Follow-Up video here on the faulty crystal - ruclips.net/video/-LfKLDuf9dk/видео.html BTW, is this a world first in electronics on RUclips i.e. a crowdsourced & AI assisted repair!
I've seen something similar myself. Heating the oscillator trimmer capacitor (MLCC) caused a non-working piece of gear to work. It wanted 18-20 pF so I replaced it with a 22pF rather than the crystal for giggles and that worked for me. My guess is than neoprene was worth a few too many pF of parallel capacitance and overdamped the circuit. If that's a thing.
Wow that’s really unusual, particularly with the USB stick crashing the whole thing as a symptom. I was out of Good Ideas myself, about to throw a logic analyzer on the thing. Kudos to your readers!
3 things to check on any microprocessor (in order) and is my checklist for new designs and fault finding ... 1) Power rail(s) present and stable, 2) Not held in reset and no glitches on the reset line or external power monitor chip (if used) releasing reset, 3) Oscillator running at the correct frequency and it's stability. The latter also includes the caps on the XTAL as some microprocessors (Atmel) don't run smoothly if the output of the crystal is anything other than near sinusoidal! Nice that you've found the issue ... simple fix 🙂 The micro was crashing because the USB clock wasn't running at a good 48MHz, so the USB stick wasn't enumerating. The code should really gracefully handle that comms failure and throw away the attempted USB attach!
Spot on, these are the absolute basics of faultfinding, the very first things to look for. After that the next thing to check is for shorted address, data and I/O lines, it's really quick to 'brush' a multimeter with fast continuity beep over a whole board and quickly identify shorts to ground/VCC, combined with the voltage/Reset/Clk checks it'll pick up a significant number of faults and save yourself a good amount of time.
The boot up also got faster after replacing the crystal, I was thinking that the boot up time was varying slightly, but I thought it was just my imagination. I have seen oscillators fail, but like you said it is rare, looks like its output was just a bit weak, pity you didn't scope it before/after.
I did off camera but the scope probe capacitance just crapped it out. It’s never a definitive check scoping xtal circuits. Oscillators are a different matter. However, plan is to breadboard the xtal, buffer the output and see what I get…….so maybe a wee follow up to this world first crowdsourced and AI assisted video……LOL!
@@IanScottJohnston A trick I use to test oscillators using a scope probe is to connect the probe through a 1-2pF ceramic capacitor, that way it doesn't load down the oscillator.
Wow, I cannot believe that fixed the problem. Congrats! I'm interested in following up on the XTAL fault by testing it on a breadboard to check if it was giving an off-spec rate. Maybe the USB code was been doing very timing-specific operations to initialize or access the USB, and being slightly off on the timing could have put it in a nonrecoverable loop. It's strange that the XTAL was the problem. Old school PCs used to have 'turbo' buttons that would actually slow down the system, right? That way, old games would run properly on newer, faster CPUs by effectively 'slowing down' the XTA. Good job on finding the fault.
Wow, I wouldn't have thought of that. Group think for the win. It would be interesting to see the scope reading of each crystal. Was it the frequency or the amplitude or the stability? Well done team internet.
Likewise. Recommend also checking the load capacitors and feedback resistor (if there is one). Very lucky (or skilled) to select a replacement crystal with the same load capacitance range.
I've never listen to a xtal problem, but it makes sense. Some uP are very sensitive to timing variation, some can work on any frequency. Very good, always learning!
I have had exactly one crystal fail on me and it was right in front of my eyes. It really surprised me. Had someone who bought one of my C64 8701 replacements have an issue with it on one C128. The original 8701 sort of worked but not reliably. Suggested he pop in a new crystal and that solved the issue. It was a SWAG on my part, but I knew my board was OK and the crustal was really the only thing left in the circuit. Guessed that the crystal was not happy with the capacitive loading and the original part loaded the crystal a bit differently.
