TF536 Rev2 in Amiga 500 Rev 5 kickstart issues

[JoeFloyd]

I have a TF536 that has been temperamental pretty much since it was installed last year. The TF536 is a self built Rev 2.

The DIP64 socket has been swapped for relocation kit. 512K ram expansion with clock is installed in the trapdoor and is recognized by the system.

https://www.ebay.com/itm/333301477068

The DIP40 PROM socket has been swapped for a ZIF socket.

This issue I'm seeing is that the OEM MASK ROM Kickstart 1.3 boots reliability with the TF536 installed using the relocation kit. Workbench 1.3 starts and software like sysinfo. Sysinfo shows no boards installed and no additional RAM.

A 27C400 mounted to an Amiga 500 Rev 5 adapter (EPROM to MASK ROM pinout converter with 4.7K pull-ups), boots fine when programmed with Kickstart 1.3 or Diagrom (downloaded directly from Chucky's site)

The same 27C400 in the same adapter with kickstart 3.1 or 3.2 fails to boot about 95% of the time. Rainbow is not always seen when resetting. Monitor output is weird. Sometimes there is no video output detected by the monitor but other times a strange distorted red or blue screen is seen. This is not the normal fault screen, but something different as far as I can tell.

A Flash based PROM replacement from GGLabs has also been tried.

https://gglabs.us/files/F2R16-A02.jpg

This Flash based ROM has the same behavior. When flashed with kickstart 1.3 or Diagrom the A500 will boot and run workbench or diagnostics without error. When flashed with kickstart 3.1 or 3.2 failure to boot and strange video output.

I programmed the TF536 CPLD with RC3 (03/2021) and did not see any difference in behavior as described above.

The relocation kit is my prime candidate at this point, but since the TF536 works with the OEM mask ROM and with a27C400 (programmed with 1.3 or diagrom), I can't really say the relocation kit is the problem. The relocation kit uses a 64 pin ribbon cable. The next thing I might try is making the ribbon cable shorter and wrapping it with copper foil.

Anyone have any experience with this kind of boot failure? Is there something I'm missing?

[terriblefire]

I have a TF536 that has been temperamental pretty much since it was installed last year. The TF536 is a self built Rev 2.

...

This issue I'm seeing is that the OEM MASK ROM Kickstart 1.3 boots reliability with the TF536 installed using the relocation kit. Workbench 1.3 starts and software like sysinfo. Sysinfo shows no boards installed and no additional RAM.

Your fastram is bad or soldered bad. Its pretty obvious. Kick1.3 wont add that ram so you wont see it with that. Stop trying different roms its got nothing to do with it.

[JoeFloyd]

Thanks for the tip. I'll double check my soldering. The TF536 did work at one point, but the board has experienced a bit of rough handling when being installed and removed.

What is the expected behavior of diagrom? Diagrom did detect the 64M of fast ram, but I didn't do a full memory scan.

Has anyone started a FAQ for the TF536?

Happy holidays.

[GadgetUK164]

I seriously suspect that "flexible relocation" kit... The fact the connections for databus and address bus are routed on PCB, through right angled connectors, over a ribbon... That's bound to be flakey and perhaps more unreliable with EPROMs.

[JoeFloyd]

I ran through all of the tests on diagrom 1.2.1

Diagrom checksum passed (64K rom checksum valid)

Manual memory test (slow) 00C00000 - 00C7FFFF No bit errors, no byte errors

Manual memory test (slow) 40000000 - 43FFFFFF No bit errors, no byte errors

Chipmem test 00000400 - 00200000 No bit errors, no byte errors

Autoconfig test sees TF536 and shows autoconfig data

IRQ test clean
CIA tests clean
CIA experimental test clean
Screen mode tests worked fine
Screen scroll test worked fine
Audio tests were fine
Keyboard works fine
I didn't run disk tests, but since the machine boots workbench from floppy, it is safe to assume floppy is

As far as I can tell the TF536 64MB of fast memory is working fine.
The data and address bus lines between the TF536 and the reset of the system are not shorted or open

Given the memory test results, it would appear that the TF536 memory is working. Is it safe to rule out soldering issues with the memory?

[terriblefire]

No. Diagrom doesnt really test the ram much. If you can use addmem to add the memory and then use Amiga Test Kit