#avr | Logs for 2012-10-04

Back
[00:43:04] <Richard_Cavell> So how come AVRs respond to so-called high voltage programming? Is there some special circuit that responds only to 12 volts instead of 5, that takes over from the normal circuit?
[01:03:26] <CapnKernel1> Richard_Cavell: Yes, there's a special circuit inside the AVR attached to the RESET pin.
[01:03:55] <Richard_Cavell> so when it detects 12 volts it says "okay, we're not operating normally here, so we'll intercept the signals and use them for programming only
[01:04:10] <CapnKernel1> Yes, it puts it into out-of-Kansas mode
[01:04:36] <CapnKernel1> Most pins of the AVR have ESD protection. The ESD protection on the RESET pin is not as strong, so that HVPP will work.
[01:05:44] <CapnKernel1> Google for AVR042 for more info about how to properly design circuitry which attaches to RESET.
[01:06:52] <Richard_Cavell> they call it RESET but it's kind of like a dead man's handle
[01:07:00] <Richard_Cavell> you don't set it to RESET, you set it to keep the chip going
[01:07:50] <Richard_Cavell> they should call it the DONTRESET
[01:17:34] <creep> watchdog timer
[01:28:08] <Richard_Cavell> not the same thing
[01:28:17] <Richard_Cavell> a watchdog timer is for when the core isn't doing anything sensible
[02:03:38] <ferdna> looking for rubber belts (timing belts?)... for my servos...
[02:03:46] <ferdna> *servos gears
[02:05:12] <ferdna> any ideas where i could get'em in assorted sizes, pitches and cheap?
[02:26:25] <Richard_Cavell> is there some sort of canonical list of which AVR8s support JTAG?
[02:26:30] <Richard_Cavell> I can't find a sensible list on line
[02:42:39] <OndraSter> I like the name of next TBBT episode: "The Decoupling Fluctuation" :D
[02:42:54] <OndraSter> Richard_Cavell, atmel studio 6 release notes
[02:43:06] <OndraSter> Dragon supports JTAG on all devices that have it
[02:43:12] <OndraSter> so you can check there
[02:43:15] <Richard_Cavell> I've been looking through them
[02:43:18] <OndraSter> (and if not, then there is jtagice3)
[02:43:28] <Richard_Cavell> so far as I can tell the smallest one that has it is xmega16
[02:43:50] <OndraSter> for debug you can use also PDI (xmega only), debugWire (megas/tinys)
[02:45:08] <OndraSter> mega32 is physically smaller. But mega32 has 1/xth of all the features of xmega.
[02:45:22] <OndraSter> just get my xboard - it fits into breadboard and has xmega on it :P
[02:45:24] <OndraSter> big ass xmega
[02:45:59] <Richard_Cavell> what does the x mean?
[02:46:08] <OndraSter> in the mega or board?
[02:46:38] <Richard_Cavell> in the mega
[02:46:49] <OndraSter> xmega is newer mega
[02:46:53] <OndraSter> with much more features
[02:47:05] <OndraSter> and double the speed (32MHz vs 16/20MHz)
[02:47:14] <OndraSter> and cheaper than regular megas even
[02:47:16] <Richard_Cavell> what kind of features? Do I need them?
[02:47:18] <OndraSter> (with comparable flash)
[02:47:20] <OndraSter> you tell me!
[02:47:37] <OndraSter> fast ADC, DAC, many PWMs, DMA, Event System
[02:47:53] <OndraSter> it puts any other chip into shame
[02:48:10] <OndraSter> only thing it lacks is ethernet
[02:48:13] <OndraSter> the only*
[02:48:39] <Richard_Cavell> to be honest I barely need the attiny85
[02:48:44] <Richard_Cavell> let alone all that stuff
[02:48:58] <OndraSter> so stop looking at xmega :P
[02:49:30] <Richard_Cavell> `well the JTAG interface would be nice
[04:45:04] <Amadiro> Looks like atmel, ti, microchip & co all send their samples from the same warehous/distributor
[04:45:20] <Amadiro> All the samples came packaged exactly the same way
[04:45:32] <OndraSter> hehe I thought it was just me :P
[04:46:06] <Amadiro> OndraSter, got a bunch of xmegas, btw, so I'll probably be playing around with those as well
[04:46:13] <Amadiro> different models than the one you use on your board, though
[04:48:36] <Amadiro> I'll probably be using the ATxmega64A3U, which seems to be pretty much the cheapest-you-can-get while still having a respectable amount of IO and 4 UARTs or more
[04:49:03] <OndraSter> same as I use, just less flash :)
[04:49:05] <OndraSter> and less RAM
[04:49:11] <OndraSter> (I use xmega256a3u)
[04:49:13] <Amadiro> well, yeah
[04:49:21] <OndraSter> peripherals are exactly the same
[04:49:35] <Amadiro> OndraSter, should I get around to actually commercializing the project, I'll probably also aim for arduino compat
[04:49:48] <OndraSter> your?
[04:49:49] <Amadiro> as in, programmable via arduino IDE
[04:49:57] <OndraSter> I HAVE GOT MY XBOARD!!!
[04:50:00] <OndraSter> COMPETITION! NOO
[04:50:08] <Amadiro> OndraSter, haha, don't worry, it's not a devboard
[04:50:14] <Amadiro> it's a solution for lighting
[04:50:27] <tld> OndraSter: congrats. :)
[04:50:27] <OndraSter> the thing is - you will have to rewrite a lot of libs since xmegas have a bit different.. everything :P
[04:50:34] <OndraSter> tld, to? :D
[04:50:40] <tld> OndraSter: you got your xboard?
[04:50:47] <OndraSter> I was the first one to ever have one
[04:50:49] <OndraSter> since I am making them.
[04:50:59] <Amadiro> OndraSter, well, I don't really need arbitrary libs to work, the only libs I expect my users to use are serial and GPIO, really
[04:51:06] <OndraSter> oh
[04:51:39] <Amadiro> serial for talking between the nodes to synchronize, and GPIO to control the lights
[04:51:55] <tld> OndraSter: Ahh, I think I got you confused with the other guy here, really looking forward to getting his.
[04:54:11] <OndraSter> which one, tld ?
[04:54:24] <Amadiro> megal0maniac_afk, presumably
[04:54:36] <tld> I *think* megal0maniac.
[04:54:40] <OndraSter> I sold my board to megal0maniac_afk :P
[04:54:46] <OndraSter> I am the maker, he is my client :P
[04:55:02] <tld> but I'm not sure, I kind of already proved I misremembered the nick. ;)
[04:55:09] <tld> yeah, I got that (now) :)
[04:55:34] <tld> hopefully, you can soon add me to your client-list, just need to do some paycheck-waiting first.
[04:56:29] <tld> what I like about it btw, is the number of intelligent IOs in a breadboard-friendly package (and breadboard-friendly == easily reusable).
[04:56:41] <tld> (intelligent IOs, as in usart, SPI etc… )
[05:00:34] <tld> Hmm. I need to mix 3.3V (radio) and 5V (LCD) in a project. In addition to those, I have components I could run as either 3.3V or 5V (avr, shiftregs for LCD, DS18B20… ). I'm thinking about where to draw the line. There's no back-talk from the shiftregs to the MCU, so it'd be convenient to run MCU at 3.3V, shiftregs at 5V, but datasheet says Vih = 3.15V with Vcc=4.5V, so low-watermark for high-IO might be too high for a 3.3V high from the MCU
[05:01:08] <OndraSter> what shiftregs are those?!