Nice one.... thoughts for possible reasons for crystal change getting it working... Fault probably reeeealy close to being either always OR never seen. ... As you observed, the glue goop on crystal insulator (or the insulator itself) becoming conductive (or residual flux?) ...Out of spec (or edge of spec) crystal (manufacturing defect as per sooooo much since Covid parts disruption)...On similar theme out of spec internal oscillator or oscillator caps ...Leaky oscillator cap that has (temporarily?) come good after soldering temps reached when changing crystal. On similar theme, heat on pins of IC, or board substrate Was there a date on any components? Extremely interesting that Heating / Cooling board had no obvious effect. Would be sorta interesting to put original crystal back and see if fault returns... (Need ideas to make short Pt3 video?) My bets are on the crystal being edge of spec or a covid parts shortage relaxion of component rejection tolerances. Thanks for video, fun to see your diagnostic process.
Power of the people ! (and maybe chatgpt) Brilliant result, rarely does the crystal itself actually cause intermittent problems (in my experience at any rate), it's usually a clock chip or a component other than the crystals drifting out of spec and affecting the timing. Nice job, very pleasing result - I love it when intermittent faults are resolved - well mostly intermittent but it did fail every time you used the USB port.
i had a usb sound card that sometimes locked up or just didnt start at all unless i pre-heated it. after a long troubleshooting and parts replacing process it turned out to be the crystal driving circuitry inside the mcu. i replaced the crystal too, it worked for a month but died again. after the 3rd crystal i tried a external oscillator feeding into xtal_in , that fixed the problem seems like the circuitry driving it degraded over time. i would try removing the crystal and feeding it from a external signal generator.
Probably the foam or the gule is resistive or the insulator loses are geting high at 2MHz. If the crystal was cracked the dry ice spray is may showing the issue well. Have a god day for you and an new RLC meter.
maybe try putting back the old crystal without the pad, and also with the pad. Glad it seems to be fixed, but I agree that it's hard to understand the repair.
Very interesting.. but there is absolutely no reasonable explanation for this error mode. Once an XTAL is oscillating, it should not stop, and obviously, the clock continued after inserting the USB stick .. or did I miss something? How should the insertion interact with the XTAL? Uf there is not enough oscillation margin in first place, it should not start oscillating, but the reverse effect will not happen. As well, it's a electro-mechanical device, so any deterioration happens mostly on the mechanical system, like the blank carrier, silver flakes, which might give rise to erroneous oscillation modes. There's also no backward influence from the program on the oscillator circuit. Maybe you can place the old XTAL on an Impedance , Gain/Phase analyser. What was the explanation of the guy who made this proposal?
Why is there absolutely no explanation? USB requires frequency to be within a certain range 0.5%, if its outside of that it then it might not work. Crystal or load caps could have put it outside of that 0.5% easily.
Congrats for the repair, Ian!, specially because I was one of the several members who suggested the xtal or the xtal oscillator. This is what I need to see in your videos: full solution!. Greetings!
Welcome in the club 😎 I had even stranger issue with x-tal even usb interface of MCU has been working only reporting garbage (did not identify proper device ID). It took me 3 weeks to identify that. And it's my board design and brand new x-tal from Farnell. That's why I suggested to check the X-tal as well.
So what frequency was the original crystal running at. Maybe the MCU was being overclocked slightly and it didn't like it. Why waste a transistor thermal pad, i would've just stuck a piece of black PVC tape over it or cut off a small piece of PET from something in the recycling . The pins would pierce through.
Fascinating couple of videos Ian. Great result. Did you get a chance to use the meter to test the values of some components, inductors, capacitors etc. Again great work.
I have had a 4th gen MB with a stuck crystal. The MB wouldn't boot and I found the crystal not oscillating, but after a diode check, ie, it put power across it, the MB starts and works fine.
What could be the problem? Is the specified frequency of the crystal out of tolerance or some kind of parasitics or leakages causing it to become unstable ?
It wouldn't measure anything in circuit so I couldn't test, i'd need to build a circuit and buffer the xtal to properly get to the bottom of teh xtal problem.
Actually I quite like it. Fast response, GUI isn’t bad to change setting, fairly intuitive. Boots up quick. Designed with production in mind because you can set alarm points on values I believe. Yep, I’d recommend it.
Fascinating couple videos Ian. Great Result. Did you get a chance to use the meter to test the values of some inductors, capacitors etc? Again great work.
