Modified ST hassles

[ralcool]

Wow, page one updates..... Here is now a bit of an index to things learnt over the course of a year or so, lots of discussions, attempts and failures.
I won't change the topic name, since we've become familiar with it, and links might fail from elsewhere. Where did we go?

Here is some links to perhaps interesting topics/short cuts during the life of this thread. (And we are not finished yet!) Some poor folks get directed to this thread and might get lost.
It somehow turned into a blog and often gets off topic. Thanks Dal for putting up with it!!

1. ROM mounted TOS/IDE board PCB

2. ROM mounted TOS/IDE board PCB with 2 TOS choices (hi/lo to A18) and programming support

3. First view of HD FDD pcb

4. How to install a 2.5" HDD under the FDD.

5. Modified 'Autoswitch' Overclock pcb developed.

6. Pic of a functional 'TomK' PS2 mouse adapter.

7. Mod a Sony MPF920 FDD for ST & HD use. And further down a complete 'kit'

8. First shot of functional Lascan screen. And below it, the 72pin RAM / 16Mhz pcb almost built.

9. How to disable original 68000 cpu Assuming you want to turn it off and mount a better or faster one......

10. First draft IDE/TOS/OC cpu mounted board

11. Experimental HighSpeed ROM function result

12. A look at the simplest TOS ROM mounting board. Sister board to the IDE/TOS adapter.

13 How the Atari looked with everything installed.

14. Amazing it actually works (old vid)

15. Bytes to change for 6ms HD FDD in TOS

16. Fit an easy Dallas RTC to your ST- support software found & Y2k ready.
update, our own member 'Spiny' tried one.... http://www.atari-forum.com/viewtopic.ph ... 50#p201305... result, Perfect. Woot!

17. Connect a MegaST keyboard to the internal ST socket

18. Finally, a proper CPU mounted IDE/TOS controller.
Prototype one, Working well for now. Need to build the companion TOS ROM mounting board for flashable multiTOS support again.


There is lots more stuff hidden among the pages, find your time and sift through it. I thought I'd make it simpler to get the good gear from page one.

This thread is heavily loaded with off topic talk, and at times contains fault finding for other members with similar hardware. I know I have sold a few kits of 'these' mods-
but the aim is only to generally document somewhat the kind of crap that happens while trying to do things yourself. (Including the blow ups!)

Nothing shown here is designed by me- but actually stuff you can find on some excellent sites that hold archives of old and useful modifcations created worldwide.
Mind you, I have altered some code or circuit details occasionally when it is required.

I seriously enjoy building circuits, and validating for myself and others- that there is some damn cool DIY stuff we can build for our Atari computers, that WORK!

Always, If you haven't noticed, I try to credit the original Author/Creator for their work. I'd prefer to build a hopefully proven existing design, than 're-invent the wheel'.
And often drivers or software is/are already written that we can use or modify to make it all happen.

If I went to commercial scale on any project I thought needed re-releasing... The original designer/author would be contacted if possible- and updates or renumeration would take place.
But basically I'm not interested- on this scale, why would you bother. Try it yourself. I might make an extra or two of something.. but that is it.

Live and learn right?

How fun is that!... (Somehow my only single working Atari puts up with all this nonsense, and continues to run.. even if some nights I go to bed thinking "I killed it this time".)

........

Now back to the original topic starting post.... ,


Hi All, New member- old time Atari ST user.

My dad bought our first ST when I was 13. I am now 33!

This is rather modifed. Firstly after a couple years, the main board was changed to non surface mounted mmu/glu chips- for various reasons. (The current board is a obvious '520STFM' origin)

Since then it has had- An AT_Speed C16 286 emulator, 2 + 2mb memory upgrade (2x 1mb hardwired 30pin simms most recently, plus a video shifter mounted 2mb kit from years ago, an added 'blitter' chip, Elco high density modual + drive, previously fitted overscan- now removed (long story), TOS 1.4 retro fitted (orig 1.09), and a Compo UK switchable TOS 2.6 rom kit. Cartridge mounted clock and debug widgit.. And a Megafile 30 purchased when new and expensive...but now mostly dead- mostly- because it will take a format for a day of around 8mb, and fail the next day. Only runs when warm.
I will consider making an IDE pcb oneday. Also the DMA, Floppy & Video shifter chips are socketed. And it now runs on an external PSU- for internal space.

