Repair of two 1050 floppy disk drives

[atariry]

I know that there are a few other threads on 1050 disk drive repair, so I hope I am forgiven for starting a new one. Hopefully some of my techniques might help others.

 

Back in November a bunch of Atari 8 bit stuff was put on Ebay UK by a seller - "ginger_colin". It was all "buy it now" and most items were "for parts or not working", even though for the computer systems he showed the computer as operational in the pictures. Someone grabbed a bargain Atari 800 for less than £100. A number of items sounded like they needed a clean. I bought a 1050 for £30, then a day or so later bought a second for £35. I bought the first to use the case for an Atari related project which I started back in 2000, but then the second just seemed too good a bargain to pass up. The second drive was badly protected (hardly any bubble wrap) and when I took the drive apart it was clear that it had taken quite a hit during delivery. I powered up drive #1, but it was completely dead. I powered up drive #2 (the damaged one) and switching on and off several times, it operated the drive once only. So I knew that I had a repair project on my hands.

 

I took both drives apart and cleaned the cases (using JIF and soap) and used a brush to get rid of fine black "dust". I suspect that the original owner was a smoker. Drive #2 had broken and cracked support pillars. The switch was bent up - quite a feat if you know the size of the 3 pins of this switch, as was the LED. The LED bent back, but the switch didn't. The tab to open the drive was completely stuck, as it was clear that the drive had been hit hard on the front. The brown front plastic surround was cracked and pieces broken off in the bottom left corner. Much later I also noticed that the front of the mech was also cracked. To his credit the seller refunded the money for drive #2 without me even asking. He was also selling a bare mechanism, and having communicated with him it sounded like this was due to him having to replace another drive (for a different buyer) which must also have been broken in the post (I assume that he must have had to replace the plastic case leaving him with a spare mech).

 

I removed both mechs and used a multimeter and scope to check the power supplies. Both drives checked out OK and the 5V looked very clean. The 6507 clocks for both drives were good and RSTB was working correctly. What surprised me was that for both drives the address lines were not toggling. This puzzled me as if the firmware was not seeing the mech I would have thought that the code would end up in a tight "error" loop so giving some 6507 address line toggling. I surmised that the CPUs have executed some bad opcodes which crashed the CPUs, possibly due to back RAM used for the stack. So I then started a more thorough examination.

 

I will post more later, so those that are not interested do not have to wade through a long initial posting.

[_The Doctor__]

because proper packing protection methods matter

 

as for the id's... deoxit or tuner cleaner and working switches may do it for you or you will have to replace the chip...

Edited January 1, 2019 by _The Doctor__

[xrbrevin]

frustrating because packing is beyond a buyer's influence

but also a shame for the hardware itself!

senseless... but 1050s do respond to kindness - hope you get them going ok

[E474]

Hi,

 

If you have a working 1050, you can swap chips in to see how well they work. Make sure you keep track of which drive the disk controllers come from, as each drive has to be adjusted to work in spec for the controllers. There's this thread on what to do if you have moved them around -

http://atariage.com/forums/topic/282626-howto-adjust-analogue-part-of-fdc-279x-in-a-1050-disk-drive/

[atariry]

Note: both drives have 2793 and tandon mechs, one has revL ROM and the other revK (FWIW).

 

I did remove all ICs from both drives, trying not to mix them . Interestingly I found in one drive, a bent pin on two of the ICs (2793 and 6810). I straightened the pins, re-socketed the chips, connected the mech, powered on, but the drive was dead. Bummer .

 

Supecting that the RAMs might be faulty and causing the strange 6507 behaviour, I took a different approach. Using a Teensy3.5 type Arduino (which has 5V tolerant pins, and plenty of them) I wired them all to the 6810 RAM chip and used an external 5V supply for power. I wrote code to thoroughly test the RAM using industry algorithms and I also checked all 6 chip selects. Both RAMs from each drive passed. The Arduino code needs tidying up but I am happy to release it, in case it might be of help. Over the next few days I repeated the same exercise for the 6532 RIOT chip, connecting PORTA and PORTB back to back. Again both chips passed with flying colours. So the problem was not caused by a faulty RAM corrupting the stack.