I was given a cassette deck in the 90's that 2 people tried to repair. It would run perfectly until about half the tape and then stop. As if the end of tape sensor was tripped. It took me quite a while to find it..... it was flux around the crystal and micro processor! Just cleaning the board with alcohol fixed it! I'd bet on the glue on that insulating pad! Anyhow, ever since those days I ALWAYS clean my work!
Follow-Up video here on the faulty crystal - ruclips.net/video/-LfKLDuf9dk/видео.html
BTW, is this a world first in electronics on RUclips i.e. a crowdsourced & AI assisted repair!
I've seen something similar myself. Heating the oscillator trimmer capacitor (MLCC) caused a non-working piece of gear to work. It wanted 18-20 pF so I replaced it with a 22pF rather than the crystal for giggles and that worked for me.
My guess is than neoprene was worth a few too many pF of parallel capacitance and overdamped the circuit. If that's a thing.
Wow that’s really unusual, particularly with the USB stick crashing the whole thing as a symptom. I was out of Good Ideas myself, about to throw a logic analyzer on the thing. Kudos to your readers!
3 things to check on any microprocessor (in order) and is my checklist for new designs and fault finding ... 1) Power rail(s) present and stable, 2) Not held in reset and no glitches on the reset line or external power monitor chip (if used) releasing reset, 3) Oscillator running at the correct frequency and it's stability. The latter also includes the caps on the XTAL as some microprocessors (Atmel) don't run smoothly if the output of the crystal is anything other than near sinusoidal! Nice that you've found the issue ... simple fix 🙂 The micro was crashing because the USB clock wasn't running at a good 48MHz, so the USB stick wasn't enumerating. The code should really gracefully handle that comms failure and throw away the attempted USB attach!
Spot on, these are the absolute basics of faultfinding, the very first things to look for.
After that the next thing to check is for shorted address, data and I/O lines, it's really quick to 'brush' a multimeter with fast continuity beep over a whole board and quickly identify shorts to ground/VCC, combined with the voltage/Reset/Clk checks it'll pick up a significant number of faults and save yourself a good amount of time.
It's *always* the last thing you try
The boot up also got faster after replacing the crystal, I was thinking that the boot up time was varying slightly, but I thought it was just my imagination.
I have seen oscillators fail, but like you said it is rare, looks like its output was just a bit weak, pity you didn't scope it before/after.
I did off camera but the scope probe capacitance just crapped it out. It’s never a definitive check scoping xtal circuits. Oscillators are a different matter. However, plan is to breadboard the xtal, buffer the output and see what I get…….so maybe a wee follow up to this world first crowdsourced and AI assisted video……LOL!
@@IanScottJohnston A trick I use to test oscillators using a scope probe is to connect the probe through a 1-2pF ceramic capacitor, that way it doesn't load down the oscillator.
@@TheDefpom Thanks for the tip
It's always the capacitors... You know the RUclipsr :) this time it's the Crystal !!! I love your repair videos.
Glad you managed to fix it. I like to watch these kind of intermittent failure videos and diagnostics 👍
Wow, I cannot believe that fixed the problem. Congrats! I'm interested in following up on the XTAL fault by testing it on a breadboard to check if it was giving an off-spec rate. Maybe the USB code was been doing very timing-specific operations to initialize or access the USB, and being slightly off on the timing could have put it in a nonrecoverable loop.
It's strange that the XTAL was the problem. Old school PCs used to have 'turbo' buttons that would actually slow down the system, right? That way, old games would run properly on newer, faster CPUs by effectively 'slowing down' the XTA.
Good job on finding the fault.
Wow, I wouldn't have thought of that. Group think for the win. It would be interesting to see the scope reading of each crystal. Was it the frequency or the amplitude or the stability? Well done team internet.
Good to see my gut feeling on the xtal as posted for episode #1 worked for you 🎉
hard to believe. make an osc with the old xtal and check freq.
call me skeptical.
Yes, I’d like to know also…….
the thing is: the disassembly and reassembly may have done the job also. but - well shit happens too.
Likewise. Recommend also checking the load capacitors and feedback resistor (if there is one). Very lucky (or skilled) to select a replacement crystal with the same load capacitance range.
I've never listen to a xtal problem, but it makes sense. Some uP are very sensitive to timing variation, some can work on any frequency. Very good, always learning!
