/diy/ - Do It Yourself

>>1729067
>I have junk, what do?
don't be this guy

>>1729086
prove it

>>1729050
I.MX RT10xx have some 500MHz options in QFP. shouldn't be much harder to work with than anything else
but you're getting to the point where leadframe inductance really starts to matter. worth considering taking that step in fabrication technology

>>1729022
>just
not just. for harder jobs it needs to be mildly activated. so add almonds

>>1715294
there are a lot of layers that your single bit goes through from kernel to port register. there's the capacitance of the cable and the low drive strength of most SoC GPIOs to think about. unless you're completely sanguine about missing bits, handshaking is probably going to be a little slow, depending on how you have the scheduler configured, and the scheduler still hates you
speaking for myself, I'd much rather see finished projects, or half-finished projects, or just-started projects, than Pic related

>>1715303
there's an interesting protocol used in the automotive industry between ECUs and sensors called SENT (single-edge nybble transfer). as the name implies, it encodes 4 bits on a variable-width pulse on a power+signal line, without stringent requirements as to regular pulse frequency. if you were to use a similar encoding with a low enough duty cycle and an emitter/receiver pair that responds quickly enough, you might be able to keep the pulses sub-threshold on average
940nm might be long enough to allow a higher duty cycle while remaining sub-threshold. photodiodes are relatively cheap but power LEDs start to get expensive at much longer wavelength

>>1603655

>>1561314
>this will become a meme
>implying it hasn't already
I gotchu fampai