#avr | Logs for 2012-04-09

Back
[01:19:20] <skorket> dammit, wanna play around with USB but don't have a 12MHz clock or any zener diodes
[01:20:12] <Casper> zener might be optional
[01:20:16] <Casper> but 12MHz is not
[01:20:47] <j4cbo> is for low-speed
[01:20:55] <skorket> I have a 20MHz crystal but the atmega328 won't go down to 3.3V at that speed
[01:27:54] <Casper> try and see if it work
[01:28:04] <Casper> it can work even if atmel say it do not
[01:28:20] <Casper> beside, some usb controller have no problem with 5V
[02:57:20] <JaggedNZ> Hi, Having an extremely fustrating time with my AVR programmer. It seems it is not reading/writing consistantly. For example it's 50/50 if I get the correct signature when I read it.
[03:00:37] <OndraSter> wrong cables?
[03:00:37] <specing> explain
[03:01:11] <JaggedNZ> Sorry, my IRC client has gone spastic BRB
[03:04:55] <JaggedNZ> Thats better :)
[03:05:55] <JaggedNZ> Anyway, I am haveing issues reading/writing to an ATMEGA8 and just wanted to check im not doing something stupid
[03:06:28] <specing> probably cables
[03:07:33] <JaggedNZ> I have MISO,MOSI,SCK,RST and VCC & GND all connected to the programmer and triple checked, a 10k Resistor between RESET and VCC and a 22uF Cap between VCC and GND
[03:08:40] <JaggedNZ> specing I'm starting to suspect that the laptop has less than realtime access on the USB port. I didn't think that was an issue with USB, but I'm pretty stumped.
[03:09:11] <specing> JaggedNZ: you shouldn't have any additional resistors on RESET line
[03:09:18] <specing> try removing it
[03:09:29] <JaggedNZ> I added it after having issues
[03:10:01] <JaggedNZ> didn't make things worse.
[03:11:43] <specing> How long are those cables?
[03:12:28] <specing> And what programmer is ihis?
[03:16:26] <JaggedNZ> about 60cm cable as supplied with programmer. Its a ZF-007 "PROGISP" chinese thing. Ordering a different one of ebay tonight, but 2weeks+ shipping time
[03:17:42] <OndraSter> 60 is a lot
[03:17:50] <Casper> maybe try to slow down the programmer
[03:18:07] <JaggedNZ> don't know if I can slow it down ...
[03:18:08] <Casper> some programmer default to a stupidly high speed, that have issue with stock clock
[03:18:22] <Casper> with avrdude I think it's the -B option
[03:18:41] <JaggedNZ> hmmm I will have to dig around and see if I can find it. Not supported by AVRDUDE >:(
[03:19:05] <Casper> o.O
[03:19:15] <Casper> a programmer not supported by avrdude??? must be really shitty
[03:19:15] <JaggedNZ> not supported by anything I can find except the manufacturers program
[03:19:27] <OndraSter> duh
[03:19:44] <Casper> don't search much...
[03:19:45] <OndraSter> let me make an ad: buy Tom_itx 's clone of AVR ISP 2 :)
[03:21:06] <OndraSter> brb
[03:21:12] <JaggedNZ> Casper, thats the pain, it's painfully similar to USBASP but has different pinout
[03:21:51] <Casper> and usbasp is known to be unreliable
[03:22:14] <JaggedNZ> heh, at least it's supported by avrdude
[03:23:23] <JaggedNZ> Anyway, I'll see if I can slow the thing down.
[03:30:15] <Metalsutton> Hello all. I have a quick question:
[03:31:41] <Metalsutton> Is there any guides on the net (i have tried looking previously), on how to implement a general button input either switch or push button? Also, if I wanted to do things like launch a certain program with a button, can this be handled in the MCU code, or does it need to be handled in the custom firmware?
[03:37:50] <SianaGearz> the question somehow doesn't make much sense. MCUs typically are harvard and only execute code from ROM, i.e. there is just one program which is the firmware.
[03:39:10] <OndraSter> if you are smart, you can implement multiple programs into one firmware and "switch" them on demand :)
[03:39:33] <SianaGearz> well they are routines, or sections, but it's still arguably one program.
[03:39:39] <OndraSter> well... yes
[03:40:36] <SianaGearz> (in particular you will usually want to have two sections, the loader which is something you can overwrite the other part of firmware conveniently with, and the rest which provides useful functionality)
[03:41:32] <JaggedNZ> THANK YOU !!! It was the $%^$%^$ cable ....
[03:43:22] <JaggedNZ> plugged the progrmmer directly into the ISP header and everything started working.
[03:43:41] <JaggedNZ> off to test my stepper motor driver!
[03:45:14] <JaggedNZ> BTW has anyone ever incountered a chip where "pin1" was flipped on the datasheet?
[03:45:20] <OndraSter> flipped?
[03:45:28] <JaggedNZ> mirror image?
[03:45:49] <OndraSter> flipped by which axis?
[03:45:49] <OndraSter> Y?
[03:45:59] <JaggedNZ> ie all the pins that should be on one side are on the other
[03:46:41] <JaggedNZ> It had me scartching my head until I reverse engineered the board I pulled the chips from
[03:46:50] <OndraSter> huh
[03:47:10] <specing> I think we are on #avr-weirdness today
[03:48:05] <JaggedNZ> The datasheet is for a slightly different package, I can only imagine that the other package mounts "upside down" or something weird like that
[03:49:47] <JaggedNZ> anyway, if anyone wants to salvage stepper drivers from old Epson printers, let me know. LOL
[03:50:09] <Tom_itx> i'm sure rue has done that many times
[03:56:41] <JaggedNZ> There's only a few documented cases I can find on the net, and mostly in spanish :S
[03:57:10] <JaggedNZ> Time to order some linear bearings >.>
[04:06:39] <OndraSter> what's it gonna become? :P
[04:07:17] <Tom_itx> a new and improved stepper driver
[04:07:39] <Tom_itx> i'm betting on reprap or cnc
[04:09:29] <OndraSter> I would build myself CNC too :)
[04:09:38] <OndraSter> but it requires some tools and skills with them :P
[04:10:02] <OndraSter> I am sure I could salvage few hundreds of steppers and such from printers and scanners I have around the house and not using them :D
[04:10:41] <Tom_itx> may as well
[04:10:50] <JaggedNZ> Yeah, CNC, but might build an extruder later on
[04:11:13] <Tom_itx> what size steppers?
[04:11:36] <JaggedNZ> just using the large ones from Epson printers to start with
[04:12:03] <Tom_itx> i'm looking at going to a 45v power supply on mine
[04:12:13] <Tom_itx> ~18A
[04:12:16] <JaggedNZ> it's going to be a pretty small/ light duty machine so hopefully they will do
[04:12:31] <Tom_itx> you will have to go quite slow with it
[04:13:01] <JaggedNZ> hmm thats one thing I like with the Epson stepper drivers (apart from the price!) ~42V @ ~2aA
[04:13:33] <JaggedNZ> aint quite ~18A though :)
[04:14:01] <Tom_itx> i have most of the parts but haven't put them together yet
[04:14:11] <JaggedNZ> Tom_itx: what kind of cutting area are you going for?
[04:15:02] <Tom_itx> http://tom-itx.dyndns.org:81/~webpage/temp/cnc/psu1.jpg
[04:15:08] <Tom_itx> i've got 3 of those now
[04:15:11] <Tom_itx> one for each axis
[04:15:19] <Tom_itx> i'm beefing up my sherline a bit
[04:15:34] <JaggedNZ> holy $**t!
[04:15:44] <Tom_itx> and getting stuff that will work on a larger mill
[04:16:37] <Tom_itx> http://tom-itx.dyndns.org:81/~webpage/boards/USBTiny_Mkii/Boxes/milling1.jpg
[04:17:09] <Tom_itx> i've replaced the steppers in that pic with a double stack
[04:17:39] <JaggedNZ> would love one of those ... they go for way to much here.
[04:17:40] <Tom_itx> wish i could afford servos
[04:17:57] <Tom_itx> err justify them
[04:19:15] <Metalsutton> Sorry, just going back to the question I asked an hour or so again. I meant program as in a windows application. Not a MCU code program.
[04:19:49] <Tom_itx> there goes the neighborhood
[04:20:57] <JaggedNZ> Metalsutton, you want to push a button connected to an AVR and have it start a program?
[04:27:42] <JaggedNZ> I guess you scared him off?
[04:51:19] <Metalsutton> im here.
[04:51:24] <Metalsutton> That is correct.
