Back
[00:20:05] <inflex> mnnnn
[00:20:09] <inflex> another design to go in now
[00:20:44] <Casper> inflex: did I asked you about bus today?
[00:22:28] <Casper> if not, do you know of an article that talk about the difference between I2C, SMBus and SPI? as in speed, bus length and number of devices?
[00:22:32] <inflex> mrmmm... oh, no
[00:22:41] <inflex> I have seen an article/white-paper
[00:22:51] <inflex> I think it was Microchip or Atmel who put it out
[00:22:59] <Casper> I did in the past too, but I can't find anything today somehow
[00:23:13] <inflex> else I suppose Wikipedia might have something?
[00:23:33] <Casper> I checked for SMBus today, couln't find fast answers
[00:23:46] <Casper> and SPI
[00:24:03] <Casper> they just used like "long bus" to talk about the length...
[00:25:48] <Casper> I'm trying to figure out if it's worth to try to do networking of my projects or not
[00:26:01] <Casper> as currently I have 3 projects that really is one
[00:26:21] <Casper> but would be fun to be able to do some query on it
[00:26:28] <Casper> hence the bus
[00:28:13] <Casper> basically, I want to make a ball valve actuator. I also want to make a timer. I also want to make a water condition detector (possibly as simple as a led and a photoresistor and a simple 2 wires in water kind of "salt detector")
[00:28:53] <Casper> my guess is: if the water is dirty, it should be blury, so the led/detector shall do the trick. Also, it should get more conductive
[00:31:18] <soul-d> what you mean with water condition detector in what medium ?
[00:32:52] <Casper> basically, there is a pit here where the french drains of the house goes. there is a pipe from the sewers that go there too to empty it. sadly, due to a 196x error in thinking, they didn't put a check valve, and the pit receive a bit of toilet water
[00:33:00] <Casper> and after a few days it start to smell
[00:33:20] <Casper> I want to make an automatic flusher
[00:33:45] <Casper> hopefully it will save on water AND odors
[00:34:10] <soul-d> so you need somthing like in the toilet stuff ? that switches on high or low level of water ?
[00:34:17] <Casper> no
[00:34:23] <Casper> on water condition
[00:34:37] <Casper> if the water is dirty, flush
[00:35:10] <Casper> first step: make a servo for the ball valve
[00:35:24] <Casper> second step: make a timer/controller
[00:35:36] <Casper> third step: water condition detector...
[00:35:39] <soul-d> k, thats a bit harder to do cheaply i guess
[00:35:51] <Casper> yeah
[00:36:42] <soul-d> thought of ph /ec meters wich probably be overkill for this job
[00:37:11] <Casper> yeah
[00:37:24] <Casper> but I considered PH a bit
[00:37:37] <soul-d> yeah problem is most meters have active stuff
[00:37:42] <soul-d> or you need a continious one
[00:37:52] <Casper> but really, what would be the best is a way to detect the bacteria level
[00:39:08] <Casper> my idea was to network all of the parts
[00:39:10] <soul-d> hard to say also define wich are bad maybe you killl or flush that one normaly kills odor
[00:39:18] <Casper> and make a network able bootloader
[00:39:25] <Casper> and debugger terminal or something
[00:39:29] <soul-d> you could
[00:39:34] <soul-d> maybe use methane or gas sensors ?
[00:40:12] <Casper> I doubt there is much methane
[00:40:26] <Casper> and anyway, it's not that important for now
[00:40:37] <Casper> the most important part is: is it worth to try to network it?
[00:42:29] <soul-d> depends that part probably will be reusable for other projects and if this measuring device is some where remote and smelly sure hook up a long cable :)
[00:44:32] <Casper> bed time, tty tomorrow
[00:45:00] <soul-d> gnite
[00:49:46] <inflex> well, over $250 of PCBs submitted now... better get one good product out of it all!
[00:53:29] <soul-d> heh, i made myself an 8bit logic analyzer :)
http://imgur.com/a/ZMLd4 will need to verify it with a a non buggy code that counts but sofar so good
[00:55:02] <inflex> nice job... what'd you do... just pull in PORTB or something?
[00:55:44] <soul-d> measuring an hc595 on avr
[00:56:12] <soul-d> wondering why breadboard is buggy with lcd screen but probably just need to clean it up and rebuild it (afther 2 years )
[01:04:42] <soul-d> http://imgur.com/a/lq6Ud current opperation of 8 bit logic and a failed test of toner transfer etching for a 16 bit version
[01:07:04] <soul-d> only need to redo the middle breadboard now but to lazy for that :)
[01:09:40] <soul-d> also need to start using all the chip lines of device's like reset and and chipselect they so stuff has less change to pickup random noise during reset or just random
[01:16:19] <soul-d> think im gonna give toner transfer up though fun if you have a quick proto to do and nothing else otherwise more hassle then worth it maybe just buy chalk paper and some photostuff for old boards
[02:05:36] <bluebie> Hey guys - I'm trying to use Timer 1 to pwm the blue channel on an RGB LED on a tiny85, but I'm having trouble getting it to work, and being confused by missing data in the datasheet and links that go nowhere
[02:06:11] <bluebie> could anyone advise me of a good example, or perhaps what I'm doing wrong here?
http://pastie.org/2819173 (bit is mapped to _BV, outputs just sets pins to outputs)
[02:11:54] <inflex> bluebie: doesn't sound like an Atmel datasheet
[02:12:00] <inflex> bluebie: which one are you reading?
[02:12:18] <bluebie> http://www.atmel.com/dyn/resources/prod_documents/doc2586.pdf
[02:12:45] <inflex> what data do you feel is missing?
[02:13:37] <bluebie> oh wait
[02:13:41] <bluebie> nevermind ._.
[02:13:48] <bluebie> I think I need a break from this
[02:13:52] <inflex> heh
[02:15:19] <bluebie> damnit
[02:15:31] <bluebie> I've been staring at this thing and puzzling over it for like 4 hours
[02:15:46] <bluebie> and minutes after I ask on irc it all starts working
[02:15:50] <bluebie> screw you universe! screw you!
[02:15:57] * bluebie shakes fist at universe
[02:17:38] <bluebie> inflex: thanks for the.. um.. magic!
[02:21:49] <inflex> np
[02:21:54] <inflex> that's usually how it goes here
[02:22:04] <inflex> I think we all can say we've done something similar
[02:27:24] <Esmil> It's called rubber duck debugging. It even has a wikipedia article ;)
[02:30:48] * h4x0r` is so fed up with size constraints on gsm applications...
[02:30:57] <inflex> ?
[02:31:26] <h4x0r`> well i want built in power, ftdi + sim holder etc
[02:33:38] <h4x0r`> avr + programming header..