It runs (now) on a LG 19" LCD through mono mode- using a cable I once made for a NEC Multisync 3d. Altered now for a 15pin vga connecter instead of the NECs 9pin plug. My old SC1224 monitor only died last week. (Maybe I look at fixing that later)

Heres a pic as it sits on the bench last week.


Mind the wiring, I was 15/16 when most of this went in...!

Most of this stuff has always worked. Some things have been a problem since new. Namely TOS 2.6, and lately the HD kit.

To explain.
1. The HD kit was okay- but since changing the FDD to a different one- it doesn't like HD mode anymore, I suspect its the drive not sensing correctly- but I am currently not that concerned. I've actually wired a bypass switch to let it run 'unmodified', and the drive is functional for normal 720k disks.
We'll tackle this problem later. Plus the ajax chip actually made it more unreliable than the old WD1772. I'll explain more later.

2. The overscan kit was removed due to strange horizontal screen shaking during FDD access. Like over 1" side to side. Ultimately turned out to be a glitch in the GLU chip solved by fitting a different part number one. But not before the overscan was sold- but I borrowed one for testing my idea. I was right- but now only the harness remains- with a bypass dil strip plugged in to reenable the cut tracks.

3. The C16 is playing up- it just will not read command.com on any boot discs anymore. Dunno why. Fix that later hopefully. Tested with PCditto, the discs are ok. Was ok b4 the 4meg upgrade... More testing required. Disable ram, remove 287 again....etc.

4. The TOS 2.6 is the MAJOR problem. And has been since day one- probably since the mid 90s. And almost never used!!!! I want to fix it now- armed with new information.

Here is the fault. (ps You can't actually see the 'kit', but its under the C16. Its cpu mounted type kit.) It has a software switchable program or accessory you run, and it reboots into 2.6. Ok so far. Everything looks normal- menus etc, you can run STinfo, and get reported the new TOS...etc

BUT, when writing to a floppy or a hdd- it corrupts file names! Randomly. Alway has. Hence why it was never actually used.

I originally though perhaps all the upgrades was causing the PSU to suffer. Okay... but after using an external ATX powersupply for some time- the problem persists!.. okay cross that theory out.

You might think the RAM upgrades are playing up- but the simms were installed long after 2.6 showed original signs of this fault.

You might think the C16 on top is causing interferance- nope removed that- still there.

High Density upgrade... maybe- but it can be bypassed back to 8mhz. And I 'think' the fault was there before the ajax was fitted too.

I now believe- after reading this board- that perhaps a 74ls74 might need to be fitted to the cpu- mega ST style. After all a blitter has been added.

Tonight I have been confirming the fault exists- copying some floppies to ram disk and back.
Using TOS 1.4, blitter on / off no fault.
Under TOS 2.6 Blitter ON, corrupted file names. eg, 'atari.prg' VS ' __ari.prg' - not when going to the ram disk- only back to fdd/hdd.
Now TOS 2.6 Blitter OFF, Seems improved. (Never actually tried this), less lower case letters.

I have a 74LS74AP sitting in my junk box, and ppero's diagram of the mega st fitment. I just need to trace where the BGACK wire leaves the blitter to the cpu- CUT the trace? , and wire in the chip. MAYBE this will help. IDEAS?!

Also the AT_SPEED instructions say to put a 220ohm resister between pin 1 of this chip and pin 12 (BGACK?) of the cpu- after relocating it for space above the cpu. Interesting....? but why? seems like a bypass to me.....

I've also noticed- the MEGAFILE 30 PSU is nearly identical to the STs, but is double the AMPS rating on the 12v rail. Easy to retrofit!, just use the old base bracket.

Thanks for reading this, I hope someone can provide some insight....

Sean

[Shredder11]

First of all a big warm welcome to the forum! You will like it here as it is chock full of ST/TT/Falcon enthusiasts armed to the teeth with great info and very helpful too. Secondly a big pat on the back for a fantastic opening post which I found really interesting.