[04:54:07] <OndraSter> I am here as well
[04:54:39] <OndraSter> you need just some kind of simple link between your AVR MCU and your PC
[04:54:46] <OndraSter> ergo connect it via some serial cable for example
[04:55:08] <OndraSter> and then some simple monitoring app that will launch the app on PC when you flip the switch on the device
[05:01:20] <JaggedNZ> Metalsutton, it might be easier to implement a USB HID device (like a keyboard) then sort it out in the OS
[05:02:49] <OndraSter> .. or that
[05:09:03] <Metalsutton> I have made the device. but i suppose its just a matter of adding code into a free I/O
[05:09:45] <Metalsutton> The device is made, however I would just like some added functionality. So instead of having to launch the program, this is handled on the device etc.
[05:09:56] <Metalsutton> and its USB.
[05:10:39] <Metalsutton> Kind of like a Soft ON/OFF as you said, monitoring program. So it also force closes on OFF. is something like this possible?
[05:14:13] <JaggedNZ> if it's already using USB I would look at adding it to that ... are you writing the USB driver too?
[05:17:31] <Metalsutton> negitive. However I will do what it takes. So if the USB driver needs to be one that isnt the norm HID ... then yes.
[05:17:54] <Metalsutton> If I cannot do it (chances are I can't) I will find someone that can.
[05:18:06] <Metalsutton> What part of NZ do you hail from?
[08:51:02] <skorket> Sorry for the vague question but is there any standard way to communicate with a microcontroller through USB? If I were to make a data logger, sensor, or any other device that I would like to communicate with, is there a 'best practice'? i.e. using v-usb and communication through a serial virtual port?
[08:52:58] <ziph> skorket: Using V-USB would come under worst practice.
[08:53:17] <skorket> xiph, oh? Why is that?
[08:53:25] <ziph> skorket: Get an AVR with an in-built USB controller and use LUFA http://www.fourwalledcubicle.com/LUFA.php
[08:53:58] <ziph> skorket: Because it fails to actually implement large parts of the USB standard and requires huge hacks both on the AVR and on Windows.
[08:53:59] <ureif> what does LUFA do over and above hardware USB ?
[08:54:03] <ziph> e.g. driver patching.
[08:54:08] <Steffanx> "Some people are not in the position to get an usb avr" ziph :P
[08:54:18] <ziph> LUFA is a driver to talk to the hardware USB.
[08:54:33] <ziph> Plus a huge set of examples and class drivers to implement things like serial out of the box.
[08:54:55] <ziph> Steffanx: Some people can't afford any electronics at all, and I think they're wasting their time trying to do it in that case.
[08:55:27] <Steffanx> ziph, that's just the standard response i get when i say people should get a avr with a 'real' usb controller
[08:55:40] <Steffanx> *an
[08:55:43] <ziph> Yeah, they're wasting their time.
[08:56:54] <RikusW> usb avrs aren't even expensive....
[08:57:08] <RikusW> and with LUFA the USB complexity is hidden :)
[08:58:01] <skorket> In my mind, if I've got the space on the chip and pins available, why not do it in software? But I think that's a seperate discussion.
[08:58:41] <skorket> So, whether I use LUFA or V-USB, is there a 'best practice' to do this sort of low bandwith communication? Treat the device as a serial device even though it's connected via USB?
[08:58:55] <Steffanx> I don't think it's about space or pins, it's more about speed and to have the time to do something else then usb only
[08:58:56] <Steffanx> :)
[08:59:33] <ziph> skorket: V-USB requires you to patch/hack the Windows serial class drivers to work.
[08:59:46] <ziph> skorket: And only works with specific versions of operating systems.
[09:00:01] <ziph> skorket: And you can't use your AVR like you would a normal IC.
[09:00:50] <RikusW> v-usb takes to many cycles for that :S
[09:01:04] <RikusW> I thought about vusb but got atmega32u2 instead
[09:01:13] <RikusW> atmega32u4 is also nice
[09:01:14] <skorket> ziph, I'm sold, LUFA sounds good, but I still would like to do a USB device that I can communicate with in some command-like interface way
[09:01:37] <skorket> ziph, and what do you mean 'you can't use your AVR like you would a normal IC'?
[09:01:39] <ziph> skorket: Using the serial class drivers is good for that then.
[09:01:51] <RikusW> LUFA does have CDC (virtual serial port)
[09:01:51] <ziph> skorket: You can't use interrupts for starters.
[09:02:17] <RikusW> skorket: vusb use almost all the avr processing power
[09:02:25] <skorket> ziph, I'm getting confused. I can't use interrupts on AVR's in general or with LUFA?
[09:02:34] <ziph> skorket: If you want to have software talk to your device though (instead of something like Hyperterm) you should consider using WinUSB/libusb instead.
[09:02:55] <ziph> skorket: With V-USB you generally can't enable any interrupts from what I understand.
[09:03:43] <ziph> skorket: LUFA/Hardware USB lets you use the AVR more or less normally except for occasionally having to call a LUFA function to check on the controller state.
[09:04:13] <skorket> ziph, first off I'm running Linux. Second off, are you saying that V-USB is too processor intensive so I can't use interrupts at all? Or that _all_ AVR software USB libraries, including LUFA, is too software intensive and can't use intterups in addition?
[09:04:23] <skorket> ah, ok, you're just further selling me on LUFA. I'm sold
[09:04:38] <skorket> I really don't care what I use, as long as it does what I want
[09:04:47] <ziph> skorket: It's both processor intensive and it probably can't have you disable interrupts at any time (and having another interrupt fire will disable interrupts).
[09:05:27] <ziph> skorket: Also, the LUFA author lurks in here. :)
[09:05:29] <RikusW> skorket: there is Atmel official CDC demo's too
[09:05:43] <ziph> skorket: And works for Atmel now.
[09:06:06] <Steffanx> Poor him indeed
[09:06:12] <RikusW> skorket: trust me, writing usb drivers yourself is time consuming and complicated....
[09:06:26] <skorket> RikusW, absolutely, I do not want to be in that position
[09:06:38] <ziph> And nearly impossible without a hardware USB Analyser.
[09:07:10] <skorket> but I'm still a little confused. Are you guys saying I should use another piece of dedicated hardware for USB? Or can I implement something that just uses an atmega with LUFA and have it work?
[09:07:16] <RikusW> I managed it with only a dragon and the atmel demo
[09:07:22] <RikusW> and some usb specs
[09:07:32] <RikusW> skorket: you need a usb avr
[09:07:41] <RikusW> like atmega32u2 or 32u4
[09:07:53] <RikusW> or at90usb1287
[09:07:56] <ziph> RikusW: Using the demo is cheating.
[09:08:02] <ziph> RikusW: ;)
[09:08:20] <RikusW> ziph: without cheating it would be nearly impossible ;)
[09:08:23] <ziph> And the vendor code is usually jaw-dropingly bad.
[09:08:37] <ziph> RikusW: I've done it on an LPC1xxx.
[09:08:38] <RikusW> ziph: I decreased usb cdc code size from 6kb to 800 bytes
[09:08:50] <skorket> ah, so you guys are saying that doing anything else aside from USB on an atmega328 (say) is nearly impossible?
[09:09:17] <RikusW> skorket: you'll have very little processing power left
[09:09:18] <ziph> skorket: It isn't sane to try to do it on an atmega328.
[09:09:39] <skorket> crap
[09:09:41] <ziph> skorket: You can also use something like an FTDI IC if you want to add USB serial to an atmega328.
[09:10:00] <skorket> yeah, I was hoping to do it in software precisely to avoid FTDI
[09:10:02] <RikusW> unfortunately all usb avr's is only available as TQFP / QFN
[09:10:05] <ziph> skorket: Although for the price of an FTDI, having an AVR with USB and an atmega328 on the board is probably much the same. :)
[09:10:10] <RikusW> but there is breakout boards
[09:10:44] <ziph> If you're still using DIP only it's about time to give up electronics too though. ;)
[09:11:11] <skorket> I'm still using through hole. I hope to change over someday soon, but I'm not there yet
[09:12:01] <skorket> I mean, is it better to have two atmega328's? One for USB and the other for the workload? Maybe have each of them communicate via SPI?
[09:12:21] <RikusW> or uart
[09:12:25] <ziph> Get a USB AVR and have it talk to the ATMEGA328 over SPI/UART.
[09:13:15] <RikusW> skorket: Tom_itx does have breakout boards for usb avr, so do I -> http://sites.google.com/site/megau2s/home
[09:13:32] <molavy> hi
[09:13:39] <molavy> i want create this
[09:13:40] <molavy> http://tuxgraphics.org/electronics/200606/article06061.shtml
[09:14:21] <molavy> but what means twisted L1 on this schematic http://tuxgraphics.org/electronics/200606/article06061.shtml
[09:14:40] <molavy> direct link: http://tuxgraphics.org/common/src2/article06061/eth-remote-device.pdf
[09:15:18] <ziph> molavy: Where does it say "twisted"?
[09:16:01] <skorket> ziph, RikusW, for the cost of the other atmega chip, the breakout board or the development kit, I might as well go with FTDI, no?
[09:16:14] <ziph> skorket: Maybe.