[02:33:54] <h4x0r`> hey
[02:34:06] <h4x0r`> does anyone know the plastic pcb circuit you find in keyboards?
[02:34:23] <h4x0r`> well is there a fabhouse that does that kind of work ?
[02:37:22] <h4x0r`> oklol
[02:56:16] <Igor2__> good morning!
[02:57:12] <Igor2__> now that i resolved all my upload problems and my code executes again, i found that whenever a function returns, the program restarts instead of continuing after the rcall instruction - i compile code with avr-gcc and the asm looks reasonable
[02:57:34] <Igor2__> i've already made sure it happens on ret, not on function entry
[02:58:02] <Igor2__> any idea what may go wrong?
[03:07:42] <inflex> lo Igor2__
[03:08:03] <Igor2__> hi inflex!
[03:08:08] <inflex> if it keeps retstarting, it's possible that the code is going through to the end
[03:08:10] <Igor2__> must be something obvious, like yesterday
[03:08:23] <Igor2__> nah, it happens exactly at the end of a function
[03:08:29] <Igor2__> i have multiple calls
[03:08:37] <Igor2__> first call reboots instead of properly returning
[03:08:54] <Igor2__> i have non-call usart writes before and after the call, so i can see what's going on
[03:09:10] <Igor2__> the function itself contains a couple of nops only, so i am sure i don't ruin the stack
[03:09:59] <Igor2__> not watchdog either
[03:12:58] <Igor2__> hmm, maybe it runs on some unhandled interrupt (but why would that happen on a ret i don't know)
[03:13:20] <Igor2__> nah
[03:13:35] <Igor2__> amikor a bbc4-en vasarnapi vallasi musor kezdodik, akkor az mr1-en valami kabare van
[03:13:46] <Igor2__> es mikor az mr1-en vallasi lofasz kezdodik, akkor a bbc4-en mar rendes musor van
[03:14:05] <Igor2__> (mert a bbc-n csak fel ora misezes van, oslt)
[03:14:16] <Igor2__> oops, wrong channel, sorry
[03:15:07] <Igor2__> i think i will hook the interrupts and see if any of the interrutps trigger
[03:18:03] <Igor2__> if i execute CLI, no interrupts will be delivered, right?
[03:20:57] <jacekowski> maybe you have usart transmitted interrupt enabled
[03:21:37] <Igor2__> but a CLI in the beginning of the code would avoid that, right?
[03:23:27] <Igor2__> i think i will send MCUSR to see if it was a real reset or just ret/jump to the wrong address
[03:24:58] <Igor2__> i get 0 in MCUSR so it's not a real reset, it more looks like jumping/returning to the wrong address
[03:25:38] <grummund> Igor2__: add a handler for ISR(BADISR_vect) to at least rule that out
[03:26:00] <Igor2__> ok
[03:29:16] <inflex> hrmm... yeah, unhandled ISRs are a common cause
[03:29:24] <Igor2__> hmm, it doesn't do what i expected
[03:29:27] <Igor2__> i mean in the lst
[03:29:53] <Igor2__> is the syntax ISR(BADISR_vect) { c-code; } ?
[03:30:33] <grummund> http://www.nongnu.org/avr-libc/user-manual/group__avr__interrupts.html
[03:30:35] <Igor2__> (sorry, i am new to avr and avr-gcc)
[03:30:42] <grummund> see "Catch-all interrupt vector"
[03:32:06] <Igor2__> thanx
[03:32:22] <Igor2__> i just forgot to #include avr/interrupts.h so it just became a function called ISR
[03:32:44] <Igor2__> ok, we can exclude interrupts, the function is not called
[03:32:59] <Igor2__> we can exclude real reset as well, as MCUSR stays zero
[03:33:20] <Igor2__> (i've tried triggering real external reset and then it got the expected non-zero value so printing of MCUSR also works)
[03:33:44] <grummund> are you capturing MCUSR correctly?
[03:34:16] <Igor2__> i think so, when i triggered external reset it got the correct bit set
[03:37:31] <Igor2__> also, if i put a character in UDR right before the wrong RET, the character is properly sent
[03:37:49] <Igor2__> if it was a real reset, it would reset the USART before it had the time to send the char
[03:40:03] <grummund> http://www.nongnu.org/avr-libc/user-manual/group__avr__watchdog.html
[03:40:16] <grummund> see there for a method of capturing MCUSR
[03:41:00] <Igor2__> ok
[03:45:25] <Igor2__> copied that code and it yields the same result (0 during my problem, 2 on external reset)
[03:56:49] <Igor2__> i am now trying to print the stack
[03:56:54] <Igor2__> (from within the function)
[03:57:13] <Igor2__> pop into a register then put the value in UDR and delay
[03:57:46] <Igor2__> and i get all zeros, which would explain why it jumps back to restart everything upon the RET
[03:59:47] <Igor2__> (if i push something, it is sent back correctly so the pop/print works)
[04:05:59] <Igor2__> heh, SPL has a pretty strange value after startup
[04:06:36] <Igor2__> sram should be between 0x60 and 0x00DF on attiny 2313
[04:07:04] <Igor2__> and i have values like 0x56 in SPL
[04:07:08] <grummund> what toolchain are you using? and has it been working before or is this the first try out?
[04:07:55] <Igor2__> first try
[04:07:59] <Igor2__> avr-gcc on Debian
[04:08:00] <doublebeta-sleep> if you're playing with the stack, do all the function in assembly
[04:08:10] <Igor2__> i am not about to play with stack by hand
[04:08:11] <doublebeta-sleep> the rest of your code can be C tho
[04:08:25] <Igor2__> rather debugging why C-only code fails on the very first RET
[04:08:56] <Igor2__> and i concluded it's because of the wrong stack pointer
[04:10:32] <grummund> compiling for the correct mcu?
[04:10:50] <Igor2__> i double checked that, but let me check once more
[04:11:12] <Igor2__> -mmcu=attiny2313
[04:11:35] <Igor2__> label on the chip: atmel 1028 attiny2313-20PU
[04:12:20] <Igor2__> the datasheet says SP is set to RAMEND upon reset
[04:12:31] <Igor2__> i think there may be 3 sources of this problem:
[04:12:43] <Igor2__> - wrong SP after reset (would be device failure, rather unlikely)
[04:12:59] <Igor2__> - C compiler sets it up after reset in the wrong way (also unlikely)
[04:13:13] <Igor2__> - my thin code that sets up the serial line accidentally overwrites it
[04:13:55] <Igor2__> is there anything else that can mess with SP?
[04:14:38] <grummund> i think you're over-analysing this
[04:14:54] <grummund> clearly something is not right
[04:15:19] <Igor2__> hehe
[04:15:29] <grummund> hook up an LED and write a simple blinky program with delay loop.