[Nikolas]

Sometimes with those st mods chips may die like mmu glue, ect..

Your st is in bad shape, i see rust allover

[wongck]

Wow you did those at 15/16..?
Must have been fun

[ralcool]

YUP!, I was young. Most of these upgrades went in before I graduated highschool. (Class of '93)- so 17 years ago. Or when I was about 16. Yes it was FUN!

The only thing added since, was the extra 2mb of simms. Nearly 2years ago. I get all nostalgic, and bring the ST onto the workbench. And PLAY!
I have done aircraft avionics for 10years, and now the Atari is a toy filled with hackable logic gates...etc.

The first ever mod was the AT_Speed C16. My English class used Word Perfect for our writing pleasure..... Basically I wrote my homework and assignments with it. School had a bunch of XT 8086s, and one 386 for printing on the only laser printer! My 286 emulator on the ST, was faster by FAR than the old mono XTs.

I convinced dad the PC-Ditto software(PC emulator) was too slow to be useful!

Then I decided I need more RAM, so I bought a 2mb kit from a local Atari freak. After a failed attempt at installing a 'Extra RAM Deluxe- simm upgrade' - in hindsite- the RAS/CAS lines of the original memory wasn't disabled- or instructed too. Strange. But I still remember attempting installing it. During getting the replacement memory kit- the same guy had TOS 1.4 roms, so I got them too.

Then the clock cartridge (Same guy MADE it!). Later the Blitter chip. (Installed by him).. then I fitted the HD modual + Ajax chip.

And finally the TOS 2.6 imported from the UK- which has always corrupted file names. Never really used.

I fitted everything except the blitter. I even still have all the original fitting instructions for the different 'kits'

I think now, even after my first post here, and reading everthing I can find- that maybe the IMP 100110-001 DMA chip could be the data corruption problem. My original MMU (also an IMP type 100109) didn't like addressing above 2mb. At one time I was running 2.5 meg- and it played up in all sorts of ways. Back to 2mb until a new MMU was found.

Or perhaps the ajax chip is being sensed by the TOS 2.06 and doing funny things on the data bus.

I have a spare broken 1040STFM here with different part # DMA (CO25913), and a normal WD1772 controller- So I'll desolder those and try them out. Also I'll remove the blitter from the socket and solder the W3/4 pins and see if that makes any difference. Before I go and muck about installing a 74LS74 like the megas. Plus the HD modual can be configured in differnt ways- My new FDD drive doesn't work like my old one.
Should fix its little bug.

Modifying stuff has been a way of life for me since, PS1 and PS2 chips- hdd upgrades to PS2s, custom ECU in my car..EFI Turbo motorcycle. all kinds of things.

My dad thinks I'm slightly mad playing with the old machine- but I still love it, and what it taught me. Custom cables et all.

[simbo]

i think you should not use the machine just now
till you unseat all the ic's that are in sockets and clean there pins
using a glassfibre pencil and remove them using the right tool for ic's
total cost is about 10 euros for the right tools
replace all the electrolytic capacitors in the psu
and also the main board bigger values
check the underside of the machine for groths of iron film etc
and im sure it will work
i think your psu would have had to work hard to cope with all the extra firmware
also tos 2 upgrades are a very fussy part within an older st and will need to work
before anything else will
so clean the pins on all the plug in ic's and i mean all
and also the sockets
i add a small smear of grease to the sockets pins {i use a good silicon grease insulative }
this reduces corrosion
if you also oil any rusy parts they will platlet heal a little and wont rot further

i think you need to move slow with this machine

a glassfibre burnish pencil
ebay free 1st class post from uk supplier
http://cgi.ebay.co.uk/FIBREGLASS-PENCIL ... 2ea9957d4d

the correct ic remover
ebay 1.99 +free 1st class post uk europe {2.5 euros or so}
http://cgi.ebay.co.uk/Motherboard-Circu ... 1e59d761b7

[AtariMega125]

Wow, thats impressive. It doesn't surprise me that it's a bit unstable. I think the 2.5 mb memory upgrade that I soldered onto the motherboard of my first 1040ST is what ultimately killed it--or at least contribuited to it's demise. On the other hand, I was playing around with 68000 assembler at the time and might have assembled something buggy that pushed the hardware where it wasn't designed to go.