[09:16:26] <ziph> skorket: But having the experience with USB AVR's and LUFA is also worthwhile.
[09:16:52] <skorket> ziph, you really think an atmega328 clocked at 20MHz wouldn't be able to do much else but USB?
[09:16:52] <molavy> from magjack (pin1) to ground
[09:17:05] <ziph> molavy: Are you asking what the symbol is?
[09:17:11] <molavy> yes
[09:17:18] <ziph> molavy: It's an inductor.
[09:18:13] <ziph> molavy: In this case it is a few windings around a ferrite bead.
[09:20:06] <RikusW> skorket: you certainly won't be able to do other realtime stuff on the m328 you use for vusb, since it is already running vusb as hard realtime code
[09:20:08] <molavy> it doesn't has more data on it
[09:20:36] <molavy> what should i do now?
[09:20:38] <ziph> molavy: There's a picture of it in "Ethernet Magnetics and Filters" on that page.
[09:20:50] <skorket> RikusW, but also with LUFA?
[09:20:58] <ziph> molavy: However if you want specifics, download and read the ENC28J60 datasheet.
[09:21:38] <RikusW> skorket: LUFA is more flexible because the realtime usb stuff is handled by a seperate usb controller on the avr
[09:22:38] <molavy> ziph, i think my magjack has internal filter
[09:23:15] <skorket> RikusW, I see, my apologies.
[09:24:17] <skorket> right, so V-USB is really the only option if I wanted to use an atmega328 for USB and you're telling me that V-USB is way to intesive to do anything else useful
[09:24:35] <molavy> another question: what means TP1 on enc chip connecion(pin5)
[09:24:45] <pc_magas> Hello what is the best way to use LCD Display with * or 4 bits?
[09:24:53] <pc_magas> Hello what is the best way to use LCD Display with 8 or 4 bits?*
[09:24:56] <ziph> You're forgetting that it isn't real USB and requires OS hacks to work. ;)
[09:25:16] <ziph> molavy: That usually means "Test Point".
[09:25:29] <molavy> what?
[09:25:46] <RikusW> skorket: you would be able to do non realtime stuff in LO-SPEED usb mode (1.5Mbps) only
[09:26:01] <ziph> A test point, where you're hook your oscilloscope or multimeter to for testing.
[09:26:02] <skorket> ziph, right. Well, only for windows. For Linux I think things work out of the box
[09:26:05] <RikusW> with usb avr you can do full speed (12Mbps)
[09:26:23] <ziph> skorket: Only because of a deficiency in the Linux USB support.
[09:26:36] <molavy> ziph,thanks
[09:26:39] <skorket> ziph, oh? whoa
[09:26:52] <skorket> RikusW, but 1.5Mps is probably plenty for what I want to do...
[09:27:38] <RikusW> skorket: via vusb it would probably be more like a 115200bps serial port, or even slower, and usb latency is bad....
[09:28:06] <RikusW> sending data byte by byte results in full speed usb only transmitting at 1kb/s
[09:31:14] <skorket> RikusW, hmm, well, I think the best way to proceed at this point is to play around with it.
[09:34:22] <RikusW> if you already have the avr's it wouldn't hurt to try vusb
[09:35:58] <ziph> Keep in mind that a baby puppy somewhere in the world dies every time you compile V-USB though.
[09:36:09] <ziph> ;)
[09:36:27] <OndraSter> I thought that every time V-USB goes through initialization
[09:36:29] <OndraSter> :P
[09:36:47] <skorket> RikusW, I'm not setup to do SMD and the breakout boards + chips seem pretty expensive. $15 for a development board vs. $3 for the chip means I can play around more stuff. Maybe, we'll see.
[09:37:13] <skorket> If it had been kittens I would not use V-USB
[09:37:27] <ziph> A soldering iron and a flux pen is all you need for SMD.
[09:37:41] <ziph> A 3x mag lens with light helps if you're over 30.
[09:38:14] <skorket> ziph, but also making the boards. I can't breadboard a project beforehand
[09:38:34] <skorket> I definitely want to go SMD but I need a way to make cheap boards...or at least have the board turnaround be minimal
[09:38:40] <skorket> as in print my own or mill my own
[09:38:47] <skorket> I'm hoping to get there, just need time
[09:38:48] <ziph> Are you in the states?
[09:38:53] <skorket> ziph, yeah
[09:39:06] <ziph> skorket: http://dorkbotpdx.org/wiki/pcb_order
[09:39:12] <RikusW> skorket: ask Tom_itx about his breakout
[09:41:31] <abcminiuser> What did I miss?
[09:41:48] <ziph> abcminiuser: We were trying to remember if it was kittens or puppies that die every time you compile V-USB.
[09:42:02] <ziph> abcminiuser: Do you remember which?
[09:42:27] <skorket> yeah, no, there's some good resources. I think a few guys either here or on #electronics also do small orders for very reasonable prices. It's not just about the price, it's about my skill level of designing boards and also turn around. Ideally I would have a setup where I could go from idea to board in under a half hour
[09:43:26] <ziph> skorket: I used to do that, but the effort involved in making the boards is such that you procrastinate longer than the turn time of a board house. ;)
[09:44:32] <skorket> ziph, yeah, I hear that, but I know that I'm going to make a ton of mistakes, so I just want something that I can do quick. Anyway, it's still through hole for me right now.
[09:44:48] <abcminiuser> ziph I thought it was monkeys
[09:45:01] <ziph> abcminiuser: Cute baby monkeys?
[09:45:05] <abcminiuser> Indeed
[09:45:12] <abcminiuser> Tasty monkeys
[09:45:27] <abcminiuser> Incidentally, latest LUFA trunk is pretty easy to jam into AS5/AS6 now
[09:45:43] <ziph> There are monkeys that are both cute and tasty?
[09:45:50] <abcminiuser> make export_tar, extract that tar out to your project directory, import the files and add the required symbols
[09:48:37] <ziph> abcminiuser: What does easter look like in Norway?
[09:52:38] <abcminiuser> Not so snowy today, it's snowed for the last week straight
[09:53:34] <ziph> Do they have eggs and bunnies?
[09:54:55] <madsy> ziph: Chocolate eggs- and bunnies, sure :)
[09:55:08] <RikusW> There is a little snow here is SA too, only one some mountain peaks I hear ;)
[09:55:31] <madsy> Some people go skiing and have a week trip to their cabin. Most people stay home and watch TV and eating goodies
[09:55:35] <molavy> pin 6 on magjack ethernet connection to nothing http://tuxgraphics.org/common/src2/article06061/eth-remote-device.pdf
[09:55:40] <molavy> ?
[09:56:43] <RikusW> molavy: it should probably look like the other one
[10:00:03] <abcminiuser> ziph, lots of chocolate all round
[10:02:13] <Steffanx> abcminiuser got a bicycle yet?
[10:05:52] <elektrinis> where have I seen you
[10:12:11] <RikusW> molavy: or maybe it shouldn't be connected
[10:23:44] <molavy> what?
[10:24:11] <molavy> but,how can i find correct situation ?
[10:59:04] <elektrinis> any body seen an FT2232 based AVR programmer with HVSP?
[11:26:25] <molavy> i want use this code but return error :http://dpaste.com/729091/
[11:27:04] <ziph> The hint is on line 3
[11:28:35] <molavy> yes, it want atmega88, but i just have atmega8a
[11:29:18] <ziph> That warning doesn't care, it just wants some CPU.
[11:33:32] <molavy> now return this error: http://dpaste.com/729093/
[11:33:36] <specing> Lol @ line 65
[11:34:05] <ziph> molavy: Do those registers exist on your particular AVR?
[11:34:23] * specing slaps molavy around with a large trout
[11:34:37] <specing> molavy: Go RTFM
[11:36:16] <molavy> i think i can't set CLK on atmega8 but atmega88 has it
[12:49:34] <OndraSter> http://www.analog.com/en/analog-to-digital-converters/ad-converters/ad9218/products/product.html
[12:49:35] <OndraSter> sexy
[12:55:25] <specing> "Low power: 275 mW at 105 MSPS per channel"
[12:55:40] <specing> Ha-Ha-Ha
[12:56:03] <specing> Im sure we could build an oscilloscope with this, am I right?
[12:57:04] * specing looks down at prices
[12:57:11] <specing> Hmm
[12:57:40] <specing> this one for 30$ + ARM is still cheaper than buying a real scope
[12:58:11] <RikusW> that would probably make a 10MHz scope or so
[12:58:15] <ureif> good luck reading that from an ARM
[12:58:17] <Tom_itx> do you need a real one?
[12:58:22] <Tom_itx> man up and get one if you do
[12:58:48] <Martyn> re
[12:59:00] <Martyn> ureif : Reading what from an ARM?
[12:59:18] <ureif> Martyn: a 105MSPS ADC ?
