Adrian is the champion of evidence based, critical thinking repairs. He has me spoiled as I cannot stand it when other online 'repair' videos start off with massive chip swapping. They teach me nothing. Adrian's video's always leave me feeling that I have learned a new skill.
You definitely get that feeling, but realistically, it's nothing you can really use in board repair. For example, I've got a cheap 386 SX-40 system which had the barrel style battery leak at some point. It was behaving very wonky. But since it's a rather no-name cheap DTK board, there are no schematics, no info, nothing about it anywhere. All I could do is probe all the exposed connections for continuity, but when everything seemed to work, what was I to do? The story ended happily, with the system just kind of deciding to work all of a sudden, but what if it didn't?
Don’t forget, you can apply this kind of diagnostic logic to other fields, depending on your hobby/job, be it software, hardware, OS/Systems or networking etc.
@@phil5564 at 37th and Rock Road? I now work for NetApp - we bought the facility from LSI who bought it from NCR. I think we sold it a few years ago and moved to KSU though! Had quite a history of storage though!
Adrian, you are a shining light in some old geek's life. I've been around computers since they first started screaming to AOL for their hourly usage. I cannot believe how incredible all of your videos are, and my god, I almost wish we could be on a repair venture together. The passion, your emotional reactions when they just "Freakin' Work." You're intoxicating, keep doing what you do, this is some of the best things to see in my life. Thank you for everything!
@@sprybug I think Adrian worked his way into my feed only about a year ago, so I'm a little newer in that regard. Anyway, probably too much to say for just a UA-cam comment, but I am just kinda running my own thing these days.
@@CaptainSouthbird I've been working on my own computer the past few years which I'm making to be backwards compatible with the TRS-80 Color Computer. Uses a 100% compatible CPU (63C09), run at 3.58mhz Turbo mode, VGA capable with several 256 color modes, SDCard, PS/2 Keyboard/mouse, 2MB SRAM max using the CoCo 3 MMU scheme, etc. I got into CPLD programming recently, and so far I got 4 CPLD's on this thing. 3 84 pin PLCC (just for VGA VIdeo!) and one 44 pin PLCC. Plus I got about 3 GALs on it too. Also will be using 4 microcontrollers to handle the more modern interfacing with the keyboard, mouse, SDCard, etc.
@@sprybug Fascinating, probably the kind of thing I wanted to work on in my 20s "just for fun." Alas, way too late for me now. As much as I admire old silicon, I have no path forward to greatly exploring legacy beyond my nostalgia.
Your joy after successful repairs is so nice to see. I don't understand much of what you do, but it's precisely because of your honest joy that I watch your videos
Commenting mostly to appease the algorithm, but I love watching the process of working through random issues and get old stuff working! Thanks for another video Adrian!
Just shows how useful schematics are, like a roadmap showing you where everything goes, and helping you find your way about until you find where the stalled roadworks are and get them back to work... :)
All these bodge wires makes for a rather pretty mess. Though the PCB designer in me screams about free hanging leads being potential EMI issues and causing all sorts of reflections and other issues. But at the very same time, this board doesn't really operate at a high enough frequency for this to be a major concern. But personally I would have cut the trace at the start and end of the bodge to ensure that as little of the "old" path isn't part of the circuit, this will ensure one gets as little weird reflection issues as possible as the signal bounces back from the end of the broken line. Another tip for working with higher frequency signals is to run a ground wire alongside the bodge (a little twist and/or glue is a simple solution for keeping them hugged together) as to provide an adequately short return path for the signal, this will greatly improve signal integrity for single ended signals. (running a ground along a differential pair can likewise help at times, but differential signals are often more robust and rarely need this, even if they can complain about going over ground plain interruptions since they do still interact with their surroundings.) However. It is important to note that one might bridge two parts of the ground plane in such a way that the ground wire suddenly sees a lot of current and burns up. My preferred 30 awg wire wrapping wire don't need much current to burn up This is rare on well designed boards but good to keep in mind. One can put a DC blocking capacitor on one end of the ground wire, but that starts being rather fancy for a bodge...