I use seltos to use TOS 2.06 in my Mega. Reading this makes me think it is the way to go. Motherboard surgery is a risky business and I'm really impressed on how well you pulled it off. You did quite a piece of work.

[ralcool]

Hi guys,

As per Simbos recommendation, I removed so far most of the socketed ICs. The TOS 2.6 board on the 68000 was rather 'white' around the pins. Also the C16 too.
Basically corrosion. So I've cleaned them up. With good old rare Freon liquid! Best stuff you can't get. I've left those boards off for now.

The corrosion could explain why the C16 is acting up. At 16mhz things move faster- and memory is stressed.

I have rewired the on board ram ras/cas disable wiring to be neater.

I've been running Memtest. As advised elsewhere on the board. After getting some slight video garbage. A minor clue here... so;
Result.... random read/write errors on the test of the same name. Most of the initial tests pass until this.

Ok, Maybe the simms (which were a pair of many in the junk box) might be faulty. So I disabled them with another ras/cas 5v high disable.

Leaving just the shifter mounted ram. Still getting errors, but now at differnet hex addresses.



So next I'll disable this ram, and re-enable the simms for another retest.

I just cleaned and swapped the MMU with the spare (same part#) before swapping banks... no improvement.

Unless just having all these memory bank still connected to the data/address bus -despite being disabled- is causing a kind of glitch. Doubt it though.

If things ever improve, I intend to make a 'lascan' for the overscan harness, and maybe use the 2.06 roms on a new board with an ide connector- just need to make a gal programmer- no big deal, this will be the 4th programmer built over the years. PIC, EEPROM, ATMEL.. now GAL.

[ralcool]

Well, suprisingly After HOURS of testing. The Shifter mounted ram must be faulty/old/worn out.

Resoldering every joint/wire, cleaning connections,etc....and multiple failed test passes. It never improved.

Last ditch attempt was setting the 2x 1Mb simms as bank 0, and it PASSES!- repeatedly! Not bad considering the spider web of fine wiring!

So now...um, might fit a old/spare 72pin 4mb FP simm for complete one piece 4mb ram solution. Shame we haven't worked out a 14Mb solution yet useless tho.
And remove the address wiring to the shifter board for decommisioning and removal.

I guess the C16s command.com load errors HAD to memory related. I'll test that theory VERY soon. (Like tomorrow)

Still doesn't entirely explain the 2.06 corruption errors, but I'll test THAT too. Maybe it will be fixed......

Wow, its 4.30am and I'm tired. And high from soldering fumes.

I COULD make a new pcb to fit simm(s) sockets around the shifter chip, but I can only make single sided boards- and still then only if I can do a negative mask/overlay.
Will end up another rat nest wiring job.

The ST lives to see another 20years...! ButI need sleep today....

peace out.

edit.
Still awake... I had a final thought. Maybe the shifter board address wiring has internal corrosion.... Happens in aircraft.
I shortened the wires that connect to the resister bank about 3mm and resoldered... Now passing successfully at 4mb. Holy cow.

Tis funny how faults hide themselves.....3 good passes so far.. I concentrated so much on the bad looking ram chip soldering, I left out other less obvious stuff.

Beware, Flux is a corrosive agent. ALWAYS remove/clean after soldering-it will fail eventually.

[ralcool]

Ok, Further update. I reswapped simms to bank 1, and shifter ram to bank 0 like original setup.

Started again to get occasional 1/2 errors on memtest, (but not the 15-30 errors like the address wiring) but noted after dec'ing the hex addresses, they were always now between 1-1.5 mb. I looked at the cas1 on bank 0 resister as suspect. perhaps from over bending resister leg.

Seems true. replaced resister and passes perfectly now. I think. Another 10-20 tests to be sure.

Testing takes so much time. But result is worth it.

C16 is now loading command.com!!!! YES!