[05:01:22] <tld> sn74hc595n, from Ti
[05:01:27] <OndraSter> huh
[05:01:37] <OndraSter> there should be zero issues IMHO
[05:01:53] <OndraSter> I am using HD44780 display which is powered from 5V but logic goes from the MCU at 3.3V just fine
[05:02:40] <tld> This is a HD44780 "or equivalent" display.
[05:03:02] <tld> I'm a bit worried that 3.3V might not read as high on the shiftregs, but I could run the shiftregs at 3.3V as well, which should be enough for the LCD.
[05:03:48] <tld> Only downside would be if shiftreg is pulling low, LCD in read-mode giving too much current over a 5V high, hurting either the LCD or the shiftregs?
[05:03:59] <tld> Which I could resolve using current-limiting resistor between?
[05:04:03] <OndraSter> why would you read the LCD? :)
[05:04:10] <OndraSter> also 595 is SIPO only
[05:04:26] <tld> Yeah, I'm not really planning to read it, was just thinking if I somehow pull it to read-mode by mistake.
[05:04:45] <OndraSter> one usually ties the r/w to LOW directly
[05:05:03] <tld> that should work. :)
[05:05:05] <tld> thanks. :)
[05:05:42] <tld> So there we have it… MCU and shiftregs both at 3.3V, LCD at 5V, r/w tied low… should work. :)
[05:05:43] <tld> thanks. :)
[05:05:49] <OndraSter> np
[05:05:58] <OndraSter> I already tried driving the hd44780 from 3v3
[05:06:18] <OndraSter> manual way though, not via the MCU directly, I was too lazy to get some code running. I just used 5V supply and 3V3 LDO
[05:06:25] <OndraSter> and manually wired all the data always :P
[05:07:00] <tld> manually wired it how?
[05:07:06] <OndraSter> 10 wires
[05:07:11] <OndraSter> going to either gnd or 3v3 :P
[05:07:19] <OndraSter> D0-D7 + RS + E
[05:07:34] <OndraSter> always set up manually all 8 data bits and pushed button on E :P
[05:07:59] <tld> but how do you get data there if you tie D0-D7 to gnd/3v3?
[05:08:09] <OndraSter> by tying them to something else :)
[05:08:15] <tld> ahh, like that. :)
[05:08:21] <OndraSter> just as MCU would output 0x3C for example, I did it manually
[05:08:22] <OndraSter> ye
[05:38:21] <tld> I'm starting to wonder if these LCDs really need 5V… at least one datasheet says 2.7 - 5.5V for operation, 3 - 11V for LCD driver power, but noting that you need 5V for 2Mhz. I'm nowhere near that, so wondering if I can simply drive it from 3.3V. Hmm.
[05:39:12] <tld> Certainly would make life easier.
[05:39:40] <Richard_Cavell> I just think it's indecisive when a datasheet gives a wide range
[05:39:48] <Richard_Cavell> My STK600 takes voltage of between 9 and 15 volts
[05:39:57] <Richard_Cavell> So does that mean 9, or 15?
[05:39:59] <Richard_Cavell> Or 12?
[05:40:23] <tld> For testing at least, I feel safer going down than going up.
[05:42:28] <OndraSter> Richard_Cavell, anything between 9 and 15
[05:42:37] <OndraSter> tld, that depends on the display
[05:42:43] <OndraSter> there are now 3v3 hd44780 LCDs on ebay :)
[05:42:48] <OndraSter> for the same price as 5V I think even
[05:43:22] <tld> slightly above the cheapest 5V ones (if you include shipping)
[05:43:31] <OndraSter> well, close enough
[05:43:51] <tld> but the one I have was listed as 5V. Just starting to wonder if there's really a difference, or if I could use 3.3V with the LCD I have (as opposed to the ones in the mail, on their way)
[05:43:53] <OndraSter> many external chips are now 3v3 only
[05:44:00] <OndraSter> tld, test it :)
[05:44:06] <OndraSter> mine didn't work on 5V
[05:44:07] <tld> on my way. :)
[05:44:09] <tld> hmm.
[05:44:13] <OndraSter> but it was as 4.5 - 5.5V
[05:44:20] <tld> strange.
[05:45:09] <Amadiro> Richard_Cavell, it typically means the board has a regulator that will accept anything from 9 to 15 volts and then convert it to the exact right thing the board needs internally (presumably 5 or 3.3V)
[05:45:39] <OndraSter> aye
[05:49:19] <TheFridge> Is it better to use AIN or DIN to read PWM?
[05:49:36] <OndraSter> it is better to use #arduino
[05:50:32] <Amadiro> theBear, AIN?
[05:51:44] <OndraSter> wait
[05:51:48] <OndraSter> is it arduino or not?
[05:52:03] <TheFridge> ok
[05:53:18] <OndraSter> what did just happen
[05:54:41] <Richard_Cavell> OndraSter: Another satisfied customer, that's all you need to know
[05:54:58] <OndraSter> mkay
[05:55:29] <OndraSter> sounds mkay
[07:05:15] <theBear> wait, 'read pwm' ?
[07:10:04] <OndraSter> yeah
[07:10:06] <OndraSter> no idea :)
[07:23:20] <Tom_itx> use the mwp pin to read it back in
[07:25:07] <OndraSter> maybe if he comes back
[07:25:23] <OndraSter> I thought that he means those analog pins on arduino so I accidentaly sent him to #arduino :D
[07:25:53] <Tom_itx> you could use input capture to read it
[07:26:19] <Tom_itx> but if it's pwm you should know what it is unless it's coming from an outside source
[07:26:35] <OndraSter> he is not here you know
[07:26:49] <Tom_itx> i'm not either
[07:28:17] <OndraSter> ah right
[07:36:55] <creep> tld<< not really, the LCD drive is probably around 12V
[07:37:14] <creep> and LCD does not take much current
[07:43:45] <tld> creep: probably bad phrasing on my part. 3-11V was probably power supply for lcd driver (I'm guessing internal step-up in the LCD module)
[07:44:08] <creep> maybe...
[07:44:20] <creep> so you can run it using a single lithium battery
[07:44:41] <creep> possibly using a 3.3V LDO regulation
[07:45:01] <tld> In either case, datasheet doesn't seem to offer any show-stoppers for using it with 3.3V instead of 5V.
[07:45:09] <tld> so testing (… getting to it… )
[07:45:26] <creep> if you really want you can make 5V from a single 1.5V battery using one transistor
[07:45:46] <OndraSter> and some magnetics
[07:45:54] <OndraSter> "transformer" you can call it if you dare
[07:45:57] <creep> so try it it won't hurt
[07:46:07] <OndraSter> creep, joule-thief-like circuit?
[07:46:18] <tld> it's just being one of those days, where the stuff I'm going to test flows nicely, but everything else keeps slightly messing up… shook up USB stack on FreeBSD-machine too badly, so seems to be partly locked up… no USB, no usbasp, no programming, no testing…
[07:46:51] <OndraSter> atmel studio 6 and 5 manage(d) to lock up its own driver quite often
[07:46:58] <OndraSter> and new devices would not be enumerated anymore
[07:47:02] <creep> OndraSter<< yeah something like that, but that joule thief thing is surrounded by huge amount of stupidity
[07:47:20] <OndraSter> not sure if it is atmel's issue in driver or the fact that intel's drivers wait for ongoing enumerations before they start another one
[07:47:34] <OndraSter> huge amount of stupidity?
[07:48:01] <tld> reminds me of what I'm seeing here actually… using USB isolated, set to full speed instead of low speed, enumeration didn't go too well, and now everything's *partly* locked up.
