#avr Logs
Nov 22 2017
#avr Calendar
12:00 AM cehteh: crystal and its caps should be also physically close to the pins (and what huge crystal do you use?) .. well in this simple stuff it doesnt matter anyway
12:06 AM thornekey: um the software is express pcb
12:07 AM cehteh: try to learn kicad
12:08 AM thornekey: yea i want to improve my pcb skills :)
12:09 AM thornekey: http://picpaste.com/pics/ex-x0Jf7yko.1511330931.png ok hows this :)
12:10 AM cehteh: awful
12:10 AM cehteh: but may work :D
12:10 AM thornekey: :S
12:11 AM thornekey: so try with the 27n cap but may not work right
12:12 AM cehteh: it possibly works, just try
12:12 AM thornekey: cool :)
12:12 AM thornekey: thanks for ur help
12:12 AM Thrashbarg: it'll be fiiiine... I've used 47pF as a temporary measure, it's fine
12:12 AM cehteh: well i dont gurantee that, and there are enoug other ways to fuck up
12:12 AM Thrashbarg: they're just load capacitors, they change the frequency of the crystal
12:13 AM cehteh: yes
12:13 AM cehteh: eventually the crystal wont swing anymore .. and the crystal driver will draw bit more current than it should
12:13 AM cehteh: but that wont be a worry here
12:14 AM thornekey: yea i only need it to work once ahaha
12:14 AM cehteh: when it works, it works, maybe not reliable, but once you programmed the avr and verified the the flash it is ok
12:14 AM thornekey: yeah
12:14 AM thornekey: :D
12:15 AM thornekey: http://www.instructables.com/id/Arduino-on-all-sorts-of-Atmels/ so see step 2. it says i have to use the corefiles and load that program onto the arduino. once i do that can i just hook it up to the computer and use any programmer software?
12:15 AM thornekey: to get the .hex on
12:15 AM cehteh: just be careful when you mess with the fuses, otherwise you brick your chip, unless you have a high voltage programmer
12:16 AM thornekey: ohk
12:16 AM thornekey: this thing is only 5v
12:16 AM cehteh: i give a shit about instructables.com .. maybe it works, it may have worked for someone else once. but rarely things there are really correct
12:16 AM thornekey: oh yeah
12:16 AM thornekey: hmm
12:17 AM cehteh: high voltage programmer is some special tool, putting 12V to the reset pin (and only there) and using some special protocol for programming
12:17 AM cehteh: that way you can recover an AVR with misprogrammed fuses
12:17 AM thornekey: ah right
12:17 AM cehteh: otherwise, wrong fuses may mean you fucked it up
12:18 AM cehteh: sometimes recoverable with some tricks
12:18 AM cehteh: but generally pita
12:18 AM cehteh: lucily thats the only way to really fuck up an avr, so be extra, double, triple careful when you set the fuses
12:19 AM cehteh: do it once, do it correct
12:19 AM thornekey: :S
12:20 AM thornekey: do u sometimes not have to set them
12:20 AM thornekey: or always
12:20 AM thornekey: what i mean is, is there a default
12:20 AM cehteh: you need to set them when you want to use the external crystal
12:20 AM cehteh: by default it runs on internal 8Mhz divided by 8 -> 1Mhz oscillator
12:21 AM cehteh: often enough i am fine with that
12:21 AM cehteh: software can switch the divider it to 8mhz. done
12:21 AM cehteh: some tinys can even run at 16mhz from internal osc and pll
12:22 AM cehteh: but the internal oscillator is quite inprecise
12:22 AM cehteh: depends on your usage
12:22 AM thornekey: gosh a lot to take in :)
12:22 AM cehteh: on the other hand it is tuneable, so with some external timing signal you can adjust/calibrate the frequency somewhat
12:24 AM cehteh: and you have 2 more io pins because the xtal pins are free then
12:24 AM cehteh: so, you get advantages and disadvantages ..
12:24 AM thornekey: hmm, so does this fuse calc help me http://picpaste.com/pics/ex-i5uIE9H3.1511331841.png
12:24 AM thornekey: i set it to 8515
12:25 AM cehteh: usually the fusecalc is correct, just dont mess with the values :D know exactly what you do
12:25 AM thornekey: ok cool :)
12:25 AM thornekey: :D
12:25 AM cehteh: some CKSEL needs to be set for the external crystal
12:25 AM cehteh: which ones .. -> datasheet
12:27 AM thornekey: o i cant see that in the datasheet
12:29 AM cehteh: you havent read it
12:30 AM thornekey: the only place it mentioned the fuses (that i could see) is pg4
12:30 AM cehteh: http://www.atmel.com/images/doc2512.pdf
12:30 AM cehteh: maybe you should look at the real datasheet, not the short overview pdf
12:32 AM cehteh: now have fun studying that ..
12:32 AM thornekey: ok so CKSEL0 is 0 (unprog).
12:33 AM cehteh: i dunno, would need to read the datasheet as well, but want to do my stuff here :D
12:34 AM cehteh: while at it .. better strive to read the whole datasheet, at least the most interesting parts, about how the pins must be connected, clocks selected, power and fuses
01:58 AM rue_bed: wait
01:58 AM rue_bed: dude
01:58 AM rue_bed: use the online fuse calculator?
01:59 AM rue_bed: why would you try to work it all out from the datasheet?
01:59 AM rue_bed: http://www.engbedded.com/fusecalc/
02:00 AM cehteh: rue_bed: suggested that, but nevertheless he should read the datasheet
02:01 AM rue_bed: sure, but having the datasheet, I usually dont trust myself not to screw up the clock and brick my avr, I let someone else screw it up
02:01 AM rue_bed: and sofar, they never have
02:03 AM cehteh: i intentionally broke one tiny13 once .. wanted to test the resetdisable fuse .. flashed test prog, disabled reset, tested something, tossed it away
02:05 AM rue_bed: cheaper to do with a mega8
02:06 AM rue_bed: how many pins can I saw off the end of a mega8 to make it a pysically smaller dip beofre I cant program i anymore?
02:06 AM cehteh: haha
02:07 AM thornekey_: i did check the calc it said SUT1, SPIEN, CKOPT, BOOTSZ1, BOOTRST all = 0.
02:07 AM thornekey_: the only think i could see in the datasheet was CKSEL0 = 0
02:07 AM thornekey_: which the calc didnt list
02:07 AM rue_bed: no use the hex codes form it so you dont scew up the avrdude command
02:08 AM thornekey_: i did use the hex DF = low CA = high
02:08 AM thornekey_: do i just use that in avrdude?
