OJ's H5 A3 Troubleshooting

[olivier.jan]

Hi everyone,
As I wanted to get my A3 running before the B1 starts popping-up, I rushed the build this week and ended with a beautiful...black screen.
After checking a bit, I realised that IC6 was not properly re-soldered which I fixed and got a beautiful.. white screen.
Some more checks later, I got proof that beer and soldering together are not a good idea cause you end up with a CPU socket at 90º
Once the socket was re-soldered in the right position I finally got a ... white screen
Strangely, while display the white screen, my monitor tells me it's NTSC, not the usual PAL I get with my STF or STE...

I tested voltage everywhere and for shorts on all extensions slots pins, clock frequencies on all the ICs, and everything looks OK so far.
Latest troubleshooting today shows there's activity but it execute a few instruction, reset, execute instructions, reset, etc...

I've tested CPU, MMU and GLUE in another machine and all are good, so maybe the ROM as I have no way to test it (AT27C4096-55, with 1.04 and EMUTOS, both gives me a blank screen).

I ordered a new ROM and drag cart from Exxos today in order to do more troubleshooting (hopefully) next week.
In the meantime, if anyone has an idea for further troubleshooting, here are a picture of the board and 2 dumps of the first cycles after reset...

IMG_5781.jpg (572.34 KiB) Viewed 437 times

Capture d’écran 2020-11-21 à 16.17.33.png (136.58 KiB) Viewed 437 times

Capture d’écran 2020-11-21 à 16.23.05.png (111.3 KiB) Viewed 437 times

[stephen_usher]

It's unlikely to be the ROM. Hopefully the Diag cart will help, especially with a terminal on the serial port.

The display will be "NTSC" until the ROM has managed to change the setting during the initialisation process.

It looks like it's reading the first few bytes of the ROM and then crashing. This could be not reading the ROM properly or not accessing RAM. Seeing as you aren't seeing what's on the other 8 data lines it's hard to see if any of those aren't working.

I've not verified the contents of those first few bytes of ROM, which should include the address of the base of the ROM.

You'd better check that all the address lines are busy.

[PhilC]

Are you saying that the H5 is resetting itself or are you resetting it? Recoding /halt is a better guide than /reset by the way.

One thing to check is to remove the CPU and check that the address and data lines are pulled up to 5v, same for R/W, BGACK, BERR etc. Its also worth checking this on the DMA and other chips to eliminate bad soldered joints.

Good luck.

[Icky]

@olivier.jan - Looking at your board JP13 is missing. You need to make sure that one is closed. This is the ROM_CE line so would explain why its not seeing the ROM

[olivier.jan]

Thanks for your replies.

Concerning JP13, it was actually plugged during my tests and got off when I removed the analyzer cables to take the picture.
I also checked all address pins and all data pins to try to spot the issue, but they’re all connected and have activity.
Tomorrow I will run some more tests, checking all the pull-ups as advised and again the data pins bit with UDS and LDS and a few other signals too.
I was reviewing the bytes captured and the sequence match the first few bytes of the rom, whic is strange as it should match only one out of two as I m only recording d0-d8...

I will also solder the cart port and serial, it will make using the diag cart a bit easier.

[exxos]

Have a look here as well viewtopic.php?f=78&t=2362

[olivier.jan]

Some progress, but not working yet.
After analyzing the bus traffic and comparing it to a proper ROM image, I confirmed the ROM was bad with two main issues.
This ROM add a TOS 1.4 FR and EmuTOS 1 (192k) on it but...both image were byte swapped. I also discovered that the TOS 1.4 FR image was corrupted while trying to burn it for an original STF.

After receiving some parts from Exxos today, I plugged a new ROM and now got garbage on the screen, which was actually better than the white screen. Newly received Diag Cart was also producing garbage, just a different kind.

I then realised that for fixe #1, after fighting with IC6 and cutting the trace underneath, I had simply forgotten to cut the bottom layer trace.
Now the diag cart starts...and stop with a very nice 'T0 MFP timer' error. So going to troubleshoot around the MFP but at least video is working.

IMG_5792.jpeg (46.4 KiB) Viewed 303 times

[PhilC]

@olivier.jan check the soldering on your MFP and its oscillator, hopefully that will let it boot further

[olivier.jan]

So, some progress here.
After checking the MFP, MFPINT line was not soldered properly, so I un soldered the whole MFP (very stupid decision) and resolder everything manually.
3 attempts later, I was still full of shorts and after removing it again, I decided to stop before damaging the motherboard and ordered a proper socket.
Socket arrived this week, soldered it and it booted perfectly with a new MFP.
Tried the old MFP (not in good shape after so much solder/unsolder) and it works.
10 minutes later after a reboot, white screen, diag cart not working,etc...

This morning, I took the logic analyzer and discovered A2 was stuck high, checked everything and there seems to be a problem with the cpu socket this time. The bin n the bottom of the MB is properl connected, but on the socket itself I have a very strange behavior where it comes and goes all the time.
Anyway, for the time being I ended up using the DIP 68000 I salvaged from the donor machine and it boots properly again.
Tests are fine, boot to the desktop too.

Now, a few more sues to tackle, floppy seems to have issues with Double sided disks, and DMA is not working properly but I’ll check both.

Still, even if there is more work to be done, I was very happy to see Bubble Bobble booting from a single sided floppy.
Thanks for all the help and advices !

[exxos]

Floppy issues , I assume by what you say you not done the fixes for it ?

Also , I don't know what sockets you used, but I only recommend 3M sockets because of all sorts of screwy socket issues cheaper sockets..