[04:15:39] <Igor2__> that worked alreayd yesterday
[04:15:42] <Igor2__> even USART is working
[04:15:49] <Igor2__> this how i can debug in the first place
[04:16:26] <Igor2__> since i use -Os, for the simple examples everything gets inlined, and no RET is executed and it doesn't go wrong
[04:16:38] <grummund> i understood that calling *any* function caused a reset upon return
[04:16:53] <Igor2__> yup
[04:17:24] <Igor2__> i can even send dozens of characters over the USART from within the function, this how i peek at the stack and SP
[04:19:34] <grummund> reduce it to the some simple code, say <20 lines LED blinker that does not work
[04:19:53] <Igor2__> and now i even checked that setting up SPL correctly before the first call, things start to work
[04:20:02] <Igor2__> ok, but how would i debug then?
[04:20:13] <Igor2__> i mean without USART, i can't look at values currently
[04:20:17] <grummund> either the LED flashes or it does not
[04:20:35] <grummund> if it fails to blink, post the code on pastbin.com
[04:20:46] <Igor2__> wait, i have an idea
[04:21:03] <Igor2__> i'll save SPL _before_ i do anything and print it after UART initialization
[04:21:19] <Igor2__> this would reveal if i messed up SPL in uart init
[04:22:52] <Igor2__> SPL is wrong right at the beginning of main() :(
[04:24:54] <Igor2__> unfortunately i need to leave now
[04:25:05] <Igor2__> thanx for all the support, will go on with this later today
[04:33:48] <Igor2__> (meanwhile i found that the init code generated by the compiler sets SPH to 0x02 and SPL to 0x5F which is clearly wrong for my device - probably i do something wrong in avr-gcc command line oslt)
[04:38:08] <Igor2__> ahh, found it, i used -mmcu for all the compile steps but not for the link step so the linker had no idea and assumed different ram size
[04:48:31] <h4x0r`> Tom_itx
[04:48:36] <h4x0r`> are you available ?
[04:49:16] <h4x0r`> I have a usart question, if I open and run 2 usarts, is possible to write the output of the second to the first?
[04:49:42] <h4x0r`> so i can use the first usart to read the outputs of both back to the screen ?
[04:56:14] <inflex> hey h4x0r`
[04:56:37] <inflex> btw h4x0r`, afaik most PCB houses will do flexible boards
[04:59:07] <h4x0r`> cool
[04:59:12] <h4x0r`> thx dude
[04:59:16] <h4x0r`> sounds nice
[05:01:36] <inflex> np
[05:04:31] <inflex> just got notice that at least 3 of my boards have now been dispatched
[05:11:29] <h4x0r`> sweet lol
[05:11:53] <h4x0r`> iw ish i could get a design that im happy with
[05:12:34] <h4x0r`> so i ship it to the fabhouse..
[05:12:58] <h4x0r`> but im going for something unique.. and it just feels like its all been done before really.
[05:15:46] <inflex> yeah, I know that feeling
[05:16:05] <inflex> I get people coming to me asking "you should make this.. it'll be really cool, no one else has done it!".... 5minutes of googling later and I find a dozen of them
[05:16:25] <h4x0r`> heh
[09:12:11] <Igor2__> now that i am over the initial mistakes, things go fast, i already like avr much more than the previous uC family i used
[09:18:26] <Tom_itx> yeah they're ok
[09:18:37] <Tom_itx> i kinda miss the old 68332 moto
[09:26:58] <karlp> Igor2__: what family was that?
[09:33:28] <Igor2__> pic16
[09:33:40] <Igor2__> main reason to switch to avr was gcc support
[09:35:13] <Tom_itx> i used CCS C and Picbasic Pro with the Pics
[09:35:27] <Tom_itx> but yeah none of their tools were free
[09:35:38] <Igor2__> i found a free compiler for PIC, called sdcc
[09:35:46] <Tom_itx> yeah that was one
[09:35:46] <Igor2__> (it's a bit wider range than just PIC)
[09:35:52] <Igor2__> but it didn't work very well
[09:35:54] <Tom_itx> there were a couple others but with limits
[09:36:19] <Igor2__> becides it was far less efficient in optimizing the code than gcc is, it often generated totally broken code
[09:43:08] <inflex> lo there Tom_itx
[09:43:18] <inflex> busy week for me... that's 7 new designs submitted
[09:43:38] <Tom_itx> lo inflex
[09:43:50] <Tom_itx> had an earthquake here last night
[09:44:10] <Tom_itx> just aftershocks felt here but i was in the basement and it moved the whole basement
[09:44:29] <inflex> woo, is that relatively common?
[09:44:34] <Tom_itx> never
[09:44:43] <Tom_itx> 1930's was the last iirc
[09:44:55] <inflex> oh okay, so you're not on a fault line
[09:44:57] <inflex> ?
[09:45:02] <Tom_itx> apparently we are
[09:45:21] <Tom_itx> i wasn't aware of it
[09:45:26] <inflex> oh okay,just very quiet one
[09:45:34] <Tom_itx> 5.6
[09:45:34] <inflex> we had a quake here earlier in the year, nothing major
[09:45:48] <Tom_itx> in OKC which is ~2.5 hrs S of me
[09:46:14] <Tom_itx> it's just odd because we never expect them here
[09:46:58] <inflex> I see it made Slashdot.org
[09:47:16] <inflex> 5.6
[09:51:02] <Tom_itx> i see you're busy making boards again
[09:52:08] <inflex> yes
[09:52:31] <inflex> just trying a shotgun approach, looking for something that'll turn into a good seller
[09:52:40] <inflex> the LOM is waiting on a new set of boards too
[11:04:16] <Casper> soul-d: poke
[11:04:45] <Casper> inflex: did you remembered where you found the bus comparison page?
[11:06:26] <soul-d> ouch
[11:08:22] <Casper> hi ya
[11:09:15] <soul-d> good evening
[11:11:09] <Casper> hmmm I wonder...
[11:11:29] <Casper> if I should make a software uart AND hardware uart...
[11:11:40] <Casper> kinda like a dual rx single tx
[11:11:47] <Casper> :D
[11:14:19] <Casper> I want to be able to do firmware update via the bus and I'm puzzled on how to do it without that the other devices decide to transmit or try to process the data
[11:14:35] <soul-d> donno depends on purpose and needs i guess
[11:14:46] <soul-d> make a reset function ?
[11:14:52] <soul-d> with interupt pin
[11:16:00] <Casper> I'm planning on having many devices, so a chip select wire is not a good idea
[11:17:47] <soul-d> cant you use i2c then or a software version of that ?
[11:18:17] <Casper> I2C per se have a too short bus length (7.6M based on wikipedia)
[11:19:19] <soul-d> talking about 7.5 meters ?