But really need another 64pin socket for clearance to refit TOS 2.06 under C16 from cpu socket. The TOS board has piggy back, but due to the tos chip mounted under pcb it won't fit without extra space provided by an extra socket on the cpu- it hits the buffer chips.... If had no sockets for TOS chips it would fit.

This is why I would like to replace the TOS board- but where to fit. Blitter location is full. Shifter has no address lines, cartridge is full. Need a ACSI port SatanDisk- but cannot buy one and/or harder to fit internal drive to external port. Solution would be a blitter chip socket piggyback mounted solution like C16 for STE cpu. Then add IDE GAL and 40 pin header.

I like Joogns simpler idea, but the diagram is a little fuzzy- and I don't know if ahdi compatible....little documentation. Why GAL if Joogn works??? 13 input nand compared to programmed GAL? Why is it better.

Getting better every day.

[ralcool]

Hi guys,

Well the memory is conclusively repaired. 4Mb worth.

The AT_Speed C16 is functioning nicely, booting both DR-DOS 5 and MS-DOS 6.22. (allthough the 286XL seems incompatible- ebay purchase)

The High Density controller is fixed. I had to wire the DD/HD sense switch directly to the modual. It didn't appear at pin 2 of the floppy- regardless of drive jumpering. But it seems to be repeatable and reliable now the controller can know the disc type. Formatting correctly to 1.44mb or 720k.

I haven't been able to try the TOS 2.06 board, I still don't have a replacement socket or DIL socket strip to add room. But using Software booted 2.6 seems to work OK.

One strange detail tho... Hardware TOS 2.06 never provided a 1.44MB format option- but soft TOS does.
Now since the STE has a jumper for this, is it possible to do something similar to the STFM? What does it actually enable? An address line...?

So happy days so far!

Sean

[jd]

Phew, need a beer after that!!

well done.

[ralcool]

Much beer was drunk....

One VERY handy feature of the Elco HD Modual- you can copy from HD internal drive to DD external- it switches FDC speed on the fly.

Most excellent. I can avoid putting some of my ancient floppies in the good internal drive and messing up the heads. So far any external drive fits my homebrew cable.

[spiny]

outstanding work

[ralcool]

Did ppera ever make a TOS patch for SuperTOS 2.06 WITH ide std/twisted support?

I have been trying SuperTOS via Seltos, and decided I generally like it. But his page only mentions patching support for 'normal' 2.06 (and the others 1.4/1.62 etc)

Despite this, the ATTACHED TOS 2.6 software bootable (auto folder) has 1.44MB support in the format menu- I'm not sure where it comes from.. but I'll try to link the file. This one actually says Tos 2.6 in the Desktop info-instead of the date. And it works... 1.4Mb formats.

Seltos loading TOS 2.06 UK.IMG or SuperTOS 2.06 doesn't have 1.44Mb support in 'Format' But....
if I boot with Ramtos7.prg in the Auto folder (with SuperTOS as RTOS.IMG or TOS 2.06 UK as RTOS.IMG in the root) it does. And installs otherwise perfectly

For some reason RamTos7b.prg says SuperTOS image not recognised.... bug?

But....
After booting with Ramtos7.prg & SuperTos or TOS2.06UK the High Density option appears, but then BOTH give a 720k format using the HD format option! Wierd. (also.Both give a CRC Chip 0 error on the first boot).

Anyway..... I have a seperate HD formatter program...... but... WTF?

But the original question remains- can I patch SuperTOS with IDE twisted cable support? & Can we fix the HD formatting option?

Sean

[ralcool]

Dunno about copyright.. but this is the AT_Speed mouse driver...

Just insert a 25-9 serial adapter..and a 9pin-PS/2 adapter into THAT.

Now a normal PS2 mouse works via the serial port.

[wongck]

Dunno about copyright.. but this is the AT_Speed mouse driver...

Just insert a 25-9 serial adapter..and a 9pin-PS/2 adapter into THAT.

Now a normal PS2 mouse works via the serial port.