 

My next approach was to dust down an old homebrew EPROM emulator. This can emulate upto a 64k byte RAM using a stack of 6264 8k RAMs. Again I connected it's control signals to pins of the Teensy and wrote the code to download a ROM image. The emulator is based on shift registers (rather like SPI) so is pretty easy to code. Of course it didn't go without a glitch. I found that I had used capacitors to slug some of the signals (I have no recollection why), and this took time to figure out. The solution was simply to snip them off! I then had an intermittant problem. There is a green LED which lights when the EPROM is being accessed, and poking on the circuit board, I could make it light brighly. Eventually I found a loose connection and I was in business. Unfortunately my header is based on the 27XXX series of EPROMs and is not compatible with the pinout of the 2332 ROM (or 2532 EPROM), so some creative wiring was needed.

 

I wrote some 6502 code (my first in over 30 years!) to simply fill the emulated ROM with NOP instructions. This was to allow me to use a scope to check the address lines.They should count up like a binary counter. I got this working pretty quickly. I then modified the code to access each peripheral chip to confirm that the chip select decodes were working. I was able to blow this code into a 2532 EPROM. Again the source and binary are available should anybody find them useful. I had to remove the NE555 reset generator IC, so that I could connect my own reset controlled by the Teensy, which I used to trigger the scope and logic analyser. In scoping the pins of the 6507 I found some unusual looking waveforms on some of the address pins. The waveforms were not simple low and high logic levels, but the logic high appeared to be modulated by lower address pins, making it look bumpy (I'll see if I have a screen grab). I was convinced that the 6507 was either faulty or that there was a short, but after a look of fruitless looking, I figured that's just how (some of) the address pins behave. Also when using the 2532 EPROM, the logic high level on pin 20 (the program pin) was really low. This pin is used as a chip select and connects to A12 (if I recall correctly). Again this appears to be normal and the system still runs (it doesn't lock up as I originally observed). I assume that the program pin is a little unusual and just heavily loads whichever IC is driving it.

 

During most of this work I have been working on drive/PCB #2 (the one which got bumped but showed a tiny bit of life). Only the 6507 and EPROM emulator were plugged in (the 2793, RIOT and 6810 were removed). I think that I briefly checked with drive/PCB #1, and it also appeared to run the EPROM emulator code correctly.

 

My plan, possibly a little ambitious, was to write some code to check out features of the floppy, with no connection to the Atari. I'm unlikely to do this now, but the idea was to have an EPROM with code to diagnose the digital circuitry and floppy operation of the 1050. I may re-visit this at a later date.

 

...to be continued.

[+Nezgar]

Thanks for the interesting read, and I look forward to hear of your further adventures. Looks like you have a knack for this stuff! I respect and admire those who are so comfortable to dive into the low level electronics and try new things like that.

 

I can see your ROM jumpers are in the JP1/JP3 positions, which is for 2332 mask ROM. Moving them to JP2/JP4 allows use of a 2732 EPROM, which might be easier for you to work with.

 

I thought that a 2532 EPROM would work in the 2332 configuration, but I recently actually tried this by programming a 2532 with the stock Rev L firmware, and it didn't work - so don't bank on that as a good test method for yourself.. Well, it would spin the disk, but that's it, which indicated it might have executed some minimal amount of code to do that, as no ROM at all does not spin the disk. Maybe an inconsistency related to your observations of pin 20 / A12. I'd be interested to hear any further observations you have on that.

 

Cheers,

Ryan

[atariry]

Hi @nezgar (Not sure if this is the correct mark-up!), many thanks for your comments. It is always nice to read that other people are interest in my postings.