This reminded me of Ben Eater's video where he deliberately tried to induce stray capacitance to mess-up the timings on one of his breadboard computers. His neater wires really did make a difference. If I recall, for 1-3 breadboards he was alright with neat wiring but spread across 6-12 boards he had to reduce the 6502 from 10MHz to 1MHz to cope. Or something. So, all that said, those few bodge wires in a small area are almost certainly not upsetting anything at 33MHz! Would be interesting to see if there's any visible difference here, however subtle, on the scope with and without the leftover longer "stub" traces though.
This is the reason why, when doing trace repair on modern boards, you should use as much of the original trace as possible, and simply bridge the broken area instead of using bodge wires.
Maybe removing the power socket for inspection/cleaning below would be worth it, possibly fix that first transaction-size trace without the bodge, and maybe the SCSI trace is visible under there as well. Could also lead to future mysteries if there is still some active goop corroding.
I recently swapped all the parts from a battery-bombed Classic to a new recreation PCB called Classic Reloaded. The old board was completely gone, so I went bananas with the desoldering and chip removal, and the power/video connector was especially difficult to clear and extract. This may be a case where it may be not worth risking doing more damage to the traces, at least not before having a clear lead on some possible issues underneath.
I was soldering out the power connectors on an industrial board the other day at work. As you know, it is quite difficult to solder them. I used a Rose alloy that significantly lowers the solder temperature and very quickly and without damage unsoldered the connector.I do the same with SMD chips. To do this, mix the Rose alloy with a soldering iron with solder on the board and then the microcircuit is very easily unsoldered with a hair dryer. Then you need to clean the tracks and solder them using the usual method.
you have FAR more patience than me. I would have nuked that thing from Orbit after part 1. Nice work I will say though glad you did this. I remember something about the Diagnostics but totally forgot about it.
I really enjoy this video. I am very happy you were able to track down the problem. I had one of those apple systems. Man I wish I had it now. I know I could maybe find one, but after my cancer surgery, it drained our savings pretty bad. Now trying to recover and its going real slow.
Good job. It was probably a good idea to check all the traces, clocks, and data lines, but following input from diagnostic tool was also quite a good approach, and is useful.
I really think you should remove that power connector and clean up underneath there. If there is some electrolyte stuck in there, it's just going to keep eating away at the board and probably break other connections in the future. Anyways, great job!
Hello Adrian, Wish I had watched your first video when it dropped, could have offered my experience. I had a very similar simasimac with a slow sad mac sound on an se30 which took forever for me to pinpoint. Eventually decided to brute force continuity check every signal line via the matrix provided in the updated SE/30 PDF schematic. My issue was basically the same - one of the SIZ signals to the glue chip wasn't connected, I bodged the offending line and it booted up fine afterwards.
Dude you are the coolest. I wished since I was a little kid that I could fix stuff like this. I suck as math so I went into IT instead but board repair is very interesting.
68k systems may be weird in term of how the CPU is wired up, but hey, what I like about some classic CPUs, up to superscalar in-order versions, is that they're fairly easier to work with providing you have decent oscilloscope and DMM (Zoyi ZT-703S works okay with the CPU around 40 MHz or less - hoping Zoyi bumps the oscilloscope bandwidth rating to 100 - 200 MHz which is doable with superscalar in-order Cortex M7 microcontroller, so it would be a bit more useful as I would want to be able to probe 100 MHz CPU master clock for recent motherboards). I like seeing classic computers getting repaired, so I come here often and watch you work your magic on the dead computers until the end that they boot up just fine.
Adrian it would be nice if we could do this to every PC's made since the late 70's and be able to replace parts. Because these computers were fun to use.
