Agranlunds TF534 Build

[agranlund]

Ok.. now we're getting into silly territory! Not bad for a 520ST

DSC_3110.JPG (76.56 KiB) Viewed 6501 times

This is with both EmuTOS and NVDI in FastRAM.

It's come quite a long way since the first time it was booted up.
Did I mention I love this? I wanted something to tinker with instead of a finished product so this is absolutely perfect.
Thank you Stephen Leary!

[IngoQ]

Whoa... I'm impressed

[DoG]

Seems that I have to finish building my 534 with these nice figures. To bad that I don't have any way to get the pak030-board in alt-ram since I don't have a frak board for it to test against this.

[agranlund]

Seems that I have to finish building my 534 with these nice figures

So far the 534 has more than exceeded my expectations
It's running stable at 40Mhz with anything I've thrown at it so far. The CPU is rated for 33Mhz but it's not really getting that hot so I will try with 50Mhz this evening.

The only real issues I have left now are:
- uIP with NetUsbee is not working when I run @ 40Mhz (only at 8Mhz - but emutos does not boot at that speed if I have FastRAM enabled)
- the IDE port on my blitter board is the same. It works only if I run the TF @ 8Mhz.

I really don't care about the IDE port on my blitter board now that I have the Terriblefire (I'm going to make a version of my blitter-board with a Ricoh realtime clock instead) - but I do use the NetUsbee for transferring files between PC<->Atari and would like to get that working at least.
However, looking at the source for EmuTos and uIP I think the issues might be related - they both perform some kind of speed tests to figure out how much delay they need to put in between various operations. My gut feeling is that this is what is causing the issues but I am not 100% sure.
Of course, the speed when going through the ST bus is held back to pretty much stock speeds so if it thinks that we're good to go for 40Mhz all the time then these delays will end up being off...

Also, the Terriblefire IDE is not working for me in Tos2.06. Not really bothered about that but I will at least try hooking up the IRQ line and see if that solves it.


I made a github account, and will spend some time reading up on how to use it. I want to make a proper pull request for my changes to go into the official firmware.

[stephen_usher]

NetUSBee is *REALLY* finicky about the system.

When I swapped the power supply on my STE from the stock to an Exxos one it stopped working. It never worked on my TT (then again lots of cartridges which use a "write-using-read" hack fail on my TT).

[DoG]

IDE is also (apparently) a bit fiddly to use. I use SatanDisk or UltraSatan with the pak030 to get HD to work. The IDE boards I got don't work with it. But Lightning ST which have IDE on it is working with Pak030. But that maybe got a own clock on it (probably).

[agranlund]

Ah, ok, thanks for the heads up! Maybe I'll just have to accept that NetUsbee is not going to work :/
I'm thinking of taking a quick stab at fiddling with the delay code in uIP but if I can't get any results reasonably quick I'll just accept defeat on that front.

Either way, it sounds like it's not a fault of the TF card as such but a common problem when using accelerators (or the TT).

[frank.lukas]

IDE is also (apparently) a bit fiddly to use. I use SatanDisk or UltraSatan with the pak030 to get HD to work. The IDE boards I got don't work with it. But Lightning ST which have IDE on it is working with Pak030. But that maybe got a own clock on it (probably).

My IDE Boards (c´t IDE or Lightning ST IDE) works fine with a PAK68/2-020at 16Mhz or even with a PAK68/3-030 at 50Mhz ...

[agranlund]

Also, the Terriblefire IDE is not working for me in Tos2.06. Not really bothered about that but I will at least try hooking up the IRQ line and see if that solves it.

Some firmware tweaks, and hooking up the IRQ line, did indeed do the trick and TOS 2.06 boots up fine from the TF534 IDE interface now


However, we need to solder a wire from the the INT2 via on the TF to ASCI pin 10 on the Atari motherboard because I am unsure how to emulate the interrupt register like the Amiga firmware does. On the Atari, that interrupt is shared with the ASCI and Floppy ports as well :/
Perhaps someone else is able to find a way to make it so that soldering that wire is not necessary for the Atari.
EmuTOS does not depend on this interrupt so if you're only interested in that you don't need to solder anything.

Stephen, on future cards, if there is room, I think it would be great to have two easy-to-solder-to vias. One for hooking up the Atari IDE interrupt and another one for ROM chipselect (most people are probably going to need rom-decoding).

[8 Bit Dreams]

Are the ST firmwares You are using on Your Guthub?
Would be glad to test them on my hardware, please let us know where we can obtain them...