Perhaps you could stress test the old crystal separately. It just may be that the old glue on the neoprene was the issue.
I have had exactly one crystal fail on me and it was right in front of my eyes. It really surprised me. Had someone who bought one of my C64 8701 replacements have an issue with it on one C128. The original 8701 sort of worked but not reliably. Suggested he pop in a new crystal and that solved the issue. It was a SWAG on my part, but I knew my board was OK and the crustal was really the only thing left in the circuit. Guessed that the crystal was not happy with the capacitive loading and the original part loaded the crystal a bit differently.
Nice one.... thoughts for possible reasons for crystal change getting it working... Fault probably reeeealy close to being either always OR never seen.
... As you observed, the glue goop on crystal insulator (or the insulator itself) becoming conductive (or residual flux?)
...Out of spec (or edge of spec) crystal (manufacturing defect as per sooooo much since Covid parts disruption)...On similar theme out of spec internal oscillator or oscillator caps
...Leaky oscillator cap that has (temporarily?) come good after soldering temps reached when changing crystal. On similar theme, heat on pins of IC, or board substrate
Was there a date on any components?
Extremely interesting that Heating / Cooling board had no obvious effect.
Would be sorta interesting to put original crystal back and see if fault returns... (Need ideas to make short Pt3 video?)
My bets are on the crystal being edge of spec or a covid parts shortage relaxion of component rejection tolerances.
Thanks for video, fun to see your diagnostic process.
Power of the people ! (and maybe chatgpt) Brilliant result, rarely does the crystal itself actually cause intermittent problems (in my experience at any rate), it's usually a clock chip or a component other than the crystals drifting out of spec and affecting the timing. Nice job, very pleasing result - I love it when intermittent faults are resolved - well mostly intermittent but it did fail every time you used the USB port.
Send the middle finger to Matrix.
btw, I had a few occasions when the crystal was the issue. On PC motherboard, and some other equipment.
i had a usb sound card that sometimes locked up or just didnt start at all unless i pre-heated it.
after a long troubleshooting and parts replacing process it turned out to be the crystal driving circuitry inside the mcu.
i replaced the crystal too, it worked for a month but died again. after the 3rd crystal i tried a external oscillator feeding into xtal_in , that fixed the problem
seems like the circuitry driving it degraded over time.
i would try removing the crystal and feeding it from a external signal generator.
Probably the foam or the gule is resistive or the insulator loses are geting high at 2MHz. If the crystal was cracked the dry ice spray is may showing the issue well.
Have a god day for you and an new RLC meter.
Wow. There's a first time for everything I guess. Like you, I've never seen a crystal fail.
I had problems with crystals and small PIC micros in the past.
I tryed a lot of different crystals e capacitors to bootup consistently.
maybe try putting back the old crystal without the pad, and also with the pad. Glad it seems to be fixed, but I agree that it's hard to understand the repair.
Fabulous video when everyone is involved love it. Question is, was it the oscillator or the insulator/glue as they do become conductive over time ??
To be sure it's the crystal you should put the original crystal back in with a different insulator and see what happens!
Very interesting.. but there is absolutely no reasonable explanation for this error mode.
Once an XTAL is oscillating, it should not stop, and obviously, the clock continued after inserting the USB stick .. or did I miss something?
How should the insertion interact with the XTAL?
Uf there is not enough oscillation margin in first place, it should not start oscillating, but the reverse effect will not happen.
As well, it's a electro-mechanical device, so any deterioration happens mostly on the mechanical system, like the blank carrier, silver flakes, which might give rise to erroneous oscillation modes.
There's also no backward influence from the program on the oscillator circuit.
Maybe you can place the old XTAL on an Impedance , Gain/Phase analyser.
What was the explanation of the guy who made this proposal?
I’ll put the xtal on a breadboard and see how it performs.
Ian..A short video must be interesting for the sake of completeness..
Why is there absolutely no explanation? USB requires frequency to be within a certain range 0.5%, if its outside of that it then it might not work. Crystal or load caps could have put it outside of that 0.5% easily.
Wow, color me surprised. I've never seen a zombie crystal before.
The plastic piece from an ordinary pin header works great to insulate xtals. I use it all the time
Unexpected to me but extremely interesting fault. A good lesson, thanks for the sharing.