 

The 2532 EPROM does work with the JP1/JP3 jumper settings. Thanks for pointing out how to use the more common 2732 EPROM. Somewhere I should have a box full of these EPROMs, but I cannot find it. I have other boxes with EPROMs of other sizes which I know exactly where they are, but 2 other boxes have gone astray. Annoyingly it means that I don't have any 2732s to experiment with. When I searched on Ebay I was shocked but not surprised at the price of 2532s, but just shocked at the asking prices for 2732s.

 

I'll be trying the "NOP" EPROM in the PCB from drive #1, and it will be interesting to see if the voltage level on the program pin is the same as on the other PCB.

[atariry]

So to continue the story... Using the 2532 "NOP" EPROM, with the reset injected externally from the Teensy3.5, everything checked with the scope looked OK. So I put a NE555 in the new socket to restore the power on reset circuit. I also put all the original chips back in, including the ROM. I didn't connect up the drive mech. Doing more checks, the power on reset was OK and clocks were OK too. This time the address bus was changing and there was a pattern suggesting that the CPU was executing some kind of loop. This was more like it, but I couldn't understand why previously in this configuration the 6507 was dead. I removed a known working Tandon mech from another Ebay 1050 purchase, however I was happy that this drive was fully working. I connected up the known good mech to the PCB from drive #2 (the one on which I had done all the work). I switched on and I saw the drive head seek and spin up and motor off. I was happy but couldn't understand why it was now working, the only major change was that the NE555 was now socketed and had a new part, however I did check the original power on circuit.

 

Now I wanted to do more checks, but there is no room on my bench for my Atari 800 and the 130XE was in pieces due to some video experiments. So I now started to write some code to make the Teensy3.5 send command packets to the drive. The protocol is simply 19200 baud "RS232" at TTL levels (which isn't really called RS232!). I connected up DATAIN and DATAOUT and had the usual serial comms confusion of which 1050 SIO pin connected to the TX and RX of the Teensy3.5. I also connected up the COMMAND SIO pin to a spare Teensy I/O. I coded the sequence to do the GET STATUS command, but (to cut a long story short) it didn't work. It was a bit of a guess, but I connected up the +5V/READY SIO pin. This did the trick and the GET STATUS command now worked. I chose this command as it does not need the mech to be connected. Some time later I did look at the commented source code of the 1050 firmware and saw where the check for READY high is done. Now I do know that the Atari 400/800 generates a clock signal which can be used by the "listener" to know when to sample each bit (I used this technique many years ago using a PIC micro) but the 1050 doesn't need or care about the serial clock and in fact the schematics (thank you Jerzy Sobola) show that this SIO pin is not connected (apart from to the other SIO connector, as a pass through).

 

I could now code up more SIO commands, particularly PUT, READ and FORMAT, When I tried the FORMAT command I saw the head seek to track 0, step from track 0 to 39, but then generated an error. Occassionally it would start to step back to track 0, but would still stop and generate an error. So now was time to get out my old original 1050 drives, both upgraded to Happy status. Well one didn't work and the other couldn't format. WTF! After much messing around I managed to get one of the drive to format a floppy. I am now suspicious of the quality of some of my floppies which haven't seen the light of day for over 30 years. I found some spare floppies and tried the repair drive again. Finally I was able to get a floppy to format. I completed coding of READ and PUT and using the Arduino serial monitor was able to play around and check that the 1050 was working. Nevertheless there were intermittant fails. Now I was using a non Atari PSU, so I switched to a genuine Atari PSU. This did appear to help (but I am not 100% certain).

 

The next evening I setup to do more work. I plugged in the Atari PSU and powered up the drive. Nothing! I toggled the power switch. Still nothing. WTF! The drive was completely dead again. So I removed the mech. The connectors are so tight and close together it is a PITA. I scoped various pins and to my surprise saw the 5V power supply going to 5V and then to 0V at a low frequency. It was a bit of a guess but I put my finger on top of each of the tall capacitors and wobbled them. This appeared to fix the 5V. I now removed the PCB from the 1050 case bottom and examined the solder joints of these capacitors. I quickly found a cracked joint on one of the pins. Now remember this was the drive that got a massive thwack in the post. I re-soldered the joint, connected up the mech and the drive now worked again (and has done ever since).

