exxos's Atari, Amiga & retro forum

Ok, the board is arrived :D
Thank you to Atarian Computing !!!

Oldskool wrote: 27 Aug 2024 19:02 Emutos buildin is also not ok. Also not with the previous fw.
Different oscillators have no effect.
Normal et4k is OK.
W32I is the same model you have

I tried to get my machine into some kind of state to reproduce what you were seeing but so far no luck with that.
It sounds like you already tried but I was going to suggest testing if the 40mhz oscillator makes a difference.

From the screenshot it doesn't look like it's missing the RDY signal at least, then you'd see larger uniform blocks being wrong like with the previous firmware (blit-errors). This looks like non-blit read or write glitches. Perhaps 68150 bus contention, or too fast back-to-back access to it.
Does it make a difference if you unplug the CF card or adapter?


I tried some stuff here:
(going to 40mhz has issues in itself and while it does work fine with the Nova driver, the EmuTOS built-in one will show artifacts - some kind of official and fully tested "slow" mode would be nice to implement at some point; probably at 32mhz bus speed for when it's tricky to get hold of 68150FN40 and/or Rev6 CPU)


These are the two I got from LittleDiode in the UK.
The 40mhz one works perfectly at 48mhz. The 33mhz one 'appears' to work but I did see some pixel artifacts rarely when using the EmuTOS driver. With the Nova driver I have not had any issues.
Saying that, I am for the most part using the 40mhz chip since I have it.

* MC68150FN40 - ZKAC9921 : 48mhz : Nova good, EmuTOS good
* MC68150FN40 - ZKAC9921 : 40mhz : Nova good, EmuTOS as expected.

* MC68150FN33 - ZKAA9738 : 48mhz : Nova good, EmuTOS some pixel artifact very rarely
* MC68150FN33 - ZKAA9738 : 40mhz : Nova good, EmuTOS as expected.

The 33mhz one from Aliexpress you kindly sent me exhibited the behaviour you reported.
The chip had some very dodgy looking pins that I tried to straighten/reshape before using it but I don't know if this made any kind of difference.
It really doesn't work well at all with the CF card @ 48mhz. Though it does go to the desktop and works fine when I unplug it.

* MC68150FN33 - ZKAA0401 : 48mhz : (without CF card) EmuTOS good.
* MC68150FN33 - ZKAA0401 : 40mhz : (with CF card) Nova good, EmuTOS as expected.

I ordered a very cheap MC68150FN40, plus an even cheaper MC68150FN33 from Aliexpress, will be interesting to test those when they arrive.
I'll have a think and see if I can send you some stab-in-the-dark test firmwares to try and pinpoint why you are getting the results that you are.
It's slightly puzzling that you're seeing the issue only with the W32i and not ET4k as well :)

Hi Anders,

Will try it in more detail.
I did some trials yesterday. (Et4k is always perfect. Tried two of them)

40 = Aliexeprss
MC68150FN40 : 48mhz : Nova good at boot, No good when using desktop, EmuTOS at boot already glitches
MC68150FN40 : 40mhz : Nova good at boot, No good when using desktop, EmuTOS at boot already glitches
MC68150FN40 : 50mhz : Nova good at boot, No good when using desktop, EmuTOS at boot already glitches

No ide = same results


33 =Littlediode

MC68150FN33 : 48mhz : Nova good at boot, No good when using desktop, EmuTOS at boot already glitches
MC68150FN33 : 40mhz : Nova good at boot, No good when using desktop, EmuTOS at boot already glitches


No ide = same results

Can try 33 or 28 as well..

Oldskool wrote: 29 Aug 2024 11:03 MC68150FN40 : 48mhz : Nova good at boot, No good when using desktop, EmuTOS at boot already glitches
MC68150FN33 : 48mhz : Nova good at boot, No good when using desktop, EmuTOS at boot already glitches

Is it no good in desktop even in Nova 2bpp mode? What if you copy emulator.bib -> sta_vdi.bib and pick the boot mode for the desktop? (don't know if that even works, have not tried)
Basically trying to narrow down what, if anything, is interfering with something and if the bootup mode is working during boot but the exact same mode is not during desktop.

I have some old ceramic MFP's somewhere and I can try to swap those in, in case there is some variation and perhaps holding onto the bus longer than expected. They would be active quite a lot during desktop. Still, I'd expect that to affect the ET4K too if that was an issue but it doesn't hurt knowing for sure.

I'm assuming you get the same result no matter which version you pick of these?
https://silicon-heaven.org/atari/nova/TT030/ET4000W32i/

Rom is latest from Github etc?
(Narnia and Iksi fw hasn't been changed since forever I believe)

My sta_vdi.inf only has the screensaver turned off and all of the other override options commented out.
It didn't make any difference for me changing the 0WS and IRQ jumpers on the w32i card (firmware is currently not acting on the ZeroWaitstate signal anyway).

Trying a slower 22LV10CQZ-30PU for Narnia did not provoke any errors. Might be worth trying the same with Iksi too. I'm not really caring how, I just want to be able to somehow get my machine to exhibit the issue you are seeing so I can understand why and fix if it's something I'm doing wrong somewhere :)

Don't think it matters in this scenario but technically you are supposed to program the 22V10C's as device "ATF22V10C (UES)" for GAL mode.
I've been running them wrongly programmed in PAL mode "ATF22V10C" up until just now.. so I guess the missing and unprogrammed fuses where not being used anyway.

Wonder if I somehow messed up and sent you the wrong firmware file. Long shot, but I can resend one this evening.

2bpp nova is not OK and maybe even worse.
1024 x768 256 looks better and has less issues (but it still has problems) vs 2bpp 320 or 640 when comparing. Although the menu at start looks ok
Crystal has no impact tested: 28, 33, 40, 48, 50
Driver version does not matter.
Nvdi installed has no effect.
Jumpers on the graphics card have no effect.
Switching around the mfp has no effect.
Contacts have been cleaned, isa position no effect..
Also tested two different cf cards.

Edit
Doom is updated correctly in a window or fullscreen.
Scrolling text is also looking pretty ok Moving windows around and the menu especially in Emutos is glitching (mint environment is maybe a bit better).

I've been wanting to get around verifying the realtime clock for quite some time now.. and it works :)
Next up, put support routines in bios and have TOS use the clock.
A little bit of the ram will be reserved for the bios and system settings but the rest could be made available as nvram for those bigger TOS's which stores language settings etc.

From uninited and stopped to inited: Power cut and restarted, still good and clock has ticked with the battery inserted:

Oldskool wrote: 29 Aug 2024 20:15 2bpp nova is not OK and maybe even worse.
...

Thank you for those detailed tests!
I'll have to hold off revisiting this on my side until the weekend or so, else it risks making the computer feel like work and I know for sure I'll just end up resenting to use/work on it.

I'm thinking a few different "what-if" firmwares to see if either of them hits the mark and can give a clue on what the problem may be.

Yes sure. It's just something to play with nothing serious :)

Oldskool wrote: 29 Aug 2024 21:58 Yes sure. It's just something to play with nothing serious :)

Yeah, don't get me wrong I really love the fact that there's more of these up and running and it is bugging me the W32i is not working on yours.
It comes down to me dealing with bug reports and reproduce-plus-fix scenarios all the time at work so I do need to be very smart about how much support for others I can actually provide during not-work without risking "burning out" on whatever the hobby project may be.

Another wavetable option now with MT32 as welll: