As a noob hobbyist I have no problem following along, you have a great way of explaining things, and you give us a great view of what you're looking at. I really appreciate it, thanks Dave ;)
I have to agree about Dave's ability to explain things. Even when explaining something I know (or think I do) he doesn't cause my eyes to glaze over and hit the left-arrow-jump. I can follow most of what he says/does and my electronics engineering expertise level is "HO trains". LOL But this is how we learn... We hang out with the smarter kids and hope it rubs off. Here a tip for you I learned just the other day. When working with a non-working power supply start by making sure it is plugged in and that after you plug it in... make sure your space heater hasn't tripped the breaker. It's never the zebras that get you it's always the horses....
i know Im randomly asking but does someone know a method to get back into an instagram account..? I was dumb lost the password. I would love any assistance you can give me!
@Atlas Stanley thanks so much for your reply. I got to the site on google and I'm waiting for the hacking stuff atm. Takes a while so I will get back to you later with my results.
Easiest way to debug shorted rails is to stick an amp or two in from an external PSU and use a thermal camera. If it's so dead-shorted that it doesn't heat up, then using a mV meter to home in on it is usually easier than doing resistance measurements as you don't have the contact resistance issue.
I've done that in a previous video. Should be secondary thing if resistance is a problem as it can potentially blow the shorted device without evidence and actually make it harder to track down the component. Not to mention it could also potentially damage something even with a low compliance voltage when other rails are powered down.
I've used this idea, which has worked great for me, www.diyrecordingequipment.com/blogs/news/15851868-5-minute-project-how-to-build-a-circuit-sniffer - ( th-cam.com/video/StUlo7MFaXw/w-d-xo.html )i would have sent 1kHz sine wave through the 5v rail and used the sniffer to detect the faulty component. The shorted component would be when the pitch is in its highest. The high pitch sound comes from the amp, in my case I used the microphone input on my laptop.
I've successfully done this on a monitor which had a shorted power rail. I don't have a thermal camera yet, but I found the bad component (a shorted capacitor) with my infrared thermometer. In hindsight, I might have found it with a visual inspection, because there was some faint visual evidence of unwanted hotness on the PCB. But I had wanted to try out the IR thermometer trick anyway. In case it's not obvious to the folks playing along at home, I applied power with an external *current-limited* power supply, with the current-limit set to a value which I hoped was high enough to make the bad component hot, but low enough to avoid blowing it open as Dave warned about.
Data I/O - engineered and built like battleships. My model 29B's (I have 3) just keep working with minimal maintenance over the last 20 years. I bought them used on eBay auctions (the shipping was more than the purchase price for one or two of them). It blows my mind that I am using equipment that I used BITD when I was a young engineer in the early 80's. I repair old arcade PCBs from the 80's and use them to program really old bipolar proms and early eproms.
hehe, ya I accidentally bought the innards of the 1981 game "astro blasters". It looked like an S-100 card cage with cards from the angle of the photo taken. It just said "Unknown electronics". It has many of the old intel 2708's on it. Classic device! I just looked up the game to see if it was any good, and boy howdy, that game was incredible! Now I want to build the rest of the system around it heh.
That was a neat way to find the fault. After I see it it's kinda obvious but until then I would not think to do this. It's amazing what you can learn from just watching people who know what they are doing for a wee while. These are the kind of videos I like watching. Always things to learn! Thanks.
Nice repair video, well done. But to get things running, the Unisite needs to boot on a master floppy or a MSM module and with all the boards and adapters plugged. Until that, the self-test led will remain on indefinitely... (Also, about the previous video, don't touch the golden pin of the adapter: you could get a "fail" during the self-test just due to that... Use a little isopropyl alcohol to clean it, in case).
In the late '80's we used one of those as a large multi-gang programmer/tester. We were in a tightly controlled industry (nuclear power plant instrumentation), and so we developers made "Golden ROMs" that were cloned en masse for production. I believe the unit handled one master EPROM and 32 slave chips. The master had to be checksummed before and after each use. The slave chips were each validated multiple times after programming, over the full rated operational voltage and timing range for the chip.
Actually, it was one of my systems at a Swiss nuclear plant that was the first commercial radiation monitoring system in Europe to detect the Chernobyl fallout. The alarm from that system caused the backup cooling system (evaporation towers) to be engaged because the primary system (river water) had higher radiation levels going INTO the plant than was coming out, which meant the Chernobyl contamination was being left on the piping, and had to be blocked.
You could also point a thermal camera at the board while trying to power it up. That works like a charm for me every time I need to find a component that's shorting a power rail to ground. It's much faster than dicking around with a multimeter. You can record video with the thermal camera, and check it later, so you can do it quickly, to make sure you don't damage anything else on the board, trying to drive current through it.