 

The Arduino software could easily be ported to an Uno or Nano. If anyone asks I will tidy it up and send a copy. One bad habit I have is never finishing anything off .

[E474]

Hi,

 

You might want to check out the drive with the Atari 1050 self test/SALT program, it's described in the 1050 Field Service Manual, and is good for checking the RPM and doing burn-in tests, etc.

[xrbrevin]

gratz!

so it was due to a heavy cap jolt in transit - thats why they add that adhesive glap in the factory. was quite a significant impact!...

run the attached software to calibrate the RPM and zero sensor

 

 

1050 Disk Diagnostics.atr

[+Nezgar]

Hi @nezgar (Not sure if this is the correct mark-up!), many thanks for your comments. It is always nice to read that other people are interest in my postings.

 

The 2532 EPROM does work with the JP1/JP3 jumper settings. Thanks for pointing out how to use the more common 2732 EPROM. Somewhere I should have a box full of these EPROMs, but I cannot find it. I have other boxes with EPROMs of other sizes which I know exactly where they are, but 2 other boxes have gone astray. Annoyingly it means that I don't have any 2732s to experiment with. When I searched on Ebay I was shocked but not surprised at the price of 2532s, but just shocked at the asking prices for 2732s.

 

I'll be trying the "NOP" EPROM in the PCB from drive #1, and it will be interesting to see if the voltage level on the program pin is the same as on the other PCB.

I tried again tonight to use a 2532 EPROM with the Stock Rev L firmware unsuccessfully in a drive configured for JP1/JP3. The drive spins, which is more than it does with a blank EPROM. I then burned the US Doubler Rev L code to another 2532, and installed the requisite doubled-RAM, and it also doesn't initialize. Both EPROM's still verify 100% on the programmer.

 

What 2532 part #'s are you using successfully? Mine are TMS2532A-45JL - meaning 450ns, which I'm thinking might be a factor here... is 450ns too slow of an access time for the 1050? A datasheet shows alternate part #'s with 250, 300 & 350ns...

[1050]

My experience is 200ns minimum for OS chips but the

black box does have some 450ns eproms and I'm willing

to suggest that this one is a particular exception.

 

One thing to note is that no matter the speed, a PC

based eprom programmer will read it and it looks like

solid data. On the Atari though solid is down to timing

and they get fuzzy near the edges. Game cartridge needs

150ns or faster. A 150ns OS may or may not work and it

might even depend on if it's tuesday too.

 

Exactly why, I wish I knew. But it does appear to me

that OS chips have a sweet spot and it's between 200ns

and 450ns. Since you are at the fuzzy slow edge I would

try a faster one.

 

I can't play, so I do appreciate reading about your

attempts and findings Nezgar. Thanks for pulling your

hair out so I don't have to.

[+Nezgar]

Today I received some more 2532 EPROM's from China in record time! I ordered on Jan 27, so an amazing 10 days compared to the 1-1.5 months I'm used to.

 

These ones are TMS2532A-25JL, so should be now 250ns instead of 450ns access time. I programmed 1 with US Doubler, and no action. (Just like the previous 450ns version) Then I programmed the stock 1050 Rev L firmware into another, and I get the drive spinning only (again, just like I saw previously with the 450ns version).

 

So, I guess it's not the speed of the chips. Maybe it is a pinout or brand difference after all, which is why I'm still curious what type of 2532 EPROM the OP, atariry, has had success with.

[1050]

Sorry to hear the bad news.

 

Verify TMS2532A is in the devices list. They do require

25 volts and sometimes 21 even with spikes won't get it done.

And then some eprom programmer manufacturers don't go to

excessive amounts of verification in their data list - errant

entries do sneak in.

 

The cleaner and most recent 1050 Service manual offered

does show without a doubt that the PROM used was a 2332

and that's not a 2532 all day long.

