In university we once had to implement a simple circuit that blinks an LED containing 2 chips. When we had everything wired up it simply didn't work at all. So we triple checked everything and in the end it turned out that we had a faulty chip. One of them was a quad NAND gate. We only needed on of those gates and exactly the one we used didn't work ^^
I would like to generalize Dave's advice. His case only covers that dual-purpose feature. My opinion is that every engineer who is using a microcontroller peripheral or any external chip, has to verify EVERY single associated configuration bit for the right value and read that piece of the documentation very carefully to operate that hardware correctly for the mode you're using it in.
Its also good to have a way to read back configuration data to make sure the commands did "stick", or even just dumping the value of the "ACK" bit somewhere in memory that you can read with a debugger. It can be a bit of a pain when a device has a write-only configuration port.
Hi Dave! Since I've found your blog about a month ago, I've been working through your videos "All willy nilly" when I saw this one. Your project curse came true for me as well. It was a simple op amp oscillator for a project and I could get the darn thing to work. Through troubleshooting and a MASSIVE oversight, I got it to work (single power supply trap). The amount I learned through the process was astounding. Thanks for the vids!
The most insidious new design issue I ever encountered was the same transistor failing twice after initially working, however, the third transistor worked and never failed in the prototype board on thru the 30th board. Patience is useless in this field as it is like so much dry tinder and is quickly consumed. The nuclear fuel of commitment is what you need whether it is a hobby or your job to get you thru the debug phase of issues you can't imagine and that no one has ever encountered before.
And that's just with voice. Experienced mastering engineers who know exactly what needs to be done when they hear a recording might process an entire album of 15 songs in 30 minutes if it's not extremely bad.
Yup and it's super easy. I tapped the audio real fast and applied a notch at 50hz and all odd harmonics. Completely gone, literally 5 seconds of processing. Maybe an extra 30 to transport back to editing software. Sync not an issue. Add 110hz HP, cut 2k+ by 5dB, 8dB noise removal (different to notch filters) algorithm, then +17dB soft-knee compression + soft limiting and BOOM, very clean, full-scale audio in about 90 seconds.
Could you not take off that audio hum by cancelling the wave – or even a simple filter? Although the awkward bit may be taking off the audio, taking it through an audio processor, and then dubbing it back on without messing the sync.
This reminds me of that case when I didn't use a pull-down resistor at the output of an optocoupler. It went straight into a chip and I got a waveform like that =/
If you turn it on and it does NOTHING, it's probably a P.S. problem, which is easy to fix. As for the learning process, it's better for it to sort-of work…eg, with distortion, intermittent cut outs, unwanted oscillations, etc.
My last project didn't work at first, so I learned what happens if I leave a CMOS input floating, a signal has too long rise time and why would I need to connect capacitors across rectifier diodes in a FWB if I wanted to use the 50Hz signal from the power transformer (because otherwise I would get a spike and the signal would be detected as 100Hz).
when i made the second prototype for my 8 digit display, i'd shifted to using a tristate output shift reg for addressing the display registers and i found all the display registers were changing and showing the same outputs, as im using the OE pin on the tristate register it makes the strobe lines on the display registers OC and with noise it was strobing the display registers, putting on some pulldown resistors helped with that
Oh man… i don't know why I've heard findtubes instead of findchips, maybe it's your accent. Don't go to what I mentioned, it's nsfw…
Everything's ok in the end.
I once had some serious problems with an Atmel flash IC in a no-lead package that was not fully solderded onto my board. So one (or possibly many) pins were only coupled by capacitance. It drove me freeking nuts finding this fault, as you could not measure the signals at the leads. The flash's basic functions like reading and writing larger data blocks seemed to work without problems but more specific stuff like reading a config register failed. After re-soldering the chip it finally worked.
@BusterBlader910 I'm an ex-electronics engineer – these days I write software. But when I started I designed analogue RF transceivers (private mobile radio systems), then went on to low-power digital RF communications, some consumer electronics design (wireless burglar alarms), then one-off electronics display systems for exhibitions and education, but then drifted into software. If you understand the theory, can build stuff that works in practice, you can do what you like.
love your debug story. nice to know the pros go through the similar bug hunts as us amateurs do, allbeit, your bug hunt involve non-amateur components.
