1090XL remake

[ivop]

52 minutes ago, reifsnyderb said:

Sorry.  I was thinking the 1090xl main board.  I was thinking that we need 1090xl main boards with PBI cables, power supply components, edge connectors, etc.  Then we can use those to make other cards.  I am seriously thinking about ordering the boards and parts myself so as to test my memory card.    ?

Sorry, too. I thought you meant development boards for plug-in cards 

 

[Mathy]

Hello guys

 

A) What do we need /HALT for on the 1090XL?

 

B) If we go from 7805 and 7812 to TRACOs, would that reduce the size of or the need for the big capacitors?

 

Sincerely

 

Mathy

 

[_The Doctor__]

A/ ANTIC banking and other support

b/you may reduce but then add to get some filtration done... they are all kinds of noisy.

[+kheller2]

3 hours ago, reifsnyderb said:

I was able to get to the atarimuseum.com 1090 schematics using http://web.archive.org 

 

https://web.archive.org/web/20210403131120/http://www.atarimuseum.com/computers/8BITS/XL/xlperipherals/1090xl.html

The 1066 didn't use the +5 from the Atari through pin 47 as per the schematics.

The 1064 does use pins 47 and 48 as per the schematics.

 

The KiCad files I used were from Ken's github and they match the schematics at atarimuseum.com (via web.archive.org)

 

While I'd love to have a confirmation from someone with the actual hardware, I think there is enough information to go with this.  We'll just need some development information available for anyone who wants to make cards for the 1090xl.

 

Well, first we need 1090xl development boards....

 

 

 

 

 

Even stranger, the 1090 schematics show that normally NC pin 37 as CardSelect and only 39 and 49 as Reserved (implying +5 600XL pin 48 isn't reserved?   But I don't see it on the schematic either. ) Interesting

 

[kenames99]

6 hours ago, reifsnyderb said:

Sorry.  I was thinking the 1090xl main board.  I was thinking that we need 1090xl main boards with PBI cables, power supply components, edge connectors, etc.  Then we can use those to make other cards.  I am seriously thinking about ordering the boards and parts myself so as to test my memory card.    ?

 

Make sure to get the latest board, it has some needed corrections. Also make sure the gerbers are current. I think i uploaded latest ones but not sure.

 

Ken

 

[reifsnyderb]

4 minutes ago, kenames99 said:

Make sure to get the latest board, it has some needed corrections. Also make sure the gerbers are current. I think i uploaded latest ones but not sure.

 

Ken

 

I just looked and I have the one from 12/28.  I'll check the newest revision.  Thanks!

[kenames99]

hi,

you might want to think a bit more. the halt signal actually opens up the door for bidirectional address bus with dma and more. the address buffer chips would need to be replaced with other buffer chips like the 74LS245 or '645 chips. this was something I was going to address in the future.

 

Ken

 

[kenames99]

On 1/8/2022 at 2:12 PM, kheller2 said:

 

Even stranger, the 1090 schematics show that normally NC pin 37 as CardSelect and only 39 and 49 as Reserved (implying +5 600XL pin 48 isn't reserved?   But I don't see it on the schematic either. ) Interesting

 

kheller2,

  hmm, the original schematic should be in my github. I will look.

 

Ken

 

[evilmoo]

On 1/6/2022 at 1:46 PM, kheller2 said:

I'm curious if they could be used with the 14xx Voice, Speech or floppy as they are parallel devices.

I think that might be part of the reason the 1090XL only had five slots.  If there were up to three devices on board the system (like there would be on a 1450XLD), then this would avoid conflicts.

[reifsnyderb]

12 hours ago, kenames99 said:

hi,

you might want to think a bit more. the halt signal actually opens up the door for bidirectional address bus with dma and more. the address buffer chips would need to be replaced with other buffer chips like the 74LS245 or '645 chips. this was something I was going to address in the future.

 

Ken

 

I was thinking of only using the halt signal for Antic banking.  But if there are other options, then, yes, bidirectional buffers would be needed.

[+kheller2]

On 11/23/2021 at 7:14 PM, kenames99 said:

Hi kheller2,

  No I do not, would appreciate a copy. Thank you.

 

Ken

 

1090RAMTEST.BIN Dump from a 2764 1090 RAM CARD Tester Rev 1.0

[kenames99]

kheller2,

  thank you! I will put it to good use.

 

Ken

 

[reifsnyderb]

With the exception of the power switch and LED's, it seems to be ok so far.  (I forgot to order the LED's and digikey didn't have the switch.)  The test points register close to +12, -12, +5, and the unregulated +10 is really about +13.5...which didn't surprise me and really doesn't matter as each card needs it's own voltage regulator anyhow.  I thought I'd need a special power supply until I realized, while ordering the parts, the 1090 power supply contains a voltage multiplier circuit.

 

Next up is to get the parts for the 2MB memory board and test this out.  Locating the parts for the 4MB board is harder due to this ? chip shortage.

 

If I get another batch of boards made I'll have to have the component values put on the board for quicker assembly.  Also, It's nice they have 2 options for buffer chips as Digikey didn't have the other chip.

 

Edited February 4, 2022 by reifsnyderb

[kenames99]

wow, that looks really nice. good work. I used a regularly stoched power switch, both digikey and mouser had stock. but these are lean times for parts. I look forward to seeing you other card there assembled and working.

