avatar_llopis

Are Version 3 Gate Arrays particularly fault-prone?

Started by llopis, 22:10, 12 July 19

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

llopis

Is it my bad luck, or are version 3 GA (the ASIC ones) really likely to die? I've only encountered a few version 3 boards, and all of them had faults I could only trace back to a faulty GA. In one case, the GA wasn't generating any video signals. In the other case, it looks like reading from RAM was busted (but ROM was fine).


If you add to that my bad luck with other ASIC chips like in the Amstrad Plus or the Spectrum +2 line... I'm hating those chips! :-b


Do you encounter many problems there? Any tips beyond reflowing the connections just in case? Or just chuck them in the parts drawer and move on?


gerald

Bad luck or not, I can't tell  ;) On the 5 I encountered only one unfixable : a data bit was stuck high at ASIC level on read from DRAM -> dead CPC.
A second one exhibited a timing problem that create artefacts on screen during read to the extended memory. That also happen on some Plus ASIC. The fix is to simply add a few Pf capa on the NCAS signal of the DRAM.
Others are fine up to now  :)
But statistically, that's a higher failure rate than 40010/40007/40008 + CRTC on other CPC.

dragon

Well they used different technology. The preasic is a sea of gates, and older gate array are classic gate array/2. Where the plus is a full custom asic.


Add that they have more gates, so they increment the heat operation.


Anyway all of them are new technology when amstrad use it.  So its beggining.

llopis

Quote from: gerald on 09:29, 13 July 19
A second one exhibited a timing problem that create artefacts on screen during read to the extended memory. That also happen on some Plus ASIC. The fix is to simply add a few Pf capa on the NCAS signal of the DRAM.
Do you have any info on that? Like how to diagnose that, and why the capacitor helped, or is that something you figure out by yourself?


I doubt this is my case here. It looks like ROM programs execute fine (memtest for example), but the memtest returns failed reads for every memory bit, whether I have good RAM ICs in place or nothing. Normally I would attribute that to the latch in the data bus, that functionality is inside the ASIC now :-( I did test that all the RAM data and address lines have correct continuity to the ASIC, so that's not it.


If you have any thoughts to further diagnose it, let me know, but I have a bad feeling about this. I don't imagine anyone has a working version 3 ASIC (with or without board) for sale...

gerald

Quote from: llopis on 14:31, 14 July 19
Do you have any info on that? Like how to diagnose that, and why the capacitor helped, or is that something you figure out by yourself?
Something I found sevelopping a RAM extension : http://www.cpcwiki.eu/forum/amstrad-cpc-hardware/cpc-ram-extension
Then, if my memory is good, @remax had the same issue on it's pre-asic 6128, which I fixed the same way.


Quote from: llopis on 14:31, 14 July 19
I doubt this is my case here. It looks like ROM programs execute fine (memtest for example), but the memtest returns failed reads for every memory bit, whether I have good RAM ICs in place or nothing. Normally I would attribute that to the latch in the data bus, that functionality is inside the ASIC now :-( I did test that all the RAM data and address lines have correct continuity to the ASIC, so that's not it.


If you have any thoughts to further diagnose it, let me know, but I have a bad feeling about this. I don't imagine anyone has a working version 3 ASIC (with or without board) for sale...
Only simultaneously looking at the RAM signals (data/address/cas/ras) and Z80 (addr/rd/wr/mreq) with a logic analyser could confirm.If you put socket for the RAM, makes sure everything is properly connected and nothing is shorted  ;)



remax

Quote from: gerald on 15:15, 14 July 19
Something I found sevelopping a RAM extension : http://www.cpcwiki.eu/forum/amstrad-cpc-hardware/cpc-ram-extension
Then, if my memory is good, @remax had the same issue on it's pre-asic 6128, which I fixed the same way.


Yes ;)
Brain Radioactivity

Bryce

The Pre-ASIC was a first run low volume part, there's a very good chance that they are less reliable than the later ASIC.

Bryce.

Powered by SMFPacks Menu Editor Mod