[07:48:17] <creep> yeah they used it as some campaign to sell useless junk that is stated to have over unity efficiency thiefing electric energy from nowhere
[07:48:28] <tld> using a FreeBSD version right after USB stack got fully replaced, and did off-spec things with it, so I have only myself to blame.
[07:48:45] <Amadiro> creep, haha
[07:49:12] <creep> there are many people who can be made to believe this kind of things if you show off something magical to them
[07:49:21] * tld wouldn't mind a 120% efficient PSU. ;)
[07:49:33] <Amadiro> creep, reminds me of those wireless chargers for cellphones
[07:49:40] <creep> "hey look at this, its working from a dead battery!"
[07:49:46] <OndraSter> mm
[07:49:48] <Amadiro> they are supposed to pick up radiation and use it to charge your batteries
[07:49:54] <OndraSter> yep
[07:49:58] <OndraSter> wifi harvesting :D
[07:50:07] <creep> haha
[07:50:14] <Amadiro> which might work if your mobile phone is a single attiny45 microcontroller
[07:50:26] <creep> TV and radio stations have this "free energy" you are talking about
[07:50:26] <tld> reminds me: wonder *exactly* how newer power-meters are measuring… would be fun to see if you could nab energy when it's not looking. not sure I'd like to go down that path though. ;)
[07:50:36] <creep> works best if tuned to resonance
[07:51:03] <OndraSter> <tld> reminds me: wonder *exactly* how newer power-meters are measuring… would be fun to see if you could nab energy when it's not looking.
[07:51:04] <OndraSter> what? :P
[07:51:12] <creep> you can really light up 40W neon lights with it
[07:51:24] <tld> WiFi doesn't put out meaningful amounts of power in the first place. Even if you could grab 100% of that energy (leaving the WiFi effectively offline), it'd still be a shit charger.
[07:51:38] <OndraSter> tld, that's why I put ":D" on the end of my sentence
[07:51:40] <Amadiro> creep, if you're right next to them, maybe -- the signal gets weaker by the inverse square
[07:51:43] <creep> but a few hundred kW radio station does
[07:52:02] <tld> but FM-radio is going away though, isn't it?
[07:52:12] <tld> so doesn't seem good for a long-term strategy.
[07:52:35] <creep> so police will send emails and skype on the radio?
[07:53:22] <tld> brb, got to reboot that router…
[07:55:01] <karlp> tld: have a look at the ade2778 or any of the other power meter ics.
[07:55:01] <creep> why is it good for the 4 pin smd quartz to have 2 NC pins diagonally?
[07:55:12] <karlp> (you don't get any "window" to grab free power or anythign)
[07:55:34] <karlp> sorry, ade7758
[07:55:45] <karlp> ST and TI both make meter ics too.
[08:32:09] <tld> Hmm. I've got a programmer with a 5V-line. It reaches target when I'm using it. If I disconnect the 5V line, connect another 3.3V-supply (referenced to same ground), it won't find target. usbasp-knockoff. Any ideas?
[08:32:33] <tld> (normally I'd think 3.3V was enough to register as high on the programmer, based on at atmega8)
[08:43:03] <karlp> tld: what's your 1.5v -> 5v with a single transistor trick?
[08:44:30] <tld> grr @ self. miswired one of the ground leads (didn't notice a split in the breadboard rail)
[08:44:56] <tld> karlp: don't have one. joule-thief should work, but never needed… so far, I've been preferring step-down with enough battery-voltage.
[08:46:12] <karlp> joule thief isn't single transistor
[08:46:50] <karlp> adding coils rapidly leaves any semblence of single transistor
[08:48:11] <Kevin`> a joule theif has no regulation
[08:48:24] <karlp> that too.
[08:49:25] <creep> one zener diode can stabilize 5.1V...
[08:49:36] <Kevin`> that's not efficient
[08:49:42] <creep> but works
[08:50:07] <creep> and the single transistor converter isn't too efficient either...
[08:50:56] <creep> up transforming, >0mA, V_CE_sat >0.3V ...
[08:51:44] <creep> drives core/transistor upto saturation/desaturation
[09:02:20] <tld> I hate this breadboard, I hate this breadboard, I hate this breadboard.
[09:06:08] <OndraSter> lol
[09:08:00] <tld> never used it before, but grabbed it now to have working-space.
[09:08:34] <tld> single rails on either side, but it's also separated into 3+4+3 on the rails, without mentioning or hinting at it. I should've checked before I started using it though.
[09:09:40] <tld> http://www.ebay.com/itm/2900-Points-Solderless-BreadBoard-with-Jumper-140-pc-/320461562033?pt=LH_DefaultDomain_0&hash=item4a9cff60b1 <-- *DO NOT BUY*
[09:10:00] <tld> I think it needs to be replaced by something like this: http://www.ebay.com/itm/2390-Tie-point-solderless-breadboard-free-shipping-/270717752784?pt=LH_DefaultDomain_0&hash=item3f080915d0
[09:11:22] <tld> http://www.ebay.com/itm/3220-Tie-point-solderless-breadboard-one-day-shipping-/320723221923?pt=LH_DefaultDomain_0&hash=item4aac97fda3 <-- better still. :)
[09:13:10] <tld> hmm, or perhaps I should just make my own, with isolated area(s) for PSU, etc…
[09:13:53] <OndraSter> strip the back
[09:15:33] <r00t|home> tld: i have that same breadboard... but it's just missing documentation really, no? (i ALSO have the supposedly better one you linked...)
[09:18:20] <tld> It's not *horrible*.
[09:18:32] <tld> I just get a better feeling with the other type (I have smaller ones of that type)
[09:18:39] <tld> easier insertion and so on as well.
[09:18:42] <megal0maniac> Breadboard missing documentation?...
[09:19:05] <r00t|home> megal0maniac: on the power rails having interruptions... should be marked on the board
[09:19:28] <tld> at least IMHO it should
[09:19:35] <OndraSter> <tld> easier insertion and so on as well.
[09:19:38] <OndraSter> that's what she said
[09:19:38] <tld> but I also should've checked first.
[09:20:54] <tld> the circuit works, and looks like I can program it without 5V from the programmer, so all worked out in the end.
[09:21:02] <tld> time to go pick up kid from school, but bbl.
[10:01:36] <megal0maniac> r00t|home: I've found it to be pretty standard, though.
[10:03:57] <Amadiro> mobile phone with activated speakers sitting on a drawing tablet = pretty weird shit happening on my computer
[10:05:22] <CapnKernel> Haha, same happens if I put my phone on the laptop trackpad
[10:05:42] <CapnKernel> Do you like to draw?
[10:07:47] <Amadiro> CapnKernel, every once in a while, but I mostly draw on pen & paper or with acryl on canvas
[10:08:15] <Amadiro> I'm not really very comfortable with the drawing tablet
[10:18:24] <tld> Problem with stuff like HD44780 lcds isn't that there isn't documentation, but that there is so *much* of it…
[10:18:52] <megal0maniac> How is that a problem?
[10:18:59] <megal0maniac> It all pretty much says the same stuff
[10:19:22] <tld> Yeah, with subtle differences and misunderstandings. ;)
[10:19:35] <tld> Don't get me wrong, too much is waaaay better than too little.
[10:19:38] <megal0maniac> And Chinglish
[10:19:51] <tld> Just a lot of "Cool! I made it work! I'l make a tutorial now!"
[10:20:33] <tld> (a lot of them are really good, but a lot of them are really bad as well… )
[10:20:42] <megal0maniac> tld: Well, I'd use that as a reference for after I'd made it work from the datasheet. Implementation examples, and such
[10:21:20] <tld> Yeah, they can be nice if you just need something quick to recall pinout and the like.