 

Ken

 

[reifsnyderb]

53 minutes ago, kenames99 said:

wow, that looks really nice. good work. I used a regularly stoched power switch, both digikey and mouser had stock. but these are lean times for parts. I look forward to seeing you other card there assembled and working.

 

Ken

 

Thanks!  I am planning to get enough parts for two 2 MB boards and going for a 4 MB Atari 800XL.  Though, I'll try it on my 600 XLM prototype board first.  If I blow something up on that I won't be as upset.  (From some prior testing the parallel port is pretty resilient as I had something else I am working on connected backwards.  lol    )

 

It would be pretty cool to have an SD card 1090 card, too.

 

Edited February 5, 2022 by reifsnyderb

[kenames99]

give me time man, just give me some time.

 

btw, I got some video out of the 80 column board yesterday. that made me happy too.

 

Ken

 

[reifsnyderb]

Just now, kenames99 said:

give me time man, just give me some time.

 

btw, I got some video out of the 80 column board yesterday. that made me happy too.

 

Ken

 

Well, I have something I am working on to connect a Teensy 4.1 to the PBI.  The first attempt was a flop.  So, I am working on it from another angle.....

 

 

 

 

[reifsnyderb]

4 minutes ago, kenames99 said:

give me time man, just give me some time.

 

btw, I got some video out of the 80 column board yesterday. that made me happy too.

 

Ken

 

This is great!  How soon until we see some pics of the 80 column board video?

[reifsnyderb]

Parts ordered for two 2MB memory cards.

[reifsnyderb]

There is a really good chance I'll have to re-design the memory cards...and not because they have a faulty design.  They are designed for 3.3 vdc Atmel ATF1504 CPLD's with a 100 pin TQFP package.  Since some of the Atmel software isn't working with ATF1504's, I've been using a program called "Digital" that allows me to design and test the CPLD before sending the result to part of Atmel's software package that does work to "fit" the design to the CPLD.  The CPLD logic works perfectly but I can't get it to "compile".  I just found out over the weekend, that Digital (which works beautifully otherwise) has a hard limit of 12 outputs.  I contacted the developer and found a way to override this...however, the "compile time", so to speak, goes up exponentially.  So far, it's been running over 3 hours.  Presently the software has consumed over 9.2GB of RAM doing it's calculations, and as time goes by I am less assured that this will work.

 

I have the boards, the parts are coming, and now this is held up by a software problem.

 

The other option is to switch to Xilinx CPLD's.  I was able to get the software working over the weekend...which takes up a huge amount of space and runs on a VM...go figure.

 

Add in another learning curve, 3 more weeks to modify, order, and take delivery of, assemble, and test the boards.

 

Just when I thought I'd have something good going by the end of the week.     ???

 

 

 

 

[kenames99]

you could try wincupl, it is noow free from atmel/microchip.

 

Ken

 

[reifsnyderb]

10 minutes ago, kenames99 said:

you could try wincupl, it is noow free from atmel/microchip.

 

Ken

 

Wincupl is the one that doesn't work.  It doesn't create the one file needed for ATMISP.  I discovered this the hard way then found Digital can create the file for ATMISP.  

 

Edit to add:  Wincupl isn't creating the JED file.  I have no idea why not.  The logs don't explain it and there isn't any applicable troubleshooting info out there.

Edited February 7, 2022 by reifsnyderb

[reifsnyderb]

I am changing the boards to use the Xilinx CPLD where a 100 pin CLPD is needed as I am out of ideas for a "quick" resolution".   ?

 

Edited February 7, 2022 by reifsnyderb

[+mytek]

Yeah I avoid designs that require CPLDs or FPGAs because the tools at least for me, have quite a learning curve associated with them, and don't always play well with Linux which is my preferred OS. It also becomes problematic when releasing DIY stuff to expect the user to do what's required to program this stuff. Yes I know you get a tremendous chip count reduction, but its just not for me. Now on the other hand my designs do like to use PIC processors, but thanks to my work with @dmsc we have a very simple and painless way to flash those with the Atari itself

 

With all that said, I did diverge in the 576NUC+ and used a couple of simpler PLDs for the MMU and EMMU, but the work had already been done by others and proven out, so it was a walk in the park for me to re-use that. And since a TL866 programmer was required to flash the custom OS ROM anyway, adding the PLDs to that list was not really seen as a big issue (same programmer and software used). Also in that project, the footprint being so small really restricted the use of a discrete off the shelf 74xxx series IC chip solution for the extended memory.

 

[reifsnyderb]

I think if there were better development tools CPLD's would really take off.  Programming a chip with dozens of gates and flip-flops is amazing.  Fixing a logic mistake in a piece of hardware with a slight code change is almost like magic.  Sadly it seems to be something of a black art.  If something like Digital would be used to program the chips and be able to connect to the chip via a USB dongle the process would be almost painless.  (For the most part, Digital is very well designed.)  The Xilinx development package looks rather painful to learn.  I'll figure it out but I am really steamed to have a working process then run into this problem.  Literally days away from having a 4MB stock Atari 800XL with a 1090 expansion connected and this happens.