STE V1.5 32MHz BOOSTER - STATUS

[exxos]

I fluxed up one row of pins on the CPU, heated them up with my gas iron, and now the board boots again! but now GB6 is like not in medium res

I mean

huh.jpg (46.57 KiB) Viewed 2359 times

EDIT:

OK, fluxed and heated up with the heatgun.. all behaving again now.. well for now.. will update post later on how long it lived this time...

[rubber_jonnie]

Damn, not an awful lot of good news then Could the CPU be heat damaged from the solder oven?

I could see it being a PCB issue, after all there were issues with the ROM switchers weren't there?

Such a shame if you can't get it going, but as ever your efforts are appreciated

[exxos]

Damn, not an awful lot of good news then Could the CPU be heat damaged from the solder oven?

I don't think so... No CPU's have have died totally.

I could see it being a PCB issue, after all there were issues with the ROM switchers weren't there?

Yeah, https://aisler.net/ only got back to me once, I don't think they even know what I am talking about with missing "inner planes" told them 3 times.. no reply for a few days now . Obviously not bothered about refund or replacing boards they clearly messed up. I have half a feeling some other boards I was having issue with, but can't just remember.

Their site doesn't play nice with some of my files and fedex shipping didn't work (though I think they might have fixed it after I told them). My files load find with OSHpark, but on aisler any tiny thing makes the upload tool freak out. I mean I had something like the dimension layer in a lib for the cartridge connector for my diag-carts, took the dimension line around the PCB. Where it "joined" in the PCB file, aisler tool cut a huge triangle out of my board from that point. like WTF ? They said it was basically my fault for a fault in the drawing.. What fault ? overlapping the dimension lines ?! I also tried butting them up to each other and didn't work either. So got OSH to do the cart-pcbs.

Overall, if my new boards work OK, then I'm not using aisler again. I normally used OSHpark, never had any issues, but used aisler as same prices but in Germany, So being closer to UK think shipping would be a tiny be faster. I have given them a fair try, but 3 strikes and they are out!

Such a shame if you can't get it going, but as ever your efforts are appreciated

Thanks..

Last nights results (still 25MHz, approx passes)
20 passes
8 passes
30 passes

I find it odd it always seems to work after a "re-solder". The soldering just can't be that bad. I think there are bad vias (or even bad copper traces) on the board and heating up the PCB warps the board a tiny amount and they make proper contact again. After maybe a few hours the board slowly straightens out and the thing breaks again.. Thats the only thing I can think of.

Its could be possible taking the GAL in and out or even pushing the board in and out of the sockets is breaking via's.. Even so, PCBs should not be that fragile.

[exxos]

So turned it on this morning, its been running almost 3 hours and at loop 47.. probably would have reached around loop 60 at 40mhz.. In anycase, its been pushing to run 15mins before... So I'm not going to touch a thing and leave it running see what happens...

If it still running later, I will bash it about a bit see if it stops working...

[rubber_jonnie]

So turned it on this morning, its been running almost 3 hours and at loop 47.. probably would have reached around loop 60 at 40mhz.. In anycase, its been pushing to run 15mins before... So I'm not going to touch a thing and leave it running see what happens...

If it still running later, I will bash it about a bit see if it stops working...

The inconsistency is very strange...

[exxos]

The inconsistency is very strange...

Got to loop 58, so 3.5 hours.. reset it and see what I get this time...

EDIT:

About loop 30 that time.. hard to keep a eye on it, so I updated GB6 to output the count to file so I don't have to keep a close eye on it..

I smacked it about, and didn't crash.. Just leaving it running to see what number it gets to this time...

[exxos]

LOOP 34 that time.. onto the next testing..

LOOP 18 65mins run time.. 36.5C (floppy 28C)

LOOP 55 33C (floppy 27C)

I'm going to start monitoring the CPU temp.. just find it odd as the day goes on and gets hotter the CPU is failing more.. but still doesn't explain why it failed in the morning when its coolest anyway.. I will take the floppy drive case as "ambient temp".

EDIT:

OK its almost like cooler temps running for longer, but will put back the 40MHz osc and I will try not to move the PCB at all...

EDIT2:

Something odd I found earlier, the OSC itself is listed as a 3.3V one.. BUT, it didn't run on 3.3V on my previous board, so thought its a 5V one and used that.. Later removed the 3.3V regulator and just ran direct to 5V... Though looking at the datasheets its a 3.3V part, so no idea why it wouldn't run on 5V before.. but its possible the OSC itself could be crapping out... So I have ordered some various 5V speed OSCs..

EDIT3

40MHz tests

16mins 16 loops - 36.5C CPU, 27c floppy case. (just went to desktop, no bombs, locked up)
92mins 30 loops - 37c CPU
13mins 5 loops - 34c CPU (even put heatsink on OSC)
60mins 20 loops - 34c CPU
222mins 71 loops - 37c CPU - 40c OSC - 39c GAL - 26c floppy case (feels a lot cooler in my room today)

Seems like temperature goes up the longer its running, which makes sense. But don't think has any relation to it crashing. I get similar with 25MHz so don't think this a CPU itself problem.

[exxos]

I've been updating the above post.. 3.7hours it ran today..

Got my proper 5V oscs and now it "quits to desktop" after a few test loops.. So maybe this is a clue.. Will investigate...

[exxos]

5V OSC

40_5v.png (4.18 KiB) Viewed 2022 times

3.3V OSC

40_3_3.png (4.21 KiB) Viewed 2022 times

Not much between them.

5V has slightly lower DC offset. 3.88V - 0.72V = 3.16V P-P.

3.3V OSC is 4.2V - 0.92V = 3.28V P-P.

Pretty much could put it down to tollerences.

I do have problems with 40MHz on the CPU buffer though as the clock is going so fast (40MHz) it hardly goes down to 1V for logic low levels. But the buffer thankfully clears all that up.

The GAL datasheet says 0.8V max for logic low level. So you would think the 3.3V one would be worse (930mV) but that one works the best.
Logic HI is >2V so assume no problem there.

I'm thinking the GAL needs a buffer also now.. I will see if I can remove the DC offset more on the 5V one and see if it helps or not...

[exxos]

Freezer spray turned up

Been spraying the 5V OSC and the min/max is about 640mV 4.00V (was 720mV 3.88V)... it run 3 full cycles, then crashed while I was trying this.. before without the spray it crashed pretty much right away..

So how it looks currently is crappy tolerances on the OSC chips