[10:21:39] <tld> But "wire four pins, use the arduino lib"-type tutorials are… sometimes okay, sometimes just noise.
[10:22:43] <edmont> hi
[10:23:17] <edmont> i'm having some problems with the ADC in atmega128rfa1
[10:24:05] <edmont> i read 638 for a 0.5 V input with the reference set to internal 1.6 V
[10:25:52] <tld> are you sure about both the 0.5V and the 1.6V reference?
[10:25:54] <edmont> it should be 1024*(0.5/1.6) = 320
[10:26:34] <tld> yeah, that's what I figured.
[10:26:35] <edmont> tld: yes, i set REFS0 and REFS1 in AMUX
[10:27:05] <edmont> and 0.5 V i measured it with multimeter and osciloscope
[10:28:08] <edmont> actually i generate them with a waveform generator
[10:28:37] <tld> Had it been a diff adc, I'd wonder if polarities could have been accidentally reversed, but…
[10:28:59] <megal0maniac> I thought the ref was 1.1V...
[10:29:18] <megal0maniac> Which still only gives 465
[10:29:33] <edmont> megal0maniac: for my MCU is 1.6V
[10:34:06] <megal0maniac> edmont: I can't find anything about an internal analogue reference on that chip. A search for "analog ref" returns nothing and it isn't mentioned in the block diagram, where it usually is...
[10:36:18] <edmont> megal0maniac: search for REFS0
[10:45:44] <megal0maniac> Weird that it isn't listed in the usual places
[10:46:23] <megal0maniac> Interestingly, what you're getting is nearly exactly double what you should have
[10:47:00] <megal0maniac> Which points to either a 0.8V ref, or something going funny in the conversion
[10:47:44] <megal0maniac> Maybe the result is getting shifted one bit to the left? 638 is an even number, so it fits...
[10:49:18] <edmont> megal0maniac: good point
[10:54:14] <SoCo_cpp_> /join #android
[10:54:32] <r00t|home> android avr port?
[11:05:37] <edmont> megal0maniac: mmm, i get odd numbers too
[11:06:09] <megal0maniac> Then that rules that theory out. Pity. I was starting to feel clever and stuff
[11:06:21] <edmont> XD
[11:07:07] <megal0maniac> Where does one start with a USB stack for xmega? Is LUFA stable enough?
[11:07:13] <Essobi> http://www.repyoblog.com/index.php/2012/10/raspberry-pi-with-avr-328p-using-spi-and-gpio/
[11:07:24] <Essobi> I'm going to put the code up somewhere tonight.
[11:07:55] <OndraSter> megal0maniac, well, not as far I know :)
[11:08:03] <OndraSter> there were notes for it in the LUFA docs
[11:08:41] <Roklobsta> ugh have any of you guys had JTAG fail to connect with a Dragin and AVRS6?
[11:08:58] <OndraSter> sure
[11:09:00] <OndraSter> but it was my fault
[11:09:09] <OndraSter> what is the error?
[11:09:10] <OndraSter> and what chip?
[11:09:15] <Roklobsta> 644PA
[11:09:20] <OndraSter> hmm
[11:09:27] <OndraSter> I was using it on mega128, mega32 just fine
[11:09:45] <OndraSter> never had any "newer" mega other than 168
[11:09:58] <Roklobsta> oh i can't rem. first time I had to hold reset and get into the fuses. OCDEN had been disabled some how. Now Dragon wont' connect at all but I can still use ISPmkII
[11:10:21] <Roklobsta> not happy, jan
[11:10:33] <OndraSter> hmm
[11:10:35] <OndraSter> no idea
[11:10:38] <OndraSter> first connect via ISP
[11:10:41] <OndraSter> and check all the fuses
[11:10:53] <Roklobsta> and MCUZone haven't made a new firmware for my mkII-CN and AVR6SP1
[11:11:16] <OndraSter> as6sp1 has the same firmwares as as6
[11:11:33] <OndraSter> at least for dragon
[11:11:36] <OndraSter> 7.18
[11:12:12] <Roklobsta> hrm
[11:12:24] <Roklobsta> it complained on mkii-CN
[11:12:36] <OndraSter> isp or jtagice?
[11:12:42] <Roklobsta> jtagice
[11:12:44] <OndraSter> ah
[11:12:48] <Roklobsta> the nice chinese one
[11:13:02] <Roklobsta> with the blue led tha burn holes in walls it's so bright
[11:13:07] <OndraSter> haha
[11:13:10] <OndraSter> blue LEDs are awful, yes
[11:13:17] <OndraSter> that's why I am using red + yellow on xboard :P
[11:13:23] <Roklobsta> the project i have now uses one. HULK SMASH
[11:13:24] <OndraSter> I originally wanted green + red, but I ordered wrong one :D
[11:13:54] <Horologium> I like my RGA LEDs.
[11:13:59] <Roklobsta> i can't talk about the project but i shake my head at the designer
[11:14:18] <Horologium> Red +/-, Green -/+, Amber AC
[11:14:47] <megal0maniac> Blue is okay if it's diffused.
[11:15:03] <megal0maniac> OndraSter: I quite like the red and yellow
[11:15:35] <OndraSter> they are on PortR..
[11:15:44] <OndraSter> I thought about putting them to PortQ on Ultra
[11:15:51] <OndraSter> to allow the use of xtals on ultra
[11:16:14] <OndraSter> I would put on something like 4MHz xtal :P
[11:19:28] <megal0maniac> I was a little disappointed, but then I realised that I'd never use external xtal
[11:24:06] <Amadiro> I really like the clear blue LEDs, they look so sciency...
[11:27:13] <OndraSter> megal0maniac, 0.5% of internal 32kHz must be enough for you
[11:28:22] <megal0maniac> OndraSter: My point exactly :)
[11:28:31] <megal0maniac> I can probably do most of what I want at 1mhz anyway
[11:29:02] <OndraSter> HEH
[11:29:05] <OndraSter> caps
[11:31:05] <OndraSter> I suggest getting some SD card adaptor
[11:31:07] <OndraSter> it will be fun on xmega
[11:31:20] <OndraSter> and connect it to UART and change it to SPI mode
[11:31:29] <OndraSter> for DMA :P
[12:06:03] <tld> avrdude: verification error; content mismatch <-- meh. :(
[12:07:51] <tld> avrdude for "At some point, you did something stupid. Go back and check everything."
[12:08:55] <Amadiro> tld, that is indeed fairly nondescript
[12:26:45] <OndraSter> dlister, in a few hours :P
[12:26:54] <dfletcher> :)
[12:27:15] <dfletcher> davelister and dlister are taken nicks of course. now nickserv hates me.
[12:27:40] <dfletcher> it's cold outside, there's no kind of atmosphere...
[12:27:50] <OndraSter> first comes first served
[12:28:43] <KrytenSeries4k> I'll just be Kryten.
[12:33:39] <OndraSterver> You still get 30 seconds :)
[12:33:43] <OndraSterver> Although...
[12:34:34] <KrytenSeries4k> 30 seconds as dave lister... it's not enough!
[12:34:57] <acerimmer> ahh here we go :)
[12:35:18] <acerimmer> just don't call me ace-hole :P
[12:36:34] <tld> hmmz.
[12:36:43] <tld> replaced chip, and it works.
[12:37:50] <tld> code worked on first try as well… must be a first for me. ;)
[12:38:40] <tld> the display however, does not want to run off of 3.3V, but works fine with 5V.
[12:38:47] <tld> Oh well, not the end of the world.