Please make videos on the other board!!!! We need to see them all through especially since we started off and skewed off to the other board in earlier videos!!!
"So to *recap* I recapped the motherboard but it came up 12V short on Data Line 31, whoops! Replaced the damaged parts but then the counter wasn't adding up causing video shenanigans. I did not SCSI what happened next but I was able to *trace* it down & oh boy was I happy cos now IT FREAKIN' WORKS!"
I can’t imagine a computer that old with a whole Gig of RAM. It cracks me up when old systems will address more RAM than any of their typical data storage options.
I was deathly curious what was under the SCSI chip. If there was still ugly goo underneath still eating away at the board, or if the cleaning got rid of that but left the bad trace there still. =)
Would seem likely there are more bad traces that haven't fully failed yet lurking on this board near the glue chip. The two bad traces connecting to the same edge of the glue chip is really suspicious. I'd guess maybe the pads are damaged or the solder has gone bad. Removing the bodges and reflowing the glue chip might fix the problem.
This was the rare case when I was ahread of you :) The binary counter was triggering twice ina cycle. So lets say it was supposed to trigger on 16, it was triggering on 11 and 16. Never was so much of your skill spent on something so weak as this board :) Nevertheless a good effort!
You can get away with running a very small program without working RAM. The 030 has cache. 256 bytes each for instructions and data. Might be why the tests showed a failure yet you could write/read individual values and run the video test. That code never really left the CPU. Modern systems rely on running a lot of code from cache before the memory controller is initialized. Fun factoid: The Apple Sound Chip buffers PCM to the DACs, and that is used for the startup chime and all sound in the OS just like in classic Macs. It also has synthesizer capabilities that were never used for anything... except the sad mac chimes! Those aren't PCM samples, they're just commands to the ASC.
FYI, the link to part 2 in the description is broken. It leads to a message that says "This video has been removed by the uploader." I found the actual video on your channel so I was able to watch it, but thought you would want to know.
Good job finding those errors in such a rat's nest of traces and chips! Seems like a typical Apple thing to do, to add a "secret" debugging interface that only their "qualified" technicians knows about.
12:40 It is pretty obvious that the first CK (1) input of the Binary Counter just does that: counts (QA(3) to QD(6)). And when it reaches 16 (or more precise every 16 clock counts) QD changes its state from 1 to 0 and feeds into CK (13). 11,085.00 Hz / 694.55 Hz = 16 (roughly). It is an overflowing counter. You could probably use this chip for two separate counters (from 0 to 15) or connect them, like it was done here, and have one counter from 0 to 255. At least that's what I assume just looking at the schematics. How many lines are there on one a screen at this device? Probably more than 16. LCTRST is then the reset of the vertical counters for the next frame. The question is: Are the output signals of the Binary Counter still okay. Maybe I'm totally wrong. PS Okay you went to the PAL chip and the LCTRST be not okay is a probable source of failure. "Double" reset of the counter would be a problem.
I never remember it, as it's a non-standard number. Apparently it's 342 pixels tall. (It's 512 wide though, which is at least a more-common binary number. You can see why they changed it to 640x480 with the colour models, it's just a little extension on both sides. But that's a slightly different aspect ratio so it's not a uniform scale from the B&W ones.)
@@kaitlyn__L It's a long time since I learned about CRTs, too. Don't remembered the exact numbers. But I think it is even 576 for PAL in Europe. In the US 480 for NTSC. However, that should both be interlaced on a regular TV. So it is every second line written and the next frame would be the other lines. (If this chips work that way, I don't know). That would only be half of the above numbers. PS 342 is really not that much. But back then, it was the world.
@@hw2508 somehow, I have tucked-away 480 visible/525 total for NTSC and 576 visible/625 total for PAL in my brain 😅 probably from all the awful home movies I made from age 6 to 14 😊
Would have been interesting to see the output of the binary counter rather than just swapping the part. Channel one on clock and channel two on each output of the counter.