02:09 AM rue_bed: do you kinow how/?
02:09 AM thornekey_: just can do it in the console
02:09 AM rue_bed: usbfuseint:
02:09 AM rue_bed: avrdude -c avrisp2 -P usb -p m32 -U hfuse:w:0xD9:m -U lfuse:w:0xE1:m
02:10 AM rue_bed: thats from my makefile for a mega32
02:11 AM thornekey_: rue_bed, btw i have a .hex file someone else made. i just want to flash it onto the atmega8515
02:11 AM thornekey_: do i need the makefile and .c file
02:11 AM rue_bed: no
02:11 AM thornekey_: cool :)
02:11 AM rue_bed: install:
02:11 AM rue_bed: avrdude -c pony-stk200 -p m32 -e -U flash:w:$(PRG).hex
02:12 AM thornekey_: that just the location of the hex?
02:12 AM thornekey_: so first set fuse bits and then flash :)
02:13 AM rue_bed: you will have to adjust things based on you hardware and stuff
02:13 AM thornekey_: its a shield for arduino uno
02:13 AM thornekey_: and using atmega8515
02:38 AM anonnumberanon: is there a way to toggle that pin with less instructions? I used avr-gcc -O0. paste.debian.net/996992/
02:43 AM Emil: anonnumberanon: in most megas you can do PINx=value; to toggle any pins where you write a 1 to
02:43 AM Emil: Also
02:43 AM Emil: never
02:43 AM Emil: ever
02:43 AM Emil: fucking
02:43 AM Emil: use -O0
02:44 AM anonnumberanon: I'm using it because I want to understand the assembly output.
02:44 AM Emil: Don't
02:44 AM Emil: use -Os
02:44 AM anonnumberanon: why
02:44 AM Emil: Because -O0 is idiotic
02:44 AM Emil: It does not generate realistic assembly
02:44 AM Emil: It does everything and I mean everything in a very stupid way
02:45 AM anonnumberanon: okay i'll keep that in mind and try -Os but I'm not there yet, still trying to understand the instruction set
02:46 AM Emil: If using the PINx method it takes a single cycle to ldi a mask to a register (plus additionally push if no space) as a setup and then it's single cycle to toggle leds with the out instruction
02:47 AM Emil: I done as PORTx^=MASK it's ldi, in, xor, out meaning three cycles and one to setup
02:48 AM cehteh: anonnumberanon: most AVR's have support for toggeling pins in hardware, you write a 1 bit into the PIN register instead the PORT register
02:48 AM Emil: Plus any push instructions if you don't have space
02:48 AM Emil: cehteh: did attinys have that too?
02:49 AM cehteh: iirc newer ones do
02:49 AM cehteh: check datasheet
02:49 AM Emil: cehteh: and iirc it was only some (relatively) newer megas hve it
02:49 AM Emil: cehteh: but I already mentioned the PINx method ;)
02:49 AM anonnumberanon: cehteh, what you said is like setbit()
02:49 AM cehteh: no
02:49 AM anonnumberanon: but you say it toggles?
02:49 AM cehteh: yes
02:49 AM Emil: anonnumberanon: setbit _sets_ bit
02:49 AM anonnumberanon: cehteh, come on now
02:49 AM cehteh: as in high becomes low and low becomes high
02:50 AM Emil: clearbit _clears_ bit
02:50 AM Emil: toggle means inverse
02:50 AM anonnumberanon: you write to 1 and that becomes 0?
02:50 AM cehteh: what do you want to do?
02:50 AM anonnumberanon: with ldi?
02:50 AM anonnumberanon: i want to toggle a pin
02:50 AM Emil: anonnumberanon: yes
02:50 AM anonnumberanon: anyway
02:50 AM anonnumberanon: ill go try it now
02:50 AM Emil: anonnumberanon: we have already given you the solution
02:50 AM anonnumberanon: with PINB0
02:50 AM cehteh: then check if your AVR supports that hardware method
02:51 AM Emil: anonnumberanon: PINB0 is not a thing
02:51 AM cehteh: thats 1 instruction for up to 8 outputs, you cant make it faster :D
02:51 AM Emil: cehteh: plus setup!
02:51 AM Emil: but that doesn't affect the O(number) of it
02:52 AM cehteh: doesnt need setup iirc byte->register transfers are immediate or?
02:53 AM Emil: cehteh: you need the mask
02:53 AM cehteh: not sure .. i'd have to check too, but anyway if thats supported in hardware you hardly can make it faster
02:53 AM Emil: that's a single ldi to setup
02:53 AM cehteh: when you are in a very performance critical thing and do that very frequently you may reserve a register for the mask
02:54 AM Emil: yes
02:54 AM Emil: I know
02:54 AM Emil: The instruction count approaches zero as toggles increase
02:54 AM Emil: approaches 1*
03:00 AM cehteh: heh just tested and gcc only optimizes that with optimization enabled of course
03:01 AM Emil: of course
03:01 AM Emil: because -O0 is stupid as fuck
03:01 AM cehteh: i thought some peephole optimizations are always on
03:01 AM Emil: It literally generates shit code
03:01 AM cehteh: and not -O0 ... just no -O option
03:02 AM Emil: cehteh: that's the same as -O1 or -O
03:03 AM Emil: Which also produces shit code
03:03 AM Emil: Prettymuch always -Os or -O3
03:03 AM cehteh: -O1 optimizes it already
03:04 AM anonnumberanon: -Os has O2, that's already a lot of optimization
03:04 AM anonnumberanon: I hope I can still understand the output with that
03:04 AM cehteh: anonnumberanon: either way .. you can toggle up to 8 outputs with 1-2 instructions when the hardware supports it and done right
03:05 AM cehteh: ldi r24,lo8(3)
03:05 AM cehteh: out 0x16,r24
03:05 AM anonnumberanon: yeah im looking through the datasheet to see if the pin register is writable
03:05 AM cehteh: PINB = 3;
03:05 AM cehteh: the question is not if its writeable but if it supports the 'toggeling' there
03:06 AM cehteh: iirc all newer AVR's do (that includes tiny25,45,85 and more) but i am not sure, go figure out
03:06 AM cehteh: maybe even the old megas dont support it
03:07 AM cehteh: or just try :)
03:07 AM anonnumberanon: "Writing a logic one to PINxn toggles the value of PORT
03:07 AM anonnumberanon: xn"
03:07 AM cehteh: :)
03:08 AM anonnumberanon: but my compiler does not recognize PINB0 as an lvalue
03:08 AM cehteh: of couse not
03:08 AM cehteh: [10:05] <cehteh> PINB = 3;
03:08 AM cehteh: for example
03:09 AM anonnumberanon: yeah but that makes no sense
03:09 AM anonnumberanon: pinb?
