|
Sagebrush posted:yeah but why would you This except unironically because the chip fast enough to run a garbage collected language will cost 10c to make and use milliwatts
|
# ? Apr 21, 2017 21:51 |
|
|
# ? May 16, 2024 13:57 |
|
it's not insane at all on gateways
|
# ? Apr 21, 2017 21:52 |
|
"the" kernel is a bit of a stretch. A savagely cut-down linux kernel running a single statically linked user space process from a ramfs linked into the kernel itself, yeah, just about. Just in case you wanted to add tens of megabytes of SRAM to your board design for absolutely no good reason.
|
# ? Apr 21, 2017 21:52 |
|
yet people keep using uclinux
|
# ? Apr 21, 2017 22:04 |
|
lol v5 c'mon
|
# ? Apr 21, 2017 22:05 |
|
arm9s are still incredible value
|
# ? Apr 21, 2017 22:08 |
|
is docker actually a (sane, working, meaninful) thing? because i got roped into thinking so, then tried to deploy something docker-based, and it was a horrorshow. which might just as well be a sign that my janitoring expertise has run its course and in fact i don't have the right skillset to manage it, but somewhat egoistically i am tempted to think it is just sort of poo poo
|
# ? Apr 21, 2017 22:27 |
|
don't mind me over here with my cores from this willenium
|
# ? Apr 21, 2017 22:29 |
|
I've been trying for three days to interface a MIPS CPU with a 'Bus Blaster' JTAG adapter and OpenOCD and haven't been successful. I'm not sure if I'm cut out for this embedded stuff after all
|
# ? Apr 21, 2017 23:38 |
|
meatpotato posted:Bus Blaster text me
|
# ? Apr 21, 2017 23:42 |
|
mishaq posted:text me uh nevermind it just started working and I don't know why it works now but didn't before. I hate that feeling.
|
# ? Apr 22, 2017 01:41 |
|
JawnV6 posted:don't mind me over here with my cores from this willenium hey the last variant was released in 2006 but really in smaller quantities (like 1k) what else can you run full linux systems on thats less than $5.
|
# ? Apr 22, 2017 02:30 |
|
Cybernetic Vermin posted:is docker actually a (sane, working, meaninful) thing? because i got roped into thinking so, then tried to deploy something docker-based, and it was a horrorshow. which might just as well be a sign that my janitoring expertise has run its course and in fact i don't have the right skillset to manage it, but somewhat egoistically i am tempted to think it is just sort of poo poo it's nice in dev but production is a horror show of incomprehensible products that are impossible to spell
|
# ? Apr 22, 2017 04:45 |
|
I want to upgrade the ROMs in my Amiga's SCSI card so it seems like a good time to pick up an EPROM burner but I don't know what to get. The ROMs are 27256s. Can I just buy a junky Chinese USB EPROM burner and an eraser box off eBay or what?
|
# ? Apr 25, 2017 07:34 |
|
i got a TL866 from ebay. they're cheap and program everything you're likely to come across. if you shop around you can usually get a few adaptors thrown in for non-DIP chips. the software isn't a flaming pile of dogshit either, which alone puts it above 95% of programmers
|
# ? Apr 25, 2017 10:49 |
|
Cybernetic Vermin posted:is docker actually a (sane, working, meaninful) thing? because i got roped into thinking so, then tried to deploy something docker-based, and it was a horrorshow. which might just as well be a sign that my janitoring expertise has run its course and in fact i don't have the right skillset to manage it, but somewhat egoistically i am tempted to think it is just sort of poo poo it's both, op. docker is a shitshow, but it's the best shitshow on the market by a mile, so you'd best just learn to live with it.
|
# ? Apr 25, 2017 14:56 |
|
Sweevo posted:i got a TL866 from ebay. they're cheap and program everything you're likely to come across. if you shop around you can usually get a few adaptors thrown in for non-DIP chips. the software isn't a flaming pile of dogshit either, which alone puts it above 95% of programmers man, that looks handy thanks!
|
# ? Apr 25, 2017 19:35 |
|
some uart questions for you folks how much accuracy is needed when matching baud rates between devices? the driver i was given uses lovely integer maths to program a fractional divider and can be off by as much as 12%. how fast can you practically run a uart?
|
# ? Apr 27, 2017 11:11 |
|
above 3% and I start getting skeptical a bog standard uart idk above 3 mbps and I'd start getting skeptical if you controlled both endpoints and routing between them you could push it dramatically faster than that
|
# ? Apr 27, 2017 15:54 |
|
thanks. i fixed things up quite a bit, now the matching error ranges from 0 - 1.5% across the 0 - 3 mbps range. can't do better since the core clock limits the sampling rate.
|
# ? Apr 27, 2017 17:01 |
|
ah, I was just thinking about this stuff. I'm trying to use an attiny as a tachometer sensor coprocessor, which means it needs a relatively accurate clock both to do the timing and to communicate with the master microcontroller. I could just stick in a crystal but I'm trying to see if I can get sufficient accuracy and stability from the internal oscillator instead. Turns out you can write to the OSCCAL register to tweak the chip frequency, and by doing that I got it down to 8.01 MHz at 5 V, so better than 0.2%, which is accurate enough for reasonable speed UART and certainly good enough for a tachometer. Only thing is I don't know how stable that is over time and with changes in temperature. Eventually this will go in a motorcycle, in a black box that may heat up to 110° in the sun and cool down to near freezing at times. I don't really care if the speedometer is out by a little bit ( legally, speedometer only has to be within -3%, +10% or something) but if it loses communication that would be annoying. I suppose crystals are like $0.35 for a 10 PPM unit but you know, I like the efficiency of doing everything in one chip. Incidentally, I was looking on digikey at crystals and there's one that is 8.000135 MHz. what would that be used in?