It might be possible to make a serial loader using a terminal emulator that's able to run scripts. Could get very fancy and have it load Intel hex files or maybe even Motorola SREC into memory using the diagnostic environment. Applications for that might be an even better diagnostic ROM,or perhaps booting from serial. I wonder if there is a way to dump the existing diagnostic environment from ROM or memory and decompile it to see how it works or if we're missing features
After Adrian’s last video I started hacking together a (so far very simple) Python script that does some of that. My goal is to have some memory read back tests by next weekend.
I think the ideal next step would be to disassemble the ROM packs from the Techstep. I suspect it implements extra tests by writing direct to memory and then telling the processor to execute it. With extra hardware like a rascsi and custom software I think you could fully test the SCSI subsystem and externally validate that the scsi controller is writing correct data out of the ports.
I was wondering if Adrian was enough of a perfectionist that he couldn't just leave the bodge wire in place; would he have to track down and see exactly where the trace was broken and solder a hair-sized wire onto it and apply more mask? I'm almost 24 minutes in and looks like he's got limits to his unnecessary excess. But there's still half the video to go and there might be a surprise part 4 of 3...
Adrian is the champion of evidence based, critical thinking repairs. He has me spoiled as I cannot stand it when other online 'repair' videos start off with massive chip swapping. They teach me nothing. Adrian's video's always leave me feeling that I have learned a new skill.
Have you ever tried applying it? It's actually damn hard!
@@jpt3640 it really is, especially if you don't have all the tools needed or good schematics.
You definitely get that feeling, but realistically, it's nothing you can really use in board repair.
For example, I've got a cheap 386 SX-40 system which had the barrel style battery leak at some point. It was behaving very wonky. But since it's a rather no-name cheap DTK board, there are no schematics, no info, nothing about it anywhere. All I could do is probe all the exposed connections for continuity, but when everything seemed to work, what was I to do?
The story ended happily, with the system just kind of deciding to work all of a sudden, but what if it didn't?
Don’t forget, you can apply this kind of diagnostic logic to other fields, depending on your hobby/job, be it software, hardware, OS/Systems or networking etc.
absolutely appreciate his grin when Adrian after a considerable what-happened-so-far goes into his signature "without further ado..."
I worked on the NCR 53C80 SCSI chip as an engineer. I am glad it was a wire fix rather than replacing another PLCC.
I wirewrapped a SCSI board prototype using a 5380, wrote 68000 assembly device drivers for it too...a long time ago
Wasn't it designed in Wichita, KS? I live not far from where the Wichita operations were, and I know they did storage controller design there.
@@douro20 The original NMOS 5380 was designed in Wichita. Other NCR SCSI controllers were designed in Colo Springs.
I'd love to hear more about the design process of these, it's capabilities and any interesting errata!
@@phil5564 at 37th and Rock Road? I now work for NetApp - we bought the facility from LSI who bought it from NCR. I think we sold it a few years ago and moved to KSU though! Had quite a history of storage though!
Adrian, you are a shining light in some old geek's life. I've been around computers since they first started screaming to AOL for their hourly usage. I cannot believe how incredible all of your videos are, and my god, I almost wish we could be on a repair venture together. The passion, your emotional reactions when they just "Freakin' Work." You're intoxicating, keep doing what you do, this is some of the best things to see in my life. Thank you for everything!
OMG. Long time no see! How ya doing? Been watching Adrian for about 5 years now.
@@sprybug I think Adrian worked his way into my feed only about a year ago, so I'm a little newer in that regard.
Anyway, probably too much to say for just a UA-cam comment, but I am just kinda running my own thing these days.