Although the datasheet for 2332 states they are pin

compatible with 2732 they don't say they are functionally

compatible with 2732 meaning the output enable is backwards.

You can see this oddity in the 1050 service manual where

the eprom is fed an inverted A12 signal for a negative chip

select on it's pin 18. But the PROM gets a positive A12

to it's pin 21 for chip enable.

 

 

This is a maskable PROM. Pin 20 and 21 are clearly shown

as being possibly either positive or negative enable.

Atari used this PROM for their game cartridges with pin

20 as a negative enable. But pin 21 could be either flavor,

when it was negative it was the low region data rom and when

it was positive enable, it was the high region rom in

Atari's two chip game cartridges. And that makes it

quite difficult to replicate quickly and easily in

eprom based boards since eproms tend to be always

negative enabled.

 

So you can cheat by borrowing the negative chip select

meant for the eprom from pin 6 of U16 LS04 and routing

to pin 21 of the 2532 via some creative jumper betwixt JP3

and JP4. You'll have to work out the exact details, but

it might be worth a shot? But since you have to use a

custom jumper scheme anyway to have hard to find 2532

eproms on board, I'm tending to think why bother, 2732

works and has documented jumpers too.

 

And then there is this I found while gathering yet more

wool from the internet sheep.

http://forum.6502.org/viewtopic.php?f=3&t=4242

 

It seems to show that 300ns works which would really be

a kick in the head after you got 250 with same results.

But he seems to have solved his own problem just the same.

Maybe his 2332 had both chip enables as negative?

 

As near as I can tell, 2532 is not the eprom to use here

to mimick this particular 2332 PROM and never was good

advice. People who say that forgot about the betwixt

JP3 and JP4 jumper they had to do? I can't know what

the problem is when that advice is given, but it seems

wrong to me.

[Mclaneinc]

I love that story Atariry, utter respect for the dedication and creativity there....Really interesting how you approached it and relayed the story..

 

Glad to hear you got a drive going...

 

As for the sellers just lightly wrapping the drives, shame on you, I brought one drive a while back off Brenski on this forum and it arrived in pristine condition and he had packaged it perfectly, even had a disk in the drive for the head...That's how it always should be but Ebayers etc can be the most lazy of folk if not downright dishonest...Thanks to the people who do it right!

[talanthalus]

I love that story Atariry, utter respect for the dedication and creativity there....Really interesting how you approached it and relayed the story..

 

Glad to hear you got a drive going...

 

As for the sellers just lightly wrapping the drives, shame on you, I brought one drive a while back off Brenski on this forum and it arrived in pristine condition and he had packaged it perfectly, even had a disk in the drive for the head...That's how it always should be but Ebayers etc can be the most lazy of folk if not downright dishonest...Thanks to the people who do it right!

My 8" drives came in recently wrapped in about two layers of bubble wrap and laid directly in the bottom of the box. Next to them was the power supply, wrapped the same way. On top, there was some busted up styrofoam. On the sides and between these heavy pieces was only the bubble wrap, so needless to say, I have drive parts. Nothing in there to protect the heads either. I can imagine not everyone has an 8" disk lying around for shipping, but with a box that big and heavy, one would at least expect it all to be packed as if they cared for it to arrive in one piece on the other side. I'm guessing these sellers must make their money on the insurance claim, because otherwise, what is the point in all of the effort if there is no care if the item arrives in working condition?

[_The Doctor__]

when will they learn for 8 inch drives you must use Styrofoam sheets and foam pads all packed tightly with card board around the drives themselves then the foam pads then the Styrofoam then the foam pad again all in a large tightly packed box, or leave them in their metal case?

I so don't understand why they resist proper packing, as they lose way more money when it arrives in pieces!

[+Stephen]

when will they learn for 8 inch drives you must use Styrofoam sheets and foam pads all packed tightly with card board around the drives themselves then the foam pads then the Styrofoam then the foam pad again all in a large tightly packed box, or leave them in their metal case?