[13:00:07] <RikusW> Martyn: would the 2 serial interfaces on that avrs of yours be capable of SPI ? that would make them quite usefull
[13:00:33] <Martyn> RikusW: Yes.
[13:00:38] <Martyn> Since they are both on CLK
[13:00:47] <RikusW> It does seem like one is shared with the usb port ?
[13:00:47] <specing> RikusW: An ARM can do 105 MSPS/s
[13:01:04] <Martyn> RikusW : Not that I can see
[13:01:09] <Martyn> RikusW : Do you have the pinout?
[13:01:27] * Martyn doesn't remember if I shared the pinout I got with you or not...
[13:01:27] <RikusW> no, but I've read the short datasheet
[13:01:39] <RikusW> no you didn't
[13:01:39] <Martyn> Oh, the summary? It doesn't have a pinout
[13:01:40] <specing> Im guessing it has parallel data connection
[13:02:03] <Martyn> RikusW : The two IO pins are completely separate from the USB differential pairs
[13:02:14] <Martyn> Although everything shares GND
[13:02:18] <Martyn> and VCC
[13:02:21] <RikusW> specing: 105MSPS would be enough for 10 samples/cycle so 105/10 = 10.5MHz
[13:02:52] <Martyn> Turns out the VCC pin is +5V tolerant (although the chip itself is 3.3v)
[13:03:13] <RikusW> Martyn: but SPI needs 3 pins.... mosi miso sck....
[13:03:24] <Martyn> sck = CLK
[13:03:43] <Martyn> or at least in this case, you can use dCLK for that
[13:03:58] <RikusW> and can it operate in master mode ?
[13:04:14] <RikusW> I guess you will need the full datasheet for that...
[13:04:28] <Martyn> Yep
[13:04:35] <Martyn> As it is, I would be happy with TWI
[13:04:39] <specing> RikusW: I am not talking about AVRs!
[13:04:41] <RikusW> Why not ask for the datasheet without the crypto docs ?
[13:04:51] <Martyn> Because they simply don't have it anymore
[13:05:07] <Martyn> as I said, they -say- the datasheet was passed on to the french company, and I belive them
[13:05:12] <Martyn> the french company refuses to even talk to me
[13:05:19] <RikusW> specing: for an analog scope you need to oversample...
[13:05:36] <RikusW> I'd say at least 10 samples per cycle
[13:06:05] <RikusW> thats nasty of the french guys...
[13:06:28] <specing> Martyn: Have you tried to talk to them in french?
[13:06:36] <Martyn> It's business... they have no requirement to give me anything
[13:06:40] <Martyn> specing, you're not helping
[13:08:54] <Steffanx> specing is right, French HATE english :P
[13:10:19] <Martyn> No, it's not a language barrier
[13:10:36] <Martyn> it's a crypto, product, marketing, export, god-this-is-obsolete-why-should-we barrier
[13:14:52] <RikusW> You might get lucky if you ask for a stripped down version...
[13:23:44] <specing> http://www.analog.com/en/analog-to-digital-converters/ad-converters/ad9283/products/product.html
[13:23:52] <specing> This is way better for our needs
[13:24:10] <RikusW> for a digital scope ?
[13:24:43] * Tom_itx gives specing a digital watch scope
[13:24:57] <OndraSter> what's wrong with the 9218? :)
[13:24:57] <RikusW> hooking it up to a hi-speed usb 3a3 xplain board might work
[13:25:04] <OndraSter> 9283 is its 8bit brother
[13:25:06] <specing> OndraSter: its expensive
[13:25:10] <OndraSter> well yes
[13:25:17] <OndraSter> that's because it is 105MSPS and 10bit (!!)
[13:25:29] <specing> 8-bit is better because we dont have to worry about the upper 2 bit
[13:25:32] <OndraSter> and actually 9283 is used in the Rigol DS1052E/1102E :)
[13:25:33] <specing> (s)
[13:25:39] <OndraSter> you can always ignore them
[13:25:46] <OndraSter> (and you ment bottom 2 bits)
[13:26:09] <specing> Yes, but that would be just wasting money
[13:26:24] <OndraSter> I lied, 9288 is in the Rigol
[13:26:26] <OndraSter> two channels
[13:26:43] <OndraSter> well
[13:26:46] <specing> hmm
[13:26:52] <OndraSter> in $10k scope it doesn't matter :)
[13:26:56] <OndraSter> hundred bucks there or there
[13:26:57] <specing> In the end, it is 10$ cheaper
[13:27:21] <OndraSter> http://www.analog.com/en/analog-to-digital-converters/ad-converters/ad9288/products/product.html
[13:27:22] <OndraSter> this is better
[13:27:25] <OndraSter> two ADCs
[13:27:27] <OndraSter> a bit cheaper per ADC
[13:27:47] <specing> yes
[13:28:00] <specing> Now we only have to find one to sell us 1-2 piece
[13:28:11] <OndraSter> mouser? digikey? :)
[13:28:15] <OndraSter> you can actually sample 2 pieces
[13:28:54] <OndraSter> hmm not on mouser appearantly
[13:29:25] <Tom_itx> non stock at digikey
[13:29:28] <OndraSter> and digikey has not that good prices compared to the AD itself
[13:31:09] <specing> costs $18 on farnell
[13:31:15] <OndraSter> which one?
[13:31:16] <specing> err, euro
[13:31:21] <specing> 9288-100
[13:31:32] <OndraSter> duh
[13:31:39] <OndraSter> if I wanted 1 - 2 pieces, I'd go for samples :)
[13:31:40] <specing> Im checking out others
[13:32:43] <specing> OndraSter: I've never ordered samples
[13:32:56] <specing> OndraSter: What is the procedure for doing so?
[13:33:24] <OndraSter> you need some non-public email
[13:33:27] <OndraSter> for me it was @ondraster.cz :)
[13:33:33] <OndraSter> then fill out your address
[13:33:35] <OndraSter> and that's it
[13:33:47] <OndraSter> I've grabbed samples from AD, TI, Maxim (Dallas)
[13:35:02] <Steffanx> you didn't abuse your uni emailaddress OndraSter ?!
[13:35:07] <OndraSter> nope
[13:35:11] <OndraSter> I am not on uni yet :)
[13:35:19] <Steffanx> Hmm
[13:35:42] <specing> Ah, so you are in highschool like me
[13:35:52] <OndraSter> finishing this year
[13:35:57] <specing> yeah, me too
[13:35:58] <smthngsmthng> highschool and in #avr? highschool me would have been very jealous
[13:35:59] <OndraSter> not sure at which age you end your high school :)
[13:36:00] <Steffanx> Your project was pretty advanced for a high school project ..
[13:36:06] <specing> finaly...
[13:36:07] <OndraSter> really? :D
[13:36:19] <specing> 19
[13:36:23] <OndraSter> same here then
[13:36:30] <OndraSter> well one of my friends doing long-term project was doing some long text abour SMPSs
[13:36:44] <specing> Im doing SMPSs...
[13:36:48] <OndraSter> and the 3rd person doing long term was programming some tamagotchi-like simple game
[13:36:56] <OndraSter> they all told me I have to go AFTER them lol
[13:37:12] <Steffanx> Hmm, i guess we finish high school way to early here..
[13:37:14] <Steffanx> As in ~17
[13:37:19] <specing> My USB battery charger just started working today
[13:37:23] <OndraSter> my project is fairly simple, just using one micro, one USB - UART converter and then shift the data through SPI :)
[13:37:26] <OndraSter> with proper timing
[13:37:44] <specing> 500 bytes flash, atmega8
[13:37:48] <OndraSter> :)
[13:37:59] <OndraSter> I've got few ideas for holiday in my head
[13:38:12] <OndraSter> with stuff like tiny13a + NRF24L01 + DS18B20 etc
[13:38:16] <specing> learning for the final exam?
[13:38:22] <OndraSter> naaa not yet
[13:38:24] <specing> :D
[13:38:27] <OndraSter> still month before it all starts LOL
[13:38:35] <specing> :/
[13:38:46] <OndraSter> specing, IIRC you are from Germany, right?
[13:38:52] <OndraSter> or was it somebody else?
[13:39:00] <Steffanx> that was amee2 k
[13:39:03] <OndraSter> oh
[13:39:05] <OndraSter> yeah
[13:39:11] <specing> no...
[13:39:26] <OndraSter> (FYI, I have fairly bad ... not fairly, but VERY bad memory for names LOL)
[13:39:30] <specing> Im two countries below Germany
[13:39:36] <OndraSter> I still have to think about names of people that I meet every day for past 4 years lol
[13:39:41] <OndraSter> sounds like Italy?
[13:39:47] <OndraSter> (I am noob in Geography as well)
[13:39:49] <specing> no, silly
[13:40:22] <OndraSter> http://binged.it/Ib4F7O
[13:40:37] <OndraSter> I see Germany -- Switzerland/Austria -- Italy
[13:40:50] <specing> Look harder..
