#avr Logs

Aug 08 2021

03:55 AM MetalSutton: Hey all. I am having some trouble getting an avr hat for raspberry pi working
04:03 AM MetalSutton: for some reason avrdude doesnt like my pinconfig file
07:36 AM specing_ is now known as specing
05:52 PM m4t: clock stretching won't work. i guess it deadlocks according to https://github.com/eblot/pyftdi/issues/163#issuecomment-587575609
05:59 PM exp: sorry to hear that
06:01 PM m4t: i will try a diode
06:01 PM m4t: but...it's pulled high
06:02 PM m4t: so that won't work, nevermind
06:04 PM m4t: ah, if only they'd routed ADBUS7 somewhere. the pitch on this is so fine that i'm sure i would short it out if i tried adding a jumper
06:10 PM exp: or bothered learning about I²C before designing their MPSSE lol :/
06:10 PM m4t: heh
06:10 PM exp: people have many criticisms of AVRs
06:11 PM exp: but I switched to using one and instantly solved these issues
06:11 PM m4t: what fw?
06:11 PM exp: LUFA and ASF
06:11 PM exp: but driving TWI manually on an AVR is easy
06:11 PM exp: especially the older ones with the primitive interrupts
06:14 PM LeoNerd: Ohgod is that the FTDI MPSSE? It's *terrible* at I²C. I mean, it's terrible period, but doubly-so for I²C
06:14 PM exp: indeed
06:15 PM LeoNerd: I don't bother with it.. for USB-attached I²C I have a real MCU running a program that acts as a bridge
06:15 PM LeoNerd: Much nicer
06:15 PM exp: yes, literally hooking any random arduino or avr up i found to be far superior than the peripheral supposedly designed specifically for it, lol
06:16 PM m4t: clock stretching does appear to be working on the shukran board
06:16 PM m4t: there's no dedicated pullup on AD7 there
06:16 PM m4t: when i say "appear" i mean i'm not actually doing any delays avr side
06:17 PM * m4t pulls out an old attiny44 board w/clock stretching enabled
06:24 PM m4t: seems to be
06:25 PM m4t: yep, it is
06:25 PM m4t: clocks out 2 bytes of 0xFF without it enabled
06:51 PM m4t: sweet, the stock i2c implementation on this new rev hw (new as of 3 yrs ago, when i bought these and stuck 'em on a shelf) already does clock stretching
06:52 PM m4t: i can turn led on, do _delay_ms(100), write the response, turn led off - still reads out the correct byte from the ftdi
06:53 PM m4t: the only gotcha on the shukran vs tigard seems to be it's required to "i2c.close(freeze=True)" before exiting
06:53 PM m4t: otherwise it gives a nack every other invocation
07:03 PM exp: interesting, i was never able to make it work reliably with a 2232
07:36 PM specing_ is now known as specing
08:00 PM m4t: hmm so this uses CKOUT as the "excitation" signal for capacitative readings
08:00 PM m4t: the dynamic range of the measurements (air vs glass of water) is negatively impacted when running at 8mhz for 3.3v
08:01 PM m4t: it DOES seem to work okay at 16mhz @ 3.3v even though it's technically overclocked / undervolted
10:22 PM rue_mohr: I'd be curious is the software imp. takes less memory than the software for the new hardware
10:29 PM m4t: cp2112 any good?
10:30 PM m4t: or ft200xd
10:30 PM m4t: hrm or https://github.com/torvalds/linux/blob/master/drivers/i2c/busses/i2c-tiny-usb.c
10:31 PM m4t: i guess that could be anything running the fw then
10:32 PM m4t: e.g. https://github.com/daniel-thompson/i2c-star