|
# ? Apr 27, 2017 17:56 |
|
Sagebrush posted:ah, I was just thinking about this stuff. I'm trying to use an attiny as a tachometer sensor coprocessor, which means it needs a relatively accurate clock both to do the timing and to communicate with the master microcontroller. I could just stick in a crystal but I'm trying to see if I can get sufficient accuracy and stability from the internal oscillator instead. Turns out you can write to the OSCCAL register to tweak the chip frequency, and by doing that I got it down to 8.01 MHz at 5 V, so better than 0.2%, which is accurate enough for reasonable speed UART and certainly good enough for a tachometer. i dont know what you are using specifically but the first random atmel datasheet i found on digikey had a nice graph
|
# ? Apr 27, 2017 18:14 |
|
a floating point glitch in the matrix
|
# ? Apr 27, 2017 18:29 |
|
Sagebrush posted:Incidentally, I was looking on digikey at crystals and there's one that is 8.000135 MHz. what would that be used in? add another 8 kHz to that and I'm sure you could use it in a calculator somewhere op
|
# ? Apr 27, 2017 19:00 |
|
does it have to be uart? just use i2c or spi
|
# ? Apr 27, 2017 19:02 |
|
uart chat: so the theory is that when the transmitter sends a start bit, the receiver should always use that edge to resynchronize itself to the transmitter. clock skew between the two and issues on the line (eg long line = slow rise/fall times) is the limit on how many bits can now be successfully transferred usually you're transmitting 10 bits because almost everyone uses 8N1, so the frame is 1 start bit 8 data bits 1 stop bit. if the timing difference between the two ends is 10% of a bit time, by the end of a 10 bit frame the timing error is 100% of a bit time, which is Bad some fancier receivers might try to resynchronize every time they see an edge and will therefore still do fine (except on bit patterns that don't have enough edges!) but you can't count on that as a transmitter take jawn's advice if you want high bit rates, at megabit rates rs232 is for chumps
|
# ? Apr 27, 2017 19:24 |
|
uart is the easiest + cheapest way to get data outta a thing into a pc at <= 3 mbps
|
# ? Apr 27, 2017 19:34 |
|
spi is nice because you get framing
|
# ? Apr 27, 2017 19:40 |
|
spi is bad because it is grossly underspecified to the point of being meaningless
|
# ? Apr 27, 2017 19:43 |
|
hifi posted:i dont know what you are using specifically but the first random atmel datasheet i found on digikey had a nice graph yeah, the problem with the internal crystal in most micros is that it drifts all over the place. you can easily trim them to <1% accuracy at a fixed temperature, but if the temperature is all over the place then so is the frequency. some micros have an internal temperature sensor, so you can trim the crystal for various temperatures and store the results in a lookup table, then have your code periodically measure the temp and adjust OSCCAL accordingly, but it'll never be as accurate as a decent external crystal - although those do also drift with temperature just not as much. Sweevo fucked around with this message at 19:52 on Apr 27, 2017 |
# ? Apr 27, 2017 19:48 |
|
BobHoward posted:uart chat: so the theory is that when the transmitter sends a start bit, the receiver should always use that edge to resynchronize itself to the transmitter. clock skew between the two and issues on the line (eg long line = slow rise/fall times) is the limit on how many bits can now be successfully transferred i2c is usually going to be slower than async serial though?
|
# ? Apr 27, 2017 19:59 |
|
BobHoward posted:take jawn's advice
|
# ? Apr 27, 2017 20:04 |
|
put in a GPS receiver and use the PPS signal to calibrate the clock, easy and no crystal required!
|
# ? Apr 27, 2017 20:38 |
|
longview posted:put in a GPS receiver and use the PPS signal to calibrate the clock, easy and no crystal required! only $30! https://www.u-blox.com/en/product/neolea-m8t-series
|
# ? Apr 27, 2017 20:48 |
|
you can get a ocxo for $30
|
# ? Apr 27, 2017 20:50 |
|
hobbesmaster posted:i2c is usually going to be slower than async serial though? it can go 400k quite easily although there's lots of devices that don't support fast mode in theory you can do a megabit but device support for that is really rare ime (idk why I've implemented the hardware and it's still the same poo poo)
|
# ? Apr 27, 2017 21:03 |
|
i think the big deal with high speed i2c is you need a lower input threshold. might need a stronger pulldown transistor as well. the pcf8574 is 100 khz only for example but it works at 400 khz as long as the bus capacitance is low enough/pullups are strong enough
|
# ? Apr 27, 2017 21:05 |
|
BobHoward posted:it can go 400k quite easily although there's lots of devices that don't support fast mode the devices I've seen that support 1m i2c have warnings about actually doing your math on the bus capacitance and pull ups instead of haphazardly hooking it up
|
# ? Apr 27, 2017 21:09 |
|
inter chip ethernet for everything imo
|
# ? Apr 27, 2017 21:28 |
|
|
# ? May 16, 2024 13:57 |
|
Bloody posted:spi is bad because it is grossly underspecified to the point of being meaningless but if you're doing your own project and control both sides of it, it's the best choice no need for super accurate clocks, no pullup nonsense burning power, zippy af
|
# ? Apr 27, 2017 22:08 |