[12:38:59] <acerimmer> for me "code works first try" means some insidious bug is hiding that won't actually appear until I try to show the device to someone :P
[12:39:31] <tld> acerimmer: Good point actually. "code works on first try" usually means "bug wasn't found first time around".
[12:39:38] <tld> this had all the complexity of "hello world" though. ;)
[12:39:47] <megal0maniac> http://pjrc.com/store/sd_adaptor.html
[12:40:02] <tld> (hello world using fresh code for talking to the LCD through shiftregs, but still… )
[12:40:29] <megal0maniac> OndraSter: Pity it's (only just) not wide enough to fit on xboard
[12:40:33] <tld> megal0maniac: cute. reminds me of openlog.
[12:41:07] <megal0maniac> tld: The usb connector looks HUGE
[12:42:10] <tld> megal0maniac: https://www.sparkfun.com/products/9530 <-- compare the microsd-sockets… openlog is perhaps larger, but not by much.
[12:42:22] <tld> megal0maniac: it comes with an atmega on it, so can actually do stuff on it's own.
[12:42:37] <tld> https://www.sparkfun.com/products/544 <-- for breadboard friendliness.
[12:43:38] <megal0maniac> Breakout, but no buffer :)
[12:50:24] <Horologium> I have one similar...just full SD
[12:50:27] <Horologium> no buffer
[12:51:57] <megal0maniac> Not entirely sure if you need it...
[12:52:13] <Horologium> if you run the microcontroller at 3.3V you don't
[12:52:26] <megal0maniac> Which I would be
[12:52:58] <Horologium> I also have a couple of 256MB SD cards that I have soldered pins to. They fit straight onto a breadboard nicely.
[12:53:53] <megal0maniac> OndraSter: I'm trying to figure out the principle of peripherals. From what I gather, a peripheral exists on a certain port. But you can assign it to any pins on that port through software.
[12:54:01] <megal0maniac> How far off am I?
[12:54:01] <tld> I didn't mean to imply buffer in the latest I linked.
[12:54:20] <megal0maniac> tld: I know. I was just pointing it out
[12:54:43] <megal0maniac> Openlog is cool, but expensive
[12:54:44] <tld> pins usually have different capabilities.
[12:54:57] <megal0maniac> tld: This is on an xmega, though
[12:55:01] <tld> I picked up an openlog, but I don't expect to get more.
[12:55:03] <Horologium> megal0maniac, are you asking for xmega?
[12:55:07] * megal0maniac nods
[12:55:38] <tld> I'm not too familiar with xmega yet. Figure I'll go down (attiny) before I go up (omega) or elsewhere (msp430, cortex)
[12:56:12] <Horologium> attiny and atmega are pretty much the same thing. attiny is just smaller and missing a couple of things the atmega has.
[12:56:16] <megal0maniac> Nor am I :P I have 4 tinys and I like them
[12:56:25] <megal0maniac> Like UART :/
[12:56:26] <Horologium> I haven't gotten to play with xmega yet.
[12:56:36] <megal0maniac> Horologium: Buy an xboard :)
[12:56:50] <Horologium> cash are tight.
[12:57:09] <megal0maniac> Save up for an xboard :)
[12:57:15] <Horologium> hope to do so for a self-christmas gift.
[12:57:35] <Horologium> closest I come to doing christmas.
[12:59:27] <Horologium> glub but these damned leg cramps are kicking my ass.
[13:12:00] <tld> Horologium: attiny is missing a few things, often a dollar of the price-tag. ;)
[13:13:19] <acerimmer> tiny's USI is upsetting me at the moment.
[13:13:26] <Horologium> hehe.
[13:13:37] <Horologium> has me many times too.
[13:16:28] <Horologium> had less problem making the attiny85 play USB with vUSB
[13:20:14] <acerimmer> well I'm also trying to do something slightly crazy. auto assign i2c addresses in a bus.
[13:20:36] <acerimmer> it's not my design. I'd probably just put a rotary switch or so.
[13:20:54] <acerimmer> but dude really wants plug and play
[13:20:57] <nomis> multiple identical devices on the same i2c?
[13:21:04] <acerimmer> yep
[13:21:21] <acerimmer> there is a trick to this that supposedly works. using broadcast mode and a GUID
[13:21:43] <nomis> fascinating. How do you want to avoid multiple unadressed devices accepting the same address?
[13:23:23] <acerimmer> master broadcasts "send me your GUIDs" because of self-arbitration nature of bus, only one response gets through. then master broadcasts "ok device with GUID [n], you have i2c address [a]". which it then self-assigns and further comms use that.
[13:23:26] <Horologium> need serial numbered devices.
[13:23:30] <tld> For one-off projects, what do you guys usually do after breadboarding it? So far, I've mostly been using "just a bunch of holes"-veroboard for semi-permanent stuff, but the wiring is a pain, and with heavier wire, it gets awfully bulky.
[13:23:43] <tld> (I know of lots of things I *can* do, just curious what people actually choose to do)
[13:23:58] <tld> ((home-made PCBs, order in ones, etc… ))
[13:24:08] <nomis> acerimmer: ah, ok. if they do have GUIDs already this seems plausible.
[13:24:16] <acerimmer> yep that's the key
[13:24:23] <Horologium> tld, depends on the project. Some I make boards for, some I use perfboard and point to point wiring with solid core cat-5 stripped out.
[13:24:42] <nomis> tld: there are nice services for hobbyists that give you professionally made PCBs for a reasonable price.
[13:24:53] <Horologium> nomis, just like DHCP IP addresses.
[13:25:12] <tld> solid cat-5 makes sense… I find 22awg too bulky, 30awg to skinny (pain to strip), and everything else hard to get (with the exception of 24awg, from ethernet)
[13:25:25] <nomis> Horologium: well, for ethernet you don't really have the same situation as a single i2c bus.
[13:25:37] <tld> nomis: yeah, I know of at least some of them.
[13:25:44] <nomis> Horologium: and there is no auto-negotiation of mac-addresses happening.
[13:26:03] <Horologium> which is why you need serial numbered devices or GUID or whatever you call it.
[13:26:45] <nomis> Horologium: yeah, but the comparison to dhcp breaks down there.
[13:26:49] <acerimmer> DHCP is a kindasorta okay analogy. at least with the whole idea of "auto assigned addresses" is concerned, if not hte details.
[13:27:45] <acerimmer> the key of course is that any number of random slaves can get plugged into the bus. we'll use flat flex cables.
[13:27:55] <Horologium> doing the same thing...devices get an address when they connect to the bus kinda thing.
[13:28:30] <acerimmer> heh that's funny Horologium. over i2c also?
[13:28:37] <acerimmer> we should share code.
[13:28:55] <Horologium> from what was asked earlier, that is what they want to do.
[13:29:03] <Horologium> normally i2c devices have a hard coded address.
[13:29:26] <acerimmer> oh I thought you said you were doing it as well. heh it's my (well, my friend's) project
[13:31:36] <Horologium> the way I would do it is to have the devices have a set address...and when connected to the bus they would respond to an occasional broadcast to that address. at that response the master would assign an address by sending a special command to that base address.
[13:31:45] <Horologium> now, this wouldn't work with multiple devices connected at once.
[13:32:08] <Horologium> it would require the addition of serial numbers at the device level.
[13:32:37] <Horologium> i2c would be harder than ethernet as it generally requires a master/slave setup.
[13:32:54] <Horologium> where ethernet sets up as multiple-master(effectively)
[13:33:22] <vectory> with ipv6 anyone can announce a random ip, needs to do the check, whether its taken already, on its own :)
[13:34:25] <Horologium> true.\
[13:34:38] <acerimmer> Horologium, http://www.microchip.com/forums/m524286-print.aspx
[13:34:44] <Horologium> with i2c you can't just connect to the bus and talk. it requires a master to tell you that you can talk.
