XEGS 128K problem

[xrbrevin]

I need some guidance from anyone with a greater understanding of the 130XE extended memory hardware

 

I have an XEGS and have done the +64k mod however, the system only 'sees' 32k of it. Its a super-reliable machine until a 128k demand is put upon it. Im getting mixed results from various software but in practice, its just not reliable.

 

the internal memory test shows all 4 banks test clean (i know its not a thorough test tho). the XE production test completes the RAM test without issue. Prince of Persia (128k only) boots up but fails part way into loading.

XRAM and shortest show only 32k extra but it tests ok on both. the SuperSALT RAM test shows various address failures but they are in different locations each time..!

 

so i have quadruple-checked the wiring and the step-by-step-guide. i even wrote my own mini-summary to make continuity testing quicker but i found nothing amiss.

 

i have cross-referenced the mod with the 130XE memory hardware schematic and it checks out.

i swapped the DRAMs so that all 4 are now identical. I tested these ICs first in my 600XL and they are fine.

i swapped the wires over on FREDDIE pins 23 and 24 but XRAM showed the same 2 banks present - i did expect it to show the other 2. Maybe this characteristic is useful for diagnosis?

 

so im thinking maybe the FREDDIE has an issue with it's extended RAM operations or maybe the EMMU but i dont have any spares - only in working machines. i do have a logic probe and 'scope but dont dig them out that often.

 

heres some pics, thanks in advance

 

p.s. i have the 4x resistors fitted under the PIA as per the guide

 

 

 

 

[1050]

Just an early report, but it seems yellow and green from Freddie are backwards at one end or the other. Higher pin count on one chip is the higher pin count on the other and yours has it backwards (high for low)?

 

And also not sure this would make any difference as I haven't looked at that aspect yet at all. Just checking the wiring so far...

 

Where is project main page at? I must have missed this one as I'm going by the photo above and a 130XE sams schematic so far.

 

Confused, never mind.

Edited June 11, 2022 by 1050
Wait, there's more

[xrbrevin]

Hi,

thanks for the reply and well spotted, after i swapped the green & yellow i tested it and got no change so ive left them there for now.

i thought this would be akin to swapping the 2x 'DRAM bank' resistors on a 16-chip 130XE but i may be wrong.

here is the info source:

 

xegs128k.zip

[1050]

Thanks for posting that.

Well I did miss that one somehow and it's a pleasant surprise to see Matt the rat connected to it from way back in the BBS days. No doubt this one was popular in certain circles. USA high school wrestlers are known by the moniker mat rats and Matt was one of us so he went by the handle Matt the rat. Doesn't sound like it, but it's coming from pure respect in the first place and second, this guy earned his high ranking on the Atari side of things too.

 

So nothing else is wrongly wired that I can pick up on at the moment.

 

It appears that there is no official XEGS version schematic either Atari or Sams which is a pity. It means we are stuck with double guessing on the XEGS. There is somewhat faulty engineering done to the 130XE as to port B pull up resistors (the four on the underside of the PIA(and elsewhere)). They are NOT needed on any of the XLs so why did Atari use them on the XE? Dunno, but in my book it's stupid to use them and just not needed. On par with the SIO "filter" caps that just get cut off all the time. The PIA has pull up resistors inside it on port B, the chip itself has provisions to power on in the high state and then the Atari OS programs port B to also be all high on top of that during boot up. There are no issues to fix here by doing that, so why they did it is beyond me. But my gut feeling is it's way too much, so I would pull those four resistors underneath and ignore the "fix" Atari has done to port B that just isn't needed at all. It is possible that soldering flux in that area might be doing things with those signal levels, so if nothing else please use a toothbrush and some strong IPA on that solder work underneath. The picture provide by Matt's team looks pretty rough in that exact area, which is why I would try it with removed resistors there and all cleaned up with IPA.

 

 

[Mathy]

Hello @1050

 

6 hours ago, 1050 said:

It appears that there is no official XEGS version schematic either Atari or Sams which is a pity.

 

Are you talking about the official schematic from Atari for the XEGS?  It wasn't there five minutes ago, but you might wanna take a look here.  (Sorry, two separate pages.  As soon as I find out how to fix that...)

 

Sincerely

 

Mathy

 

[1050]

Hi Mathy,

 

Yeah, that'll work, thanks. Don't worry about the page layout too much, it's a common thing to need a table top sized document when reading the fine print on these anyway. Not as bad as the 130XE version, but I can only read it in places where I already know what it says, they didn't try very hard did they?

 

Where you been keeping this one at? Is there a story about how you came to be keeping it to yourself all this time?

[xrbrevin]

I think the next course of action is to substitute the EMMU for a GAL16V8 & 74LS95 arrangement. The XEGS doesnt have a location for the 'LS95 so i have some spaghetti-wrangling to do, after i reconcile the XE schematic variations!...

an untidy photo will follow but hopefully also a functional RAM mod ?

Edited June 12, 2022 by xrbrevin
syntax error at line 10

[+kheller2]

On 6/12/2022 at 7:01 AM, xrbrevin said:

I think the next course of action is to substitute the EMMU for a GAL16V8 & 74LS95 arrangement. The XEGS doesnt have a location for the 'LS95 so i have some spaghetti-wrangling to do, after i reconcile the XE schematic variations!...

an untidy photo will follow but hopefully also a functional RAM mod ?

Did you ever get this to work?  I would think a 128K XEGS would be something that has been done before by folks, something I might want to do.

[xrbrevin]

its a tried and tested mod - attached is the info

thus far i have surmised the EMMU is at fault in my installation. i dont fancy trying the GAL16/74LS variant now, instead i am seeking a real EMMU to keep it minimal inside.

altho only this morning i found a github page that uses a GAL22V10 and no TTL chip so when ive finished sorting 2x CPCs and an electron, i'll get back on it.

 

github EMMU link

 

 

xegs128k.zip

[Mathy]

Hello guys

 

6 hours ago, xrbrevin said:

altho only this morning i found a github page that uses a GAL22V10 and no TTL chip so when ive finished sorting 2x CPCs and an electron, i'll get back on it.

 

github EMMU link

 

The chip is called CO25953, not C25853 as it's repeatedly called on github.  And the info they found on the internet probably comes from my site.

 

Sincerely

 

Mathy