From what I can see in the documentation (and I may be off the mark - I admit), the self test light stays lit until it sees a disk. I'm guessing the test routines are actually part of the OS bootstrap.
haha - I was expecting it to be an AVX cap, but it's close enough. I could be wrong, but I think STC was "Standard Telephone & Cables", which used to own the old Nortel site in Paignton before it turned into Nortel itself. The AVX Tantalum division was literally building alongside the Nortel building. I worked at both places, so this vid is a bit of a weird coincidence. The Tantalum stuff could be a tad scary in the factory, as they had a sintering oven which used to work on the verge of catching fire and exploding every ten minutes. They said that if it ever did go up, it would probably take the rest of the building with it, so there were daily fire alarm tests and drills. There was also the slight risk of a Tantalum dust explosion, but they didn't really have many processes where there was a lot of dust floating around, luckily. I worked on the machine that used to pick up large metal pallets of Tant caps, then dip them all into various process tanks. Lots of potassium permanganate etc. (nicknamed "Mango"). The pallets would then go through 24-hour bake, so there were two large conveyor ovens which extended through the dividing wall. The machine took up two rooms which were about 30ft long. It was like a giant "soft toy crane". The software wasn't particularly clever though, so if the pallets ever got out of sequence, you'd have to hit the E-stop, figure out where each pallet was in the two rooms, use the manual controls to lift them onto a trolley (or physically lift them), then put them in the right places again. It could have done with a few more sensors as well, as it didn't know when things like a riser platform was in the up or down position, so it often used to drop the pallets from a height of about 8 feet onto the floor. lol The pallets themselves were about 3ft by 2ft, with welded frame which weighed maybe 8Kg each, so they used to make quite a lot of noise, and wake everyone up in the canteen next door. A relatively interesting job for the first few weeks, but I don't think I'd want to do the four 12-hour shifts again (four days on, four days off).
i worked next door at Nortel making a type of remote mobile/landline telephone system (FWA or FRA as it was known). The whole site is now a community college. Small world!lol
Fascinating! I work with or have worked with many ex-STC, AVX, Nortel, Bookham and Oclaro employees and visited (the late) Syntech so have seen the site - at least what was left of it. Cheers!
Great stuff. Nice to hear from some fellow worker drones. lol Yeah, I worked at Bookham's on the old site for about 6 months in the fibre dept. (S-Building? 2005?) Then worked at Syntech after that, when it was at Aspen Way, and then again for all of 2011 when it went to the old AVX building. Real shame that Nortel closed, as it was a good place to work, and it was obviously one of the largest employers in the bay. There were rumours that Microsoft might buy the old buildings, but it never came to be. I worked at actual Nortel from 1999 to 2001, and in "The Great White Elephant" (Nortel House) for most of the second year. Some great people there too. We got the work done (and passed test occasionally. lol), but had a proper laugh as well, especially when having to wear the full "bunny suits". :p
+ElectronAsh I really enjoyed that explanation! I have often wondered how basic electronic components are made. I can't seem to find good videos covering the actual process for making old-school non-SMD caps, transistors, resistors etc.
That is what pc hard drives must do, the hard drives internal cpu must detect power loss and reset the head to its default position with the last bit of charge to avoid data loss and damage to the hard disk.
Given you are getting messages from the tty port, I assume the 68k is resetting and fetching code from those EPROMs. ....but the lack of further output means it is wedged or in an endless loop with no messages to say what is wrong. Maybe use a logic analyzer to see the access to the EPROMs? May get a clue to see if it is stuck or see what sequence is occurring?
My favourite way to track down a short was to apply a couple amps from a current limited supply and find the hot thing Saved a 3,000$ laptop from a shorted ceramic cap near the docking connector
Ya, good luck finding all the GALs and PALs and trying to figure out how they were programmed using only your knowledge of logic design and the surrounding circuitry...barf.
I used one of these a few years ago (also had the smaller 3900), from memory it looks for a terminal at boot/power-on as part of self test (already suggested previous comments I think?) we used a VT340 terminal to operate ours and later moved to using terminal emulation in Windows Hyperterminal as it had the option to emulate the VT340.
That was what I was saying for the last few minutes of the video...how can it possibly self test accurately if it isn't fully assembled as if it were ready to use? The answer is it simply won't. It's probably waiting for a signal ACK line to toggle that never toggles.
I didn't see the first "Tear Down" video on this one, but would LOVE to see a completed repair on it! I wonder if the problem could be a corrupted "self test file", sort of like having a "IF THIS, THEN statement that isn't completed anymore, and it is still waiting on what "then" is. I enjoy your programs!
Pin 2 on RS-232 is called TX full stop. TX is an output at the DTE, an input at the DCE. The signal name on pin 2 is never different, it is always called TX. This can be confusing because of MCU and UART pin labeling, which always has TX as chip output and RX as a chip input. The signal name on such a chip only matches the RS-232 signal names when implementing a DTE. The chip signal names and RS-232 names will be swapped, e.g with a net names across the RS-232 level shifter, when the chip is in a device implementing a DCE.
around 10:00 that reminds me of the grid of transistors. Moving around the grid gives different resistances. I didn't see boards like that in the past. neat.
