FluffyRedLobster's build

[FluffyRedLobster]

Hi all,

I know the TF534 is not too far away but I've been obtaining parts for a 530 build for a while now and would like to get this one over the line before tackling a 534

My first two attempts both died a horrible death while I was trying to figure out SMD soldering (I know, I should have practiced on a test board...well I did, but not a small enough one...)

My third board fared much better and got as far as successfully programming both CPLDs. It however came to end when an attempt to remove the 030 socket due to a damaged pin resulted in a couple of pads being lifted. Just as well I got a lot of spares!

So I'm now onto my fourth board and it is so close... It boots into Workbench and runs stable with the RAM CPLD erased so I figure the bus side of things is good. With the RAM CPLD programmed (RC2 firmware), I cannot get to the kickstart screen. DiagROM detects the board in 24-bit space and shows it as 2MB in the detailed autoconfig page, but at the bottom of the screen it shows 512kb of fast RAM - not 2MB. So here's a question...

If a do a manual memory test in diagrom from 00200000 - 0027FFFF, I get 512Kb tested all good. 00280000 - 00300000 is intermittent Good / Bad / Dead blocks. 00300000 - 037FFFF is all good, and 00380000 - 003FFFFF is Good / Bad / Dead. So my question is, is this a valid test of the memory on the 530? And does each range represent a separate chip? I'm wondering if I may have cooked a couple of the memory chips when removing them from board 3... Any help or advise gratefully appreciated!

Here's the token pic (sorry it's not an official board, my first two were, may they rest in peace)

Cheers,
FluffyRedLobster

[Homer79]

Untitled.jpg (26.31 KiB) Viewed 6090 times

is that a short between the pins

[FluffyRedLobster]

Hmm, I thought I'd gone over the memory with a fine tooth comb but you're right that does look suspicious. I was probably going snowblind after looking at it for so long. I'll get the microscope back on it.

Might be a little while before I can make any further progress. I made the mistake of not powering off the A500 before removing the 530 today (once... out of the dozens of times I've done it today alone) and it really didn't appreciate it. I'm currently getting nothing but a black screen and random ascii characters over serial when trying to boot diagrom - I'm thinking I may have fried the CPU so need to source another one to test. The list of casualties in this project is starting to get... embarrassing

Thanks for chipping in Homer79! (Thanks for supplying my second two boards too )

[Homer79]

Quick tip for finding bridges, get a bright light. i.e phone torch and put it under the board.
you can spot them a mile away

[FluffyRedLobster]

Some progress

Replacement 030 arrived from China - popped it in, no dice. So I started checking everything on the board and replacing components that either didn't check out, or I didn't have the equipment to verify, one at a time. The voltage regulator was putting out about half the voltage it should have been so that got replaced, as did the bus cpld as a precaution as I'd messed with it so much.

Nothing changed until I replaced the capacitors under the ram and ram cpld, after which it popped back in to life, even with the old 030 I'm not sure if this made a difference, but I made sure to use X7R caps (as prescribed for the 534) rather than spares I had from the 530 build, which were X6S.

It's now booting into diagrom and workbench with both cplds programmed with rc2 firmware and detecting 2mb fastram.

Memory tests of the fastram in diagrom and systest are both failing though Next (hopefully last!) job is to check all contacts on the ram cpld and memory chips. Hoping that will sort it and I don't need to order any more memory...

[FluffyRedLobster]

Success!

After a day of faffing about, I have the 530 booting into WB3.1 alongside a Boobip 2MB expansion. IDE is working, 030 running at 50MHz using the 17/11/2018 beta firmware. This seems to be the most compatible for me. RC2 feels snappier in workbench but WHDLoad slaves tend to hand on the splash screen.

The only slight snag is that if I enable cpu fastrom I get a lot of instability - freezing when opening workbench windows etc. For now I will live with that turned off. I've got a 534 that needs building now anyway

Gratuitous sysinfo pic attached (this was with cpu fastrom enabled, I get about 9500 dhrystones with it turned off).

IMG_0183.jpg (86.4 KiB) Viewed 5975 times

Thanks to terriblefire for making this stuff - great fun!

[FluffyRedLobster]

I'm after a little help (sorry for the length of this)...

Due to slight OCD / perfectionism, I really want to get cpu fastrom working on my 530. It's sooo much faster in workbench and, well the fact that I can't get it to work is bugging me I could just start building a 534, but...

So here's the setup and symptoms...

A500 Rev6A board
2mb boobip memory expansion with gary adapter and 0.5mb chipmem mod. Total of 4.5mb ram (I did remove this entirely, no difference)
TF530 with 50Mhz 030, 40Mhz 882 and a 50Mhz crystal (tried two 50Mhz 030s, no difference) oh and alliance ram as per the bom
Firmware 17/11/2018 for both bus and ram - tried all of them from RC1 onwards, 17/11 seemed most stable
Kickstart 3.1 / Workbench 3.1
Transcend 4gb compact flash connected via a Startech ide / cf adapter (~1mb/sec with fastmem off, 6.5mb/sec with it on says sysinfo. maxtransfer set)
Tried the cpu fastrom command as well as mufastrom from mmulib (together with the new mmu.lib)

With fastrom off, it's stable aside from some keyboard weirdness (occasional flashing caps lock and keys stop responding). I think this an issue with the keyboard though as I used to get it before the 530 was installed. Workbench works, whdload works, amigatestkit memory tests run happily for ages (well 10 minutes) without issue.

With fastrom in user-startup (either cpu fastrom or mufastrom) the first time I try to open the dh0 window in workbench, the window opens, no icons appear and I get the stopwatch timer forever (although I can still move it around the screen). If I soft reboot, the cf disk is no longer detected. Hard reboot, it comes back.

If I boot without fastrom in startup and execute it manually, navigating workbench is fine, but if I run amigatest kit memory tests, it stops whilst filling the fastram for the first time - again cursor can be moved but the system is totally unresponsive. Hard reboot needed to make ide come back again.

When I first got the 530 assembled I was running workbench 1.3, which went unresponsive if fastmemfirst was in startup-sequence. 3.1 doesn't have that of course.

I'm foxed. Any suggestions from any of the more experienced faces would be greatly appreciated!

Cheers,
Chris

[dalek]

Two things I'd try:

1. Add pullup resistors if you are using EPROMs: http://eab.abime.net/showpost.php?p=118 ... count=1973
2. Try running the whole thing at a lower frequency: say 33MHz or 25MHz (as a troubleshooting step, which may show issues with your CPU/FPU actually being able to run at 50MHz)

[FluffyRedLobster]

Great suggestions, thanks dalek. I never would have made the link to try the pullups.

I have a 25Mhz oscillator so can try that straight away, and I'll order some pullups. Got to place an order for ram for the 534 anyway.

Thanks again!

[FluffyRedLobster]

Tried a 25Mhz oscillator. A bit more stable with cpu fastrom enabled - navigating workbench didn't freeze up. However memory tests in amigatestkit still froze when filling fastram.

Ordered the necessary resistors, will try that next