@@CaptainSouthbird I've been working on my own computer the past few years which I'm making to be backwards compatible with the TRS-80 Color Computer. Uses a 100% compatible CPU (63C09), run at 3.58mhz Turbo mode, VGA capable with several 256 color modes, SDCard, PS/2 Keyboard/mouse, 2MB SRAM max using the CoCo 3 MMU scheme, etc. I got into CPLD programming recently, and so far I got 4 CPLD's on this thing. 3 84 pin PLCC (just for VGA VIdeo!) and one 44 pin PLCC. Plus I got about 3 GALs on it too. Also will be using 4 microcontrollers to handle the more modern interfacing with the keyboard, mouse, SDCard, etc.
@@sprybug Fascinating, probably the kind of thing I wanted to work on in my 20s "just for fun." Alas, way too late for me now. As much as I admire old silicon, I have no path forward to greatly exploring legacy beyond my nostalgia.
Sir. You are a modern day wizard. I love your break downs and showing your troubleshooting techniques.
This is 'FREAKIN' AWESOME!' I just love the process of getting to the conclusion of what wasn't working so it can be fixed.
I don't know what makes this particular series so special but it's really just been so much fun to follow. Thanks, Adrian!
Your joy after successful repairs is so nice to see. I don't understand much of what you do, but it's precisely because of your honest joy that I watch your videos
Commenting mostly to appease the algorithm, but I love watching the process of working through random issues and get old stuff working! Thanks for another video Adrian!
I love the analytical approach, instead of just throwing out random ICs. Exercises critical thinking, logic, and analysis
Just shows how useful schematics are, like a roadmap showing you where everything goes, and helping you find your way about until you find where the stalled roadworks are and get them back to work... :)
Really awe-inspiring work. No board left behind!
Just imagine all the great games and productivity software you can run on the SE/30. That high-res screen looiks amazing.
Congrats Adrian! Thanks for the journey!
Adrian never gets bored. He makes his own adventure 😂
All these bodge wires makes for a rather pretty mess. Though the PCB designer in me screams about free hanging leads being potential EMI issues and causing all sorts of reflections and other issues. But at the very same time, this board doesn't really operate at a high enough frequency for this to be a major concern. But personally I would have cut the trace at the start and end of the bodge to ensure that as little of the "old" path isn't part of the circuit, this will ensure one gets as little weird reflection issues as possible as the signal bounces back from the end of the broken line.
Another tip for working with higher frequency signals is to run a ground wire alongside the bodge (a little twist and/or glue is a simple solution for keeping them hugged together) as to provide an adequately short return path for the signal, this will greatly improve signal integrity for single ended signals. (running a ground along a differential pair can likewise help at times, but differential signals are often more robust and rarely need this, even if they can complain about going over ground plain interruptions since they do still interact with their surroundings.)
However. It is important to note that one might bridge two parts of the ground plane in such a way that the ground wire suddenly sees a lot of current and burns up. My preferred 30 awg wire wrapping wire don't need much current to burn up This is rare on well designed boards but good to keep in mind. One can put a DC blocking capacitor on one end of the ground wire, but that starts being rather fancy for a bodge...
This reminded me of Ben Eater's video where he deliberately tried to induce stray capacitance to mess-up the timings on one of his breadboard computers.
His neater wires really did make a difference. If I recall, for 1-3 breadboards he was alright with neat wiring but spread across 6-12 boards he had to reduce the 6502 from 10MHz to 1MHz to cope. Or something. So, all that said, those few bodge wires in a small area are almost certainly not upsetting anything at 33MHz!
Would be interesting to see if there's any visible difference here, however subtle, on the scope with and without the leftover longer "stub" traces though.
This is the reason why, when doing trace repair on modern boards, you should use as much of the original trace as possible, and simply bridge the broken area instead of using bodge wires.
This channel is gold, keep em coming Adrian, you rock!
You're a genius. I love how deep you delve into these things. I've been learning so much.
I love watching your work, very thorough and very informative, thanks.
Thanks!
Maybe removing the power socket for inspection/cleaning below would be worth it, possibly fix that first transaction-size trace without the bodge, and maybe the SCSI trace is visible under there as well. Could also lead to future mysteries if there is still some active goop corroding.