New technique for finding shorts is to use your power supply and supply small limited current to the 5v shorted rail and use a FLIR to find the component warming up.
I so love tear-down and repair vids like this! This is a bit outside my skillset I would say, more just find bad e caps with my ESR or completely nuked components, but I certainly hope to be someday. I look forward to part two!
That tantalum cap at 11:00 I spotted because of the colour, it stood out as the bad one to me even before you tested it, it is slightly darker which happens when they get hot from shorting.
Waveform card is obviously a precision adjustable power supply, that can run/test the chip at exact voltages, such as max/nom/min from datasheet. All the pin drivers may be powered from it too. And to avoid switching noise, it's probably all linear regulation, dumping all the excess voltage into the heatsink. Next test: Plug in a module to see if that's the unwanted condition leading to the RTOS shutting down and telling you "Shutdown complete, it is ok to POWER DOWN now".
Great job, Dave! I've already learned a lot from you and yet there is still more to learn. And I always enjoy gorgeous old electronic coming back to live as well.
It needs a boot disk to complete self test Dave. Check the data io yahoo group for a replacement hard disk module. Boots much faster and you can have the last released device library and software. I have had a few of these.
any switches or jumpers that could enable a diagnostics mode? what about that 'remove for kernel test thing'? You could do some very basic checks on the 68k like looking for activity on DTAK and BERR during start up which might indicate logic faults.
Yeah, it would nice to get something at least, like the last self test mode it was in or something. It seems like one of the self tests is locked up. The power interrupt obviously hardware overrides any software lockup.
Hi Dave, a couple ideas that may have already been mentioned: connect the serial handshaking lines appropriately, connect one of the programming adapters that goes on top of the unit, and try putting any floppy in the drive since somebody mentioned it may never leave self test without one. HTH.
Don't you love it when a 30 cent part kills a $30,000 piece of gear. [prices approximate] :) Looking forward to part 2 where you might complete the repair.
OK! I found you! It is the next video, duh!. (I just watched #1060 and I am two years late. Better late than never! If this is the first video you are seeing on this, watch #1060 first. What a gorgeous device.
Love these videos. IMHO you're just like the The Serviceman in the old Electronics Australia magazine. That was always the first article I read in each issue.
Fair enough. I've seen a number of them fail short (and occasionally open) in 70s and 80s computers, and I am usually reluctant to put tantalum caps back in unless there is a specific reason for using them.
Tantalum also happens to be a conflict mineral, so people's hands get chopped off and child slaves work in mines to produce our capacitors. en.wikipedia.org/wiki/Coltan_mining_and_ethics
I never met a Tantalum cap that didn't make me suspicious. I try to replace them with some other formulation cap whenever I find a bad one if possible. I sleep better that way.
If you are afraid of tantalums, look at polymer tantalums, I believe they fail open circuit. There is some cases where a ceramic cap would actually make design of for instance some switchmode power supplies more complicated due to having lower esr, so you need to adjust the loop compensation. (What alan said)
Shawn White Wow that's an amazing trick! Thank you. On 2-Layer boards i just cut the Power trace whenever it branches on the board to narrow down where the fault is. Then i suck out caps and chips until the short is gone. Something my dad told me which is quite true is that doing something, even if it has no system in it is better than just sitting infront of the board doing nothing and thinking about the (your) mistake :D
Don't need to desolder the whole thing. Just cut one of the legs. In this case it worked out, but in some cases that will safe a lot of time finding the right cap.
I never thought of using your resistance method for finding a short. But then, I never owned a milliohm meter so it was not an option for me. This was 30 years ago when I still played around with electronics.
Are modern tantalum caps any better than the old 80's ones, or are they likely to go the same way in another decade or two? Or maybe the modern tantalum caps are even worse than the old ones?
20:58 Can anybody tell me what was used to decode the serial data? Thanks. EDIT: Ah, OK, I guess it was the scope. I'm not used to the capabilities of an expensive scope.
Hmmm ... did plugging it into the PC enable DTR at the RS-232 port? Some cables just run 3 wires. It could be that self-test does not properly start if DTR is not high when the box is powered up. DSR will typically be asserted by a box (not necessarily the type of unit at hand) so a bridge between the two as well as CTS/RTS does all the necessary hardware "handshake" to let data flow. Did you chuck out your serial breakout box?
Personally, the resistance sniff approach hardly ever works for me because of capacitors disturbing the resistance numbers. Maybe I just need a better multimeter.
Hmm, this reminds me a lot to my repair of the first MK1 synth i got, it tried to power up, had also shorted tatalums and, after removing these, it got stuck in the selftest, as the main CPU crashed, everytime...And it got soundcards inside, about 12 total, each card has two Voice CPUs on it, as well as tantalums and DACs and analog stuff. I narrowed it down to a faulty voicecard, by just putting two (this is the absolute minimum the OS needs to have to boot up) in and done a powercycle. May be this it´s the same here with all these 4Ch pindriver cards....may you can test one by one and may you find a non working one...