I so don't understand why they resist proper packing, as they lose way more money when it arrives in pieces!

Never under estimate the sheer stupidity of the general public?

[+Nezgar]

So you can cheat by borrowing the negative chip select

meant for the eprom from pin 6 of U16 LS04 and routing

to pin 21 of the 2532 via some creative jumper betwixt JP3

and JP4. You'll have to work out the exact details, but

it might be worth a shot? But since you have to use a

custom jumper scheme anyway to have hard to find 2532

eproms on board, I'm tending to think why bother, 2732

works and has documented jumpers too.

 

1050, I really appreciate your 'knowledge dump' here! Read it many times I tinkered away at it again today, mapped out a bit using the DMM and took my logic analyzer out of the packaging for the first time. The schematics are really hard for me to understand...

 

My observations:

 

JP1 connects CPU pin 16 (A11) to 2332/2532 ROM pin 18 (A11)

JP2 connects CPU pin 16 (A11) to 2732 EPROM pin 21 (A11)

JP3 connects CPU pin 17 (A12) to 2332/2532 ROM pin 21 (CS2)

JP4 connects (not sure CPU A11? DMM cont. test fail, maybe because the NOT gate in the schematic?) to 2732 EPROM pin 18 (A11)

 

looking with the logic probe with the original 2332 in place:

2332 pin 20 (CS1) is a solid LOW

2332 pin 21 (CS2) has both low/high signals/frequencies, changes depending on drive behaviour (door closed, motor stop, etc)

 

with the 2532:

2532 pin 20 (PD/PGM) is a solid LOW

2532 pin 21 (Vpp) shows mostly high, +neg with some activity at power-on (locked up/inoperable state)

 

I put the 2332 mask ROM into an Atari 8K cart, and dumped it with DOS, and I can see that the LOW 4K is all FF, and the HIGH 4K has the 1050 ROM data.

With the 2532 EPROM in the Atari 8K cart, it dumps as 2 repeats of the same 4K. (which also confirms the EPROM reads OK in more than just my programmer!)

 

Looking at the memory map of the 1050 in the Altirra hardware reference manual, I see the ROM is mapped to the top $1000-$1FFF, everything else is in the bottom $0000-0FFF.... Seems to support that the 1050 uses the "A12" line to turn the ROM on and off, but the 2532 EPROM is ignoring this, and staying on all the time, conflicting with the RAM.

 

Maybe that A12/chip select/negative chip select needs to be mapped to the 2532's OE pin 20 instead.

[1050]

1050, I really appreciate your 'knowledge dump' here!

Almost honored to help when I can. Your projects and

interests are best information and in general really

good stuff.

 

My observations:

Are a bit confused at best..

many false conclusions, the PROM and EPROM are fed different

signals with each jumper. The only one that doesn't change

is pin 20 tied to ground via R83 for both.

 

I put the 2332 mask ROM into an Atari 8K cart, and dumped it with DOS, and I can see that the LOW 4K is all FF, and the HIGH 4K has the 1050 ROM data.

With the 2532 EPROM in the Atari 8K cart, it dumps as 2 repeats of the same 4K. (which also confirms the EPROM reads OK in more than just my programmer!)

If the 2532 was working as it should, you should have

had the reverse situation where the low area read good

data and the high was all FF. In agreement with your

next to last statement that this data is clobbering RAM

because the 2532 isn't shutting up to let others talk

when that's required. Conclusion is your last statement

is worth trying even when that means more goobering with

lifting leg of R83 and supplying pin 20 with output from

pin 6 U16.

JP1 yes

JP2, JP3, JP4 no - open

You should tie pin 20 to pin 21 and run them both from

pin 6 U16.

 

Texas Instruments was found to be lacking with their

TMS2716 offering in a similar manner that totally ticked

off everybody. Result is no TMS2716 exist nor the datasheet

from TI. And TI reissued a suitable eprom as the TMS2516,

perhaps starting the entire line of 25xx morphed 27xx's?