03:09 AM anonnumberanon: what would it make sense
03:09 AM anonnumberanon: why i mean
03:09 AM anonnumberanon: oh i see
03:09 AM anonnumberanon: but that would be like PORTB, rather
03:09 AM anonnumberanon: PORTB = 3;
03:09 AM cehteh: no
03:10 AM cehteh: that would set the bits, not toggle
03:10 AM anonnumberanon: yeah i know
03:10 AM rue_bed: if you write to the input only port, it toggles the pin
03:10 AM rue_bed: (under the new core arch)
03:10 AM anonnumberanon: but which pin are you writing to with PINB?
03:11 AM cehteh: pinb is the whole port, up to 8 pins
03:11 AM cehteh: you have to provide a mask of the bits to toggle
03:12 AM cehteh: =3 for example toggles the lower two outputs
03:12 AM cehteh: and leaves the upper 6 untouched
03:12 AM anonnumberanon: ok that works
03:13 AM Emil: anonnumberanon: don't you read what I said? :D There's no such thing as PINB0
03:13 AM anonnumberanon: well, i'll rephrase, that compiles
03:13 AM cehteh: PORTB = 6; for(;;) PINB = 3; -> 00000110 , 00000101, 00000110, ...
03:13 AM Emil: cehteh: don't forget DDRB
03:13 AM anonnumberanon: yeah
03:14 AM cehteh: assumed that, just showing what happens when you toggle data
03:15 AM anonnumberanon: I can use that and try different optimization flags now.
03:16 AM Emil: anonnumberanon: the best you can achieve is ldi, out
03:16 AM anonnumberanon: Emil, okay. Hopefully that's what I get :)
03:17 AM anonnumberanon: let's say with -Os, if not then I need to do it by hand
03:24 AM cehteh: what do you try to do anyway?
03:25 AM cehteh: bitbanging some bitstream? or changing multiple outputs at once? why is it so performance critical?
03:26 AM Emil: cehteh: probably just interested and learning
03:32 AM anonnumberanon: wow
03:32 AM anonnumberanon: yeah Emil that's what i got, ldi and out
03:34 AM Emil: ;)
03:35 AM anonnumberanon: which are both 1 clock..man
03:36 AM anonnumberanon: yeah cehteh I am learning about how to make many PWMs without jitter
03:37 AM anonnumberanon: an no interrrupts, PLEASE don't bring that up :)
03:37 AM Emil: eh?
03:37 AM Emil: Why not?
03:37 AM Emil: You can't have pwm without jitter without interrupts or hardware, really
03:37 AM anonnumberanon: they take too long for this
03:37 AM Emil: wtf wat
03:38 AM anonnumberanon: yes i tried interrupts and NOPs and interrupts took longer
03:38 AM anonnumberanon: i could not reduce my PWM wavelength with them
03:38 AM anonnumberanon: only NOPs worked
03:39 AM anonnumberanon: but it may have been my testing method
03:39 AM Emil: NOPs?
03:39 AM Emil: You don't need no stinking NOPs
03:40 AM daey: neither nops nor interrupts affect the hardware pwm output :')
03:41 AM daey: thats the whole point of having it
03:41 AM anonnumberanon: so with interrupts I noticed that the wave was nicely jitter-free but could not be made small
03:42 AM anonnumberanon: i was trying for exactly 1 microsecond but it did not make it
03:43 AM anonnumberanon: i was doing an overflow interrupt and setting the counter to something like 250 out of 256 so that it would only count 6 times before overflowing
03:43 AM anonnumberanon: (at 16 megahertz)
03:44 AM cehteh: interupts give jitter, because they can only happen when instructions complete (and some instructions taking 2 cycles) even without disabling interrupts that may jitter
03:44 AM anonnumberanon: daey, yes but there are not enough hardware PWM pins
03:44 AM Emil: anonnumberanon: you really canä't
03:44 AM cehteh: when you really want to do that in software you need a constant time loop
03:45 AM cehteh: thats almost impossible to do reliably in C
03:45 AM Emil: anonnumberanon: your only option is to buy a harware pwm chip that you talk to with some protocol
03:45 AM cehteh: at least when you have to check inputs and branch
03:46 AM Emil: anonnumberanon: if you can dedicate your avr to pwm then it's somewhat possible
03:46 AM cehteh: but i'd recommend to do hardware PWM as well
03:46 AM cehteh: use an avr with enouh pwm channels or external pwm controller
03:46 AM anonnumberanon: yes cehteh i also tried that, it didnt work (the constant time loop), i didn't try it in assembly though
03:47 AM cehteh: when you cant do that, then you picked the wrong chip for the job
03:47 AM cehteh: the constant time loop will work for sure .. if done right
03:47 AM anonnumberanon: with any chip this would not work?
03:47 AM cehteh: not that i recommend that but it can be done that way
03:48 AM anonnumberanon: I mean I've tested it working with a constant pulse width. But obviously I don't want a constant pulse width.
03:48 AM cehteh: but hardware pwm is much easier and you have the mpu to do other more generic useful stuff like fetching data which should be pwm'ed
03:48 AM daey: a for(;;) { toggle_pin; } should give you a constant frequency
03:48 AM anonnumberanon: daey, I want variable frequency though
03:49 AM cehteh: use hardware
03:49 AM cehteh: if you dont get it working with hardware pwm then you doing it wrong, in software it makes hardly sense in almost all cases
03:49 AM daey: anonnumberanon: thats possible but you would have to know exactly how long every task in your program needs. so its unrealistic
03:50 AM cehteh: and some frequencies are 'impossible' and you need to approximate then, depending on your clock frequency, dividers and counter's top
03:50 AM cehteh: if thats not ok, -> you picked the wrong hardware
03:51 AM daey: trying to build an lcd driver?
03:51 AM anonnumberanon: because i need from 500 to 2500 microseconds i figured if i can have it down to the microsecond then the rest is just a multiple of it
03:52 AM anonnumberanon: daey, no, a servo driver
03:52 AM cehteh: how many channels?
03:52 AM anonnumberanon: 18
03:52 AM anonnumberanon: 17, really...
03:52 AM Emil: eh?
