Hacker News Reader

Comment by rasz | Hacker News Reader

You are the problem if you wait two years for F7 instead of retooling.

>We have open orders for thousands of microcontrollers from NXP and Microchip. The problem extends far beyond processors. Texas Instrument power switchers used throughout the industry are in very short supply; we use them for pulling down 15V supplies to 3.3V or to generate +/- and 48V phantom power rails. We also cannot get some ESD protection parts, power MOSFETs and a number of somewhat obscure analog switching ICs. We’re also paying 30-40% more for metal cases and packaging.

Meanwhile less reputable Chinese branded replacements are plentiful and cheap.

View on HN

> obscure analog switching ICs

This one was a surprise to me when I went looking. I'm building a custom audio processor for someone and I couldn't find something as simple as a 2-up dual mux in stock anywhere. I ended up finding a handful of DIP-package dual multiplexers in a drawer in my lab and using those. Luckily it occurred to me to check availability before doing a PCB layout and I only needed 3, but seriously! A basic analog switch was the last thing I though I'd have trouble finding.

> less reputable Chinese branded replacements are plentiful and cheap.

Do they work? What's the low reputation due to?

We have tested a Chinese replacement for a TI op amp SKU: it was terrible.

You got lucky. Sometimes part evaluation comes out all nominal within datasheet spec, until 6 months later it isnt because someone found a "clever" optimization at the fab, or ran out of particular part and decided to temporarily substitute, cha bu duo strikes again. Even Apple is not immune https://www.macrumors.com/2022/05/04/apple-drops-boe-display...

> Meanwhile less reputable Chinese branded replacements are plentiful and cheap.

Not on the power switchers, oddly.

DC-DC Buck/Boost regulators have been phasing in and out of existence like ghosts. Yes, even the Chinese ones. Even worse, changing a buck/boost regulator generally has lots of knock-on effects--the inductors change size so you just wasted a bunch of inductor stock, your swicthing fdrequency changes so your EMI is all over the map, etc.

Retool for what? How do you know whatever you choose will be available for production?

We buy what’s available in stock, design with these parts small runs(500-2000 units). And then the run in 3 months is being made with completely different parts that were available. It’s emergency solution. But it’s better than being out of business.

I'm out of my depth here, but I'm under the impression that synthesizers just don't work that way. My source is a housemate of an old friend, whose profession was restoring a certain sort of mixer popular with house DJs, for whatever that's worth.

Analog synths, at least, just won't sound right if components get swapped. The TR-808 was discontinued, and replaced with the 909, because a critical component went out of production, and it wasn't possible to deliver the sound without it.

It's the difference between a wood shortage for a cabinetmaker and for a luthier. A cabinetmaker can use mahogany where he would use ebony, a luthier is out of business if she can't get certain woods until she can again.

This could be totally inapplicable to modern synths? I wouldn't bet on it though.

The two parts mentioned were an ARM board and DSP. Those are easily retooled for and not like an opamp or something very particular.

> Analog synths, at least, just won't sound right if components get swapped.

I think this is true in some cases for particular analog synths. But my impression is that most synths aren't using unobtainium chips like this. I think you generally can redesign the circuit using alternate components and get to a fairly close sound. But that redesign process is iterative and time-consuming.

By analogy, imagine there a shortage in "compiler invocations". You have a giant program written in one language, but you can't run a compiler on it. Could you port it to another language without a shortage? Sure, but it's a ton of work, and if you end up in another language that has its own compiler invocation shortage, you're no better off.

Both Curtis and SSM chips used in synthesizers like the Sequential Circuits Prophet-5 in the 1970s have being cloned and are in production.

See https://electricdruid.net/analog-renaissance/

>… a luthier is out of business if she can't get certain woods until she can again.

I learned the other day there are people who’s job is to troll antique shops and other related places looking for old furniture to turn into tone wood for the luthiers.

Dont go for fancy new parts, use what Chinese use at volume.

Thats a good way to get a reputation for gear that dies on you. Instrument components need to be extra robust, they are put through a lot of abuse.

Plus, pro audio gear needs to be on the same order of reliability as industrial equipment. For instance, an artist that has a piece of gear fry has potentially wrecked a contracted performance (which can affect revenue as well as reputation). This could also heppen during recording, which ruins very expensive studio time that may be pre-paid for. For most of the software devs here, having a computer fry would be at most a 1 day loss of productivity and a neglible effect on overall company earnings. Not so true when you have a booked venue and your amp or synth just blew up during sound check hours before the show opens.

It's honestly not that dissimilar to how industrial equipment can cause huge revenue effects from a failure due to lost productivity.

And what happens when your Chinese parts are stuck in China for weeks due to their zero-COVID policy?

Reply on news.ycombinator.com