[11:19:34] <Casper> I want longer wire
[11:20:30] <soul-d> donno maybe CAN bus ? but only know of it in name
[11:20:43] <Casper> can bus seems to be a good candidate
[11:21:16] <Casper> twisted pair, 1km at 40kbit
[11:21:20] <soul-d> http://www.softing.com/home/en/industrial-automation/products/can-bus/more-can-bus/bit-timing/practical-bus-length.php?navanchor=3010538
[11:21:26] <soul-d> not to bad if thats correct
[11:23:45] <Casper> wikipedia seems to show that it could be exactly what I want
[11:24:21] <Casper> harder to work with due to arbitraction, so need can enabled avr or software can
[11:24:50] <Tom_itx> hardware would be easier
[11:25:00] <Casper> yeah
[11:25:00] <soul-d> yeah better have a hardware do stuff for you i guess although maybe for understanding or practice while waiting on your chip
[11:25:12] <Casper> but I currently have no avr with hardware can
[11:25:41] <Casper> now... how to connect to computer...
[11:25:59] <Tom_itx> ask anonimasu
[11:26:02] <Tom_itx> he's done can
[11:26:15] <Casper> he's not here
[11:26:24] <Tom_itx> seattlerobotics emc etc
[11:27:39] <Casper> ow wow... digg is so late on news!
[11:28:10] <Casper> "African E-mail Scammers Now Demanding Ransom for People's Missing Loved Ones" ← I got some of those letters over a year ago, the local news talked about it over half a year ago
[11:28:16] <soul-d> probably just a converter dedicated chip with both can and somthing else uart i2c
[11:28:52] <Casper> yeah that is what I was thinking to do
[11:29:00] <Casper> I wonder if can work well on phone wire....
[11:31:39] <soul-d> http://copperhillmedia.com/cannewsletter/PDF/CAN-Wiring.pdf
[11:31:45] <soul-d> seems to be good on that
[12:34:08] <Casper> maybe I should just make my own protocol and bus...
[12:38:55] <karlp> yeah, that sounds like a good idea :)
[12:39:38] <Tom_itx> breadboard a firmware solution
[12:39:45] * TwisteR right how is working on Pelco-D protocol realization to use with custom device on RS-485 bus
[12:41:16] * karlp was just about to mention rs485
[12:41:27] <karlp> multidrop, long distances
[12:45:01] <karlp> can gives you multi master though
[12:45:06] <karlp> if that's important.
[12:47:16] <grummund> Steffanx: one for your next project? -
http://www.austriamicrosystems.com/eng/led-driver/AS1130
[12:47:39] <Steffanx> Nah, it's a matrix
[12:47:52] <Steffanx> I've seen a similar chip somewhere
[12:48:07] <grummund> no good?
[12:48:18] <Steffanx> I don't want to use a matrix :)
[12:48:33] <Tom_itx> mmm i get emails about drivers quite often
[12:48:43] <grummund> why not?
[12:49:12] <Steffanx> If I do.. i only need a 4x4 driver
[12:49:57] <grummund> ok
[12:50:26] <Steffanx> And I'm still not sure if i'll really build it
[12:50:44] <Steffanx> I'm at my "can i really use this"-point again
[12:51:09] <Steffanx> costs<>usability :)
[12:55:12] <Casper> I think I'll go with can bus...
[12:55:25] <Casper> if I can find a canbus terminal :D
[12:55:32] <Casper> else I'll have to make one
[12:56:00] <Steffanx> Unless those chips are cheap-ish grummund
[12:56:04] <Steffanx> but guess they're not
[12:56:16] <grummund> i didn't look even if they are available
[12:56:41] <Steffanx> The one i've seen was similar, but it forgot the bookmark the page
[12:56:42] <grummund> 12x11 could also be 4x11 RGB ;)
[12:57:27] <Steffanx> RGB === expensive :(
[12:58:43] <grummund> seems no-one has that IC, sorry ;)
[14:24:32] <Casper> I have a good question for ya guys, let's say you have a network of software addressable address devices
[14:24:55] <Casper> and you have an address conflict... how can you deal with that? so you can reassign one device to a new address?
[14:25:59] <JanneP> it they're software assignable, why not?
[14:33:08] <Casper> JanneP: if you have 2 devices on address 1, how can you tell which one need to change to address 34?
[14:37:01] <JanneP> without local access, it may not be possible
[14:37:45] <Casper> yeah that is what I'm worried, unless there would be a serial number or something, that avr do not have
[14:38:39] <mrfrenzy> have them both change to a random address?
[14:39:09] <Casper> let's say you have an 8 bits address, and 240 devices
[14:39:41] <Casper> there is an high chance that they both go to another conflict, so now you went from 2 to 4 conflicts
[14:39:50] <mrfrenzy> two ways:
[14:40:17] <mrfrenzy> 1) list all the legal random adresses to choose from, tell all devices with address 1 to move to one of these legal adresses
[14:40:45] <mrfrenzy> 2) let the device pick a random address, probe the network and ask if it is free, if not: continue choose another random address
[14:41:42] <Casper> if only avr had a build in serial number
[14:42:17] <JanneP> you could always hardcode your own serial number?
[14:42:26] <JanneP> if that feels like a good solution..
[14:42:40] <Casper> yeah, but again prone to human error :D
[14:43:05] <mrfrenzy> look at the dhcp protocol for example, that's where I got the above ideas
[14:43:42] <mrfrenzy> if no dhcp server replies, the client will try to use it's old address, if it is not free, pick a random address, if that is not free, continue with a new random address etc
[14:44:11] <Casper> yeah, might as well do that
[14:45:04] <mrfrenzy> if all your devices power up at the same time, you need a communication protocol with collision detection
[14:45:10] <mrfrenzy> or a random delay before choosing adresses
[14:45:28] <Casper> I'ld use can bus, which already have collision detection
[14:45:44] <Casper> as those who transmit a 0 take priority
[14:45:54] <mrfrenzy> great
[14:46:02] <JanneP> something like the broadcast search on 1-wire bus could also work
[14:46:10] <Casper> yeah
[14:46:13] <JanneP> where all the devices are polled on startup
[14:46:20] <Casper> I wonder how long 1 wire bus can go...
[14:46:36] <JanneP> I've tested some ds18b20 sensors on 100m cable..
[14:46:43] <JanneP> don't know for longer than that
[14:47:13] <Casper> what worry me is that it will be a kind of star network, which don't seems to work well :D
[14:47:24] <JanneP> but if you want robust long distance bus i'd go for rs 485
[14:47:35] <Casper> rs485 don't work for star
[14:47:39] <Casper> as most don't ...
[14:47:50] <Casper> as they need a terminating resistor
[14:48:55] <Casper> over the years I'ld like to add some home automation here
[14:49:16] <JanneP> unless you would build a hub of some sort..