[13:41:29] <OndraSter> .. still Italy
[13:41:53] <Steffanx> SE ..
[13:42:09] <Steffanx> or E
[13:42:44] <OndraSter> .. I don't need to know it :D
[13:42:48] <Steffanx> Yes you do
[13:43:43] <RikusW> specing: to the east or west ?
[13:43:49] <specing> right
[13:43:53] <Steffanx> eat
[13:43:55] <Steffanx> st
[13:48:58] <RikusW> it wouldn't be hungary ? :-P
[13:49:30] <specing> eh...
[13:51:38] <RikusW> or is greece a bit too far down ? ;)
[13:51:45] <OndraSter> :D
[13:52:02] <OndraSter> anybody feeling like doing board for this? :)
[13:52:02] <OndraSter> http://cz.mouser.com/ProductDetail/Cypress-Semiconductor/CYRF8935A-24LQXC/?qs=2UFnnMkojnWbhSIKe5uwJWf7fEJ31XPNekrT%2f3%252bZ6lY%3d
[13:52:06] <OndraSter> "how hard can it be!"
[13:52:29] <RikusW> specing: would down mean north or south to you ?
[13:52:54] <specing> -.-
[13:53:38] <j4cbo> OndraSter: oh it's only qfn
[13:53:49] <OndraSter> .. that's not the problem
[13:53:52] <OndraSter> the problem is that it is RF board :)
[13:53:59] <specing> OndraSter: Output power: 0 dbm
[13:54:09] <OndraSter> 1dBm actually
[13:54:13] <specing> -.-
[13:54:20] <OndraSter> plenty :)
[13:54:25] <specing> Thats like one meter or ... ?
[13:54:28] <j4cbo> eh, they give you a reference design
[13:54:36] <OndraSter> j4cbo, reference schematics
[13:54:37] <OndraSter> not board
[13:54:43] <Steffanx> Just get a cheap nrf24lxxx from ebay OndraSter :P
[13:54:51] <Steffanx> Different chip, but also ~2.4GHz
[13:54:54] <OndraSter> Steffanx, sure, but I would prefer to build something on the main board itself
[13:55:04] <j4cbo> "RF path: Adhere closely to the recommended reference design
[13:55:05] <j4cbo> circuit."
[13:55:14] <OndraSter> oh
[13:55:26] <j4cbo> under "Recommendations for PCB Layout"
[14:33:48] <Martyn> Back
[14:34:15] <Martyn> The interrupt issue has a solution. The FAE and I tried two of the workarounds, and one of them does allow all the timers to be setup and fire correctly
[14:34:28] <Martyn> There is a bug in the state engine used to setup interrupts on the SAM3U
[14:34:35] <Martyn> (actually on all the SAM3 based chips)
[14:34:50] <Martyn> A bug report and errata will be published
[14:40:06] <Steffanx> How nice
[14:54:03] <Martyn> Very!
[14:54:12] <Martyn> Considering I had bought $1k worth of processors
[14:54:25] <OndraSter> only $1k?
[14:55:03] <RikusW> well if there is a bug in their processors it is in their best interest to find it....
[14:55:16] <Steffanx> Yeah, only in their interest
[14:55:30] <Steffanx> Now you have x devices with a bug
[14:57:15] <Steffanx> Be happy about the fact there's a workaround Martyn :)
[14:58:51] <specing> Martyn: How many chips is this?
[14:59:03] <specing> And why do you need them for?
[14:59:16] <OndraSter> isn't he building the new supercomputer for NASA?
[14:59:18] <OndraSter> or was it somebody else? :)
[14:59:58] <Steffanx> You don't use cm3 for that
[15:19:41] <hetii> Hello :>
[15:20:04] <OndraSter> ey
[15:20:20] <hetii> Q: is it possible to use enc28j60 +avr+ tda1543 to have some streaming player ?
[15:20:49] <district> hetii: what audio format?
[15:21:04] <district> the avr won't have enough juice to decode mp3
[15:21:15] <district> there are mp3 decoding chips about though
[15:21:21] <OndraSter> yeah
[15:22:18] <OndraSter> also, I am not sure about the I2S throughouput on regular mega
[15:22:22] <OndraSter> probably xmega would be better choice for that
[15:22:36] <OndraSter> or if you would manage to tie I2S into some existing hardware peripheral
[15:22:48] <j4cbo> i've done enc28j60+avr
[15:22:55] <hetii> yhy, ok so other use case this enc28j60 + some uC + tda1543 and the point is to have server for pulseaudio so use it to play sounds from my laptop :)
[15:23:08] <j4cbo> it's pretty tight if you want TCP
[15:23:38] <OndraSter> xmega :)
[15:23:48] <OndraSter> they are not as expensive as they look actually
[15:24:01] <hetii> but where will be the problem ? with enc28j60 or avr ?
[15:24:10] <OndraSter> shouldn't be
[15:24:10] <district> hetii: i'd go with an arm cortex m- of some description
[15:24:12] <j4cbo> hetii: you'd be far better off with an arm
[15:24:23] <district> they'll have hardware i2s and much better number crunching ability
[15:24:27] <j4cbo> get one of these: http://mbed.org/
[15:28:04] <hetii> check, it to but its powerfull but not cheap :)
[15:29:35] <Martyn> No, you certainly don't use Cortex M3's for a supercomputer
[15:29:50] <Martyn> you DO however use Cortex A9-Quad and Cortex A15 eight core chips :)
[15:30:25] <Martyn> district : You know, now you're going to make me want to make an mp3 decoder that works in 20MIPS... just cause!
[15:31:57] <essobi> What's up peebles.
[15:32:19] <RikusW> arm discussions that you just missed ;)
[15:32:23] <Martyn> *sigh* Yep, math can't work out .. no way you can do mp3 decoding in 20MIPS 8 bit
[15:32:30] <Martyn> yep yep yep
[15:32:37] <Martyn> I'm just happy to ==have== a work-around
[15:32:46] <Martyn> so I don't have to switch from SAM3U to STM32
[15:32:48] <Martyn> or SAM4S
[15:32:58] <hetii> to be onest i have already board with atmega128 so i just consider to add to it the enc28j60 and then check what could be possible ??
[15:33:19] <essobi> RikusW: Awwww.
[15:33:33] <RikusW> zlog
[15:33:49] <essobi> RikusW: I've got an ep9302 I've was considering digging out to play with again.
[15:34:03] <RikusW> essobi: whats that ?
[15:34:09] <essobi> Seems I've lost my prototype 328p board in the move. :(
[15:34:20] <essobi> RikusW: http://www.cirrus.com/en/products/ep9302.html
[15:34:28] <essobi> ARM920T
[15:34:52] <RikusW> can run Linux, nice
[15:35:24] <district> Martyn: you won't succeed ;)
[15:35:28] <essobi> RikusW: And a few BSD variants
[15:35:40] <district> Martyn: it's hard enough even with 60MIPS and 32-bit :)
[15:35:53] <RikusW> what does it cost ?
[15:37:18] <essobi> RikusW: Ehh, bare chip is $5 last time I saw it. I paid like $105 delivered for this olimex eval board...
[15:37:24] <essobi> Got lots of bells on it.
[15:38:34] <RikusW> seems like eval boards can be expensive
[15:39:01] <RikusW> I've seen a minimalist m168 board selling for !! $62 !!!
[15:39:07] <RikusW> thats a ripoff
[15:39:48] <essobi> lulz... yea.
[15:40:11] <essobi> Still thou... saw the ard uno with an ethershield built in for around $60... IDK if that's worth it thou.
[15:40:23] <osterhase> well, the engineer wants a piece of the pie
[15:40:24] <essobi> I mean.. I wouldn't be using the actual ard framework.
[15:49:16] <hetii> ok Other subject :) Maybe some of you see somwhere some implementation of FFT that is used to control external devices by whistling ?
[15:49:22] <Martyn> osterhase : Of -course- the engineer would like to be paid .. sheesh
[15:49:33] <Martyn> The average board I make takes me between 100-200 hours of design and testing
[15:49:48] <Martyn> and that's assuming I have the advantage of having a reference design to work off of
[15:50:10] <osterhase> then price depends on how much are produced
[15:51:50] <specing> Martyn: 100-200? what kind of boards are this?
[15:52:51] <Martyn> specing : QuadCopter UAV boards
[15:53:21] <Martyn> (and 100 hours of work is only a little over two weeks of 8 hours a day .. not a lot)
[15:53:55] <Martyn> If you can do the design work faster (and better) I'd like to see it.
[15:54:31] <Martyn> The average little throw-away board I make for a conference still tends to take me a couple weeks to design.