I bet this issue was to do with a faulty rail array, considering it worked fine beforehand. Just put it under a thermal camera, the faulty rail shouldn't be hard to notice.
In the 90's my father tried to trouble shoot a computer that wasn't booting up. His first thought was that the BIOS halted because we couldn't hear the floppy-drive, but after a month of BIOS-fixes it was discovered that the floppy-drive in question had a defective disk-inserted sensor. The BIOS was fine, but the perceived lack of a floppy-disk prevented the floppy-drive from even spinning up. That might be happening here.
Makes sense to me. I just had an Amiga 500 where this disk present switch was totally corroded internally. The disk drive itself looked good. Only disk changes were not detected. The A500 is actually from the same time period.
Heh, thankfully the C64 uses some cheapy parts(literally an LED and photo resistor for write protect detection, and didn't care if you changed the floppy or not. It just assumed the disk was there. Lesson learned...use tape drives on your IBM and Amiga in the future.
Any pointers to the mean to triangulate a dead component like you did? I'm watching the video for the second time and I have trouble seeing which pins you mesure and how you managed to close on the dead component. A document on the general theroy behind this (or even a book maybe) would be great. Thanks!
Nice short trace. Looking forward to cliffhanger resolution. I remember a repair where a self test hung in a loop and I could never resolve why. Hope you have better luck!
As a noob hobbyist I have no problem following along, you have a great way of explaining things, and you give us a great view of what you're looking at. I really appreciate it, thanks Dave ;)
I have to agree about Dave's ability to explain things. Even when explaining something I know (or think I do) he doesn't cause my eyes to glaze over and hit the left-arrow-jump. I can follow most of what he says/does and my electronics engineering expertise level is "HO trains". LOL But this is how we learn... We hang out with the smarter kids and hope it rubs off.
Here a tip for you I learned just the other day. When working with a non-working power supply start by making sure it is plugged in and that after you plug it in... make sure your space heater hasn't tripped the breaker.
It's never the zebras that get you it's always the horses....
I had a zebra or two in my day...where it was one of the chips it couldn't be.
i know Im randomly asking but does someone know a method to get back into an instagram account..?
I was dumb lost the password. I would love any assistance you can give me!
@Salvador Harrison instablaster :)
@Atlas Stanley thanks so much for your reply. I got to the site on google and I'm waiting for the hacking stuff atm.
Takes a while so I will get back to you later with my results.
Easiest way to debug shorted rails is to stick an amp or two in from an external PSU and use a thermal camera. If it's so dead-shorted that it doesn't heat up, then using a mV meter to home in on it is usually easier than doing resistance measurements as you don't have the contact resistance issue.
I've done that in a previous video. Should be secondary thing if resistance is a problem as it can potentially blow the shorted device without evidence and actually make it harder to track down the component. Not to mention it could also potentially damage something even with a low compliance voltage when other rails are powered down.
There you go: th-cam.com/video/FlO-sLfZCoc/w-d-xo.html the short was in a capacitor :-)
I've used this idea, which has worked great for me, www.diyrecordingequipment.com/blogs/news/15851868-5-minute-project-how-to-build-a-circuit-sniffer - ( th-cam.com/video/StUlo7MFaXw/w-d-xo.html )i would have sent 1kHz sine wave through the 5v rail and used the sniffer to detect the faulty component. The shorted component would be when the pitch is in its highest. The high pitch sound comes from the amp, in my case I used the microphone input on my laptop.
I've successfully done this on a monitor which had a shorted power rail. I don't have a thermal camera yet, but I found the bad component (a shorted capacitor) with my infrared thermometer. In hindsight, I might have found it with a visual inspection, because there was some faint visual evidence of unwanted hotness on the PCB. But I had wanted to try out the IR thermometer trick anyway. In case it's not obvious to the folks playing along at home, I applied power with an external *current-limited* power supply, with the current-limit set to a value which I hoped was high enough to make the bad component hot, but low enough to avoid blowing it open as Dave warned about.
Spoiler alert!
Data I/O - engineered and built like battleships. My model 29B's (I have 3) just keep working with minimal maintenance over the last 20 years. I bought them used on eBay auctions (the shipping was more than the purchase price for one or two of them). It blows my mind that I am using equipment that I used BITD when I was a young engineer in the early 80's. I repair old arcade PCBs from the 80's and use them to program really old bipolar proms and early eproms.
hehe, ya I accidentally bought the innards of the 1981 game "astro blasters". It looked like an S-100 card cage with cards from the angle of the photo taken. It just said "Unknown electronics". It has many of the old intel 2708's on it. Classic device! I just looked up the game to see if it was any good, and boy howdy, that game was incredible! Now I want to build the rest of the system around it heh.
