(no subject)

My job title is validation engineer. That means I design hardware and software for testing chips in high power edge situations that our automated testers that verify chip functionality can't manage.

What I actually do, ever since my coworker died in a car crash in November, is 80% of what he used to do, which is take (usually hand-drawn) large circuit drawings from coworkers and convert those into real hardware. Make sense of the schematic, draw it, push that into a program where that is turned into artwork that shows the actual copper traces across fiberglass, send all that artwork off to a company that makes circuit boards, go purchase all the parts needed, send that off to another company that assembles parts onto boards, and document all this through an arcane system that makes sure everyone gets paid.

The problem area is the 'purchase all the parts needed' step. We have a list of maybe a quarter million parts we're allowed to use, because the company has validated that those parts have usable simulation data, three-dimensional models, and a reputable supply chain. The component houses we buy from may list 10M parts or more. The problem is that I have to find the intersection of those two lists. In the past this was merely tedious. But there's something going on in the electronics industry right now: there are massive parts shortages. I'll go to Digikey and check a resistor and they say they have 2 million in stock, and then I go check another and they say that they have none in stock and it's going to take a month before they have any. Some parts are listed as not shipping for 52 weeks, which is the maximum anyone will quote. (In the past, when things like this have happened, about six years later there's a big anti-trust lawsuit.)

For one of my boards, I needed a really fancy transistor. The two usual places I check both said "we don't have any and won't for 28 weeks." The (comparatively) underdog outsider says they have 140 in stock. That's possible: that's so few, no company doing production orders is going to want that. I ordered 20. Usually we receive stuff in three days. Four days came and went and I emailed them, and, even though they *still* list 140 in stock, they're all "oh, no, those are backordered 28 weeks" but because of our weird procurement system there's no way for them to let me know.

I also need a fancy inductor, which I ordered using their cute order form that has me enter my shipping address. The company instead googled our company name, found an address for a research group that hasn't been in business for 5 years, and sent the order there, and when it was returned, a guy at the distributor through whom I ordered it, who knows me well, found it sitting on a shelf and physically drove it up to me. (He works about 50 miles away, and he is a salesman so he tried to get us interested in some other stuff: this is less valiant than it sounds, but it's still valiant.)

So that was a lot of my day. Late afternoon, I had at least addressed, if not solved, most of the issues, and was moving onto my next problem, of modifying 80 identical circuits on a massive board we use for stress testing at high temperature.
You can see a shiny little snippet of silvery wire, that I'm adding in by hand, as part of the rework at one point. The huge pinkish semicircular object at the bottom is part of one of my fingernails, to provide scale.

So that's going to take a while.

My coworker T walked in and said "can you help me with a problem?"
This took a lot of drawing out before I understood the scope of the problem.
Three years ago we built a chip with a really neat function where it automatically calibrates itself as it powers up. Most people loved it, but didn't actually buy any. One company bought a lot, and then they bought a lot more, and now they're asking for massive quantities, but they *hate* the calibration function because they want to do their own calibration function, and they want each chip to always come up in exactly the same state, which it won't if it's self-calibrating. They're buying so many chips, it was worthwhile to build a separate run of chips that has that calibration function turned off.
The modified chips come in Friday morning. Our people-who-talk-to-customers said we can ship them out Friday to the customer. My manager says it is crazy to not test that the fix we made actually works. So there's a four hour window during which we need to test the parts, and that includes: unpacking them, getting them through the production test system, packing them back up, and shipping them off, and within that window, my manager wants the validation program I wrote to be run.

I'm not going to be at work Friday. I'm taking care of Monty who just got out of surgery. My coworker T isn't going to be there: he's going to his grandmother's funeral. My manager is in Germany. The product and program manager (inconveniently the same person, which we usually try to avoid) is hanging out at the hospital because his wife had surgery today.

We grabbed our brand new product engineer trainee, who works right across the hall from me, and said "hey, want to do some product engineering?"
In an hour I put together a test rig to run three systems in parallel, with a (hopefully) simple enough user interface for her to run it and analyze the data. We had her go through running the program and figuring out all the glitches. But, oof, it was a stressful way to end the day, desperately trying to manage resources to do things like find the address to which we should send them (the project manager knows, but he's incommunicado) and find someone who has the skills to swap chips on the boards I set up (there are only three of us that are really good at soldering, and two of us are gone tomorrow) and get physical resources like the temperature chamber this is running in set up and clear of time conflicts.

I'm kind of glad I have the next couple of days off.