03:52 AM Emil: that's super easy with interrupts
03:53 AM cehteh: i'd use a lot attinys (or whatever has 16bit timers) each one only driving as much servos as it can (2 per timer)
03:53 AM cehteh: making it easy
03:53 AM cehteh: then generate the servo signal in hardware and use the mpu for communicaton uart/i2c/spi to send data to the servos
03:54 AM cehteh: with some efforts you can make smart servos this way, use the adc to measure current, include calibration curves, maximum speeds, acceleration/deceleration, endpoints and more
03:55 AM anonnumberanon: ok but what about the case where 2 servos on the same chip have pulse width of 1 microsecond of difference, I'm trying to reduce time so that I won't end up having one pulse at "555" and one at "556" which wave will stop at 555 and 559
03:55 AM anonnumberanon: if we consider that the first servo just finished servicing an interrupt and that took time, then the other one interrupts late, so at 559 instead of right after 556
03:56 AM cehteh: servo signals are not totally simple because they need this short exact pulses and lots of pause
03:56 AM cehteh: but it is doable
03:56 AM cehteh: you dont need to do one interrupt per servo
03:57 AM cehteh: when the time is to short for interrupt handling you can do a busy wait for some microseconds
03:57 AM cehteh: but .. again .. try to make the signals in hardware
03:57 AM anonnumberanon: you mean hardware pwm or external chips?
03:58 AM cehteh: hardware pwm
03:58 AM cehteh: with 16bit counter
03:58 AM cehteh: thats the lazy way :)
04:00 AM cehteh: what servo protocol do you want to speak? good old 50hz? 400hz .. something else?
04:00 AM anonnumberanon: 50 hz
04:02 AM cehteh: so pick a atiny with 2x 16bit counters .. caclulate which divider/counter TOP you need to get the best resolution at 50hz .. done
04:03 AM cehteh: (or even a smaller atmega)
04:04 AM anonnumberanon: why the 16 bit counters?
04:04 AM cehteh: because you will need the resolution
04:05 AM anonnumberanon: yes but doesn't a 16 bit counter have less resolution than an 8 bit counter?
04:05 AM cehteh: huh?
04:05 AM anonnumberanon: wait im tired
04:05 AM anonnumberanon: but yes
04:05 AM cehteh: one cycle = 20ms, with 16bit that 65536 steps (at best, didnt do the exact math)
04:06 AM Emil: anonnumberanon: servo signal is 50Hz
04:06 AM cehteh: you are only intersted in the first 1.5ms there
04:06 AM anonnumberanon: the 8 bit counter would overflow after 256 cycles right?
04:06 AM Emil: anonnumberanon: 500u is your _minimum_ signal on time
04:06 AM Emil: and maximum is 2.5ms
04:06 AM cehteh: i am just trying to go the lazy way
04:06 AM Emil: And 0.5ms is really fucking short, are you sure about that
04:07 AM Emil: because it's usually 1ms-2ms
04:07 AM Emil: Range
04:07 AM Emil: anonnumberanon: so if you are making a servo driver you aren' thinking clearly because with interrupts it works jsut fine
04:07 AM cehteh: that old servo signal sux .. you may check if your servos support 400hz, many do
04:07 AM Emil: That's only 4 times as fast
04:08 AM cehteh: yes still ok
04:08 AM cehteh: anyway for the lazy way without interrupts only hardware pwm
04:09 AM Emil: And your minimum trigger is 500u which means 8000 cycles at 16MHz
04:09 AM Emil: and maximum is 2500u
04:09 AM Emil: which means 40000
04:09 AM Emil: cycles
04:09 AM Emil: So interrupts are more than enough here
04:10 AM cehteh: yes there are other approaches
04:11 AM cehteh: the 20ms timeframe doesnt need to be exact, only the high signal must
04:11 AM cehteh: so you can also focus on that
04:11 AM anonnumberanon: i was under the impression the 20ms windows needs to be accurate
04:12 AM Emil: servos don't really need super accurate timing
04:12 AM cehteh: nop
04:12 AM cehteh: anythings between 80-22ms works for sure
04:12 AM cehteh: 18
04:12 AM cehteh: prolly even more error
04:12 AM anonnumberanon: on the servo is it an ADC that reads voltage made by the pwm, or actually a pwm reader? unclear on that part
04:13 AM cehteh: but the short peak defining the servo position needs to be accurate
04:13 AM cehteh: depends on the servo
04:13 AM cehteh: the 50hz protocoll originates from 60 years ago and was all analog in the beginning
04:14 AM cehteh: nowadays with asics and digital servos things are internally handled differently
04:14 AM cehteh: you shouldnt care
04:15 AM anonnumberanon: i don't have digital servos
04:16 AM anonnumberanon: digital means you communicate with them with serial
04:16 AM anonnumberanon: analog means with pwm
04:16 AM Emil: pls
04:16 AM Emil: no
04:16 AM Emil: stop
04:16 AM Emil: that's so wrong it hurts
04:16 AM anonnumberanon: lol
04:16 AM Emil: your basic servo controller counts time
04:17 AM Emil: It's not ADC based
04:17 AM anonnumberanon: ok
04:17 AM Emil: the only ADC in servo controllers is reading the feedback from the motor, if it's voltage divider based
04:17 AM cehteh: digital servos use the same protocol
04:18 AM cehteh: there are (rare) servos which use sbus or digital protocols but thats another thing
04:18 AM Emil: there are uart servos, sure
04:18 AM anonnumberanon: a "digital servo" is what is known as a servo to talk to with I2C
04:18 AM Emil: or any protocol
04:18 AM Emil: anonnumberanon: pls stop
04:18 AM cehteh: no
04:18 AM cehteh: its just that the controller internaly is a microcontoller and positioning it digitally
04:19 AM cehteh: they speak the same old 50 or newer 400hz protocoll
04:19 AM Emil: "protocol" is also a stretch
04:19 AM cehteh: whatever
04:19 AM cehteh: PCM :)
04:19 AM Emil: PWM
04:20 AM cehteh: whatever
04:20 AM anonnumberanon: cehteh,
04:21 AM anonnumberanon: how would you run an I2C servo by talking to it with a PWM lol
04:21 AM anonnumberanon: you wouldn't
04:21 AM cehteh: the 'digital servos' sold as 'digital servos' are NOT i2c
04:22 AM anonnumberanon: well this is a slippery slope because many PWM servos are sold as "digital" and they don't have micros in them
04:22 AM anonnumberanon: just chips
04:23 AM anonnumberanon: not digital chips
04:23 AM anonnumberanon: it's a marketing lie
04:23 AM cehteh: servos usually have some asics .. analog or digital
04:24 AM cehteh: https://hobbyking.com/de_de/corona-digital-servo-1-8kg-0-12sec-9g.html
04:24 AM cehteh: -- for example .. is a 'digital' servo .. but speaks the 50hz pwm
04:25 AM cehteh: possibly also 400hz
04:40 AM anonnumberanon: I'll need to get my oscilloscope out to continue this adventure.