That was a neat way to find the fault. After I see it it's kinda obvious but until then I would not think to do this. It's amazing what you can learn from just watching people who know what they are doing for a wee while. These are the kind of videos I like watching. Always things to learn!
Thanks.
Nice repair video, well done. But to get things running, the Unisite needs to boot on a master floppy or a MSM module and with all the boards and adapters plugged. Until that, the self-test led will remain on indefinitely... (Also, about the previous video, don't touch the golden pin of the adapter: you could get a "fail" during the self-test just due to that... Use a little isopropyl alcohol to clean it, in case).
In the late '80's we used one of those as a large multi-gang programmer/tester. We were in a tightly controlled industry (nuclear power plant instrumentation), and so we developers made "Golden ROMs" that were cloned en masse for production. I believe the unit handled one master EPROM and 32 slave chips. The master had to be checksummed before and after each use. The slave chips were each validated multiple times after programming, over the full rated operational voltage and timing range for the chip.
Sooo.... You recon you sold some of those chips to the Chernobyl plant?... lol
Actually, it was one of my systems at a Swiss nuclear plant that was the first commercial radiation monitoring system in Europe to detect the Chernobyl fallout. The alarm from that system caused the backup cooling system (evaporation towers) to be engaged because the primary system (river water) had higher radiation levels going INTO the plant than was coming out, which meant the Chernobyl contamination was being left on the piping, and had to be blocked.
remove the kernel test jumper look for output.
You could also point a thermal camera at the board while trying to power it up.
That works like a charm for me every time I need to find a component that's shorting a power rail to ground.
It's much faster than dicking around with a multimeter. You can record video with the thermal camera, and check it later, so you can do it quickly, to make sure you don't damage anything else on the board, trying to drive current through it.
From what I can see in the documentation (and I may be off the mark - I admit), the self test light stays lit until it sees a disk. I'm guessing the test routines are actually part of the OS bootstrap.
Nah, he put a disk in there he said for one of the tries. What he didn't do was finish assembling it! :O
Awesome Dave! you finally got your hands on an interesting repair, can't wait for the next one
Tantalum friends are very common to go happy like that :-D
i'd replace ALL of them without even powerin up :D
haha - I was expecting it to be an AVX cap, but it's close enough.
I could be wrong, but I think STC was "Standard Telephone & Cables", which used to own the old Nortel site in Paignton before it turned into Nortel itself.
The AVX Tantalum division was literally building alongside the Nortel building.
I worked at both places, so this vid is a bit of a weird coincidence.
The Tantalum stuff could be a tad scary in the factory, as they had a sintering oven which used to work on the verge of catching fire and exploding every ten minutes.
They said that if it ever did go up, it would probably take the rest of the building with it, so there were daily fire alarm tests and drills.
There was also the slight risk of a Tantalum dust explosion, but they didn't really have many processes where there was a lot of dust floating around, luckily.
I worked on the machine that used to pick up large metal pallets of Tant caps, then dip them all into various process tanks. Lots of potassium permanganate etc. (nicknamed "Mango").
The pallets would then go through 24-hour bake, so there were two large conveyor ovens which extended through the dividing wall. The machine took up two rooms which were about 30ft long. It was like a giant "soft toy crane".
The software wasn't particularly clever though, so if the pallets ever got out of sequence, you'd have to hit the E-stop, figure out where each pallet was in the two rooms, use the manual controls to lift them onto a trolley (or physically lift them), then put them in the right places again.
It could have done with a few more sensors as well, as it didn't know when things like a riser platform was in the up or down position, so it often used to drop the pallets from a height of about 8 feet onto the floor. lol
The pallets themselves were about 3ft by 2ft, with welded frame which weighed maybe 8Kg each, so they used to make quite a lot of noise, and wake everyone up in the canteen next door.
A relatively interesting job for the first few weeks, but I don't think I'd want to do the four 12-hour shifts again (four days on, four days off).
i worked next door at Nortel making a type of remote mobile/landline telephone system (FWA or FRA as it was known). The whole site is now a community college. Small world!lol
Fascinating! I work with or have worked with many ex-STC, AVX, Nortel, Bookham and Oclaro employees and visited (the late) Syntech so have seen the site - at least what was left of it. Cheers!
Great stuff. Nice to hear from some fellow worker drones. lol
Yeah, I worked at Bookham's on the old site for about 6 months in the fibre dept. (S-Building? 2005?)
Then worked at Syntech after that, when it was at Aspen Way, and then again for all of 2011 when it went to the old AVX building.
Real shame that Nortel closed, as it was a good place to work, and it was obviously one of the largest employers in the bay.
There were rumours that Microsoft might buy the old buildings, but it never came to be.
I worked at actual Nortel from 1999 to 2001, and in "The Great White Elephant" (Nortel House) for most of the second year.
Some great people there too. We got the work done (and passed test occasionally. lol), but had a proper laugh as well, especially when having to wear the full "bunny suits". :p
ElectronAsh
+ElectronAsh I really enjoyed that explanation! I have often wondered how basic electronic components are made. I can't seem to find good videos covering the actual process for making old-school non-SMD caps, transistors, resistors etc.