Just came here to link this video to someone, wow, can't believe how long ago this was and how much the blog has changed!
best Speech ever!!! if my amplifier fails wait it to your mail!!
thank you ))
In university we once had to implement a simple circuit that blinks an LED containing 2 chips. When we had everything wired up it simply didn't work at all. So we triple checked everything and in the end it turned out that we had a faulty chip. One of them was a quad NAND gate. We only needed on of those gates and exactly the one we used didn't work ^^
I would like to generalize Dave's advice. His case only covers that dual-purpose feature. My opinion is that every engineer who is using a microcontroller peripheral or any external chip, has to verify EVERY single associated configuration bit for the right value and read that piece of the documentation very carefully to operate that hardware correctly for the mode you're using it in.
That ground loop hum makes my head hurt..
Watching all of them 🙂
Thanks
Its also good to have a way to read back configuration data to make sure the commands did "stick", or even just dumping the value of the "ACK" bit somewhere in memory that you can read with a debugger. It can be a bit of a pain when a device has a write-only configuration port.
Problem like that always happened with me ,simple problem like current sinking sourcing,interfacing hardware stuff etc..
Can you please make video on problem that you have faced at some time,that will be helpful.
Hi Dave! Since I've found your blog about a month ago, I've been working through your videos "All willy nilly" when I saw this one. Your project curse came true for me as well. It was a simple op amp oscillator for a project and I could get the darn thing to work. Through troubleshooting and a MASSIVE oversight, I got it to work (single power supply trap). The amount I learned through the process was astounding. Thanks for the vids!
The most insidious new design issue I ever encountered was the same transistor failing twice after initially working, however, the third transistor worked and never failed in the prototype board on thru the 30th board. Patience is useless in this field as it is like so much dry tinder and is quickly consumed. The nuclear fuel of commitment is what you need whether it is a hobby or your job to get you thru the debug phase of issues you can't imagine and that no one has ever encountered before.
And that's just with voice. Experienced mastering engineers who know exactly what needs to be done when they hear a recording might process an entire album of 15 songs in 30 minutes if it's not extremely bad.
Yup and it's super easy. I tapped the audio real fast and applied a notch at 50hz and all odd harmonics. Completely gone, literally 5 seconds of processing. Maybe an extra 30 to transport back to editing software. Sync not an issue. Add 110hz HP, cut 2k+ by 5dB, 8dB noise removal (different to notch filters) algorithm, then +17dB soft-knee compression + soft limiting and BOOM, very clean, full-scale audio in about 90 seconds.
Could you not take off that audio hum by cancelling the wave – or even a simple filter? Although the awkward bit may be taking off the audio, taking it through an audio processor, and then dubbing it back on without messing the sync.
This reminds me of that case when I didn't use a pull-down resistor at the output of an optocoupler. It went straight into a chip and I got a waveform like that =/
Youtube Captions 0:21 – part number of partial condom on
Youtube captions 0:00 – hi welcome to the eighty below on highest digest and this is excited about
If you turn it on and it does NOTHING, it's probably a P.S. problem, which is easy to fix. As for the learning process, it's better for it to sort-of work…eg, with distortion, intermittent cut outs, unwanted oscillations, etc.
@Pentium100MHz Good work. Next try floating the power pin and see what happens!
My last project didn't work at first, so I learned what happens if I leave a CMOS input floating, a signal has too long rise time and why would I need to connect capacitors across rectifier diodes in a FWB if I wanted to use the 50Hz signal from the power transformer (because otherwise I would get a spike and the signal would be detected as 100Hz).
when i made the second prototype for my 8 digit display, i'd shifted to using a tristate output shift reg for addressing the display registers and i found all the display registers were changing and showing the same outputs, as im using the OE pin on the tristate register it makes the strobe lines on the display registers OC and with noise it was strobing the display registers, putting on some pulldown resistors helped with that
Oh man… i don't know why I've heard findtubes instead of findchips, maybe it's your accent. Don't go to what I mentioned, it's nsfw…
Everything's ok in the end.
I once had some serious problems with an Atmel flash IC in a no-lead package that was not fully solderded onto my board. So one (or possibly many) pins were only coupled by capacitance. It drove me freeking nuts finding this fault, as you could not measure the signals at the leads. The flash's basic functions like reading and writing larger data blocks seemed to work without problems but more specific stuff like reading a config register failed. After re-soldering the chip it finally worked.
@BusterBlader910 I'm an ex-electronics engineer – these days I write software. But when I started I designed analogue RF transceivers (private mobile radio systems), then went on to low-power digital RF communications, some consumer electronics design (wireless burglar alarms), then one-off electronics display systems for exhibitions and education, but then drifted into software. If you understand the theory, can build stuff that works in practice, you can do what you like.
love your debug story. nice to know the pros go through the similar bug hunts as us amateurs do, allbeit, your bug hunt involve non-amateur components.