04:40 AM anonnumberanon: I am bed now by all.
04:41 AM cehteh: osc wont do the work :D
04:42 AM cehteh: i dont own one
04:49 AM nuxil: i got one. oscillioscope is a handy tool. although i want a logic analyzer, i think its more handy when working with chips.
04:52 AM cehteh: and costs only a fraction of an osci
04:53 AM nuxil: yea.
04:53 AM cehteh: these $15 saelae clones are already useful
04:53 AM nuxil: hmm. 15 bucks
04:54 AM nuxil: where do you see them for 15bucks?
04:54 AM nuxil: cehteh, do you have a cheap one like that and its working ? :p
04:54 AM cehteh: yes
04:55 AM cehteh: amazon ebay whatever
04:55 AM Emil: cehteh: 15 dollaroos?
04:55 AM Emil: That's quite a lot
04:55 AM nuxil: Emil, it is? it sounded verry cheap to me.
04:55 AM cehteh: https://www.amazon.de/AZDelivery-Logic-Analyzer-kompatibel-Version/dp/B01MUFRHQ2/
04:55 AM cehteh: 12Eur
04:56 AM cehteh: $15 was just a guess
04:56 AM Emil: >amazon
04:56 AM cehteh: i bet elsewhere you get them cheaper than amazon
04:56 AM cehteh: but .. $15 is already cheap enough imo
04:56 AM Emil: less than 5 dollaroos from ebay
04:56 AM cehteh: https://www.amazon.de/iHaospace-Channel-Analyzer-Support-support/dp/B071LFZRTZ/
04:57 AM cehteh: yeah
04:57 AM cehteh: i dont use ebay
04:57 AM nuxil: nor do i
04:57 AM Emil: https://www.ebay.com/itm/191685084604
04:57 AM Emil: cehteh: wtf why not
04:57 AM Emil: nuxil: you are missing out
04:58 AM Emil: eBay is gold mine for electronics hobbyist
04:58 AM Emil: s
04:58 AM Emil: And most of Amazon sellers just resell from eBay
04:58 AM Emil: eBay has realy good customer protection also
04:58 AM Emil: The only issue with eBay is the long shipping times
04:59 AM Emil: but that's the only tradeoff
05:09 AM nuxil: i think i will try and order that think. its only ~6bucks
05:09 AM nuxil: not worlds end if it dosent work.
05:15 AM Emil: from ebay?
05:15 AM nuxil: yes
05:15 AM Emil: if it does't work just claim that you didn't get it
05:15 AM Emil: and you'll get your money back
05:18 AM nuxil: alright. its on its way..
05:18 AM nuxil: will get it next year
05:19 AM nuxil: Delivery estimate: Fri. Dec. 8 - Thu. Jan. 18
05:19 AM nuxil: :\ thats some gap there
05:19 AM Emil: haha :D
05:20 AM Emil: Make a reminer on your calendar for 19.1. to check if you got it
05:20 AM Emil: reminder*
05:20 AM nuxil: yea. will do :D
05:21 AM nuxil: but im not gonna go nuts for 5.5bucks. that like 50 Norwegian Kr. not even enough for a beer on the pub.
07:48 AM kl0wn is now known as _kl0wn
07:48 AM _kl0wn is now known as kl0wn
08:20 AM kl0wn is now known as _kl0wn
08:26 AM _kl0wn is now known as kl0wn
08:55 AM rue_mohr: if you used aliexpress you wouldn't need the reminder, they built in a timer and TELL you what the remaining time is
08:55 AM rue_mohr: and sellers dont get the money till you say your good
09:01 AM rue_mohr: so aliexpress deals with issues itself, instead of handing them over to the financial service
09:52 AM Emil: rue_mohr: ali is shit compared to ebay
09:53 AM Emil: rue_mohr: aliexpress has shit customer service
09:53 AM Emil: for example, you buy chips from ali and cant test them in the two weeks time
09:53 AM Emil: tough luck
09:53 AM Emil: if they are broken
09:53 AM Emil: on ebay no problemos
10:18 AM LeoNerd: I tend to treat Ali as place of last resort... if it's all junk then meh...
10:18 AM LeoNerd: I only buy stuff there if I literally can't find it anywhere else at all
10:30 AM polprog: damnit UPS
10:30 AM polprog: yesterday didnt deliver because there was noone at home
10:30 AM polprog: nobody shoed up today
10:30 AM polprog: showed*
10:30 AM polprog: muh parts
10:35 AM Emil: polprog: were you at home?
10:35 AM Emil: Call and say that you were home the whole time waiting for a call or ring but no one showed up
10:36 AM Emil: Because delivery agents are known to fuck the customers with "attempted delivery but no one was home"
10:37 AM polprog: i wasnt
10:37 AM polprog: i expected it to be delibered today
10:40 AM Emil: ah
10:41 AM polprog: delivered*
10:41 AM polprog: shit i need a new keyboard ;)
11:03 AM polprog: self written libraries ftw: https://twitter.com/mikelectricstuf/status/932914982924189696
11:13 AM emil is now known as Emil
12:27 PM day__ is now known as daey
12:29 PM polprog: yiss the parcel came
12:30 PM polprog: now gotta redo the pcb
12:31 PM polprog: i made the vias too small :(
12:38 PM HighInBC: oh dear
12:38 PM HighInBC: they not connecting through the middle?
12:41 PM polprog: no, the fab guy told me he cant make them this small
12:41 PM polprog: can i make vias in the pads? just want to be sure
12:41 PM polprog: also what's up with polymer filling the vias
12:41 PM polprog: ?
12:42 PM HighInBC: I have made vias inside the pad
12:42 PM HighInBC: as for the solder mask over the vias they are called blind vias and you want them for some cases and not others
12:42 PM HighInBC: I cover them by default but not when I want to use them as probe points
12:42 PM polprog: ok
12:42 PM nuxil: hey polprog
12:42 PM polprog: hey nuxil
12:43 PM HighInBC: you should be able to get design rules from your fab and enter them into your software
12:43 PM nuxil: may i ask. where do you get your pcb's done ?
12:43 PM HighInBC: it is fairly crucial
12:43 PM polprog: HighInBC: i forgot about this rather crucial part of the design process this time ;)
12:44 PM HighInBC: oh they are not called "blind vias" that is something else
12:44 PM polprog: or better said, i assumed an easily checkable thing
12:44 PM polprog: nuxil: local pcb fabricator in poland
12:44 PM nuxil: do they ship to private persons outside poland ?