[14:50:11] <Casper> yeah...
[14:50:26] <Casper> but I think 1 wire do not need such thing
[14:50:37] <Casper> and the slow speed could be fast enought
[14:52:20] <JanneP> 1 wire doesn't go too well with star either, as far as i understand
[14:52:27] <JanneP> with short cable runs it may be ok
[14:53:00] <Casper> in the short term, it could be very easy to do normal bus
[14:53:07] <JanneP> but 1 wire hub ought to be cheaper
[14:53:26] <Casper> in the very short term, it's a 3 devices network
[14:54:17] <Casper> then I might add some "high tech" stuff, like turning on the lights when someone is in the stairs
[15:05:35] <_abc_> Hello. Is FT232 based USB2serial converter based bit banging fast on ponyprog serial dongle style programmers?
[15:09:23] <Tom_itx> compared to what?
[15:10:07] <_abc_> Compared to other serial usb converters and to a raw native serial port
[15:10:45] <Casper> I doubt it bitbang well
[15:11:30] <_abc_> You doubt but I ask because it is rumored to do so quite well.
[15:37:00] <Stipl> helloooooooooo
[15:37:04] <Stipl> *echo* helloooo
[15:41:48] * _abc_ echoes a slap towards Stipl
[15:42:43] <Stipl> hey abc
[15:42:49] <Stipl> do u understand C code?
[15:42:55] <Stipl> and PIC microcontrollers?
[15:43:23] <_abc_> This is #avr
[15:43:33] <_abc_> And yes
[15:43:49] <Stipl> heh
[15:43:54] <Stipl> ok
[15:54:02] <_abc_> Sorry
[15:56:40] <Stipl> did u get my msgs? _abc_
[15:57:13] <_abc_> Yes and I said sorry.
[15:57:17] <_abc_> Can't help you.
[15:58:21] <Stipl> ok
[15:59:44] <ferdna> http://i.imgur.com/TFKaE.jpg
[16:21:23] * specing smacks Stipl for not knowing how to search for channels
[18:21:15] <Casper> wow... can protocol is really... inneficiant...
[18:21:42] <Tom_itx> but widely used
[18:22:11] <Casper> yeah... 44 bits for the header... to carry 64 bits of data....
[18:22:54] <Tom_itx> works in noisy environments
[19:23:44] <Casper> not sure that can bus is apropriate for my purpose...
[19:23:54] <Casper> anyway
[19:24:28] <ishmal> hi all
[19:24:47] <ishmal> hey, i was directed here from #arduino for this question: is it more efficient to do pop r0, pop r1, ... pop r32 ... OR ldm --sp r0-r32 ?
[19:25:54] <Landon> ldm? I'm not finding that instruction in the instruction set pdf
[19:26:11] <ishmal> maybe that's the wrong opcode. let me check
[19:26:27] <ishmal> might be a macro
[19:26:41] <Landon> http://www.atmel.com/dyn/resources/prod_documents/doc0856.pdf is what I'm looking at
[19:26:44] <ishmal> this is inline assembly... just a sec
[19:27:31] <Landon> and you can count instruction size and cycles for your definition of efficient :
[19:28:13] <Casper> look like it's a macro... look at the generated code
[19:28:15] <ishmal> hmm. now where did i see that jewel? :)
[19:28:23] <ishmal> ah in freertos souce
[19:28:31] <LikWidChz> HOLAH!
[19:30:04] <Valen> can is electrically robust
[19:30:11] <Valen> and really designed for short messages
[19:30:15] <Valen> like turn ac on
[19:30:28] <Valen> or, car is crashing, kiss ass goodbye
[19:30:50] * Landon finds that difficult to do in a car
[19:30:51] <ishmal> yes, it's in freertos's portmacro.h. stuff like, ldm sp++, r0-r7
[19:31:10] <ishmal> so i am guessing that the inline asm for avr has a macro for that
[19:33:17] <LikWidChz> anyone know if it would be challenging to write some code that converts PWM to a serial string perhaps like read 7 pwm values and put them out a single pin??
[19:33:52] <Casper> I think the can bus is a good thing for me, but not the protocol
[19:34:07] <Valen> what are you trying to do?
[19:34:28] <LikWidChz> valen is that a question directed at me?
[19:34:39] <Valen> @ Casper
[19:34:49] <Casper> LikWidChz: seems like you want to reset a timer, count and send the value before resetting it... shall be easy
[19:35:02] <Valen> use ICP port for added nifty
[19:35:10] <Valen> s/port/pin
[19:35:44] <Casper> Valen: I'm planning to do a few projects, home automation. I also want to be able to monitor the data for debugging, remote control and also firmware upgrade
[19:36:01] <Valen> check rs 458 or 485, cant rember which
[19:36:08] <LikWidChz> well casper... I have a rc transmitter and would like to have some intermediary device read the data before it gets converted to serial since serial would be a lot eaysier to deal with them 7 individual pwm values... either that or PPM to serial
[19:36:10] <Valen> also ethernet is becoming more feasable
[19:36:55] <Casper> LikWidChz: so basically an "analog to digital servo"
[19:37:26] <Casper> Valen: yeah ethernet is also a possibility, but the modules are "expensive" and then have to deal with tcp and all... not fun
[19:37:50] <Casper> Valen: what bug me of bus is that you can't go star network
[19:38:02] <Valen> you can get AVR's with built in ethernet, just add magnetics
[19:38:14] <Valen> you can do one of those on a star
[19:38:19] <Casper> yeah, but then have to code a tcp stack and all...
[19:38:32] <Valen> its probably off the shelf
[19:40:04] <LikWidChz> well.... not reallly... so I have this device by lynxmotion.com called the Pheonix its an 18 servo hexapod walker... totally geeky, totally cool, in the beginning the device was controlled by a teatherd ps2 controller which was great, but I wanted wireless the 30mhz basic AtomPro has a function called pulsin which reads a single line and converts it into a value 1000 - 2000 with 1500 or 1.5ms being centered... long story short, havin
[19:41:33] <LikWidChz> so either someone make some avr code for me, or I learn C++ and do it on one of the smaller usb boards.... just not sure how feasable every option is
[19:42:14] <Casper> your big line got cut
[19:42:20] <LikWidChz> jigga what?
[19:42:33] <LikWidChz> but but the thing was tugging on it it was a marlin or something
[19:42:35] <Casper> [...]long story short, havin
[19:42:50] <LikWidChz> do I need to further explain myself?
[19:42:52] <Casper> but yeah it's simple
[19:43:06] <LikWidChz> so this is what I made... one sec
[19:43:18] <Valen> you want to a) use C not c++ ;-P
[19:43:21] <LikWidChz> http://jreise.de/PPM/R617FS.html
[19:43:43] <Valen> b) look at examples using input capture on timers to do your reading
[19:43:55] <LikWidChz> does it need to be in assembly?