[15:54:39] <j4cbo> that seems like a while o_O
[15:54:45] <Martyn> The one I made for ToorCon was easily four weeks
[15:55:01] <Martyn> j4cbo: As I said, feel free to design a 4-layer board faster :)
[15:55:23] <_hetii_> :(
[15:55:29] <Martyn> I just have started to really loathe people when they say "Components cost X! Why doesn't board cost Y?"
[15:55:42] <Martyn> And the answer is "Because I have to make a living"
[15:56:06] <Martyn> Also, boards have a base cost that you can't get around from the manufacturer
[15:56:10] <Tom_itx> i generally say it takes about 3 revs to get one rihgt
[15:56:24] <Tom_itx> rev2: right*
[15:56:32] <Martyn> For example .. every SINGLE new run of a board I make at e-Teknet (like Gold Phoenix) has a $75 setup fee, a $200+ stencil fee, etc..
[15:56:53] <Martyn> Tom_itx: I'm proud to say that I tend to get things right on the 1st rev, by being careful and paying attention to detail
[15:56:53] <Tom_itx> you should try capnkernel
[15:57:00] <j4cbo> Martyn: 4-layer boards are easier; less fucking around with power traces :V
[15:57:10] <Martyn> j4cbo: And more fucking around with EMI
[15:57:12] <OndraSter> I am proud to say that I tend to get things right on the 1st rev without being careful and ignoring details
[15:57:13] <Martyn> there's always a tradeoff
[15:57:28] <OndraSter> :P
[15:57:45] <OndraSter> lucky me, doing simple stuff
[15:58:01] <Martyn> OndraSter : that's a Good Thing(tm)
[15:58:08] <OndraSter> :D
[15:58:09] <Tom_itx> barring design changes
[15:59:25] <Tom_itx> how large are your boards Martyn?
[15:59:29] <Tom_itx> on average
[15:59:32] <Martyn> Quite small, for the most part.
[15:59:35] <district> Martyn: oh you do quadcopters? awesome!
[15:59:46] <Martyn> The UAV boards tend to be around the size of 1/4 eurocard
[15:59:48] <Tom_itx> you should try capnkernel once
[15:59:56] <Martyn> (about the size of an Arduino in fact)
[16:00:19] <Martyn> The SAM3U breakout board is 5cm square
[16:00:27] <Martyn> but it's not very complex..
[16:00:51] <Martyn> My little ARM motherboards that can use ATX power supplies, and that use OMAP4 processors, are Mini-ITX sized
[16:00:57] <Tom_itx> he uses the same mfg as itead and that other chinese broker
[16:01:02] <Martyn> Those are by far the most complex, and have PCIe bus
[16:01:11] <district> fancy!
[16:01:24] <Martyn> Tom_itx: I use e-teknet because they have had the best price and quality for me for a long time
[16:01:31] <Martyn> it's also the one with the most layers -- 12
[16:01:32] <district> Martyn: how many layers?
[16:01:38] <district> oh well there you go ;)
[16:01:49] <Tom_itx> i'm not sure the layer restriction he has
[16:01:50] <Martyn> district :mostly due to the ballout of the BGA part
[16:02:08] <Tom_itx> iirc he has about the smallest trace i've seen
[16:02:21] <essobi> Martyn: URL?
[16:02:29] <essobi> Martyn: For the ARM boards.
[16:02:53] <Tom_itx> bga makes layers add up quick
[16:03:35] <Tom_itx> i haven't done any on the hobby level yet
[16:04:03] <OndraSter> also you can't visually check for soldering problems on BGAs
[16:04:05] <OndraSter> without xray
[16:04:13] <OndraSter> and honestly - who has xray at home? :)
[16:04:32] <Tom_itx> nobody probably should
[16:04:33] <Martyn> essobi :
[16:04:44] <Martyn> essobi: We're still setting up the website -- www.antscale.com
[16:04:55] <Martyn> the 1st prototype boards came back April 3rd
[16:05:19] <Martyn> Commercial availability (and finished website!) will be ~May 5th
[16:05:33] <Martyn> Assuming everything passes final verification
[16:05:54] <Tom_itx> how's it compare to the hammer board etc?
[16:05:57] <district> Martyn: what are the quad core ARM chips?
[16:06:12] <essobi> Martyn: Any idea what the prices are going to be on the market? Can I PM you?
[16:06:14] <Martyn> district: OMAPO
[16:06:21] <Martyn> OMAP 44xx
[16:06:30] <Martyn> essobi: Sorry, that's still under wraps
[16:06:49] <Martyn> essobi: It will be very, very comptetitive with other ARM boards
[16:06:50] <Tom_itx> do you use GP for production?
[16:07:07] <Martyn> Tom_itx: No. All production of these boards is done in the USA, in Denver
[16:07:16] <Tom_itx> AP
[16:07:18] <Martyn> To meet security and validation requirements
[16:07:45] <Tom_itx> or advance
[16:07:48] <j4cbo> iiiiiiinteresting
[16:07:51] <Martyn> Can't say :)
[16:07:53] <Tom_itx> i can't ever remember which on is in denver
[16:08:16] <Martyn> Tom_itx: It's not Advanced or AP.. there are lots of manufacturers in Denver/Broomsfield/Boulder :)
[16:08:21] <essobi> Martyn: PM!
[16:08:24] <Tom_itx> i suppsoe
[16:08:35] <Tom_itx> those are the most common
[16:09:12] <Tom_itx> i had one here locally and didn't even know it
[16:09:15] <Martyn> Sure, for short and small runs :)
[16:09:16] <Tom_itx> until they closed
[16:09:22] <Martyn> I like Advance for some things
[16:09:24] <Tom_itx> i got all their excess board material
[16:09:37] <Martyn> but believe it or not, I still get most of my quick-run stuff done by PDX and midnight maker
[16:09:42] <Tom_itx> ~2 pickups full
[16:10:02] <Martyn> okay, I need to go pick up some test leads .. back in a while guys :)
[16:11:11] <essobi> Tom_itx: nice!~
[16:43:35] <ircWALK> how hard would it be to take apart a scientific calculator, and replace the circuitry with my own
[16:44:15] <OndraSter> yesterday's question? :)
[16:44:17] <ircWALK> so basically use the existing buttons/screen - but recreate the electronics inside
[16:44:24] <OndraSter> you tell us! :P
[16:44:27] <OndraSter> take one apart..
[16:44:29] <OndraSter> and see :)
[16:44:41] <ircWALK> im askin again coz i got 1 answer yesterday and it was "try it and see"
[16:44:46] <OndraSter> :D :D
[16:44:53] <OndraSter> I'd say quite hard
[16:44:57] <OndraSter> you are really limited with space
[16:45:08] <OndraSter> and you have no info about how it is wired up
[16:45:25] <OndraSter> no info about the display etc
[16:46:40] <ircWALK> why do you need to know how its wired up? wouldn't you be doing all the wiring up by yourself? all you are doing is using some buttons/screen/case that already exist
[16:46:45] <ircWALK> instead of moulding it into shape yourself?
[16:48:37] <OndraSter> oh
[16:48:46] <OndraSter> I thought you would just "replace" the IC
[16:48:48] <OndraSter> somehow
[16:48:57] <OndraSter> get rid of the black block
[16:49:37] <ircWALK> no. im effectively saying instead of buying buttons/screen/case from a manfacturer - you extract them from a calculator you buy in shops
[16:49:48] <ircWALK> and then completely remake everything else
[16:50:15] <ircWALK> so atleast it saves on the costs of moulding plastic and printing ink onto plastic and bla bla
[17:38:39] <essobi> ircWALK: yea.. sounds possible if you can source a calculator that you can figure out the driver for the LCD and keypad interfaces to hookup to whatever mcu you want to use.
[17:50:53] <Kevin`> ircWALK: the type of lcd that calculators use can actually be probed moderately easily
[19:53:20] <poincare101> Any idea how these guys (http://supermechanical.com/twine/) got WiFi under $100?
[20:09:14] <CapnKernel> You can buy a PCI Express WiFi card for under USD20. That means the parts are around USD10, tops.
[20:11:43] <Casper> but I have high doubt on the months claim on 2 AA batts
[20:11:48] <Casper> heck
[20:11:49] <Casper> worse
[20:11:52] <Casper> on 2 AAA batts
[20:13:06] <CapnKernel> Example: http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en548014
[20:13:30] <CapnKernel> The WiFi is not constantly on. Those chips are designed to wake up, send a few bytes, then go back to sleep.
[20:13:52] <CapnKernel> You'll note there's nothing on the supermechanical web page that says you can contact the box. The box contacts you.
[20:14:21] <CapnKernel> Those microchip modules are USD12.50 QTY 5000
[20:15:08] <CapnKernel> http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=2885&param=en547232
[20:16:11] <CapnKernel> You'll note the send-data conditions are all things you can get an interrupt on. The processor and radio are asleep until that happens.
[20:16:34] <CapnKernel> So running on two AAA batteries is quite feasible.