12:44 PM HighInBC: tented vias is the term
12:45 PM polprog: not sure about that
12:45 PM polprog: but there are plenty of fabricators out there
12:45 PM LeoNerd: "blind via" - via that doesn't entirely penetrate the board, hits only a few layers from one side
12:45 PM polprog: ^^
12:45 PM HighInBC: yup
12:45 PM LeoNerd: "tented via" - via with soldermask over the top to seal it in
12:45 PM nuxil: anyone you would reccomned ?
12:45 PM HighInBC: terminology, half the battle
12:45 PM LeoNerd: They're independent things, so a via could be either or both
12:45 PM polprog: nuxil: i heard about oshpark, but you'd have to ask
12:46 PM HighInBC: I use oshpark for small boards
12:46 PM LeoNerd: I don't believe OSHpark can do blind vias
12:46 PM polprog: nuxil: where are you from by the way?
12:46 PM nuxil: norway
12:46 PM HighInBC: they do tented, I have never ordered a 4 layer board from them so no idea about blind
12:46 PM LeoNerd: I've got one product on the 4layer. Their fab doesn't do blinds
12:47 PM LeoNerd: Tenting is fine, because that's just a matter of not making holes in the soldermask where the vias are. Fabs don't tend to need to do anything special to support that
12:47 PM LeoNerd: Though best to keep the drill size small; too big and the mask might sag into the hole
12:47 PM nuxil: polprog, and trying to find some in this contry is out of the question.. it be to expensive-
12:48 PM polprog: there are also some china fabs like dirtypcbs but im not sure how long you wait
12:48 PM polprog: i wait a week for a soldermasked 2 layer board
12:49 PM HighInBC: yes I have had larger tented vias not fully cover the hole
12:49 PM LeoNerd: I'm generally disappointed by dirtypcbs. They're OK-ish for internal things, larger sizes.. Their silkscreen text printing is terrible, I wouldn't want to sell boards based on those
12:49 PM HighInBC: I use smaller vias and when I need more current multiple smaller vias
12:49 PM nuxil: polprog, do you mount the parts yourself or do they do it (is it a option you can have)?
12:51 PM polprog: i mostly order 3 pcs of the board so that would not calculate well for price ;)
12:51 PM polprog: i can solder smd easily
12:51 PM nuxil: i was watching some EEVblogs and he said that many pcb manufacturers would mount parts for you if you wanted.
12:51 PM day__ is now known as daey
12:51 PM polprog: yeah, if you pay and provide additional files
12:51 PM polprog: they would
12:51 PM nuxil: not sure if its a common practice since i never had someone else making the pcb
12:52 PM polprog: but unless you need to make a lot it's just better to do it yourself
12:52 PM polprog: it's very common practice in medium and large scale prod
12:52 PM polprog: sometimes you send the parts and pcb to another company that does the assembly
12:53 PM polprog: oh snap i have to replace the resistor network for a larger one :(
12:53 PM nuxil: heh. oki
12:53 PM polprog: to small clearance :(
12:54 PM nuxil: you desierve a smily song. :)
12:54 PM nuxil: https://www.youtube.com/watch?v=1a9tzTlwyKk
12:56 PM polprog: :)
12:57 PM polprog: bad: http://picpaste.com/Przechwycenie_obrazu_ekranu_2017-11-22_19-55-25-en63jnH7.png
12:57 PM polprog: red arrows point at bad clearance on resistor networks, right footprint is 0603 resistor
12:57 PM polprog: too bad i have the networks on my desk, they arrived an hour ago
12:58 PM nuxil: lol. so redesing time ?
12:58 PM polprog: yeah, pretty much
12:59 PM polprog: i have to fit 10-odd traces in between 7 segment displays. the hieght of the right part is 8.1 mm
12:59 PM polprog: between the square pads of the 7 segs
01:00 PM nuxil: what foorprint is your resistors ?? 0402 ?
01:00 PM nuxil: or 1206
01:00 PM polprog: those are 4x 0402
01:01 PM polprog: the one the yellow arrow points at is 0603
01:01 PM polprog: tiny stuff
01:01 PM nuxil: yea
01:01 PM nuxil: 402 is like 0.6mm
01:01 PM nuxil: padlength
01:01 PM polprog: this is the size im considering wearing a face mask, one unintentional sigh and all your resistors are all over the bench :P
01:02 PM polprog: 0603 = 2.4 x 1.4 mm, about that
01:02 PM polprog: size
01:03 PM polprog: it's... 6 x 3 mils ?
01:03 PM polprog: no too small
01:04 PM polprog: 60 x 30 mils
01:04 PM polprog: DIP pin pitch is 100 mils
01:05 PM polprog: hmm 8 0603s should fit... more space for routing
01:07 PM nuxil: what program do you use for the pcblayout ?
01:08 PM nuxil: pcbnew 4.0.7?
01:08 PM polprog: Kicad, so pcbnew
01:09 PM polprog: 4.0.7 as well
01:15 PM Ameisen: I kind of want to make an AVR in FPGA
01:15 PM Ameisen: which would make a faster AVR than any actual AVR
01:16 PM Ameisen: polprog - I studied history as my major
01:16 PM Ameisen: sometimes I dick around and pretend its 1910. Polish and French people get mad at me.
01:16 PM polprog: Ameisen: oh, what got you into the compiler word :P?
01:16 PM polprog: brb, supper
01:16 PM polprog: history is cool ;)
01:16 PM Ameisen: I'm an engineer.
01:17 PM Ameisen: also, no offense, coupled with the fact that i can neither spell nor pronounce most polish city names
01:17 PM Ameisen: the German versions are pretty easy in English
01:17 PM Ameisen: :|
01:17 PM Ameisen: likely due to the fact that they're closely related and share the same orthography
01:17 PM Ameisen: I only figured out how to pronounce Wroclaw like, a few years ago
01:18 PM Ameisen: And... unsurprisingly, it's not pronounced at all how I presumed. Closer to Vratswav
01:18 PM Ameisen: Latin or German is easier, though.
01:19 PM Ameisen: Latin versions are Bratislavia/Vratislavia (based on Wroclaw), Breslavia (based on Breslau), German is obviously Breslau.