[13:34:58] <acerimmer> the big yellow box that starts "The process can be fully automated..."
[13:35:02] <acerimmer> near the bottom
[13:35:26] <acerimmer> that 6 step plan is what I'm trying to implement
[13:35:35] <acerimmer> USI is fighting me a bit but I think it might be possible.
[13:35:56] <Horologium> ayup.
[13:36:02] <Horologium> similar to what I was saying.
[13:36:35] <Horologium> what tiny are you using?
[13:37:04] <acerimmer> 25s are the ones I have here for testing
[13:37:13] <acerimmer> pretty tight, may get bigger ones for production
[13:37:38] <Horologium> aahh. was thinking if you were using one of the bigger ones, jump up to an atmega88
[13:37:57] <Horologium> full usart, i2c, and spi implementation separately.
[13:38:03] <acerimmer> actually that doesn't work
[13:38:06] <acerimmer> tried it
[13:38:25] <Horologium> why not?
[13:38:40] <acerimmer> the trouble comes when replying to the broadcast. replying to broadcasts just doesn't work, period.
[13:38:46] <acerimmer> they all reply in lockstep
[13:38:55] <acerimmer> so the response is a wired or
[13:39:16] <Kevin`> tjqat
[13:39:17] <acerimmer> USI at least lets me get some control bit by bit, I can bail out if the line doesn't match the value
[13:39:29] <Horologium> aahh.
[13:39:37] <Horologium> have you tried software i2c implementation?
[13:39:42] <acerimmer> :(
[13:39:56] <acerimmer> I think if it's going to be software I'll do one wire or something
[13:40:15] <acerimmer> software i2c slave is a hard thing to write
[13:40:21] <Horologium> so steal it.
[13:40:29] <Horologium> procyon avrlib has one in it.
[13:40:50] <acerimmer> have to be able to sell these
[13:41:49] <Horologium> pascal stang is a good guy to work with...he is the author of avrlib
[13:42:26] <Horologium> so, let's see what else we can come up with.
[13:42:52] <Horologium> guessing this needs to be a small implementation, physically and software wise, yes?
[13:47:46] <Horologium> is there a problem with having unique IDs on the devices?
[13:48:14] <acerimmer> no I just made that part of the `make program` command. it stamps each one.
[13:48:41] <acerimmer> anyway, I nearly got this I think
[13:48:52] <Horologium> then why not do that as the i2c bus address?
[13:49:06] <acerimmer> just need to spend another few hours or so and haven't had time yet
[13:49:20] <acerimmer> because a GUID is ... big
[13:49:20] <Horologium> oh, wait, forget I said that.
[13:49:27] <acerimmer> :P
[13:49:36] <Horologium> limited addresses on one bus.
[13:49:40] <acerimmer> it would collide 1 in 255 times
[13:50:26] <acerimmer> so several bytes are transfered during that first stage
[13:50:40] <acerimmer> actually I think the libuuid I used is like 16 bytes or something
[13:51:25] <Horologium> to make your arbitration a little easier if you have multiple devices on the bus, base a delay on the GUID before it responds.
[13:51:51] <acerimmer> yeah but that also collides 1 in 255 times or more
[13:52:04] <acerimmer> so it just pushes the problem down the rabbit hole a bit :P
[13:53:02] <acerimmer> actually yeah I did try something similar. can't remember if that was with the atmega88 or the tinys
[13:53:10] <acerimmer> wasn't very reliable
[13:56:26] <Horologium> have you thought of SPI with addresses?
[13:56:47] <acerimmer> what about all the CS lines?
[13:57:00] <Horologium> forget those.
[13:57:09] <Horologium> use addresses.
[13:57:32] <acerimmer> check out "Daisy chain SPI configuration" on http://en.wikipedia.org/wiki/Serial_Peripheral_Interface_Bus
[13:57:38] <acerimmer> I have thought about that one
[13:57:47] <acerimmer> but... ugh :/
[13:58:31] <acerimmer> just looking at the wiring diagram gives me a headache. and it only has 3 components.
[13:59:29] <Horologium> one master device, bunch of slave devices, on spi bus....no daisychaining, just connected to the comms pins.
[13:59:45] <Horologium> data is sent on the bus from the master.
[14:00:08] <acerimmer> oh and just tie SS low?
[14:00:24] <acerimmer> like broadcasting and then slaves drop off when the address doesn't match?
[14:00:30] <Horologium> by sending a command with an address attached, your GUID, then only the device with that GUID need respond and the others stay quiet.
[14:00:43] <acerimmer> needs start and stop conditions
[14:01:06] <acerimmer> but yeah maybe
[14:02:10] <acerimmer> I donno I still think USI can be forced to work. there's a mode that does an interrupt for every bit out. so I test the wire against the output value and if they don't match this slave lost arbitration and bails until next start condition.
[14:02:32] <acerimmer> but to be quick enough I think this needs to be ASM
[14:02:45] <acerimmer> so I've been putting it off. I'll try again this weekend.
[14:02:46] <Horologium> the slaves don't do anything to the bus until they are commanded to send back data on SPI as I recall. So, if they do not see their own ID in the transmission then they just don't respond.
[14:02:57] <acerimmer> I understand that
[14:03:08] <acerimmer> so how does a slave know when to start listening again?
[14:03:12] <Horologium> looking at the protocol now to see if that would work.
[14:06:39] <Kevin`> it's not normal for the data pins on spi to be floating
[14:07:01] <Kevin`> you COULD do it though. wouldn't be standard spi behavior
[14:07:41] <Horologium> have the non-active devices watch all 3 pins for a stable position...all 3 low or all 3 high for a certain amount of time...when that happens they all go into listen-for-id mode again.
[14:08:09] <acerimmer> yikes you've completely gone off the rails now Horologium ;)
[14:08:16] <Horologium> probably.
[14:09:18] <Horologium> I know I've seen this done somewhere.
[14:09:22] <Horologium> just trying to find it.
[14:13:03] <Horologium> DUH...nevermind....the pain meds are making my brain mushy.
[14:13:31] <Horologium> it was rs232 9-bit bus I was thinking of that did command/data with addresses.
[14:14:42] <Horologium> and the USI can do 9-bit
[14:15:56] <Horologium> acerimmer, are you getting the USI to talk i2c?
[14:16:38] <acerimmer> yes. as I said, that 6 part plan on http://www.microchip.com/forums/m524286-print.aspx is what I'm implementing. or trying to.
[14:16:58] <acerimmer> so it already has start/stop conditions, addressing, etc
[14:17:04] <OndraSter> megal0maniac_afk, well
[14:17:04] <OndraSter> no
[14:17:08] <OndraSter> peripherals have preset pins
[14:17:18] <OndraSter> the fact that almost all ports have the same peripherals on all the ports is other thing though :)
[14:17:33] <acerimmer> Horologium, the main thing that's missing is as I said the trouble with atmega88 is - replies to broadcasts aren't exactly "normal"
[14:17:44] <Horologium> yeah.
[14:17:50] <acerimmer> but the USI seems flexible enough that I might get away with it
[14:18:34] <Horologium> my brain keeps going back to the software implementation. But I am rather sick and twisted that way.
[14:20:35] <OndraSter> I have twisted mind
[14:20:41] <OndraSter> which is twisting my body
[14:20:54] <OndraSter> (hands, while connecting some circuit)
[14:20:56] <OndraSter> ((for pervs)
[15:32:27] <Horologium> watching Arena
[15:32:32] <Horologium> this movie is seriously screwed up.
