spot"..



The track where the yellow arrow is, is where the majority of the heat seems to be coming from. Either side of that track and the temperature dropped sharply by like 10c ! It does feel a little warmer there as well. But the thing is, that corner of the IC is the hottest part as well. Its like 10c hotter than the other corners of the chip.

So I think the huge copper area being heated up underneath the chip, in that particular place, its all adding up.. Just a case of "radiated heat" heating up that track on the PCB.

I think I may suggest fitting a heatsink to the PLD. Is not strictly required, but it will help alleviate the hotspot on that corner of the chip. I will also see if I can do something about the top layer copper fill and give it a bit more area to dissipate some heat that way..

Is the filled area ground? Do those bright green vias take it to a bigger groundplane inside or on the back or is it alone there? It seems to make a loop under the IC and back out again but I think this is irrelevant if it's tied to another plane on the back.

BTW you don't need to go for a fancy thermal camera - I have an old Android one (SeekThermal) which is pretty low res but for a few years now its been possible to plug these into a PC on a USB extension cable and use 3rdparty/opensource software to capture and clean up the live image, much better and lower noise than the original Android apps were for the same device. I'm sure there was thing about it on EEVBlog at the time (maybe 10 years ago!)

I actually wrote my own for MacOS a few years ago using LibUSB which did the same thing because I hated the Android app so much it drove me to do my own - but I definitely saw some decent looking Windows stuff for this which supports a lot of different cameras and I would probably just try that now.

@dml The vias under the PLD connect it to GND, but it's not a thermal path.

I redid a lot of it last night.

It just gives that heat source a little bit of somewhere to go..

Been doing some tweaks to the next revision board. Nothing in particularly critical. But it seems the jellybean LM1117 doesn't have great transient response times. There are some people moaning on Reddit about it in fact.

So I have swapped it out for a "more modern" TI part. It has twice the current capability and should have a lot better transient reaction times. Mostly it's just a experiment to see if I can figure out why fast slew does not work..

exxos wrote: 31 May 2025 13:23

coonsgm wrote: 31 May 2025 12:40 Working like a charm! I'll do some more thorough testing. Unfortunately, the new STE I have coming won't be here until early next week (darn shipping delays).

:bravo:

Hopefully @Badwolf will have some better news next...

Sorry. Mad weekend. I had only ten minutes to flash and try CURRENT13 a couple of nights ago and I'm afraid it's bad news with hangs (interrupt LED flicker stops) fairly frequently. Sometimes can get to desktop by bypassing mass media but sometimes the hang is really early in the boot process.

Felt connected to floppy/ASCI access so I think often the first boot sector seek was the trigger. I initially suspected IDE so removed that and added ASCI. Made things worse. At one point I ended up on a desktop where even the floppies hadn't been recognised.

I've had no time to pursue further, I'm afraid, but I can't offer any good cheer just yet.

BW

@Badwolf

Might be best to send me your board to try here. I've got 4 boards tested on 3 motherboards and no issues now.

Though I only used floppy and IDE so far. ACSI I haven't had chance to test fully yet. But not seen any problems there so far.

I was able to get a fair amount of testing in.

I was able to connect and boot from both IDE and ACSI and copy large amounts of data to and from my Ultasatan to the IDE drive. This was done under both TOSEC and EmuTOS.

STDOTT boots under EmuTOS but not TOS. Software demos and games I can typically run with accelerators run fine: Ultima's, Civilization, Frontier, Dungeon Master all run from the IDE.

coonsgm wrote: 03 Jun 2025 11:46 I was able to connect and boot from both IDE and ACSI and copy large amounts of data to and from my Ultasatan to the IDE drive. This was done under both TOSEC and EmuTOS.

Great! Assume with current13 firmware ?

STDOTT boots under EmuTOS but not TOS. Software demos and games I can typically run with accelerators run fine: Ultima's, Civilization, Frontier, Dungeon Master all run from the IDE.

Yeah, I don't think DOTT runs in TOS for some reason. Maybe someone else can verify this on a stock machine.

exxos wrote: 03 Jun 2025 11:50 Yeah, I don't think DOTT runs in TOS for some reason. Maybe someone else can verify this on a stock machine.

It's modern software custom written by Anders with exactly ST536s in mind. It works in all configurations. If it's not working: something's wrong.

BW