[19:44:29] <Casper> depending on the number of inputs, it's easy. use a timer (probably want to go with 1 count = 1ns), take note of the timer value when it go high, take note when it goes low, do the math (beware of warp over), output the result
[19:44:53] <Casper> can be done in C
[19:46:07] <Casper> btw, in case you didn't know, servo pulses are 1-2ms secs long, hence the 1000-2000 count out
[19:46:29] <Casper> but can be slightly longer or shorter due to crappiness in the circuit
[19:47:34] <LikWidChz> yeah I know this about the pulses, my transmitter a Futaba T7C has the ability to be out of spec when it comes to pulses they drop down to about 850 to 2250ms if like I wanted greater accuracy
[19:48:17] <LikWidChz> -mS and put in the correct value, but you know what I mean
[19:48:53] <LikWidChz> ... so do I want to do this with a single atmega chip? or do I want like a mini usb arduino
[19:49:16] <LikWidChz> I have a spare atiny 2313
[19:49:53] <LikWidChz> but the link i posted above the device captures 4 of the 7 channels and outputs a single PPM data line....
[19:50:15] <LikWidChz> at this point it would be kinda dumb going from pwm to ppm to serial with 2 different devices..
[19:53:07] <Casper> as I said, it's not that hard, the coding is probably very easy to do
[19:53:21] <Casper> the hardest part is to get the timer to run at the right speed
[19:53:34] <Casper> or do the math to output the right count
[19:53:38] <Casper> but math is cpu intensive
[19:54:16] <LikWidChz> so I mean if I know what the center time is, so if the timer counts a value like 234234 ticks cant I just make that 1500 and do the division/subtraction and whatever to get the right or sorta right value out?
[19:55:00] <LikWidChz> i mean so many clock ticks will = a single uS right?...
[19:56:09] <LikWidChz> honestly THE only thing this device will be doing is just that capturing data and spitting it out
[19:56:37] <LikWidChz> maybe light an led when its done a single pass....
[20:02:29] <Casper> yes you could do the math
[20:02:42] <Casper> but remember that division is expensive
[20:04:00] <magkas> hello i am designing an avr board with LM7805 regulator for power supply. Because i want to put an ISP header to use the isp option do i have to put a diode on the LM7805 output for the ISP voltages not to burn the LM7805 during the programming or there is no problem? Here is the scheme
http://img228.imageshack.us/img228/1157/screenshot20111107at351.png
[20:04:26] <LikWidChz> Casper... well what is better really a single avr chip or an arduino?
[20:05:28] <Casper> there is no difference
[20:06:03] <Casper> arduino is a stupid library that claim to make it easier while it just make it harder
[20:06:08] <LikWidChz> well what specific avr are we talking about? cause the most amount of data I would need would be 9 reads a second which seems like is only possible
[20:06:18] <Casper> it give cute name, but make everything else more complicated
[20:06:21] <LikWidChz> heh
[20:06:26] <LikWidChz> funny stuff man :)
[20:06:38] <Casper> servo is 50Hz
[20:06:44] <Valen> his application sounds pretty light, he could even do it as floats and it wouldn't be an issue
[20:06:57] <Casper> and I think someone benchmarked it at 8000 division/second
[20:07:07] <Casper> so yeah, won't be an issue to even do float division
[20:07:21] <LikWidChz> well I code basic a bit so honestly it sounds like it would be less then 100-150 lines of code
[20:07:37] <Casper> bbs
[20:07:53] <magkas> any answer for me?
[20:08:26] <LikWidChz> magkas, that voltage regulator which one is that?
[20:08:44] <Landon> 7805 is 5v
[20:08:56] <LikWidChz> when you do ISP what voltage does it rise up to?
[20:09:23] <magkas> i program it via stk500
[20:09:33] <magkas> and the VCC is 5Volt for the ISP
[20:09:54] <magkas> so it is like applying +5Volt at the LM7805 output pin
[20:09:55] <LikWidChz> how are you going to burn up that chip then?
[20:10:11] <LikWidChz> I dont think that will damage it
[20:10:20] <Valen> putting 5v on the output of a 7805 will damage it
[20:10:23] <LikWidChz> but your idea of a diode would fix that I would think right?
[20:10:32] <magkas> i don't know i am asking if it is going to be damaged if i supply +5Volt at its output pin
[20:11:00] <LikWidChz> I think putting a diode on the output pin would be safe way to correct that problem
[20:11:17] <LikWidChz> holy shit those are 18 cents each
[20:11:18] <Valen> it would but your avr will be getting 4.4v as vcc
[20:11:51] <LikWidChz> well you could also put a diode in series with that output pin to vcc since a diode drops .5volts... or was it .6 or .4... it drops something I know!
[20:12:15] <Valen> I wonder if you could stick another on the gnd reference of the 7805 to cancel it out
[20:12:36] <magkas> well here i found a design of an avr development board and he doesn't use a diode while he is using the LM317 for power supply
[20:12:36] <magkas> http://www.sparkfun.com/datasheets/AVR/avr-p40-8535-sch.gif
[20:13:00] <LikWidChz> my atiny 2313 operates at about 4.2 - 5.8 volts
[20:13:12] <LikWidChz> before the code on it says "NO" and fails to run anymore
[20:13:15] <Valen> probably doesn't use icsp for power
[20:13:25] <Landon> http://www.national.com/ds/LM/LM340.pdf has a few application hints, maybe one of them will give you an idea
[20:13:30] <Valen> LikWidChz: they have these things called datasheets for telling you stuff like that
[20:13:37] <LikWidChz> hahahaha :)
[20:13:54] <LikWidChz> yeah yeah... its fun learning the hardway sometime
[20:14:13] <magkas> i searched the lm7805 for something like this but i didn't found anything
[20:14:33] <LikWidChz> well again, the price of the component is very cheep you could assemble one and give it a try
[20:15:08] <LikWidChz> any time I start a project ill buy extra parts, and those voltage regulators are pretty slick... although at 1amp they get pretty hot if you pull that
[20:15:23] <LikWidChz> for little shit like avr and less then a 100 or so ma they can be left un sinked
[20:22:38] * Landon enjoys getting things right the first time, every time :P
[20:24:44] <karlp> Casper:
http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en538609
[20:24:49] <karlp> add a serial number to anything....
[20:25:25] <karlp> currently, my method is just to use the low byte of an eui48 as the device address.