Was about to comment that as well.
If two traces are broken somewhere, what are the chances more are faulty and we just didn't notice effects yet.
I recently swapped all the parts from a battery-bombed Classic to a new recreation PCB called Classic Reloaded. The old board was completely gone, so I went bananas with the desoldering and chip removal, and the power/video connector was especially difficult to clear and extract. This may be a case where it may be not worth risking doing more damage to the traces, at least not before having a clear lead on some possible issues underneath.
The link to part 2 in the description seems to been fed 12 volt as well. 😅
Just another testament to how back in the day they made stuff to be fixed. Awesome fix!
I *love* catching the 3rd part of something; makes it feel like Tenet when I go back and watch parts 1 and 2!
This has been the best series in ages, kudos
YESSSSSSSSS!!!!!!! I LOVE THIS SERIES! Thanks Adrian!
and another mainboard fixed and ready to work... Great job
I was soldering out the power connectors on an industrial board the other day at work. As you know, it is quite difficult to solder them. I used a Rose alloy that significantly lowers the solder temperature and very quickly and without damage unsoldered the connector.I do the same with SMD chips. To do this, mix the Rose alloy with a soldering iron with solder on the board and then the microcircuit is very easily unsoldered with a hair dryer. Then you need to clean the tracks and solder them using the usual method.
Two excellent, very satisfying fixes 😀😀😀
you have FAR more patience than me. I would have nuked that thing from Orbit after part 1. Nice work I will say though glad you did this. I remember something about the Diagnostics but totally forgot about it.
What an amazing repair! Thanks so much for sharing.
By far, Adrian is the smartest guy on UA-cam. What a beautiful mind!!
> What a beautiful mind!!
I'm always creeped out when people say that about someone. You sound like Dr Frankenstein on his quest hunting for brains.
Nice work Adrian! Your repair video put me in a good mood. Always good to see old hardware given new life.
Again, another excellent example of the use of logic and good troubleshooting technique.
Don't let this guy discover Minecraft redstone.
I really enjoy this video. I am very happy you were able to track down the problem. I had one of those apple systems. Man I wish I had it now. I know I could maybe find one, but after my cancer surgery, it drained our savings pretty bad. Now trying to recover and its going real slow.
Great to see you back to basics, Adrian!
Better than "How To Basic" where he would've thrown eggs at those once non-working motherboards.
That was a brilliant repair series.
man, your dedication and patience is unmatched... probs
This series is up there in legendary status with the P20!
Good job. It was probably a good idea to check all the traces, clocks, and data lines, but following input from diagnostic tool was also quite a good approach, and is useful.
I really think you should remove that power connector and clean up underneath there. If there is some electrolyte stuck in there, it's just going to keep eating away at the board and probably break other connections in the future.
Anyways, great job!
Hello Adrian, Wish I had watched your first video when it dropped, could have offered my experience. I had a very similar simasimac with a slow sad mac sound on an se30 which took forever for me to pinpoint. Eventually decided to brute force continuity check every signal line via the matrix provided in the updated SE/30 PDF schematic. My issue was basically the same - one of the SIZ signals to the glue chip wasn't connected, I bodged the offending line and it booted up fine afterwards.
Imagine how long it would take for an SE/30 to boot up with 1GB of RAM and the stock ROM!
Real determination. Nice work!
LOL I can hear it in your voice Adrian. As soon as you said the intro line, there's just despair. :)
Genius Adrian, great job!
Dude you are the coolest. I wished since I was a little kid that I could fix stuff like this. I suck as math so I went into IT instead but board repair is very interesting.
