TF330 Akiko Rev 0 Issue (SOLVED)

[alenppc]

I have a PAL unit.
I will burn a DiagROM 1.1 tonight.
I will test the memory from $40000000 to $44000000.
The IRQ tests passed, I don't think the CIA test will pass from memory (I didn't think the CD32 had CIAs - or maybe you implemented them).

When trying to boot up with a normal ROM the screen sometimes goes from black -> grey, but I can't get to the boot menu with both mouse buttons pressed down.

With the jumper set to the DISABLESPEED the unit will start, but there are no boards detected in the boot menu (I assume this is normal).

The board takes a few seconds to boot, so black > grey is normal also, then the standard 3.1 IDE check delay comes.

The jumpers don't work like that anymore since the latest firmware, you have to close the bottom two sections of Disable and Speed in order to enable the card.

[anton.gale]

I installed a DiagROM 1.1B, I like the enhancements!
CIA-B does fail & IDE doesn't appear to work.
The auto-config does not repeat anymore and reports the correct amount of RAM, but an incorrect amount of RAM is detected on start-up. The 'test detected fastram' seems to want to start at $7F000000 instead of $400000000. If I manually test the ram a bad block is detected (see picture).
Are there any particular pins on Akiko that I need to look at, or should I replace Akiko?

CIAB Failed

20190306183559.png (94.73 KiB) Viewed 4179 times

sysinfo.png (135.55 KiB) Viewed 4179 times

autoconfig.png (80.68 KiB) Viewed 4179 times

ide.png (39.71 KiB) Viewed 4179 times

badram.png (109.89 KiB) Viewed 4179 times

[terriblefire]

I installed a DiagROM 1.1B, I like the enhancements!
CIA-B does fail & IDE doesn't appear to work.
The auto-config does not repeat anymore and reports the correct amount of RAM, but an incorrect amount of RAM is detected on start-up. The 'test detected fastram' seems to want to start at $7F000000 instead of $400000000. If I manually test the ram a bad block is detected (see picture).
Are there any particular pins on Akiko that I need to look at, or should I replace Akiko?

20190306183559.png
sysinfo.png
autoconfig.png
ide.png
badram.png

The DiagROM fastram detection code is still suspect on the CD32. Just use complete memory detection (option 6 in memory tests i think).

This looks healthy except for the CIAB stuff. The RAM is looking fine and gets corrupted in the ZII config area (which is probably expected if Autoconfig hasnt run). If you run autoconfig and configure the cards then this disappears. Even so i'd expect it to boot with that much RAM.

If you are able to reflash the CPLD i'll come up with a patch for you. Dont replace anything. I will bet you if you run diagrom without the TF330 those tests will pass.

[terriblefire]

I made a firmware update with 2 tiny tweeks.

1. The Clock Domain cross has 2 possible values. Both work for me but i've put the other one in this test.
2. The ZII area is fully decoded (this is probably a waste of CPLD real estate in the long run).

https://www.dropbox.com/s/8m90s1mtgbsb5 ... a.zip?dl=0

[alenppc]

My rev3 boards have arrived today! yay!

I built one and flashed it with the firmware from this thread and it works all the same as the other one, but of course there are no issues on my machine. Just thought I'd mention.

[anton.gale]

I flashed the TF330 with the firmware you kindly modified. The correct amount of RAM now displayed in DiagROM1.1 and a 'Complete Memory Detection' shows the correct address range and size.

The unit will still not boot, but I assume you expected that as the CIA-B issue and IDE drive interrupt doesn't trigger. The drive light does come on when this test is run.

If anyone has any pointers as to how I can fix the CIA-B issue I am all ears. Thank you again for your help so far.

Memory Detection

20190307195818 - memdetect.png (45.56 KiB) Viewed 4125 times

64MB Fast Check

20190307200020 - goodmem.png (60.95 KiB) Viewed 4125 times

[terriblefire]

Ok the RAM is fine and the CPU runs... here is what I think is happening. The CIA timer timeout for something is not triggering and thus it waits forever.

I suspect some physical fault like DSACK1 being shorted to DSACK0.. either on the CD32 or the card. Or possibly no contact between DSACK0 and the CD32.

Can you run the CIA tests 20 times and see if the results are consistent. If they are then it’s a physical fault. If it’s not consistent it’s maybe a clock switch issue.

Also run the graphics tests. Tell me if any crash the machine

[anton.gale]

All of the graphics tests run fine without issue.
DSACK0 & DSASK1 have ~1K Ohm of resistance between them and they buzz out all the way to the edge connectors after going through two 27ohm resistors.
I have run the tests well over 20 times.

[terriblefire]

All of the graphics tests run fine without issue.
DSACK0 & DSASK1 have ~1K Ohm of resistance between them and they buzz out all the way to the edge connectors after going through two 27ohm resistors.
I have run the tests well over 20 times.

Is this on the TF330 alone or connected to the CD32. I think the issue is on the CD32. There is a separate set of DSACKs to the 030. They might be shorted too.

[anton.gale]

Without, the issue exists on the vanilla CD32. I guess I've never noticed because the unit does boot and play games with the CIA-B issue.