Hey Dave.
The video is older but:
Will there be part 2?
I would really be interested in whether you can repair the device. Greetings from Germany
i am sure he threw this old board out by now. -sorry
maybe he did but then got deleted/censored?
That is what pc hard drives must do, the hard drives internal cpu must detect power loss and reset the head to its default position with the last bit of charge to avoid data loss and damage to the hard disk.
Given you are getting messages from the tty port, I assume the 68k
is resetting and fetching code from those EPROMs. ....but the lack
of further output means it is wedged or in an endless loop with no
messages to say what is wrong. Maybe use a logic analyzer to see the
access to the EPROMs? May get a clue to see if it is stuck or see what
sequence is occurring?
My favourite way to track down a short was to apply a couple amps from a current limited supply and find the hot thing
Saved a 3,000$ laptop from a shorted ceramic cap near the docking connector
That was my next step. Done that in a previous video somewhere.
Not recommended unless you can't track it down resistively.
Yep, blowing tantalum caps up is fun, but I don't think I'd like to do that on a rather rare board like this.... :)
Ya, good luck finding all the GALs and PALs and trying to figure out how they were programmed using only your knowledge of logic design and the surrounding circuitry...barf.
I used one of these a few years ago (also had the smaller 3900), from memory it looks for a terminal at boot/power-on as part of self test (already suggested previous comments I think?) we used a VT340 terminal to operate ours and later moved to using terminal emulation in Windows Hyperterminal as it had the option to emulate the VT340.
Self test failure due to modules not being plugged in ?
TheDefpom I know my data I/o 212 won't pass self test if the module isn't inserted.
That was what I was saying for the last few minutes of the video...how can it possibly self test accurately if it isn't fully assembled as if it were ready to use? The answer is it simply won't. It's probably waiting for a signal ACK line to toggle that never toggles.
I would like to see dave check back on that. I hate it when stuff is that close to be working!
But it looks like all the modules were reinstalled, no?
So glad you're sharing the repair with us! I was hoping you would after the original video!
I liked the way you diagnosed that shorted cap. Great job Dave.
Ya, no kidding, I learned a thing or two about tracing bad components by watching this, and also reading the comments!
I didn't see the first "Tear Down" video on this one, but would LOVE to see a completed repair on it! I wonder if the problem could be a corrupted "self test file", sort of like having a "IF THIS, THEN statement that isn't completed anymore, and it is still waiting on what "then" is. I enjoy your programs!
Pin 2 on RS-232 is called TX full stop. TX is an output at the DTE, an input at the DCE. The signal name on pin 2 is never different, it is always called TX.
This can be confusing because of MCU and UART pin labeling, which always has TX as chip output and RX as a chip input. The signal name on such a chip only matches the RS-232 signal names when implementing a DTE. The chip signal names and RS-232 names will be swapped, e.g with a net names across the RS-232 level shifter, when the chip is in a device implementing a DCE.
Yes, DTR and RTS signals can be blocking the output from the RS-232 port
around 10:00 that reminds me of the grid of transistors. Moving around the grid gives different resistances. I didn't see boards like that in the past. neat.
I have learned something new again from you. Thank you Dave.
New technique for finding shorts is to use your power supply and supply small limited current to the 5v shorted rail and use a FLIR to find the component warming up.
Yes, I like that kind of repair videos much more than a auto routing soft test. Thumbs up on this one :)
I so love tear-down and repair vids like this! This is a bit outside my skillset I would say, more just find bad e caps with my ESR or completely nuked components, but I certainly hope to be someday.
I look forward to part two!
That tantalum cap at 11:00 I spotted because of the colour, it stood out as the bad one to me even before you tested it, it is slightly darker which happens when they get hot from shorting.
i know them only as getting very dark -after getting very bright. :-) often they burn like a matchhead when they short out
TheDefpom You did it again. I spotted that tant bugger, but you beat me to it LOL. Regards Chris
Waveform card is obviously a precision adjustable power supply, that can run/test the chip at exact voltages, such as max/nom/min from datasheet. All the pin drivers may be powered from it too. And to avoid switching noise, it's probably all linear regulation, dumping all the excess voltage into the heatsink.
Next test: Plug in a module to see if that's the unwanted condition leading to the RTOS shutting down and telling you "Shutdown complete, it is ok to POWER DOWN now".
The RTOS only shuts down when I hit the power switch.
Great job, Dave! I've already learned a lot from you and yet there is still more to learn. And I always enjoy gorgeous old electronic coming back to live as well.
Dave This upgrade is for a later video after unit is fix, is to replace the disk drive with a floppy/diskette emulator.
It needs a boot disk to complete self test Dave. Check the data io yahoo group for a replacement hard disk module. Boots much faster and you can have the last released device library and software. I have had a few of these.
Can i buy one?