68k systems may be weird in term of how the CPU is wired up, but hey, what I like about some classic CPUs, up to superscalar in-order versions, is that they're fairly easier to work with providing you have decent oscilloscope and DMM (Zoyi ZT-703S works okay with the CPU around 40 MHz or less - hoping Zoyi bumps the oscilloscope bandwidth rating to 100 - 200 MHz which is doable with superscalar in-order Cortex M7 microcontroller, so it would be a bit more useful as I would want to be able to probe 100 MHz CPU master clock for recent motherboards).
I like seeing classic computers getting repaired, so I come here often and watch you work your magic on the dead computers until the end that they boot up just fine.
Adrian it would be nice if we could do this to every PC's made since the late 70's and be able to replace parts. Because these computers were fun to use.
Another mission for you, most excellent!
Please make videos on the other board!!!! We need to see them all through especially since we started off and skewed off to the other board in earlier videos!!!
Apple Macintosh slogan “The Power to be your Best”
The Power to be a beast (like Adrian).
"So to *recap* I recapped the motherboard but it came up 12V short on Data Line 31, whoops! Replaced the damaged parts but then the counter wasn't adding up causing video shenanigans. I did not SCSI what happened next but I was able to *trace* it down & oh boy was I happy cos now IT FREAKIN' WORKS!"
the bus originally was busted
spoilers!!!
That is incorrect you're mixing two different repairs together
@@malanvogt ah, mixing two different audio mixers then?
Wayne & Garth:
“We’re not worthy
We’re not worthy!”
I love my Macintosh Color Classic from 1993. It’s Thomas the train trying to keep up with the Win-Tel and UNIX computers of the 1990’s
Another great fix.
What a great run! The ups! The downs...., what a ride. Fiction seems so pointless these days.
Good work m8.Not the easiest piece of diagnostics.
Fantastically satisfying. 😍
A required upgrade for the basement would be a microscope to repair damaged traces instead of running botch wires.
Bodge
I'd love to see you install A/UX on one of these SE/30's. :)
I really like those 302's
Super nice video!
“Control, we still have the flashing screen”
22:36 To ease debugging, I recommend looking at wiring matrices right side up. 😏
what a ride. Thanks.
I can’t imagine a computer that old with a whole Gig of RAM. It cracks me up when old systems will address more RAM than any of their typical data storage options.
Well done !!!
thanks for video
I was deathly curious what was under the SCSI chip. If there was still ugly goo underneath still eating away at the board, or if the cleaning got rid of that but left the bad trace there still. =)
best repair video series ever!
Would seem likely there are more bad traces that haven't fully failed yet lurking on this board near the glue chip. The two bad traces connecting to the same edge of the glue chip is really suspicious. I'd guess maybe the pads are damaged or the solder has gone bad. Removing the bodges and reflowing the glue chip might fix the problem.
This was the rare case when I was ahread of you :) The binary counter was triggering twice ina cycle. So lets say it was supposed to trigger on 16, it was triggering on 11 and 16. Never was so much of your skill spent on something so weak as this board :) Nevertheless a good effort!
Super job.
Estoy enganchado a esta SE/30, a ver q pasa en este y la puedes reparar!
You can get away with running a very small program without working RAM. The 030 has cache. 256 bytes each for instructions and data. Might be why the tests showed a failure yet you could write/read individual values and run the video test. That code never really left the CPU. Modern systems rely on running a lot of code from cache before the memory controller is initialized.
Fun factoid: The Apple Sound Chip buffers PCM to the DACs, and that is used for the startup chime and all sound in the OS just like in classic Macs. It also has synthesizer capabilities that were never used for anything... except the sad mac chimes! Those aren't PCM samples, they're just commands to the ASC.
FYI, the link to part 2 in the description is broken. It leads to a message that says "This video has been removed by the uploader." I found the actual video on your channel so I was able to watch it, but thought you would want to know.
that is a big Chassis
what tenacity!
looks like she needs a replacement Gas Tank
I enjoyed it so much, I even hit the thumbs up twice.. oh, wait...
Joke aside, this was an awesome repair series, I absolutely enjoyed watching it!
