How about trying this.
To eliminate what is failing due to heat, when its failed, they freeze spray on the chips one by one whilst resetting the computer to see which one helps it recover by cooling it manually, then aim at working out why that chip is failing.
derkom's attempt
Re: derkom's attempt
If it ain't broke, test it to Destruction.
Re: derkom's attempt
Didn't even know there's flux that isn't. It certainly seems to do the job.
Yeah, I was considering that, or possibly the opposite direction and selectively cooling parts of the board with icepacks to see if I can get it to where it won't crash.
Re: derkom's attempt
So after yesterday's improved performance after cleaning up the underside of the GLUE socket, I decided to give the whole board a good cleaning, by bathing it in IPA for an hour, spraying it down with water, bathing it in IPA again, spraying it again, and finally rinsing it in IPA before hanging it up to dry.
After it dried, I booted it up to run GEMBench (off the exxos diag cart). I realised as it was booting that I'd forgotten to replace the DMA buffer chips after cleaning, but decided "I don't care, because it's going to crash soon anyway." Turns out it ran all night, having completed 110 GEMBench loops by this morning (previous record was seven). I re-added the buffer chips this morning (with Icky's patented inverted inverter bodge) and ran another 29 GEMBench loops with no crash. Not entirely an unexpected result, since cleaning did seem to help yesterday, but still kind of surprising that that seems to have fixed it. The flux gunk really was becoming conductive (or perhaps, as suggested by a friend of mine, capacitant) when it warmed up.
The only thing left to test at that point was DMA function, and I've found that it is booting off ACSI just fine, and copying back and forth between floppies and HDD just fine, so the inverter bodge is also working.
I think that makes the system finally 100% functional.
47 days. Can you beat that and win the boobie prize, @exxos?
After it dried, I booted it up to run GEMBench (off the exxos diag cart). I realised as it was booting that I'd forgotten to replace the DMA buffer chips after cleaning, but decided "I don't care, because it's going to crash soon anyway." Turns out it ran all night, having completed 110 GEMBench loops by this morning (previous record was seven). I re-added the buffer chips this morning (with Icky's patented inverted inverter bodge) and ran another 29 GEMBench loops with no crash. Not entirely an unexpected result, since cleaning did seem to help yesterday, but still kind of surprising that that seems to have fixed it. The flux gunk really was becoming conductive (or perhaps, as suggested by a friend of mine, capacitant) when it warmed up.
The only thing left to test at that point was DMA function, and I've found that it is booting off ACSI just fine, and copying back and forth between floppies and HDD just fine, so the inverter bodge is also working.
I think that makes the system finally 100% functional.
47 days. Can you beat that and win the boobie prize, @exxos?
Re: derkom's attempt
@derkom it might be worth checking what temperature you are using on your soldering iron. Maybe you are burning the flux with it being too hot?
I had that problem on the Alpha when I'd soldered one of the sockets in at 400C when I should have used 300C for the flux cored solder that I use. (Had forgot to turn it down from doing some desoldering)
I had that problem on the Alpha when I'd soldered one of the sockets in at 400C when I should have used 300C for the flux cored solder that I use. (Had forgot to turn it down from doing some desoldering)
If it ain't broke, test it to Destruction.
Re: derkom's attempt
Good thought @PhilC. The flux is definitely burning a bit, as I can see the carbonisation, which probably does create some conductivity.
I did my initial soldering at 300°, but when reflowing went up to 350°, which was perhaps a mistake.
I am learning a lot from this process.
I did my initial soldering at 300°, but when reflowing went up to 350°, which was perhaps a mistake.
I am learning a lot from this process.
Re: derkom's attempt
Good hint, I always use 350°C, maybe that's why I have such a hard time cleaning the flux residues...