Maybe try re-seating all the socketed IC's?
any switches or jumpers that could enable a diagnostics mode? what about that 'remove for kernel test thing'? You could do some very basic checks on the 68k like looking for activity on DTAK and BERR during start up which might indicate logic faults.
Yeah, it would nice to get something at least, like the last self test mode it was in or something. It seems like one of the self tests is locked up. The power interrupt obviously hardware overrides any software lockup.
There is that "remove for kernel test" jumper array. Maybe that'll get you more info on the terminal connector.
DextersLab2013
Man - you and your diag switches and Bus Errors. lol
I sense another IDA Pro session here.
LOL, it's all your fault Ash!
If he doesn't assemble it completely, he won't get the ACK from the devices it is supposed to test. It should be assembled as if it were ready to use.
Perhaps it is looking for a boot disk for the test program.
zx8401ztv That's my thoughts too, like an old IBM AT pc.
but it didn't even check the floppy drive, u would hear it
Douglas Walrath you have a point.
But it could be depending upon a disk sense line ... Hmm.
No disk? No spin-up ...
i doubt it
With this insight I'mma go check all my 'dead' equipment for tantalum caps.
"$30,000.00 in 1986 had the same buying power as $67,664.41 in 2018"
Hi Dave, a couple ideas that may have already been mentioned: connect the serial handshaking lines appropriately, connect one of the programming adapters that goes on top of the unit, and try putting any floppy in the drive since somebody mentioned it may never leave self test without one. HTH.
Don't you love it when a 30 cent part kills a $30,000 piece of gear. [prices approximate] :) Looking forward to part 2 where you might complete the repair.
by finishing putting the last few parts on top he somehow forgot... lol
Yep I've been there, Quick tip, stick the cap in a blob of bluetack, = 3rd hand
You had a result, happy days
OK! I found you! It is the next video, duh!. (I just watched #1060 and I am two years late. Better late than never! If this is the first video you are seeing on this, watch #1060 first. What a gorgeous device.
Dave, in case you missed it, in the other video someone mentioned that they still buy these back and refurbish them as there is some demand for them.
Super interesting troubleshooting! These old machines intrigued me.
Love these videos. IMHO you're just like the The Serviceman in the old Electronics Australia magazine. That was always the first article I read in each issue.
I went binge reading that column at the local college library!
Where is part 2?
You are so entertaining. Thank you. You should have more comments but I guess there aren't enough nerds in the world.
A short on the board! Nice!
I'm glad I watched this, once again very informative! :)
I'm sharing this one around!
Have you ever gotten this fully working? Maybe a revisit, update video??
Love your work...it has taught me a lot so far
Can you talk about why you wouldn't replace that power supply decoupling tantalum cap with a ceramic or an electrolytic?
No real reason, just a warm fuzzy.
Fair enough. I've seen a number of them fail short (and occasionally open) in 70s and 80s computers, and I am usually reluctant to put tantalum caps back in unless there is a specific reason for using them.
Tantalum also happens to be a conflict mineral, so people's hands get chopped off and child slaves work in mines to produce our capacitors. en.wikipedia.org/wiki/Coltan_mining_and_ethics
I never met a Tantalum cap that didn't make me suspicious. I try to replace them with some other formulation cap whenever I find a bad one if possible. I sleep better that way.
If you are afraid of tantalums, look at polymer tantalums, I believe they fail open circuit. There is some cases where a ceramic cap would actually make design of for instance some switchmode power supplies more complicated due to having lower esr, so you need to adjust the loop compensation. (What alan said)
NICE ! I love you video, mate, can't wait to go back to OZ
It's looking for the boot disk, 99% sure it has a disk sense line and does not leave self test until it sees one
That was a great troubleshooting tutorial Dave. Thanks for another repair video. Always great to see the master at work.
dave you should give that cap a proper send off and hook it up to the mains, that would also be an example to the other caps, so they'll behave
Hook it up to a current limited 5v voltage source and pour some alcohol on it. It worked for Louis Rossmann it'll work for you :D
Using a simple meter worked just fine.
Shawn White Wow that's an amazing trick! Thank you.
On 2-Layer boards i just cut the Power trace whenever it branches on the board to narrow down where the fault is. Then i suck out caps and chips until the short is gone. Something my dad told me which is quite true is that doing something, even if it has no system in it is better than just sitting infront of the board doing nothing and thinking about the (your) mistake :D
There's a video of Jim Williams fixing a friend's Tek 465 and the culprit was also... a tantalum capacitor.
Don't need to desolder the whole thing. Just cut one of the legs. In this case it worked out, but in some cases that will safe a lot of time finding the right cap.
Are there any memory modules in there that are wiped by UV? It was mailed overseas, so it was xrayed? Just a WAG of course.
x ray normally does not erase UV EPROMs. at least not when people tried to use it to erase plastic cased OTP parts.
Thank You, Steve Leibson! (You have just been added to my dictionary). :-)
I never thought of using your resistance method for finding a short. But then, I never owned a milliohm meter so it was not an option for me. This was 30 years ago when I still played around with electronics.