01:21 PM Ameisen: Another latin form is Budorgis (no idea the origin). Old English would be Breslu, Middle High German was Prezla (became Preßlau and then Breslau), Hungarian is Boroszlo (based on German), Czech is Vratislav (Czech orthography is closer to German orthography than Polish, due to Habsburg influence), Hebrew is Vrotslav, Yiddish is Bresloi (Yiddish is a German dialect), Silesian German is Brassel
01:21 PM Ameisen: (my personal favorite), and Silesian [Slunsk] is Breslawia/Bryslal.
01:22 PM Ameisen: I liked studying history and linguistics
01:22 PM Ameisen: fun seeing how city/region names evolve
01:22 PM Ameisen: since cities that went from Polish -> German -> Polish originally had Old Polish names, which morphed over time into Germanic names, and then were readapted into Polish which often Polonized the German name instead of using the older Slavic name.
01:23 PM Ameisen: though the proposed Polish name of Brandenburg (Brennaborg) is sort of uninventive.
01:41 PM polprog: wow
01:41 PM polprog: im amazed
01:42 PM polprog: or Ameised ;)
01:45 PM nohitzzz: Ameisen it has been done numerous times
01:46 PM Ameisen: changing names>
01:46 PM Ameisen: yes
01:46 PM Ameisen: polprog - so, if I ever say Bromberg or Breslau, it's not because I hate you
01:46 PM Ameisen: it's because I cannot remmeber how to spell Bydgosztzs
01:46 PM Ameisen: it has a bunch of consonants after one another that don't make sense to me.
01:46 PM Ameisen: and my last name is Polish, to
01:46 PM Ameisen: too*
01:47 PM Ameisen: my great aunt always calls me dupa
01:47 PM Ameisen: my other great aunt was ashkenazi from Prussia, and spoke Yiddish
01:47 PM Ameisen: I have complicated heritage
01:47 PM Ameisen: surprisingly, the German and Polish halves of my family don't get along that well
01:48 PM Ameisen: we have no idea what the ORIGINAL name of, say, Breslau was
01:48 PM Ameisen: before the Poles/early West Slavs were there, there were Germanic tribes (West and East Germanic). Before them, Celts
01:48 PM Ameisen: before them, Old Europeans (pre-Indo-European tribes)
01:48 PM Ameisen: probably spoke Rhaetic or such (similar to Etruscan)
01:48 PM Ameisen: but the names are lost to history
01:49 PM Ameisen: though most of those cities have been settled for thousands of years
02:07 PM polprog: haha
02:07 PM polprog: interesting
02:11 PM papirkopi is now known as grafi_
02:12 PM Emil: polprog: vias inside pads works but you should check the connection
02:12 PM Emil: when assembling
02:17 PM polprog: ill keep that in mind
02:18 PM polprog: lol my pcb looks greaqt
02:18 PM LeoNerd: Vias in pads usually suffer the problem of "solder theiving"... the solder drains away down the hole rather than onto the leg of the part
02:18 PM polprog: i managed to squish 19 signal lines in an 8mm passage
02:19 PM polprog: LeoNerd: those will be soldered manually
02:19 PM LeoNerd: Ah that's not so bad then
02:19 PM Emil: polprog: pictures pls
02:19 PM LeoNerd: Though... if it's large enough to make easy soldering by hand, why does it need via-in-pad?
02:19 PM polprog: Emil: when it's finished
02:19 PM Emil: polprog: no
02:19 PM polprog: ill post a screenshot
02:19 PM Emil: polprog: right now
02:19 PM polprog: okay
02:19 PM Emil: of the empty pcb
02:19 PM Emil: :D
02:20 PM polprog: http://picpaste.com/Przechwycenie_obrazu_ekranu_2017-11-22_21-19-48-Pp1KVz3a.png
02:21 PM polprog: i have to redo the traces
02:21 PM polprog: they were placed to see if they wil fit
02:21 PM polprog: straighten some out
02:21 PM polprog: other touches
02:21 PM Emil: Noo
02:21 PM Emil: I want a pic of the actual pcb
02:21 PM polprog: ah
02:22 PM polprog: im redoing it cos the fab said they dont make vias that small
02:22 PM Emil: Also
02:22 PM Emil: why
02:22 PM Emil: is there a gab
02:22 PM Emil: gap*
02:22 PM polprog: i had to move the cpu part to make space and didnt draw the edge ;)
02:22 PM Emil: on the border
02:22 PM Emil: polprog: blasphemy
02:22 PM polprog: actually not
02:23 PM polprog: the left part will dangle on thin traces
02:23 PM polprog: it's art
02:23 PM Emil: yes
02:23 PM Emil: This pleases me
02:36 PM fooman2011: Hello. Could you please tell me what is the difference between interrupt on a "standard" pin and on a pin marked as "INT0" or "INT1" ?
02:36 PM * polprog should be doing homework now
02:36 PM polprog: and i will probbaly take a break
02:38 PM LeoNerd: fooman2011: PCINTs can only tell you of a change, in either direction. INT pins can be either-edge triggered, just one edge (low->high or high->low), or low-level triggered
02:39 PM fooman2011: LeoNerd: Thanks you :)
02:45 PM antto: what's the smallest (pin-wise) atmega?
02:47 PM LeoNerd: I can't think of anything lower than the 28pins
02:47 PM LeoNerd: Smaller tends to be tiny
02:51 PM antto: atmega8HVA?!
02:59 PM antto: oh, there are more
02:59 PM antto: which atmega would you pick with considerations for cheapness and future obtainability
03:00 PM antto: i need basically just one UART and 3 pins, 5V
03:01 PM LeoNerd: You probably want an ATtiny841 then
03:01 PM LeoNerd: That's my 'tiny of choice these days. It's kinda like a mini version of the ATmega328
03:01 PM antto: i don't have a programmer for the attinys
03:01 PM LeoNerd: Two UARTs though :)
03:01 PM LeoNerd: It's just regular ISP
03:01 PM antto: is it?
03:01 PM LeoNerd: The same RESET + MISO/MOSI/SCK
03:01 PM LeoNerd: Yup.. many of the tinies are
03:02 PM LeoNerd: It's only the *really* small 6pin ones that differ
03:02 PM antto: oh
03:02 PM polprog: aaand done
03:02 PM LeoNerd: I use these aaaall the time and I only have an ISP programmer
03:02 PM antto: technically i have a parallel port.. stk200 ;P~
03:02 PM antto: so it's pretty much avrdude doing the bitbanging
03:03 PM antto: ..not a "proper" programmer
03:03 PM polprog: wonder if i could make a pic programmer this way
03:03 PM polprog: or does it need some 12v line i've heard?