[20:25:37] <karlp> and check that you don't have dups beforethe devices ship out
[20:26:04] <karlp> but we're only expect 10-15 out of 247 possible address (modbus has a few reserved addresses in the 8 bit space)
[20:26:44] <karlp> if you want to get closer to 240/247, then yeah, as someone else said, the one wire address resolving algoritim is probably pretty good
[20:37:42] <Casper> karlp: I find it sad that atmel didn't put a serial number on their avr...
[20:37:56] <Valen> i believe its an option
[20:37:59] <Casper> but I have another idea on how to deal with it
[20:38:11] <Casper> Valen: nope, the s/n is only software
[20:38:14] <Casper> nothing hardware
[20:38:45] <Valen> you want them to burn a serial number into every parts silicon?
[20:38:50] <Valen> nobody does that
[20:39:11] <Valen> things that have serial numbers have them stored in some form of prom
[20:39:20] <Casper> but I might reserve a space in the address for the "brand new and conflict"
[20:39:42] <Casper> Valen: they could have put one in the eeprom or something
[20:39:49] <Casper> or a special area eeprom or whatever
[20:40:07] <Valen> as i said i believe you can order them with some form of either serial or guid
[20:40:53] <Casper> Valen: I kinda doubt it, as they have no reserved space for it, but if you make them program the device for you then they can make the serial
[20:41:06] <Casper> but that's only for bulk purchase
[20:41:22] <Valen> so you can in fact *order* it
[20:41:33] <Valen> with sufficent moq
[20:41:51] <Valen> thing is why bother doing that, 99% of applications have no need for it
[20:42:03] <Casper> well, except for the usb
[20:42:13] <Casper> and even the usb one don't have serial
[20:42:13] <Valen> and since everything needs to be programmed anyway just put it in then
[20:42:20] <Valen> they don't need a serial number
[20:42:30] <Casper> if you plug 2 of them yes they do
[20:43:07] <Valen> doesn't negate my point
[20:43:12] <Valen> you need to flash the chips anyway
[20:43:16] <Casper> yeah
[20:43:48] <Casper> hmm might actually make a mix of "ip" and CAN for my protocol...
[20:43:56] <Valen> there is no requirement to have a serial number
[20:44:07] <Valen> i don't believe
[20:44:11] <Valen> for usb
[20:44:47] <Casper> iirc, if you have 2 or more, the serial need to be different, else the driver won't be able to proprelly expose the 2
[20:44:59] <Casper> but I might be wrong
[20:45:02] <Casper> anyway
[20:45:38] <Casper> so... if I make my own protocol, I might as well do it with a 16 bits address space and reserve like 100 address for new and conflicting devices
[20:46:57] <Casper> priority, from, to, length, data, crc, EOF
[20:47:37] <Casper> man that will be a pita to start :D
[21:11:31] <inflex> lo folks
[21:17:48] <Valen> ahoyhoy
[21:23:00] <inflex> how's things?
[21:29:27] <Valen> procrastionational
[21:29:37] <Valen> youtube = the debil
[21:31:28] <Casper> hi inflex
[21:31:43] <Casper> inflex: finally, I think I'll go with a personal variation of can bus
[21:32:05] <inflex> Casper: oh well, at least it'll work :)
[21:32:28] <Casper> inflex: I like canbus electrical, but I don't like the official protocol
[21:33:49] <Valen> if your going to use can stick with the protocol, then you can use off the shelf stacks
[21:34:08] <Valen> also they have got stuff in hardware for it i believe
[21:34:27] <Valen> also keep in mind that can AVR's probably don't have the trancievers onboard
[21:34:35] <Valen> (the electrical driver bits)
[21:35:15] <Casper> thing is, can bus is just broadcast, it do not have a mean to direct the info to one particular controller
[21:35:16] * inflex goes looking for a beefy MOSFET that can handle a few hundred watts of dissipation
[21:35:21] <inflex> might go with multiple in parallel
[21:35:33] <Valen> thats generally the way its done
[21:35:35] <Valen> Casper: so?
[21:35:52] <Casper> I want to talk directly to one of them
[21:35:53] <Valen> if its home automation how much data do you need to say "open the blinds"?
[21:35:59] <Valen> why
[21:36:04] <Casper> debug and firmware update
[21:36:09] <Valen> so?
[21:36:26] <Valen> its like saying ethernet isn't good enough because you want to talk to one computer
[21:36:37] <Casper> ethernet have a destination field
[21:36:43] <Valen> (ethernet is a broadcast protocol btw, it just doesn't look like it anymore)
[21:36:47] <Valen> so does can
[21:37:06] <Casper> I do not see any source or destination in the protocol posted on wikipedia
[21:38:21] <Valen> the identifier
[21:39:01] <Casper> yeah that could be a source
[21:39:05] <Casper> but no destination
[21:39:12] <Valen> use it as the destination
[21:40:45] <Casper> but then you can't know the source :D
[21:41:11] <Valen> put it in the data
[21:42:13] <Casper> it's only 8 bytes long
[21:42:16] <Valen> you generally run a protocol ontop of the can bus itself
[21:42:22] <Valen> opening blinds still isnt hard
[21:42:36] <Valen> if it can run the ABS in your car it should be good enough for home automation
[21:42:39] <Casper> will check...
[21:43:07] <Casper> but can could work as it maybe
[22:06:13] <amstan> is there something better than simulavr that's free?
[22:06:21] <amstan> simulavr is throwing me a bunch of decoder.h:59: WARNING: Unknown opcode: 0xffff
[22:06:27] <amstan> and not sure what to do about them
[22:07:16] <inflex> gharrr... using a ganged array of MOSFETs just won't work nicely
[22:07:22] <inflex> because I can't heatsink the bastards properly
[22:07:26] <amstan> here's my attempts so far:
http://pastebin.com/zhGjadtz
[22:10:11] <Valen> inflex: why not?
[22:10:48] <inflex> how the heck are you supposed to manage 300W dissipation from a D2PAK SMD device?
[22:11:05] <inflex> I mean, 50W, yeah, maybe 100 with a heatsink on top... but dang, 300W
[22:11:09] <Tom_itx> liquid nitrogen
[22:11:43] <Valen> you put 6x in and do 50W per device with active cooling
[22:11:53] <Valen> keep in mind the datasheets lie
[22:11:55] <Valen> alot
[22:12:15] <Valen> like the little * on the To220 package fets that says basically the legs melt off at 50A
[22:12:32] <Valen> even though they spec the parts to 200A
[22:12:49] <inflex> yea,
[22:13:09] <inflex> basically I'm trying to make a variable load
[22:13:31] <inflex> which is nothing more than a MOSFET driven by an opamp monitoring the source voltage relative to ground
[22:13:35] <Valen> don't use the fet as the load
[22:13:40] <inflex> (with a shunt between source and GND)
[22:13:46] <inflex> no, it's perfectly fine to do so
[22:13:54] <Valen> put a power resistor + inductor and PWM as the load
[22:13:56] <inflex> much simpler than a chunk of bloody resistors
[22:13:58] <inflex> can't PWM this load
[22:14:06] <Valen> PWM *Faster* ;->
[22:14:06] <inflex> has to be constant
[22:14:23] <Valen> inductor on the high side of the switch with a phat cap?