Ingo
| Atari 1040STE@32MHz | Amiga 1200 (ACA1220) | Atari 800XL (U1MB, SIDE2) | Atari 130XL (Sophia DVI) | C64 (1541 Ultimate II, Rev3 RFMod Replacement) | TI 99/4A (F18A, 32k, FlashROM 99) | Sinclair ZX Spectrum 128 (Stereo, DivMMC) | Amstrad CPC664 (512k, M4 Wifi) | ... |
| Atari 1040STE@32MHz | Amiga 1200 (ACA1220) | Atari 800XL (U1MB, SIDE2) | Atari 130XL (Sophia DVI) | C64 (1541 Ultimate II, Rev3 RFMod Replacement) | TI 99/4A (F18A, 32k, FlashROM 99) | Sinclair ZX Spectrum 128 (Stereo, DivMMC) | Amstrad CPC664 (512k, M4 Wifi) | ... |
Re: derkom's attempt
BREAKING THE H4 WITH A GOTEK
Since my H4 has been working entirely too well lately, I decided it was time to figure out how to get the HxC firmware Gotek drive's OSD working.
So I soldered in the necessary wires for the OSD on the Gotek (forgetting that the H4 has a composite sync generator, so I didn't actually need all three of these)...
I decided the best place to pull the csync and to return the output video was on the underside of the board, and went through one of the holes for the HD15 video connector I'm not using...
And wired it all in semi-nicely on top (I need to get a longer floppy cable)...
For a little while, all was looking good...
But then all of a sudden, poof, I've lost my red channel (which is the one Gotek was wired to)...
Ignore the lack of drive icons there. This photo was taken after I completely removed the Gotek.
It is plausible that the Gotek's OSD output has somehow toasted the red channel on the DAC? I plan to scope this tomorrow and see where the red is actually disappearing, but I'll give you guys some overnight time to think about this.
I don't particularly look forward to soldering that DAC for a third time, but I guess I will if it's toast. If it is indeed plausible that the Gotek OSD has killed part of the DAC, could this be prevented by sticking a diode inline with the output resistor on whatever channel I'm sticking the OSD on? Or is there something else wrong with my approach? (Yes, I've verified that the monitor and cable are fine, and that the R line is still connected from the monitor port to the DAC.)
Thoughts welcome!
Since my H4 has been working entirely too well lately, I decided it was time to figure out how to get the HxC firmware Gotek drive's OSD working.
So I soldered in the necessary wires for the OSD on the Gotek (forgetting that the H4 has a composite sync generator, so I didn't actually need all three of these)...
I decided the best place to pull the csync and to return the output video was on the underside of the board, and went through one of the holes for the HD15 video connector I'm not using...
And wired it all in semi-nicely on top (I need to get a longer floppy cable)...
For a little while, all was looking good...
But then all of a sudden, poof, I've lost my red channel (which is the one Gotek was wired to)...
Ignore the lack of drive icons there. This photo was taken after I completely removed the Gotek.
It is plausible that the Gotek's OSD output has somehow toasted the red channel on the DAC? I plan to scope this tomorrow and see where the red is actually disappearing, but I'll give you guys some overnight time to think about this.
I don't particularly look forward to soldering that DAC for a third time, but I guess I will if it's toast. If it is indeed plausible that the Gotek OSD has killed part of the DAC, could this be prevented by sticking a diode inline with the output resistor on whatever channel I'm sticking the OSD on? Or is there something else wrong with my approach? (Yes, I've verified that the monitor and cable are fine, and that the R line is still connected from the monitor port to the DAC.)
Thoughts welcome!
Re: derkom's attempt
Looking over the STV8438CV datasheet this morning, I see nothing about any protection circuitry on the analog outputs, but...
[bunch of old post edited out here, because it became irrelevant]
I just realised that the Gotek was wired to the green channel, not the red one. (Damn mirror image port/plug problems!) So I have some investigating to do.
[bunch of old post edited out here, because it became irrelevant]
I just realised that the Gotek was wired to the green channel, not the red one. (Damn mirror image port/plug problems!) So I have some investigating to do.
Re: derkom's attempt
A little bit more information on my confusing new problem. Here's the diag cart test pattern:
So it looks like I'm missing part of my blue. Hrm.
I have swapped out the Shifter, and it isn't that. Will keep investigating.
So it looks like I'm missing part of my blue. Hrm.
I have swapped out the Shifter, and it isn't that. Will keep investigating.