STE V1.5 32MHz BOOSTER - STATUS

[exxos]

Been busy all day making up 8 more. Three of them are the red boards. I need to program the GAL's and solder them in yet, but currently the first red board is now on test.. So far so good..

IMG_3595.JPG (115.63 KiB) Viewed 3662 times

[PhilC]

Thumbs up for getting it done.

Thumbs down for setting off my OCD with the photo

[exxos]

So 2 boards work, the others don't

One board I am testing seems like before, runs at 8MHz fine, anything else, nope. I guess my bad for only testing the CPU's in the STE MB before soldering them onto the booster boards (again) . Though I never had this problem before with 100's of CPU's I have tested. They all worked fine in my V1 STE booster, but they simply wouldn't work at all if they were faulty, not even at 8mhz. So I binned those. But these current CPU's, it's like 90% of them just refuse to run faster than 8MHz. They are supposed to be 20MHz parts.. but again.. China crap I guess.

I will see if my V1 STE machine is still running and do some tests on this batch of CPU's. Will have to "retire" the 8MHz only CPU's as a new part in my store at a cheaper price. I guess at the end of the day, this is overclocking CPU's and "mileage may vary" but this is just ridiculous when they won't even run at 16MHz I can only assume these CPU's are really old, and actually more like 10MHz parts restamped as 20MHz parts or something.

[troed]

I can only assume these CPU's are really old, and actually more like 10MHz parts restamped as 20MHz parts or something.

Yeah restamping has become a huge issue when sourcing parts. I bought a bunch of ST 27C256 EPROMs a while back and even though all of them were clearly stamped as being M27C256 (ST) a bunch of them were id:d as being 27C256 by different manufacturers and some were 27C512 and even 27C128.

I could sort of understand it when it came to the 68060 since those are expensive, but 32KB EPROMs .. ??

[exxos]

Yeah, I have had 1MB ROMs which I know are TI parts, but stamped as ST parts. Same with GAL's, all stamped as 7ns parts, can be anything, generally 12ns or 25ns etc. Speed of the GAL isn't much important on this project, but obviously the CPU's are.

EDIT: Obviously the 060 CPU's have been a nightmare for me. Also the DRAM I used for my 4MB kits, I went though a span where they were all 1MB parts, not 2MB, and some where less than that. I ended up buying them from mouser at like £10 a pop instead.

In fact a lot of china stuff isn't so cheaper anymore. With all the fakes, things like the DRAM work out more cost effective for me to buy parts from mouser instead. Be it a higher price, but bulk buying DRAM from China is a huge gamble, and basically "junks" a whole DRAM kit along with it when they don't work.

[exxos]

I have changed the socket just to rule out iffy connections... This is a LCS socket, and fits the current booster perfectly as well.

I have tested 10 CPU's in the V1 booster (32MHz) and they all seem to work fine

The booster in the image works fine in 8MHz mode, but in 36MHz, nope, just 2 bombs on power up.

So I will swap this CPU out and swap it for the ones which have been tested just.. Then see what happens...

IMG_3597.JPG (141.66 KiB) Viewed 3723 times

IMG_3599.JPG (135.02 KiB) Viewed 3723 times

[exxos]

So I tried a 16MHz osc.. nope.. tried 32mhz.. nope.. put back 36mhz and it tried to boot, put a 40mhz one in .. nope..

Put back one of my test boards and its working fine, so its not like anything on the STE is iffy..

Then just I had a realisation.. These GAL's are "new" ones which came last week... So I tried one of these in my test board and are having issues with bombs.. So something is up with these GAL chips.. I don't know what, as they actually verified fine...

So I will make some tests between the 36Mz clock input and output and see what the delays are like across the 2 GAL's...

EDIT:

"BAD GAL"

a1.png (6.36 KiB) Viewed 3715 times

" GOOD GAL"

a2.png (6.02 KiB) Viewed 3715 times

So the good GAL is as it supposed to be 7ns part, the bad GAL looks to be a 10ns or 12ns part. This should not really affect the booster in any case..

Looking at the actual output voltages, on the bad GAL, it is a fraction lower for some reason. So I am thinking this lower voltage is what is upsetting the booster logic...

More investigation work to do

[exxos]

POS GAL chips. Will have to program all the GAL's I have left and try them all in my test board to see if they actually work or not.

The original GAL's I was using I brought like 200 of them, and I have like 5 left now. It's why I brought some more from another seller, only these ones are junk.

So I will just have to see how many GALs work in the booster, put them in my store, and then it will have to wait before any more are built until I can obtain some GAL's which actually work.

[Icky]

POS GAL chips. Will have to program all the GAL's I have left and try them all in my test board to see if they actually work or not.

The original GAL's I was using I brought like 200 of them, and I have like 5 left now. It's why I brought some more from another seller, only these ones are junk.

So I will just have to see how many GALs work in the booster, put them in my store, and then it will have to wait before any more are built until I can obtain some GAL's which actually work.

Not only Bad DMAs but now Bad GALs.

This sucks when you buy a batch of chips and they are not as specified.

[exxos]

Not only Bad DMAs but now Bad GALs.

This sucks when you buy a batch of chips and they are not as specified.

Indeed. I am testing all the ones I can find currently. It seems the case of one works and 2 don't so far.

I don't really understand what the actual problem is with these chips either. The voltages and speed ratings they are outputting is all within tolerances of what I tested while ago.

I would have thought they would have failed verification if there was something wrong with them. But I guess there still could be internal faults which do not show up under verification. I mean verification really only checks all the internal "fuses" are blown correctly I assume. There would not be any way to know if the internal flip-flop logic is functioning correctly for example.

Now makes me wonder how long I have actually had this problem for