HSB240 coprocessor 68881 info needed.

[guus.assmann]

Hello,

My HSB240 has two empty sockets, one for an 68881 and one for the decoder this needs.
So I would like to fill the sockets and already have the 68881.
All I need now is the decoder Gal or the Jedec file for this.
Does anybody have this available?

BR/
Guus Assmann

[frank.lukas]

Try to find Mr. Neumann or Mr. Heyer and contact them ...

[Greenious]

Actually adding a FPU is not very difficult, if you can trace what signals goes to the decoder GAL, it should not be very hard reverse engineering it.

[guus.assmann]

Well, I got a nice mail from mr Heyer. But he had no info.
So I started myself.

First, there's an 74F20 that has it's output(6) to pin 2 of the GAL.
Inputs on this are A23, A22, A21 and A20 on pins 1,2,4,5 respectively. (Other half F20 not used)
The Gal has it's pin 19 connected to the 68881 CS, with a 10K pull-up.
Gal pins:
1 = A5 : 2 = 74F20-out : 3 = A6 : 4 = A19 : 5 = A18 : 6 = A17 : 7 = A16 : 8 = A7 : 9 = A8\ : 10 = GND
11 = A9 : 12 = NC : 13 = A10 : 14 = A4 : 15 = A12 : 16 = A13 : 17 = A14 : 18 = A15 : 19 = CS-68881 : 20 = VCC
So all I have to do now is enter the I/O into a Gal programming prog and add the equassion.
The latter is simple:
/CS = /74F20 & A19 & A18 & A17 & A16 & A15 & A14 & A13 & A12 & A11 & /A10 & A9 & /A8 & /A7 & A6 & /A5
But making 11 and 13 to 18 inputs may be hard. (Had a problem in the past)

I guess a 16V8 with 7nS will do the trick.
BR/
Guus Assmann

P.S. I will post the results when I'm done.

[guus.assmann]

After a lot of setbacks, I finally managed to get a Jedec file.
It's been made with Hatari emulator and the GABI software from Ralf Zimmermann.
The GAL does give the correct reaction on my testboard.
But in the HSB it doesn't work.
I may have destroyed the 68881, it's been in the socket 90 degrees counter clockwise.
Not sure if it can surrvive this. The Power Supply didn't start.
With the chip seated properly, it still doesn't work, but the Supply starts correctly and the computer runs just fine.
I hope I can find another 68881 or test it in another way.

DEC68881.zip

[guus.assmann]

Today, I successfully tested the HBS240 with an 68881 processor.
Gembench 3 and 4 do give the expected high speed results for "flooting point math"
Sysinfo doesn't see the processor.

Anyway, this test confirmes that I got the Gal right. I've used a GAL16V8 with 5nS.
And I guess the clock to the FPU is 16Mhz. Maybe I'll overclock it as well. Just because it can be done.
BR/
Guus

[JamesG]

Today, I successfully tested the HBS240 with an 68881 processor.
Gembench 3 and 4 do give the expected high speed results for "flooting point math"
Sysinfo doesn't see the processor.

Anyway, this test confirmes that I got the Gal right. I've used a GAL16V8 with 5nS.
And I guess the clock to the FPU is 16Mhz. Maybe I'll overclock it as well. Just because it can be done.
BR/
Guus

Hello Guss.

I am following in your footsteps to sort a GAL for my HBS240 FPU. I am finding mostly that the GAL16V8 chips aren't made anymore (I am in the UK), or at least GAL16V8 with 5nS. Have you tried any other spec of GAL chip that successfully worked please? Or does anyone else have an idea which GAL chip spec I could try? I know this is an old post but I only came back to the 16-bit world 3 years ago. Thanks in advance. James

[frank.lukas]

... you can get Lattice 16V8 Gals from Ebay (China). I have had good experiences with them but direct from China from Aliexpress often get fake chips.

[mpattonm]

I would strongly suggest otherwise. Its hit and miss in both cases, even with sellers you know and you bought from previously. These ICs often ship from some common warehouse and sellers have no control over what they ship. Save yourself from frustrasion, avoid getting an e-waste.
Mouser still has some stock of Microchip (Atmel) 16v8s: https://eu.mouser.com/c/semiconductors/ ... es/?q=16v8

edit: Its also worth mentioning, depending on what programmer you have, you might have trouble programming GAL16v8s too. There are many variants that vary in number of fuses and programming sequence, newer programmers might not be able to program them correctly. For example, very common TL866-II programs/verifies(!) them seemingly OK, throwing no error, but the result is silicon with all outputs in high-Z state. Same with 400+ EUR SP610, but it at least throws "Invalid ID" code, bogus by the way. Both of these list them as supported. My ancient pinMaster48 from now defunct company mqp, on the other hand, programs them just fine and produces valid PLD.
ATF16v8 can be programmed even on TL866-II AOK.

So it is a side note to take: the fact you have working simulation but failing product could also be the result of failed programming.

[JamesG]

... you can get Lattice 16V8 Gals from Ebay (China). I have had good experiences with them but direct from China from Aliexpress often get fake chips.

Thank you Frank for your advice again. I had been looking at Ali Express too!! Regards James

[JamesG]

I would strongly suggest otherwise. Its hit and miss in both cases, even with sellers you know and you bought from previously. These ICs often ship from some common warehouse and sellers have no control over what they ship. Save yourself from frustrasion, avoid getting an e-waste.
Mouser still has some stock of Microchip (Atmel) 16v8s: https://eu.mouser.com/c/semiconductors/ ... es/?q=16v8

edit: Its also worth mentioning, depending on what programmer you have, you might have trouble programming GAL16v8s too. There are many variants that vary in number of fuses and programming sequence, newer programmers might not be able to program them correctly. For example, very common TL866-II programs/verifies(!) them seemingly OK, throwing no error, but the result is silicon with all outputs in high-Z state. Same with 400+ EUR SP610, but it at least throws "Invalid ID" code, bogus by the way. Both of these list them as supported. My ancient pinMaster48 from now defunct company mqp, on the other hand, programs them just fine and produces valid PLD.
ATF16v8 can be programmed even on TL866-II AOK.

So it is a side note to take: the fact you have working simulation but failing product could also be the result of failed programming.

Thank you MPattonm for the great detail. This does help lots. The good news is, I don't have a 'modern' programmer yet, so I am looking at the TL866-II.
I do have a 1988 EP 16/512 which I hope works (hope I can get adapters for it), but as you may guess I am a beginner for chip burning. However, it's something I'm eager to learn. I figured if I can understand chips and what they do, I'm making progress on understanding the Atari ST series a lot more.
Putting in an order at Mouser. Any other info from anyone with experience or tips/recommendations would be great. James