This might be the first "AI assisted electronics repair video" - albeit with the great work of the comment-section :)
Oh, never thought of that…….I think you are right! I will change the thumbnail image 😀
Great stuff =D I guess its a reminder to all of us - check your voltages, then clocks!
Congrats for the repair, Ian!, specially because I was one of the several members who suggested the xtal or the xtal oscillator. This is what I need to see in your videos: full solution!. Greetings!
Wonder if it was the crystal or the glue on the neoprene?
What about testing the crystal , and an Ohm meter on the insulator ?
Thanks again
I did check the insulator off camera.......it didn't appear to give any resistance, so probably not the issue.
Great!!!👍 I love your repair videos.
if you also noticed the meter is now updating the display faster
Really! I never noticed. Thanks.
As one of of the people that suggested the crystal, I'm happy that my gut feeling was right.
Welcome in the club 😎 I had even stranger issue with x-tal even usb interface of MCU has been working only reporting garbage (did not identify proper device ID). It took me 3 weeks to identify that. And it's my board design and brand new x-tal from Farnell. That's why I suggested to check the X-tal as well.
So what frequency was the original crystal running at. Maybe the MCU was being overclocked slightly and it didn't like it. Why waste a transistor thermal pad, i would've just stuck a piece of black PVC tape over it or cut off a small piece of PET from something in the recycling . The pins would pierce through.
You got it.
Fascinating couple of videos Ian. Great result. Did you get a chance to use the meter to test the values of some components, inductors, capacitors etc. Again great work.
Exactly same results using my test capacitors, so all good.
@@IanScottJohnston Brilliant.
I have had a 4th gen MB with a stuck crystal. The MB wouldn't boot and I found the crystal not oscillating, but after a diode check, ie, it put power across it, the MB starts and works fine.
What could be the problem? Is the specified frequency of the crystal out of tolerance or some kind of parasitics or leakages causing it to become unstable ?
It wouldn't measure anything in circuit so I couldn't test, i'd need to build a circuit and buffer the xtal to properly get to the bottom of teh xtal problem.
@@IanScottJohnston hope you'd make a follow-up video on that👍
I wouldn't suspect a crystal in very last try... I have the same unit and it works well for a year now. Quite pricey anyway, got it second hand.
So it probably was glitching after all. Not on the power rails but on the clock line, maybe?
Congratulations Ian! Nice machine. I am sure it will perform perfectly now.
Ian, what’s your overall opinion on this model lcr meter?
Actually I quite like it. Fast response, GUI isn’t bad to change setting, fairly intuitive. Boots up quick.
Designed with production in mind because you can set alarm points on values I believe. Yep, I’d recommend it.
Thanks! Looking for a bench meter…this one is on the list, I appreciate your evaluation!
Fascinating couple videos Ian. Great Result. Did you get a chance to use the meter to test the values of some inductors, capacitors etc? Again great work.
My guess would have been the USB interface chip. The crystal is a bit of a surprise. I am glad you found it.
Wow, this was like being in a repair cafe, the whole community talking and making suggestions, very interesting format
Let me be crystal clear, that was really cool.
Crowdsourced and AI aided troubleshooting.
You are on the cutting edge of technology Ian.
I suggested it was to do with 'MCU loading', and I was thinking in the right direction. Good fix.
good it is fixed did you check the crystal that you removed and I suspect it was the adhesive pad 🙂
I can feel smug now as I was one of the ones that suggested the crystal in the comments of the previous video.
I was given a cassette deck in the 90's that 2 people tried to repair. It would run perfectly until about half the tape and then stop. As if the end of tape sensor was tripped. It took me quite a while to find it..... it was flux around the crystal and micro processor! Just cleaning the board with alcohol fixed it! I'd bet on the glue on that insulating pad! Anyhow, ever since those days I ALWAYS clean my work!
Fantastic result, well done Ian!
I think it was the glue.😊
Well that was no how I thought this would end ! I've never had a dead crystal before.
The crystal wasn't dead - otherwise the Microcontroller would not run at all. Probably the frequency was off for the reasons Ian surmised.
Very strange ...
wow! very nice job
So far, lookin good.