[15:43:15] <Amadiro> why's that
[15:43:56] <Horologium> have you seen it?
[15:44:24] <Amadiro> I've never even heard of it
[15:44:34] <Horologium> me either till I found it online a couple weeks ago.
[15:44:45] <Horologium> just now watching it.
[15:44:51] <Horologium> came out last year.
[15:44:53] <Amadiro> the title cover looks incredibly cheesy
[15:45:05] <Horologium> it is a B+ grade movie at best.
[15:45:12] <Horologium> acting is mediocre.
[15:45:18] <Horologium> story line is,,,totally fucked up.
[15:45:54] <Horologium> this group kidnaps people and puts them up on video online to fight to the death.
[15:45:55] <Amadiro> plot sounds pretty generic
[15:46:02] <Amadiro> Just read the wiki article
[15:46:15] <Amadiro> I've seen at least 3 other movies that have that exact plot
[15:46:42] <Horologium> I got it because of Samuel L. Jackson as the main badguy.
[15:48:01] <Amadiro> he is a pretty badass dude, yeah
[15:50:59] <Horologium> and so far he doesn't even fight or do anything but play mastermind.\
[20:52:38] <Essobi> How big is the SPI buffer on an AVR?
[21:05:08] <damjan> Essobi: 1 byte I think?
[21:16:13] <Tom_itx> 1 byte
[21:16:43] <Tom_itx> i've used some chips that let you set the length
[21:37:37] <Sur3> Yay my AVR 7-Segment Thermometer is working ^^
[21:37:43] <CapnKernel> yay
[21:38:01] <Sur3> but i seem to often need recalibrating it
[21:39:15] <Sur3> anyway does someone here know about transistors, is it a difference either i connect the payload to the emitter or the kollektor?
[21:41:34] <Sur3> i think the only difference is that the base-current also powers the payload when i connect it to the emitter instead of the collector, is this right?
[21:42:51] <CapnKernel> One can't generalise without knowing whether you're talking about an NPN transistor, or a PNP transistor
[21:47:03] <Sur3> ah ok, so its base current on emitter for npn, and on collector for pnp?
[22:01:31] <CapnKernel> Sur3: Could be good to read the wikipedia article
[22:01:37] <Casper> Sur3: if you adapt the base drive circuit, then it do not make a big difference, however in some config it can be hard to impossible to get the lowest possible drop
[22:05:31] <Sur3> because i already burned two npn-IGBTs -.-
[22:06:47] <theBear> say npn, if you put the load on the emitter, the voltage the base needs to achieve X amount of on-ness/current varies, in some circuit (especially linear) this makes things 'trickier' than using the collector side for the load
[22:07:31] <CapnKernel> the important thing is not the voltage, but the Vbe current
[22:07:34] <Sur3> ahh thx theBear :)
[22:07:50] <CapnKernel> Vce current is proportional to Vbe current
[22:08:07] <CapnKernel> Silly me mixing up symbols
[22:08:11] <theBear> CapnKernel, i didn't mean to say the voltage is important, just that instead of maybe .8v for saturation, you need the load voltage PLUS that
[22:08:14] <CapnKernel> Ice proportional to Ibe
[22:08:26] <theBear> and of course the load voltage is proportional to current and source voltage :)
[22:08:40] <CapnKernel> Yes. The voltage drop across the load can make it more difficult to get a suitable Ibe current
[22:08:56] <Sur3> also i wonder if i want to make an h-bridge i need to connect the payload both sides, so would it be better to use 2 npn and 2 pnp for the h-bridge?
[22:09:01] <CapnKernel> Sur3: There may be circuits where you actually want that. It's not really possible to generalise
[22:09:09] <theBear> it's true
[22:09:13] <CapnKernel> Sur3: If you want to make an H-bridge, better to use a chip or mosfets. Really.
[22:09:37] <theBear> for reference hanging off an npn emitter is a 'voltage follower' of sorts, for the very reasons i just described, and it wouldn't have a name if it wasn't useful :)
[22:09:47] <CapnKernel> If you have four switching elements, then yes, you want two of each complementary type
[22:09:47] <Richard_Cavell> Good afternnon everyone!
[22:10:01] <theBear> goodmoaning :)
[22:10:03] <Sur3> but i want to make an igbt-h-bridge because of the higher frequenzy
[22:10:26] <theBear> err, what ? igbts are more about stupid high voltage/current than freq
[22:10:48] <theBear> like 100a@600v kinda stupid
[22:11:39] <Sur3> not stupid if you plan on controlling a tesla coil ;)
[22:12:23] <CapnKernel> Sur3: If you're not that travelled with transistors, but you want to do fancy stuff, might be worth looking at/getting someone else's design
[22:12:27] <theBear> maybe, just remember you're using them for voltage/current ratings, not 'cos they have some amazing hi freq performance other devices don't have
[22:13:32] <Sur3> hm, well i need both cause if i dont get about 100kHz then the PWM-Sound would be bad ^^
[22:14:02] <theBear> lol, be ambitious much on first project ?
[22:14:14] <Sur3> jep xD
[22:14:15] <theBear> and 100khz isn't hi freq
[22:14:27] <theBear> it's 'barely' past audio :)
[22:14:41] <Sur3> ^^
[22:15:54] <Sur3> but i realize the prob now, i used 4 npn so i need to get two pnp to complete the bridge ^^
[22:17:20] <theBear> that alone should blow things up, unless you were turning high and low on at the same time on the same side of the bridge, that'd do it... and what you got for a drive circuit ? probably worth having something to be SURE you never got both high and low in a pair blowing each other up
[22:18:26] <Sur3> i just connected the bases to my atmega pins for switching them
[22:19:10] <theBear> ummmmmm, i'm guessing you using more than 5volts for high side... you gonna want some kinda high side driver
[22:19:31] <theBear> and you wanna be SURE that the software NEVER turns them both on at once, including during turn-on, which MAY not be possible like that, not sure
[22:20:24] <Sur3> so i probably set a short delay between on states where both are off
[22:21:33] <Sur3> high side driver? you mean 5V Vbe is not enaugh?
[22:21:36] <theBear> erm, probably, with switching capacitances and stuff...
[22:22:09] <theBear> how do you achieve 5vbe when a igbt is connected to 12v and a load ?
[22:22:32] <theBear> and it will also depend a lot on the devices, it might not be enough
[22:23:34] <Sur3> connected to 12V? where did you get this voltage?
[22:25:28] <Sur3> so u mean i need to put a transistor before the igbt, like a darlington?
[22:26:04] <CapnKernel> A little knowledge is a dangerous thing.
[22:31:10] <Sur3> well my knowledge is not that little, i just dont understand semiconductors ;)
[22:32:08] <theBear> where you getting 5v from ? you know a even a small ss tesla coil needs some amount of current/power right ?
[22:32:20] <theBear> more than a modicum, as it were
[22:32:41] <theBear> and that switching a big fat primary like that off a 5v rail will freakout your micro
[22:33:41] <Sur3> ah ok but i thought IGBTs have insulated gates to not burn the controlling micro?