[22:14:36] <Valen> (so the spikes dont eat yer fets)
[22:15:39] <Valen> if its a load like that I'd bust it off a PCB and solder the fets to a nice hunk of copper and heatsink it (with a fan)
[22:16:17] <Valen> or look at something like an isotop fet
[22:27:44] <inflex> well, I don't see the problem with just using the opamp drive method - gives me more precision too
[22:28:05] <inflex> since I can drive it with a 12bit DAC
[22:28:37] <Valen> nothing bad about it, just trying to work around the problem
[22:28:46] <Valen> you probably want some older fets lol
[22:29:04] <Valen> the new ones probably go from off to on with too little voltage swing ;->
[22:29:06] <inflex> oh okay, the problem really was just a heatsinking one and keeping the MOSFETS on the same mounting heatsink
[22:29:38] <Valen> we used To-220's and made a heatsink to bolt the fets to
[22:29:47] <Valen> you could go something like a victor speed controller
[22:29:48] <inflex> yep
[22:30:07] <Valen> http://www.robotcombat.com/store_ifispeedcontrollers.html
[22:30:08] <inflex> well, was thinking of TO220's, though they're limited to about 100~170W
[22:30:16] <Valen> crap photos
[22:31:02] <Valen> but 9 fets in a square vertically mounted with a fan above them blowing through
[22:34:11] <inflex> eh, no heatsinks on them
[22:34:50] <Valen> they don't but they have active cooling and lots of fets
[22:34:52] <inflex> but you see though, the design focus for an ESC is different, I _want_ to consume watts, where's in the ESC you're going for minimal wattage dissipation
[22:35:02] <Casper> ain't to220 limited to 25W?
[22:35:15] <Valen> both have the same issues though
[22:35:20] <Valen> you both want to get rid of heat
[22:35:23] <Casper> to3p/to218? are 125W...
[22:35:26] <inflex> so, while those FETs are going to be pushing out 5~10W each, where's mine will be stomping for 50~100W
[22:35:41] <Valen> those fets are running at 100A or so each
[22:35:55] <Valen> 3khz PWM
[22:36:19] <Valen> what is the total load you want to make?
[22:36:26] <inflex> yeah, don't want to use PWM, this is meant to be a DC load for testing lipo cells for iR
[22:36:32] <Valen> I know that
[22:36:41] <Valen> the PWM adds to the heat load
[22:36:51] <Valen> bigass fets is my bag baby ;->
[22:36:55] <inflex> heh
[22:37:09] <Valen> 600A H-bridges ftw
[22:37:30] <Valen> 24 fets and 13A fet drivers
[22:38:03] <inflex> insane
[22:38:06] <Valen> whole thing was smaller than your fist too
[22:38:13] <Valen> no photos allowed unfortunatly
[22:38:16] <inflex> might go with something like a TO-3PBL package (looks like a flat-wide TO220
[22:38:24] <Valen> look at isotop
[22:38:32] <Valen> if they are still what the cool kids use
[22:38:45] <Valen> isotop + CPU fan = win?
[22:40:15] <inflex> ha, they're about 10x more expensive
[22:40:27] <Casper> I love to3p and to218
[22:40:28] <Valen> yeah but you only need 1
[22:40:46] <inflex> well, yeah... but I was only planning on 2 or 3 of the others like TO3P
[22:40:46] <Valen> and the wiring is really easy ;->
[22:41:26] <inflex> heh I hate wiring! :p I think TO3P might be a reasonable compromise
[22:43:13] <Valen> url for to3p part?
[22:43:20] <inflex> http://au.element14.com/fairchild-semiconductor/fqa85n06/mosfet-n-to-3p/dp/9846093
[22:43:35] <Valen> interesting
[22:43:59] <inflex> yeah, untl yesterday I always thought TO3 was those twin-lobed metal cans
[22:44:12] <Valen> I thaught it was that big round metal can
[22:44:18] <Valen> 10C piece sized
[22:44:27] <Valen> 3 wires out the bottom
[22:44:33] <inflex> yea, that's just TO3 only, as opposed to TO3P
[22:44:48] <Valen> well they added a letter, might as well change the whole thing
[22:45:41] <inflex> scary thing is, they're in some ways MORE expensive than the ISOTOPs (the TO3)
[22:45:49] <Valen> raining so hard i cant see houses across the street
[22:45:56] <inflex> send it up here
[22:46:15] <inflex> My favourite MOSFET package for what I do a lot of is the SOT663 (or is that 669)
[22:46:30] <inflex> anyhow, looks like a SO8 on the source/gate side, but it's a flat-tab on the drain side
[22:46:32] <inflex> very flat
[22:46:42] <Valen> yeah i've seen those
[22:46:46] <Valen> crap its getting heavier
[22:46:55] <Valen> and heres the hail
[22:47:19] <inflex> nice
[22:47:31] <inflex> but the insane thing is the low Rds they're getting...down to 1.05mR
[22:47:48] <inflex> http://au.element14.com/nxp/psmn1r2-30ylc/mosfet-n-ch-30v-100a-lfpak/dp/1895403
[22:48:00] <inflex> going to use a 4 of those on my 100A R/C power switch
[22:48:30] <inflex> won't require any heatsinking at all
[22:49:59] <inflex> they'll dissipate 1W/pc and most of that will go in to the copper flood
[22:52:57] <inflex> oh gooooodie... someone just wanted to order 6 of my mAh counters
[22:59:31] <Casper> how's the selling of those products? enought to make a living of it or barelly covering your hobby?
[23:06:35] <ziph> Yay.
[23:06:57] <ziph> Finally back from the horror weekend.
[23:07:59] <ziph> inflex: I'd love to get some finite element analysis software for heat transfer, but I haven't seen anything open source.
[23:27:34] <inflex> hrror w/e ? what went wrong?
[23:42:19] <ziph> inflex: Backed in to a silly girl using a car park like a race track, managed to miss a sign for the only 80 zone (and a 1Km long one at that) in a 200 Km long 110 zone.
[23:43:46] <CapnKernel> ziph: Fun.
[23:43:57] <CapnKernel> Change your nick to "Lucky"
[23:44:29] <ziph> Oh, in a rental car.
[23:45:04] <ziph> And the place I usually use with $0 excess changed their contracts to completely exclude animal damage (just in case you
[23:45:34] <ziph> 're driving somewhere in Australia that doesn't have Kangaroos and cattle I guess), so I had to use an agency with an excess.
