[dmitrygr]

I hope there is a follow up where the LED is ground down and a PICKIT2 is used to read out the code :)

[farkanoid]

PicKit2 and MPLAB8 made MCU programming a pleasure. Everything was so damn snappy and responsive. Feels as though everything went downhill with the advent of PicKit3 and MPLABX.

Edit: The site only exists on the wayback machine now, but there was a hate page [1] for the PicKit3 posted on Dave Jones' twitter some time ago

[1] https://web.archive.org/web/20180423225612/http://www.fuckit...

[joezydeco]

I don't buy the PIC chip theory since that means it's either programmed in-situ which seems imposible, or they're ordered with premasked or preprogrammed ROM which is hellishly expensive.

[crote]

It could also be programmed during production. Clearly the LED manufacturer has the equipment to do wire bonding, so why not use the occasion to program it too?

It'd be a bit comparable to the test and assembly process of the WS2812B, see [0] @ 2:30 or 6:10.

[0]: https://www.youtube.com/watch?v=pMjhJ9kcaU4

[joezydeco]

Anything is possible, I suppose, but burning EEPROM is an order of magnitude slower than anything these robots are doing.

[crote]

I'm not too worried about that. The slowest part of EEPROM & flash operation is erasing an already-used page - and that's not the case here.

Chips like these only have 1kB of flash - I'd be surprised if you couldn't program them in less than 100ms.