Really interesting mate, I can learn so much from you! Thanks
Are modern tantalum caps any better than the old 80's ones, or are they likely to go the same way in another decade or two?
Or maybe the modern tantalum caps are even worse than the old ones?
You might try to put in the proper floppies. I believe that the self test finishes up AFTER the floppies are read for the programming software.
20:58 Can anybody tell me what was used to decode the serial data? Thanks.
EDIT: Ah, OK, I guess it was the scope. I'm not used to the capabilities of an expensive scope.
Nice vid. I like to see these trace & fix vids. :) But your other vids are great too! Thanks!!
You just make a shit ton of money, GOOD JOB! OLD AND GOLD! GOING W3ST
9:19 holy smokes that is a lot of stuff
Check all the other tantalum caps on the waveform board for shorts.
I did and said that in the video.
Sorry I forgot.
Where is repair Part 2?!?
I noticed the module connector plugs don't have their round screws you took out. Plug in the modules as well, and see what happens.
Hmmm ... did plugging it into the PC enable DTR at the RS-232 port? Some cables just run 3 wires.
It could be that self-test does not properly start if DTR is not high when the box is powered up. DSR will typically be asserted by a box (not necessarily the type of unit at hand) so a bridge between the two as well as CTS/RTS does all the necessary hardware "handshake" to let data flow.
Did you chuck out your serial breakout box?
Yes, DTR and RTS signals can be blocking the output from the RS-232 port
Yay a repair video! I’d love this to program my arcade and pinball stuff
Personally, the resistance sniff approach hardly ever works for me because of capacitors disturbing the resistance numbers. Maybe I just need a better multimeter.
5:17 Ohww... Thats such a mood
Where is the rest of the repair?
That thing runs on VRTX? I've programmed VRTX-based software into EPROMs using one of those programmers!
Whoa . . . turn back the hands of time. I used VRTX on the Pegasus F-1 flight at Orbital Sciences many many moons ago.
Any harm in hooking it up to a current limited supply to heat the component up? Then use a thermal camera to see it.
Tantalums, and RIFAs. The bane of retro computer repair.
Check all the TO-220 devices too for shorts.
What a nightmare, no SMDs. Back to old school days.. :)
Didn't know about the 'triangulation' technique. I'll try that on my recently fried GPU!
Can't wait for part 2.
Hmm, this reminds me a lot to my repair of the first MK1 synth i got, it tried to power up, had also shorted tatalums and, after removing these, it got stuck in the selftest, as the main CPU crashed, everytime...And it got soundcards inside, about 12 total, each card has two Voice CPUs on it, as well as tantalums and DACs and analog stuff.
I narrowed it down to a faulty voicecard, by just putting two (this is the absolute minimum the OS needs to have to boot up) in and done a powercycle.
May be this it´s the same here with all these 4Ch pindriver cards....may you can test one by one and may you find a non working one...
I bet this issue was to do with a faulty rail array, considering it worked fine beforehand. Just put it under a thermal camera, the faulty rail shouldn't be hard to notice.
Any updates on this?
Did you ever get this thing working?
Amazing, very didactic too. Kudos
Where's part 2 ??
I would have quickly checked the other tantalums while I had it open...
In the 90's my father tried to trouble shoot a computer that wasn't booting up.
His first thought was that the BIOS halted because we couldn't hear the floppy-drive, but after a month of BIOS-fixes it was discovered that the floppy-drive in question had a defective disk-inserted sensor.
The BIOS was fine, but the perceived lack of a floppy-disk prevented the floppy-drive from even spinning up.
That might be happening here.
Makes sense to me. I just had an Amiga 500 where this disk present switch was totally corroded internally. The disk drive itself looked good. Only disk changes were not detected. The A500 is actually from the same time period.
Heh, thankfully the C64 uses some cheapy parts(literally an LED and photo resistor for write protect detection, and didn't care if you changed the floppy or not. It just assumed the disk was there. Lesson learned...use tape drives on your IBM and Amiga in the future.
hi i a new sub ty u for makeing a grate vid it cool see old stuff fix up and u see how long a way computer have come ty u chris
This was a legit way to debug stuff.
I haven't needed your micro amp converter but I would definitely buy a milliohm box if you made one.
"haych" gets me every time
I noticed the non-skippable ad, is it working? You made that look too easy brotha.
Maybe probe clocks/ICs for activity.
Would you upload a tutorial on smell tests?
Any pointers to the mean to triangulate a dead component like you did? I'm watching the video for the second time and I have trouble seeing which pins you mesure and how you managed to close on the dead component. A document on the general theroy behind this (or even a book maybe) would be great. Thanks!
It's good education videos these repair ones... Thanks :-)
Nice short trace. Looking forward to cliffhanger resolution. I remember a repair where a self test hung in a loop and I could never resolve why. Hope you have better luck!