Is the video of part 2 no longer available? The link in the descriotion doesn't work
what happened to part 2?? it´s gone
Good job finding those errors in such a rat's nest of traces and chips! Seems like a typical Apple thing to do, to add a "secret" debugging interface that only their "qualified" technicians knows about.
Best unsolder the bad 393 before you forget it's there 😉
just drawing an X is ok, the computer might come in handy later even if the video is a bit wonky
12:40 It is pretty obvious that the first CK (1) input of the Binary Counter just does that: counts (QA(3) to QD(6)). And when it reaches 16 (or more precise every 16 clock counts) QD changes its state from 1 to 0 and feeds into CK (13). 11,085.00 Hz / 694.55 Hz = 16 (roughly). It is an overflowing counter. You could probably use this chip for two separate counters (from 0 to 15) or connect them, like it was done here, and have one counter from 0 to 255. At least that's what I assume just looking at the schematics. How many lines are there on one a screen at this device? Probably more than 16. LCTRST is then the reset of the vertical counters for the next frame.
The question is: Are the output signals of the Binary Counter still okay.
Maybe I'm totally wrong.
PS
Okay you went to the PAL chip and the LCTRST be not okay is a probable source of failure. "Double" reset of the counter would be a problem.
I never remember it, as it's a non-standard number. Apparently it's 342 pixels tall.
(It's 512 wide though, which is at least a more-common binary number. You can see why they changed it to 640x480 with the colour models, it's just a little extension on both sides. But that's a slightly different aspect ratio so it's not a uniform scale from the B&W ones.)
@@kaitlyn__L It's a long time since I learned about CRTs, too. Don't remembered the exact numbers. But I think it is even 576 for PAL in Europe. In the US 480 for NTSC. However, that should both be interlaced on a regular TV. So it is every second line written and the next frame would be the other lines. (If this chips work that way, I don't know). That would only be half of the above numbers.
PS 342 is really not that much. But back then, it was the world.
@@hw2508 somehow, I have tucked-away 480 visible/525 total for NTSC and 576 visible/625 total for PAL in my brain 😅 probably from all the awful home movies I made from age 6 to 14 😊
@@kaitlyn__L 😅 Funny what information the brain remembers.
Would have been interesting to see the output of the binary counter rather than just swapping the part. Channel one on clock and channel two on each output of the counter.
woohoo! double victory!!! :D
WeeHee indeed! Great work and great video; thanks!
Series? No, my friend, it's a SAGA!
It might be possible to make a serial loader using a terminal emulator that's able to run scripts. Could get very fancy and have it load Intel hex files or maybe even Motorola SREC into memory using the diagnostic environment. Applications for that might be an even better diagnostic ROM,or perhaps booting from serial. I wonder if there is a way to dump the existing diagnostic environment from ROM or memory and decompile it to see how it works or if we're missing features
After Adrian’s last video I started hacking together a (so far very simple) Python script that does some of that. My goal is to have some memory read back tests by next weekend.
I think the ideal next step would be to disassemble the ROM packs from the Techstep. I suspect it implements extra tests by writing direct to memory and then telling the processor to execute it.
With extra hardware like a rascsi and custom software I think you could fully test the SCSI subsystem and externally validate that the scsi controller is writing correct data out of the ports.
Good show! But you'd never make it as an apple authorized repair center. :P
yep looks like it must be Oil
I was wondering if Adrian was enough of a perfectionist that he couldn't just leave the bodge wire in place; would he have to track down and see exactly where the trace was broken and solder a hair-sized wire onto it and apply more mask? I'm almost 24 minutes in and looks like he's got limits to his unnecessary excess. But there's still half the video to go and there might be a surprise part 4 of 3...
Dr Adrian 👨🎓
get the Bugs for me
OK, so who is gonna load doom on this mac via the diagnostics port?
I can just call you Hot Cheeto Face 🙂
Félicitations !
😄