[22:34:15] <Sur3> but i could put an optocoupler between
[22:36:31] <Sur3> probably i should start with an on/off coil with just one igbt, seems much easier than the h-bridge :-/
[22:55:35] <theBear> erm, i'm not real sure what kinda gates they have, but it's not about the micro, it's about 'turning on' the igbt
[22:56:32] <theBear> or off... i mean, if you use a pnp for high side, you need almost high v+ to turn it off, and anywhere low to turn it on, but the same thing n on low side, you need nothing to turn it off, of 5v to turn it on (assuming 5v is enough for that device)
[22:57:31] <theBear> also i'm no expert on ss tesla design, but i suspect it would be much more 'tesla-ish' if you just yoink the coil with a single device ignition coil style, or even a single device on either side for 'bridged' ignition coil driving
[22:59:36] <Sur3> ah so u mean gnd is not enaugh and i need to convert +5V to -5V for switching? i thought that gnd is like -5V if if use the 5V as new GND...
[23:04:49] <theBear> what ? no !
[23:05:07] <Sur3> well tested just on off switching a led with IGBT over atmega at base directly workes as it seems
[23:05:18] <theBear> i mean that usually you would be using something higher than 5v for your 'power' rail/supply, and as a side note, i'd be surprised if igbts work at 5v
[23:05:43] <theBear> how many volts do you see from igbt to the opposite rail ?
[23:06:33] <Sur3> i dont measured, just seeing the led blink in atmega pin frequency
[23:07:28] <Sur3> used 5V on both sides in this setup, next step is testing to switch higher Vec..
[23:07:53] <theBear> can you measure tho ?
[23:08:09] <theBear> oh, so you DO have higher vcc you are using ?
[23:09:39] <Sur3> well tested to switch 230V rectified a few days ago but that burned my IGBTs
[23:10:13] <Sur3> but normally they should be ably to switch it, so it was my wrong setup
[23:10:27] <theBear> oh dear, i really worry that you are going to kill yourself
[23:11:44] <Sur3> well i touched 230V years ago, that alone wont kill me that fast even thoug i dont plan on touching high end.. ;)
[23:13:43] <Sur3> could the just rectified but not smoothed voltage also be a problem for the IGBT, so do i need a Condensator for Smoothing high end Voltage and could that be the reason the IGBTs burned?
[23:15:32] <theBear> if they are 400v igbts, 230vac rectified is close to their limit, caps would make the circuit work better, but shouldn't hurt the igbts without, but you probably did something silly like turn both igbts on
[23:16:15] <Sur3> no its 230vdc but going up and down
[23:16:21] <CapnKernel> Sur3: Really dude, start simple and work up.
[23:17:09] <Sur3> anyway they are states 600V 40 A
[23:19:59] <Sur3> ahh ok sorry i read you wrong ^^
[23:20:26] <Sur3> why would be 230Vac rectified be at the limit of 400V IGBTs?
[23:22:58] <theBear> 230 * 1.41, if you do a halfwave doubler it's WAY over 400v
[23:22:59] <creep> h
[23:23:18] <Sur3> ahh k, right thank, forgot that ^^
[23:23:29] <creep> 400V IGBT?
[23:23:36] <creep> i see 600V ones all over
[23:23:42] <Sur3> 600V
[23:23:47] <theBear> i regularly see 400 or 600v i'm sure
[23:23:59] <creep> 400 are mosfets
[23:24:13] <theBear> you sure ? even 1st gen igbts ?
[23:24:22] <Sur3> thought about buying a 1200V one at ebay but want to get the 600V one work first
[23:24:34] <creep> Sur3<< why do you want 1200V igbt?
[23:24:49] <theBear> he's going for the BIG darwin award <grin>
[23:25:01] <Sur3> because they will not get that hot
[23:25:08] <creep> lold.
[23:25:15] <creep> ok good playing
[23:25:24] <theBear> and heat err, isn't a reason
[23:25:33] <creep> (they will get hotter)
[23:25:36] <CapnKernel> A little knowledge is a dangerous thing.
[23:26:05] <theBear> and i got a lot, just imagine how dangerous i am <grin>
[23:26:17] <creep> wondering what does this have to do with avrs
[23:26:22] <Sur3> no i thougth 1200V one would be more robust and not getting that hot when i drive them with about 600V things
[23:27:07] <Sur3> @creep: i control them by an ATmega
[23:27:21] <theBear> it may be more robust, but that's no reason to be less hot
[23:27:41] <theBear> the heat is mainly generated (assuming relatively fast switching) by the series drop * current
[23:27:50] <creep> higher boltage device has larger VCE_star btw
[23:28:01] <creep> and they are slower
[23:28:20] <creep> but they will be definitely colder...
[23:28:23] <Sur3> ah ok
[23:29:06] <CapnKernel> Watch out for icicles.
[23:29:08] <creep> Sur3<< so how you drive the IGBT ?
[23:29:38] <creep> 1k ohm 0805 resistor?
[23:29:46] <creep> and 5V acr pin?
[23:29:50] <creep> *avr
[23:30:17] <Sur3> @creep i directly connected the base to an avr-pin
[23:30:25] <creep> oh uh
[23:30:51] <creep> you have quite a bit of money to throw out
[23:31:26] <Sur3> s/base/gate/
[23:32:58] <Sur3> @creep, yep already burned 2 igbt
[23:33:23] <creep> sauare you a programmer? or jst hobby tinkering?
[23:33:32] <creep> *Sur3
[23:34:16] <Sur3> programming is my hobby
[23:36:17] <creep> Sur3<< what is the project?
[23:36:30] <creep> 400V assuming 230VAC something
[23:39:15] <Sur3> hm my led is not realy bright, probably i need to remove the resistor between avr and igbt-gate... *thinking*
[23:40:31] <Sur3> nope, seems the IGBT itself eats most of the 5V Vec
[23:41:02] <theBear> yep, and you wanna keep a resistor, they're good on gats
[23:41:04] <theBear> gates
[23:41:39] <Sur3> yep read that thats why i put it there ^^
[23:42:43] <creep> 400V leds ahhhhh
[23:42:59] <Sur3> anyway i should use a cascade two double the gate-voltage
[23:43:10] <Sur3> s/two/to
[23:43:21] <creep> 5V VCE?
[23:43:35] <creep> an IGBT has 2.3-5V VCE_sst
[23:43:59] <creep> why don't you get some smd mosfets ?
[23:44:01] <Sur3> yep right now, because the 230V burned two IGBTs so i now first test with 5Vce
[23:44:30] <creep> those will switch nicely 3-20V things at 8A
[23:44:57] <Sur3> @creep: i want them for a tesla coil ;)
[23:45:33] <Sur3> playing sounds over pwm ^^
[23:45:35] <creep> resonant mode?
[23:45:58] <creep> without basics you will not be able to make it
[23:46:16] <creep> it will just thrash all your IGBTs
[23:47:35] <Sur3> well thats why i asked for help with the IGBTs ;)
[23:47:56] <Sur3> i think i already made some progress ^^
[23:48:04] <creep> but after you are able to drive an IGBT it will blow up because of incorrect drive
[23:48:56] <creep> what control method would you use for the resonant tank to be able to modulate with audio?
[23:49:23] <Sur3> well, i want to use the ATmega pwm pins
[23:49:34] <creep> (PWM will not do)
[23:50:24] <creep> or you are not building an SST, just some high voltage noise maker
[23:51:13] <Sur3> huh? sure pwm is the way to go, when putting the 'noise' on and off with i frequency pwm i get a sound
[23:52:03] <creep> you don't have resonance, cause large radio interference, and it is dangerous because of the high voltage and power level
[23:52:42] <creep> not really a beginner project
[23:53:07] <CapnKernel> creep: He doesn't want to hear that.
[23:53:56] <Sur3> u think i still need a resonant circuit? the main coil itself is an inductance and i just need to get a collector in row i think
[23:54:46] <Sur3> s/collector/condensator/
[23:57:17] <creep> Sur3<< no you don't need a resonant circuit