[20:17:43] <CapnKernel> Me, I'd like to know how they do power supply. Run directly off the 3V? Small SMPS? What about the SMPS efficiency?
[20:18:16] <j4cbo> surely an SMPS
[20:18:21] <learningc> Hi CapnKernel :)
[20:18:33] <j4cbo> probably 90%+ efficient
[20:20:17] <Casper> so
[20:20:22] <Casper> what would be the power used?
[20:20:25] <Casper> 100mW?
[20:21:11] <CapnKernel> learningc: hi
[20:21:39] <Roklobsta> morning
[20:21:44] <CapnKernel> The problem is even if your chips have microamp quiescent current, your SMPS is still running
[20:22:24] <j4cbo> CapnKernel: not necessarily.
[20:22:42] <Roklobsta> in order for the control loop to be stable I think all SMPS need a minimum load.
[20:22:58] <Casper> hmm 700mA 2.4V... that's 1.68Wh, at 100mW it would be about 17 hours of wifi transmit
[20:23:08] <CapnKernel> It's all about duty cycle - transmit at 100mW for 30 seconds every day, and you're then looking at very low wattage, over time.
[20:23:09] <j4cbo> there are a number of ways to arrange for the SMPS to be totally off most of the time
[20:23:17] <Casper> so probably 8 hours for the whole box
[20:23:39] <Roklobsta> sure you can toggle the shutdown pin on many SMPS chips
[20:23:40] <CapnKernel> Some sort of low-leakage storage cap to give you the quiescent current?
[20:23:48] <CapnKernel> Post-SMPS storage cap?
[20:23:50] <Casper> if it power up every 5 minutes for 10 seconds... that would be...
[20:24:03] <CapnKernel> 3% duty cycle
[20:24:39] <Roklobsta> caps are a bit crappy. use a battery where you can. maybe a rechargeable button cell could do what you want.
[20:24:53] <j4cbo> no...
[20:25:05] <Casper> at 10 secs/5 mins, 200mW, that would give a theorical 8.4 months if I didn't messed up my math?
[20:25:08] <Roklobsta> my last job had a so called supercap on it which could only sustain the RTC for a week or 2.
[20:25:30] <CapnKernel> That sounds fine. Top it up every day, when the voltage drops
[20:25:34] <j4cbo> no
[20:25:38] <j4cbo> that is totally unncessary
[20:25:42] <j4cbo> you just pull the shutdown pin of the smps high
[20:25:50] <CapnKernel> and?
[20:26:13] <CapnKernel> How do you keep your micro/radio powered?
[20:26:25] <Roklobsta> you need a battery.
[20:26:26] <Casper> learningc: hi
[20:26:29] <j4cbo> then the smps stops totally - and you have a tiny cap on the shutdown pin to keep it disabled
[20:26:46] <CapnKernel> j4cbo: How do you keep your micro/radio powered?
[20:26:50] <Roklobsta> or a big cap which you keep topped up
[20:26:55] <j4cbo> CapnKernel: when idling, you don't...
[20:26:58] <CapnKernel> I'm asking Mr. No.
[20:27:16] <CapnKernel> You need to, or you'll have no deep sleep mode to wake from.
[20:27:20] <Roklobsta> Haha, I didn't know Tony Abbott was an AVR hacker.
[20:27:29] <CapnKernel> I was thinking PRECISELY the same thing
[20:27:37] <j4cbo> nah, just kill the rail totally
[20:27:43] <j4cbo> keep only the rtc powered, nothing else
[20:27:58] <CapnKernel> j4cbo: You need the micro to be powered (and in deep sleep) so you can respond to external events.
[20:28:01] <j4cbo> (the rtc can run straight from the AAs; typically rtc chips draw some nanoamps and can run down to a volt or so)
[20:28:21] <j4cbo> CapnKernel: not if you poll
[20:28:25] <CapnKernel> Roklobsta was talking about an RTC, as an example of current draw on a supercap. I'm not.
[20:28:43] <CapnKernel> Hey, this guy's suggesting using polling!
[20:28:52] <Roklobsta> oh the supercap was replaced with a battery as 1-2 weeks wasn't good enough.
[20:29:11] <j4cbo> wake the switcher and uC up for a few microseconds every few seconds.
[20:29:19] <CapnKernel> I'm saying it could be, if you have a mechanism to top up the supercap
[20:29:20] <j4cbo> beats keeping a switcher going all the time
[20:29:30] <CapnKernel> Charging batteries is lossy.
[20:29:38] <Roklobsta> what is the primary source of power that SMPS is feeding off?
[20:29:42] <CapnKernel> Nobody is talking about keeping a switcher going
[20:29:44] <CapnKernel> 2 AAA cells
[20:30:05] <CapnKernel> We're talking about how to avoid constantly running a switcher, but still having power for a micro
[20:30:05] <j4cbo> CapnKernel: so, you're saying run the micro straight off the AAAs?
[20:30:10] <CapnKernel> No
[20:30:21] <Roklobsta> depending on the drop a low dropour linear with a shutdown pin might be just as efficeint as an SMPS
[20:30:36] <learningc> Casper: hi
[20:30:48] <CapnKernel> I mean you could, if the micro was friendly to it, but assuming the micro is designed to run at 1.8V or 3.3V or something, you need a switcher
[20:31:08] <Casper> learningc: http://img521.imageshack.us/img521/6374/img3246l.jpg :'(
[20:31:22] <CapnKernel> j4cbo: Why do you need to "wake the switcher and uC up for a few microseconds every few seconds"?
[20:32:00] <Casper> the uC could actually be run off the battery when in sleep mode
[20:32:20] <Casper> since it's basically 0Hz, it do not need much voltage
[20:32:24] <CapnKernel> Batteries are variable voltage, not all micros are as tolerant as the AVR
[20:32:37] <Casper> not all true, but many are
[20:32:44] <j4cbo> so, here's what i'm envisioning: 3.3v rail that everything runs from; a little ARM or something and a wifi module
[20:32:44] <Casper> beside
[20:32:45] <CapnKernel> You may be confusing voltage with power
[20:32:54] <Casper> who say that the micro is actually running?
[20:33:03] <CapnKernel> I didn't.
[20:33:07] <j4cbo> and some sort of RTC (i've seen ARMs with a separate power domain for the onboard RTC)
[20:33:09] <Casper> there could be a kind of RTC that wake the microcontroller every x seconds
[20:33:11] <CapnKernel> I said it was powered up, not running
[20:33:32] <CapnKernel> Powered up, and in deep deep microamps-type sleep
[20:33:37] <CapnKernel> Ready to respond to an interrupt
[20:33:49] <learningc> Casper: what those parts for?
[20:35:05] <Roklobsta> capn: if you are happy for the micro to run at the lowest voltage and corresponsing frequency you could get away with a micro running off 3.3 to 1.8V off AA batteries.
[20:35:23] <Roklobsta> you don't have to run the thing at max freq.
[20:35:43] <Roklobsta> the project i am working on now has an AVR at 3.4V and 11MHz.
[20:35:44] <CapnKernel> There are good reasons to run it at maximum frequency
[20:36:02] <CapnKernel> With maximum frequency, you get the job done faster, and you can go back to sleep sooner.
[20:36:14] <Roklobsta> partly because the designer didn't want to support an extra 5V rail
[20:36:40] <CapnKernel> That's one of the things the Linux power management guys worked out
[20:36:46] <Roklobsta> true. slower clock rates used less energy for the same amoutn of work though
[20:36:57] <CapnKernel> No
[20:37:09] <CapnKernel> Slower clock rates use less energy for the same amount of time
[20:37:31] <learningc> what would be the fastest algorithm to queue a stream of 11 bits into an array of 8bits datatype?
[20:37:54] <CapnKernel> learningc: Can you tell us more about the problem you're trying to solve?
[20:39:00] <CapnKernel> If you have a deep sleep mode, and an intermittent workload, you're better off running at full speed, finishing quicker, and going back to sleep sooner. Uses less power overall.
[20:39:36] <learningc> CapnKernel: I have an array of char (8 bits) that must hold a stream of bits coming in, but the stream packets come in 11 bits
[20:39:39] <Casper> learningc: my car :(
[20:40:12] <CapnKernel> learningc: Can you describe the problem at a higher level?
[20:40:20] <Casper> why can't you make them 16 bits?
[20:40:34] <Landon> lose a little bit of space, in exchange for a little bit of sanity
[20:40:38] <Landon> Casper++
[20:44:57] <Roklobsta> use 16 bits. if it makes you feel warm and fuzzy use bitfields. like "int Data : 11;" The compiler will allocate 16 bits though.
[20:45:55] <Roklobsta> do you want to pack all the data into 11 bits with no waste of unused 5 bits in 2 bytes?
[20:47:01] <Roklobsta> in that case you will have to write your own bit packer using modulos of the word count to figure out where the 11 bits you are after lie in a pari of bytes.