03:03 PM LeoNerd: Most MCUs tend to be some sort of synchronous serial
03:03 PM antto: doesn't the MCRL use high voltage during programming?
03:03 PM LeoNerd: So generally anything can bitbang those, if you're happy to take the hit at being slow
03:11 PM antto: the microchip website poped up a thing asking me "How are we doing?"
03:11 PM antto: >:)
03:12 PM antto: i'll run out of symbols
03:42 PM mmfood: if one gets a verification error after uploading with avrdude, does that mean that the chip is trash? I tried another chip and it worked fine.
03:43 PM polprog: does the failure always happen at the same adress?
03:43 PM polprog: address*
03:43 PM mmfood: it would appear so
03:43 PM polprog: might be deab flash
03:43 PM mmfood: byte 0x0006
03:43 PM polprog: dead*
03:47 PM mmfood: ok, good to know
04:00 PM polprog: Reminder that good old parport gan be used like a gpio: https://www.youtube.com/watch?v=EtX3p4HFbtc
04:25 PM aarcane: So I'm doing some RMS calculations... I gather the values and square them in the ADC ISR, then mean and root them in a timer ISR, as well as clearing the ADC counter and accumulator...
04:27 PM aarcane: But I've gotten the following strange behaviour, and I hope someone could explain why. Given float sig = ADC-511/*Center it about zero*/ the following bit gives me final results in the 50K range that *decrease* with current: sum += sig*sig; but if I replace that with sum += abs(sig)*abs(sig), it works perfectly and gives sane results. Can anybody explain why?
04:28 PM fooman2011: Hello. Could you please tell me if an activated pull up resistor of an AVR is consuming current ?
04:52 PM polprog: fooman2011: yes it is
04:52 PM LeoNerd: When not driving against the input, I wouldn't imagine so
04:52 PM LeoNerd: When it is, then sure. it has to provide that current
04:52 PM polprog: aarcane: why float
04:53 PM * LeoNerd generally avoids floats on AVR
04:53 PM aarcane: polprog, because it makes it explicit that the value is 32 bit instead of assuming it's the 64 bit normally associated with doubles.
04:54 PM polprog: int32_t
04:54 PM LeoNerd: Except one tiny bit of my voltmeter where I have to calculate dBV values; that requires a log10 function
04:54 PM LeoNerd: But that's only in the display path, not the critical measurement end
04:57 PM fooman2011: Typically, when using I2C between an attiny and an RTC, the SDA and SCL are both connected to VCC through a pullup. Is it possible to replace VCC by an attiny pin defined as output high ?
04:57 PM mmfood: I am playing around with PWM and managed to get one port working. But I thought I could use two different ports with the same counter? I tried this, but I only get a pwm on the OCR1A port, not the OCR1B
04:57 PM mmfood: https://paste.fedoraproject.org/paste/YykHjjEbk33lNhXiePvWBA
04:59 PM aarcane: LeoNerd, also, squaring and rooting often results in float values, so it's fairly simple. I'm going to figure out what bits I can convert to int32_t once I get it working, but the final sqrt needs to be in float, so the sending of data should also be in a float-friendly format.
05:00 PM LeoNerd: squaring each sample individually shouldn't
05:00 PM aarcane: sorry, polprog ^^
05:00 PM LeoNerd: I'd square samples and average them while still in integer, then do the final squareroot at the end.. possibly still in integer too
05:00 PM LeoNerd: integer roots are still fine
05:00 PM polprog: after all they are all just series of 1s and 0s
05:00 PM polprog: in the memory
05:02 PM mmfood: found the error :) forgot to set the compare output mode
05:03 PM LeoNerd: Ahyes, the many and varied stages to timers
05:03 PM aarcane: the sqrt(f) function takes and returns double(float). The final result will probably be in mA instead of A, so I'll probably cast it up from int and back down to int... but I really just want to know why the number was somehow buggered up in the sample above
05:03 PM polprog: garbage in garbage out
05:04 PM fooman2011: Sorry, maybe my question is stupid :/
05:04 PM aarcane: polprog, well duh. where's the problem though?
05:10 PM polprog: aarcane: sorry i misread something
05:10 PM polprog: try to dump the values via uart
05:10 PM polprog: check stuff that's easy to check
05:10 PM polprog: dont make too much assumptions
05:12 PM polprog: goodnight
05:57 PM mmfood: are there differences in how the attiny84 handles pwm and timers compared to atmega328p? I can't find anything in the datasheet. All the register names seems identical. Yet when I try to program the attiny it doesn't seem to work
05:57 PM mmfood: I changed the ports for the DDR ofcourse but otherwise it is the same code?
06:25 PM fooman2011: Hello. In the DS1337 datasheet (http://docs-europe.electrocomponents.com/webdocs/124d/0900766b8124d616.pdf) it says that a pullup must be put between the INTA pin and the microcontroller. But what is the value of this pullup ? I would like to have the lowest possible current consumption. Is it possible to put a 4.5Mohm resistor ?
06:32 PM mmfood: fooman2011: are you using it in your application? It says it can be left floating otherwise
06:40 PM fooman2011: mmfood: yeah i'm using it
06:40 PM fooman2011: to wake up my microcontroller
06:51 PM mmfood: https://electronics.stackexchange.com/questions/23645/how-do-i-calculate-the-required-value-for-a-pull-up-resistor#23646
06:51 PM mmfood: fooman2011:
06:53 PM mmfood: seems to depend on your microcontroller. What values is considered a logic high and what the input leakage current will be will determine the highest value on the pullup
06:55 PM mmfood: so VCC - I_leakage / R > V_high I guess
06:57 PM Emil: mmfood: rtfm. The register descriptions are 100% clear
06:57 PM fooman2011: mmfood: thanks :)
06:57 PM Emil: fooman2011: 4.5Mohm is a great source of noise
06:58 PM mmfood: Emil: rtfm?
06:58 PM Emil: and might (definitely is) too great for it to function properlu
06:58 PM enh: Anybody knows what happened to Lambda_Aurigae?
06:59 PM Emil: you only use energy when pulling it low (so during when talking to the device)
06:59 PM enh: He has not been here for a while now
06:59 PM fooman2011: Emil ? It' not pullup for I2C, it's pullup for INTA pin
07:00 PM fooman2011: the INTA pins is always floating except when the alarm is notified.
07:00 PM fooman2011: so the pullup is almost always used
07:00 PM fooman2011: right ?
07:02 PM enh: Almost two weeks off
07:18 PM Emil: funky1: pullups (or pulldowns) dobt consume energy
07:18 PM Emil: unless you fight against them
11:39 PM day__ is now known as daey