H5C1- firST STeps

[JezC]

Looking back you posted this..

But SIDE 8 ?!

Ah, just found the photo of the screen on my phone...should be 0 rather than 8 - sorry!

[exxos]

Ah, just found the photo of the screen on my phone...should be 0 rather than 8 - sorry!

oh.. Going to say I have never seen that problem before Thought you copied and pasted it via a terminal on a PC or something not actually typed it out..

What does TOS do when trying to access the floppy ?

I don't really get how it could come bus error accessing the floppy.. Address on the bus must be bad for some reason. Its trying to access the floppy registers, but ending up on a invalid address somewhere in the memory map which is invalid causing GLUE to issue BERR.

I think your floppy problem is more of a symptom than a actual fault...

Have you tried running GB6 from the diagnostic cart ? It would help test the machine out to make sure the bus is stable.

[JezC]

What does TOS do when trying to access the floppy ?

It was over 2 years ago that I last tried it but IIRC it bombs (can't remember how many, I may have slept since then ) when I try to access the floppy but is fine with no floppy in the drive.
I hope to get it set up and repeat the test later this weekend (after I drop my son off to start Uni)...

I don't really get how it could come bus error accessing the floppy.. Address on the bus must be bad for some reason. Its trying to access the floppy registers, but ending up on a invalid address somewhere in the memory map which is invalid causing GLUE to issue BERR.

I think your floppy problem is more of a symptom than a actual fault...

Have you tried running GB6 from the diagnostic cart ? It would help test the machine out to make sure the bus is stable.

I will run GB6 and see if it throws up any other errors (didn't even think about running it from the diag cart before... ) over the weekend or early next week, then report back.

[JezC]

Ok, just cleared the 'operating table' sufficiently to put the H5C1 on it & connect things up to test a bit more...

With the modified TOS 1.04 (shows RAM - 4M - & number of floppies - 1 - at boot) & no floppy in the drive, it boots to the desktop OK. If I then insert a floppy & open drive A I get 4 bombs & the machine is locked (reset or power cycle to recover it I think - not tried the keyboard reboot when in this state though).

During the boot process, I can hear the drive being accessed (no floppy) & the floppy Led comes on. When at the desktop I can insert the floppy in the drive & the floppy Led then goes out - still no crashes at that point.

I will try to dig out he diag cart later (I remember that locked up too) and also set up the logic analyser on the working H4 to capture the activity when that accesses the drive to open the drive window, then repeat with the H5 & compare the two.

I'm expecting the chip select to work but maybe an address or data line is faulty.
I can't test DMA (yet) as there is no connector on the ACSI port...unless I can connect that nifty new SC HDD from the @exxos store to the internal Mega ST ACSI header...?

[exxos]

With the modified TOS 1.04 (shows RAM - 4M - & number of floppies - 1 - at boot) & no floppy in the drive, it boots to the desktop OK. If I then insert a floppy & open drive A I get 4 bombs & the machine is locked (reset or power cycle to recover it I think - not tried the keyboard reboot when in this state though).

I think the floppy drive is a symptom at this point. There is some sort of memory corruption going on either with ROM or RAM.

I don't know if you have tried swapping between the PLCC and DIP CPU ?

I'll definitely continuity check all the CPU lines out to a different socket. You could have issues with LDS,UDS for example. Indeed continuity check all the MMU & GLUE lines back to the 68K socket.

I can't test DMA (yet) as there is no connector on the ACSI port...unless I can connect that nifty new SC HDD from the @exxos store to the internal Mega ST ACSI header...?

I don't think anyone has even tested that header yet.. I wouldn't know how to hook it up off-hand. Easy just to solder the DMA connector on though I don't think you are going to have any better luck with the DMA port over the floppy anyway.

[JezC]

@exxos Thanks - I did run the memory test from the diag cart & no errors there but can check it again just to be sure nothing else has gone wrong.

I'm pretty sure I swapped all the chips between this H5 & my working H4 (at least once) so it's most likely either a socket or PCB fault.

I haven't soldered in any DIP sockets for the CPU yet so I'm stuck with the PLCC CPU for now (but probably a relatively easy task & it will have something in a 68k socket - either an accelerator or an IDE interface...once I have the basic operation working).

Ok, so I'll re-run the memory test & then check the connections for the CPU as the first steps & report back (hopefully before the weekend).

On a happier note, I've been using the Exxos PSU in this H5 (well, it avoided pulling a PSU out of another working system...) and so far it's been running much quieter than before

[JezC]

Right, so I found the diag cart & ran the RAM test - 4M found, no errors reported.

I tested the ROM as well but it reports bad CRCs on 4 of the 6 ROMs (but given that it's a patched TOS1.04 I'd be more surprised if it worked...

I then tried to set up the diag cart for YAART ST...computer boots to the desktop but I didn't see anything on the cart when I opened it (select a drive, install a new drive & assign it as 'c', open c but it's empty) so I'm probably doing something wrong there...

Don't think I'll have time to solder in a DIP CPU socket before the weekend

[exxos]

I then tried to set up the diag cart for YAART ST...computer boots to the desktop but I didn't see anything on the cart when I opened it (select a drive, install a new drive & assign it as 'c', open c but it's empty) so I'm probably doing something wrong there...

As long as you install lowercase 'c' for the cartridge it should show files there..

I can only think it's falling on 8bit stuff. Which is why I wonder if UDS LDS is broken between the CPU and MMU & GLUE or something. You can still beep out the PLCC CPU to a 68K socket without having to install the DIP socket.

[JezC]

As long as you install lowercase 'c' for the cartridge it should show files there..

Yes, I tried that but may have made a mistake...will try again tomorrow (and on a working system to rule out the H5 board).

I can only think it's falling on 8bit stuff. Which is why I wonder if UDS LDS is broken between the CPU and MMU & GLUE or something. You can still beep out the PLCC CPU to a 68K socket without having to install the DIP socket.

Well, once I have it working on another system I'll be able to confirm where the issue lies (me or the H5...or both!)

On a totally different topic, I've just managed to get my Greaseweazle v4 set up on my Win 10 box...and it seems to have read an Atari ST floppy ok...I think!

[JezC]

I can only think it's falling on 8bit stuff. Which is why I wonder if UDS LDS is broken between the CPU and MMU & GLUE or something. You can still beep out the PLCC CPU to a 68K socket without having to install the DIP socket.

Ok, so I finally managed to check the connections between the PLCC and one of the DIP sockets... the 8MHz clock was about 67 ohms (which makes sense given each has a 33 ohm series termination resistor ) but the rest were all under 2 ohms (most under 1 ohm) so they seem to be connected at the CPU end...
Some pins did need a fair bit of pressure to show the connection, so a removal & re-seat of the PLCC CPU might not be a bad idea...

I guess I can follow the control signals through to prove the floppy accesses are being triggered as expected next...unless anybody has any other ideas?