[20:47:25] <Landon> heh, even worse, in a triplet of bytes
[20:47:38] <rue_house> !seen speedevil
[20:47:38] <tobbor> SpeedEvil was last seen in ##electronics on Dec 02 18:06 2010
[20:48:02] <learningc> CapnKernel, Casper: yes, actually I'm holding the 11 bits data in 16 bits datatype, but I'm not sure how to break it into pieces and fit them into 8 bits or less
[20:49:34] <learningc> ok for example the first 11 bits, I need to break it into 8 bits and 3 bits. The first 8 bits go into the first char of the array, the 3 bits left go into the second char of the array, which is notleft with 5 bits
[20:50:06] <Casper> why don't you use 16 bits array? save lots of headache
[20:50:41] <learningc> casper, how? put 11 bits into that 16 bits array?
[20:50:53] <Roklobsta> capn: yes you're right. duh. lowering the *voltage* reduces power. I guess what I am trying to say is that to maximise the useable battery voltage range just run the CPU at the lowest speed so it will work reliably at 1.8V.
[20:51:28] <Roklobsta> landon: true a triplet.
[20:52:32] <learningc> Casper: then I still need to fill the remaining 5 bits in that 16 bits array :(
[20:52:47] <Roklobsta> leanrinc: do you care to waste 5 bits?
[20:52:50] <Casper> why fill?
[20:53:04] <Roklobsta> is it a problem?
[20:53:24] <Roklobsta> it might be if you have bandwidth or storage issues.
[20:55:54] <learningc> Roklobsta: It's encoding QR codes, and the characters are encoded successively in 11 bit and need to be added heads to tails, then finally the stream broken down into chunk of 8 bits to be fed into an error correction scheme of a sort. I would love to find an easier way too
[20:56:24] <Landon> ah
[20:56:24] <CapnKernel> Ahh, see that's useful information, that's the requirements!
[20:56:42] <Roklobsta> right so yo uDO need to pack the data.
[20:56:43] <CapnKernel> My suggestion: If you go for 12-bit, you can store two values in three bytes with not much pain.
[20:57:52] <learningc> 12 bit? one char andd a half?
[20:57:56] <learningc> *and
[20:58:37] <Landon> could you just store the character encoded in the 11-bit and then feed 8 bits of that at a time into the error correction? saving the remaining bits until you have enough to send two bytes
[20:58:44] <learningc> I don't see if it will make things easier
[21:00:30] <Casper> is the error correction on a per character basis? or multi-character basis?
[21:00:59] <learningc> Landon: the problem is that sometimes, it will need say 5 bits from a first 11-bit character and 3 bits from the next 11-bit to make 8 bits
[21:01:18] <Landon> keep a 2 byte buffer around for that
[21:01:39] <Landon> and keep track of the bit you're at
[21:01:56] <Casper> learningc: you don't make things clear. why do you insist on storing as 8 bits?
[21:02:18] <Casper> when a 16 bits would hold all without any headache?
[21:03:19] <learningc> Casper: 8 bits length "codeword" is needed to apply error correction scheme afterward
[21:05:02] <learningc> Casper: the error correction scheme is reed-solomon something
[21:05:16] <Casper> learningc: but on a per character?
[21:06:22] <learningc> ok, I'll try to explain it better, two characters are combined to make an 11-bit combined character
[21:06:50] <learningc> then you add them all head to tail
[21:07:09] <learningc> then you break downn the stream in 8-bit pieces
[21:07:24] <Casper> the way I see it: unsigned int temp = 0; while (i = 0; i < 11; i++) { temp = (temp << 1) | newbit; }; data[number] = decode(temp);
[21:07:46] <learningc> then the error scheme is apply on each 8-bit chunk
[21:12:25] <Casper> still not following...
[21:12:55] <learningc> array data is a char?
[21:13:19] <Landon> 11 bits expands to 2 characters, but that's not necessarily aligned with the error correcting chunks
[21:13:50] <Casper> I don't see how 11 bits can expend to 2 bytes
[21:14:05] <Landon> it's a QR code thing
[21:15:05] <Casper> ah I see
[21:15:18] <Casper> it's a special character set with reduced bits
[21:15:37] <Casper> 45 symbols
[21:15:50] <learningc> yeah, I foresee a lot of bit shifting :/
[21:16:27] <Casper> actually a lookup table
[21:16:36] <Casper> Alphanumeric encoding mode stores a message more compactly than the byte mode can, but cannot store lower-case letters and has only a limited selection of punctuation marks. Two characters are coded in an 11-bit value by this formula:
[21:16:36] <Casper> V = 45 × C1 + C2
[21:16:39] <learningc> I'm looking for a fast algirithm and can queue 11-bit packets into 8bit array basically
[21:16:53] <learningc> and = that
[21:17:00] <Casper> get 11 bits, decode, store...
[21:18:38] <learningc> Casper: what do you mean?
[21:18:39] <Landon> learningc: I'm really thinking you just need a 2 byte array if you want to do error correction, but it will definitely be a lot of bit shifting
[21:19:06] <Casper> the problem is.. the 11 bits is after the ECC code
[21:19:31] <Landon> ah, I need to go look at the pretty pictures again
[21:20:22] <Roklobsta> capnkernel: for power consumption it's probably not appropriate to compare a system like linux with DRAM to a fully static system like an AVR where the P=CV^2f formula dominates.
[21:20:56] <learningc> Landon: actually, it's error encoding, I need to store them all in a bit linear array afterward to place them in the QR matrix
[21:21:07] <learningc> bit = big
[21:21:24] <Landon> oh, I thought you were decoding a QR code
[21:21:48] <CapnKernel> Roklobsta: I wish I wasn't on my 100Mb/month wireless dongle, I'd show you the paper. It was specifically related to cellphones, where minimising overall power consumption is vital.
[21:21:48] <learningc> Landon: sorry, I forgot to specify encoding :)
[21:21:55] <Landon> looked up and you did specify it
[21:22:58] <Roklobsta> capn: a cellphone is more complex than an AVR system usually and has DRAM that needs refresh cycles
[21:23:12] <learningc> Landon: I guess not well explained then...
[21:23:31] <Roklobsta> it's not a stright comparison. I do beleive you though for a linux type system.
[21:23:39] <Landon> nah, I'm just really good at reading 75% of the content
[21:24:35] <Roklobsta> lordy, i thought you were back in .au anyway on real broadband.
[21:35:19] <CapnKernel> Roklobsta: Google for "race to idle", esp this: http://mjg59.livejournal.com/88608.html
[21:35:39] <CapnKernel> I'm at my mum's place, and she doesn't have internet.
[21:35:59] <CapnKernel> It's a way of avoiding my kids during school holidays, so I can get work done :-)
[21:41:19] <Casper> man
[21:41:41] <Casper> the stealerships are running ads on tv saying they ain't more expensive than elsewhere
[21:42:02] <Casper> ... there is a reason why they are called stealership...
[21:42:48] <Casper> sway bar link kit (basically a long bolt, some washer, a nut and some rubber piece), JC Cote: 19.85/pair, stealership: 37$/ea
[21:42:49] <Casper> caliper, JC: 2x 75$, stealership: 283 + 304$
[21:42:49] <Casper> rotor (brake disk), JC: 2x 40.40, stealership: 2x 242$
[21:42:49] <Casper> brake pad, JC: 42.13/set, stealership: 192$
[21:42:49] <Casper> outter tie rod (a stearing component), JC: 53.79$, stealership: 196$
[21:42:51] <Casper> !!!
[21:44:24] <Roklobsta> capn: btw, twelstra now gives you 200MB/month for $10 on prepaid
[21:48:07] <inflex> Roklobsta: that's about 5minutes on FB
[21:48:32] <Roklobsta> yah but you said you were on a 100MB plan.
[21:51:45] <Roklobsta> some of the carriers like optus give free FB on wireless plans. anyway.
[21:53:20] <Tom_itx> FB ain't worth any bandwidth imo
[21:54:01] <Roklobsta> FB doesn't use any bandwidth at my house coz I have no freinds.
[21:54:23] <Casper> FB use even less bw here than you
[21:54:30] <Casper> because I don't even have a FB account
[21:57:30] <Roklobsta> reading those articles still shows that you can't really compare a complex linux system. it's not quite apples and apples.
[21:59:12] <Roklobsta> especially when a PC has lot sof extra crap like USB and hard disks needing constant polling and so on.
[22:31:30] <CapnKernel> My ISP gives me 100Mb a month free wireless. Each additional 100Mb, $2
[22:47:37] <Casper> 100MB... man... just find a restaurant... lots give free wireless... and that would be what... few secs of slow downloading...
[22:48:05] <CapnKernel> And the obligation to buy something
[22:48:16] <Casper> just park outside for a minute :D
[22:48:55] <CapnKernel> I'm not like that.
[22:49:18] <Casper> get a coffee, download 1G :D