I dunno. A bit before my time even though I was alive,

I wasn't 'plugged in'.

But I do know that often the datasheet has OE for CS and

the other way round and good things can happen when we swap

them despite what the datasheet says each one does and why.

Supposedly OE is faster, but when it doesn't work at all then

we are forced to goober. Depending on the exact end point for

pins 20 and 21 on the game cartridge, I'm wondering if a solder

blob on the 2532 across those pins wouldn't restore the missing

functionality there as well? Its a very commonly done thing to

tie both OE and CS lines to each other especially as third

party projects. Even in these datasheets we don't really

get a good sense of which is which really, they call them

all manner of things so standardization wasn't even a thing

yet in this regard. Later on, the nomenclature becomes quite

clearly adhered to as well.

 

In researching this I came across a board produced by

ELEKTOR. It's a 2532 to 2732 converter. 2532 goes in the

left side or 2732 goes in the right side. And the other

side gets pins so it can be used either way depending

on which side gets the pins or chips. Don't stare at it

too long, just glance at it from time to time is how I

manage those things. It can be crazy making otherwise.

 

[_The Doctor__]

simplicity

[+Nezgar]

Conclusion is your last statement

is worth trying even when that means more goobering with

lifting leg of R83 and supplying pin 20 with output from

pin 6 U16.

JP1 yes

JP2, JP3, JP4 no - open

You should tie pin 20 to pin 21 and run them both from

pin 6 U16.

 

1050, big thanks for your fact checking of my writings along the way. I'm happy to report this was the solution! I'm not a fan of soldering directly to the pins of chips, so I used an intermediary socket. I also didn't want to solder directly to pin 6 of U16, so I started poking around with the DMM continuity test and was happy to find that this pin goes to the top of JP4, which was a mystery to me where it went before! So it just means that the wire from pin 20+21 just needs to go into the top of JP4.

 

Here's pics of my quick hack that confirmed it works:

 

 

In researching this I came across a board produced by

ELEKTOR. It's a 2532 to 2732 converter. 2532 goes in the

left side or 2732 goes in the right side. And the other

side gets pins so it can be used either way depending

on which side gets the pins or chips. Don't stare at it

too long, just glance at it from time to time is how I

manage those things. It can be crazy making otherwise.

I really should have kept most of these posts over in my original topic where I started pondering this 2532 challenge. I actually found and acquired a 2532->2732 adapter similar to what you mentioned, and that's how I successfully programmed these 2532's with my TL866. Even though the datasheet programming voltage is 25V, it actually works with the TL866's maximum output of 21V. (There are other pinball and commodore forums I previously read about that working)

 

http://atariage.com/forums/topic/285971-2532-eprom-uses-adapter-for-tl866/

 

I also acquired some smaller adapter PCB's from a fellow named B. Brzezicki that make a 2732 work as a 2532, and had the exact same symptoms now that I think about it in the 1050, and when inserted into a standard Atari cart PCB, which has to be the locked low OE pin.

 

Anyhow... it was interesting, at least the 2532's are 'possible' to be used in a 1050, but not as a drop-in replacement for a mask ROM like I wondered. 2532's can also make 4KB mask ROMs in stock Atari cart PCB's.... (repeated twice)

[+Nezgar]

Oh, and I just thought to try my previously programmed 450ns 2532's, and I found they work just fine in the 1050 with this adapter as well. One programmed with stock 1050 Rev L, the other with ICD US Doubler.

 

[1050]

Nice wrap up and conformation of the issue. Bad datasheet

misspeak about OE and CS crisscrossed function - AGAIN.

 

Even Atari didn't have the full scoop on it or they

never would have tied OE to grounded pin permanent.

 

The struggle is real. Makes victory even sweeter.

Glad I could help.

[+kheller2]

Scratches head..

 

Isn't the 2332/2532 the same pinout as the masked ROM, so you would leave the four jumpers set to ROM: Only short JP1 and JP3.

2732 requires the flipping of two lines, hence the jumpers: only short JP2 and JP4