That be a serial mouse, which is still obtainable nowadays, fi you look hard enough.
There are several serial mouse drivers available for Atari.
I was using one of them years ago

[ralcool]

With regard to TOS 2.06. I am trying to understand the mechanics of the physical rom size. 1Mb roms come in 28 AND 32pin flavours

2 chip <1.4 is supplied on 1M (128x8) roms. As does the STEs 1.6/1.62 Both have 28pins. The image for 1.4 is 196k. So fits into 256k of space provided.
The output files when split are 2x 131072 bytes. For hi/lo.

TOS 2.06 original image is 262144 bytes, and the output files are the same as 1.4s for example. 131072bytes.

Sure they use the 32pin version of the 1Mb chips- but does it have too when 28pin versions are possible. (Not fitting into 6 x 32k / 27c256 chips makes perfect sense- 192k is < 256k. Not big enough)

WHY can't we just burn the output files onto a normal 2chip set?

I have tried to read everything on this board- but don't see the understanding or the answer. Why the hassle if the (2chip TOS1.4) roms are large enough.
Can 2.06 only be 'addressed' on 32pin?

[ralcool]

I think I get it now.. the differing TTL/Gal logic selects address lines off the cpu that allow the roms to sit/map into the correct address space they're designed for.

I wonder tho how one could remove the need for two physical roms- and use just one. The Mario Becroft ide adapter appears too.

Looking at the ST schematics- one would have to combine the UDS/LDS lines to the OE pin of the roms.

What other uses could there be for the now unused ROM0/1 lines......

[Shredder11]

This is a great thread by the way!

[ralcool]

Thanks, I'm just documenting getting the old girl back in service. Perhaps someone might find it interesting.

Last night I completed patching TOS 1.4 with pperos IDE Twisted , um patch. The used his rom splitter to seperate it into 6 parts.

Then I also used his splitter to create 6x TOS 1.2 images to compare to some TOS 1.2 roms sitting in an ST I use for spares. 100% identical.
Interestingly, I score 6 dvds some years back .. The ultimate ST/STE /Falcon collection, and it has almost every TOS image on it.

So I extracted/split the TOS 1.2 uk file, and compared it to my spare roms.... like I said 100% match!

So after erasing the spares in my UV device- I reburn them with the patched 1.4.. Testing of these will continue today in my ST. See if it boots.
When my 29F010 flash roms arrive- I 'll burn the patched 2.06 files (or Supertos for the interum until I get to making the ide controller)

There are 109 TOS images on this dvd- including EmuTOS, Kaos, and what appears to be every language of TOS 1.0,1.2,1.4,1.6,1.62,2.05,2.06,3.06,4.00,4.02,4.04 & 4.92. Ru,Fr,Ch,De,US,Uk,It.

[ralcool]

Anybody know why there appears to the two versions of UK TOS 2.06. Normal and [a]

These files compare differently in HxD. But appear identical run under STEEM.

I've attached them for your reverse engineering pleasure.

[ralcool]

This thread appears to be getting some views- what are you all thinking... ?

Anyway- Tonight I tested (With Steem & Seltos) that SuperTOS seems OK to patch with the Pperos 2.06 IDE patch. It appears most of SuperTOSs reworking was later in the image- so the position of the jump fitted fine, and the rest of the patch data was able to go in the usual position.

After a CRC reset I'm ready to burn it when my new flash roms arrive.

I was able to confirm the TOS2.06 roms that came with my upgrade board wasn't the [a] version either.

But the suprising thing is.....

SuperTos compared with HxD appears to be identical to the [a] version tos I attached. So WHO wrote it.. Atari? Is this the correct SuperTOS 2.06. Certainly full windows drag works.

I guess the makers of the DVDs I got, added the modified file and renamed [a]

[Shredder11]

Sounds like your DVD would make a useful torrent download for the people here

[simbo]

this is a complex machine
mostly dependant on software to activate addins

my best advice is to remove both ram upgrades the 2.5mb shifter??
and the home brew 2mb 32 pin simms for 2MB ram

look to here and do a 72 pin update to 4mb or 1mb

credits to many others and alison
http://www.atari-forum.com/viewtopic.php?f=74&t=18614

ive done 12 machines st too 4mb or 1mb on switch!

i think your upgrades as haphassard???

however maybe not !!!

adding satandisk
and leaving all alone maybe for you
is maybe your way,,