Chipageddon - has anyone had to build their own version of a generic component?

[e100]

For example, CAN transceivers are currently out of stock everywhere.
Are people building their own replacements from scratch using discrete components?

[moffy]

Had to do it once for an out of date CMOS logic package, rewired a similar IC into a socket and plugged it in. Not the neatest solution but saved several thousand dollars for a replacement monorail motor drive.

[AndyC_772]

No - but that's not to say I've not had to make design changes because of component availability or lack thereof. In fact I've not done much else all year.

CAN transceivers that will operate from a 3.3V supply were completely unavailable last time I checked, but if you have 5V available in your system, then there are options that you can buy off the shelf today.

Remember that your microcontroller may have 5V tolerant pins, even if that's a feature you didn't ever think you'd use.

[Electroplated]

Does this count as duplicating a generic:

I designed ( don't laugh ) an entire fake fire bed in a fake electric fire, one of those that used a small motor to turn a rotary reflector, got fed up with replacement motors buzzing so designed a log effect bed that's lit by LED's, those are in four rows fade up and down, I planed to use one of my remaining ATMEGA328P.

Come the 'great reset' and I decided I should keep my few ATMEGA328P for better things, yeah I know, there is some knock off ones about but I don't touch knock off.

So after scratching my grey beard I decided to go ol skool and redesign using 2 quad op amps, a few k30 FET's ( have a pile of them from years ago) 4 lm555 ( because I use those in almost everything ) a few tip31a's and a simple linear psu. the 555's provide pmw to the tip31 led drivers that are configured to provide a simple CC supply, the op amps generate a slow triangle that controls the 555 pwm via a fet stage. It works but not as good as planned due to the fact there is not as much randomness as there would be with a ATMEGA328P, best I could do was mix the LED's up a little.

Next, though not really due entirely to the 'great reset', I was asked to repair an old bench power supply that used a LM723 as a voltage reference that was a dead duck, knowing the LM723 is almost unobtainable I took the repair on as a fun project, once I had found out what the actual reference voltage was, no diagrams or service info so did the big sums going by the circuit around the LM723, I then created a small pcb with a tl431 ref and a few transistors, the pcb fits there the LM723 was, the psu sprang back to life, only needs a to twiddle a few presets so the output matched the V/A printed on the back plate.

There's a ton of other stuff I've had to either scrap till the 'great reset' is over or redesign using whats available, its becoming rather tiresome.

--
 

[ajb]

Maybe not quite 'generic', but I needed a protected high-side switch for a project relatively recently.  While those aren't quite jelly-bean, normally there are a number of off-the-shelf options available.  I ended up using a 7905 to provide a 5V supply referenced to the +24V supply to run an op amp as an active current mirror to shift the current sense signal down to ground and provide some gain in the process.  Could've used a 24V-capable op amp instead, but there are a lot more ~5V parts that can hit the top rail, so it was worth the extra regulator to have some more supply options.  From there, the ground-referenced and amplified signal went into one of the MCU's on-board comparators, with its DAC channel internally routed to provide a programmable trip point.  The comparator output was then internally routed to a timer's fault input for semi-hardware-based handling of overload conditions. 

Ended up working quite well, except it turns out the MCU's comparators are pretty slow--still just fast enough for my needs, but only barely.  I would totally use that solution again, but with faster external comparators instead of the internal ones.