| Author |
Message |
|
|
 Posted: Aug 05, 2012 - 12:02 PM |
|
Joined: Aug 05, 2012
Posts: 1
|
|
i want to be able to burn 10 Atmega328P-PU microcontrollers at once, ive tried to do it with the help of another Arduino Geek, he knows his stuff but i dont and i tried 'daisy chaining' the wires that go to load the bootloader from the Arduino as ISP bootloader instructions and just came out with errors saying something like:
Yikes! The device signature is invalid, check all connections etc etc...
and i was trying to burn two Atmega Chips using one RBBB set up to use the 'burn bootloader' function in the Arduino IDE and i could burn to only one chip but never the 2nd chip even though i duplicated the wires from the RBBB to both chips in separate ZIF sockets with their accompanying 16 mhz crystals and capacitors...
Really been knocking my head against this one for the last 3 months and my brains totally overheated now, lol !
im buying 100 MCU's at a time and just want a faster way of burning the Arduino UNO bootloader onto them as burning only 25 chips takes me 1 hour, so 100 chips takes a damn-good 4 hours which is extremely tedious and just takes too flamin long...
please please please could anyone help with a plan or a schematic to burn 10 atmega328P-PU MCU's in ZIF sockets all in one go ????
This way i'd be done burning within an hour or less i assume, its killing my time to keep having to do this one-by-one and ive scouredd the internet through google and still coming up empty for many MONTHS so far...
Is there any way anyone could possibly point me in the right direction towards my goal please please please ???!!!???
Additional Information:
i also used 2 ZIF sockets connected to the same 'Arduino as ISP' function to load the bootloaders onto two chips, left the 1st ZIF socket empty (this is the one where i have no problems burning the bootloader)...
and then the 2nd ZIF socket that i had 'Daisy-Chained' to the connections that went to program the original ZIF socket and thats where i had the problem!
ALSO - Someone said in a forum post:
You'll have to consult the ISP manual for the AVR microcontroller. My guess is if you cascade the spi MOSI and MISO pins from one device to the next, and you make your own custom ISP software, you may get away with it.
There ARE custom burners for industrial quantities of chips.
Please help as it takes 4 hours to burn the UNO bootloaders and another 4 hours to test each individual one with the Blink-Sketch to verify the Bootloading has completed ! A Whole 8 HOURS !! I know this kind of operation could be done in probably an hour or two if a could burn 10 Atmega chips at once !!!
many thanks in advance for any and all suggestions but i must only be interested in the model number MCU of ATMEGA328P-PU and only the UNO bootloader, many thanks again in advance! |
|
|
| |
|
|
|
|
|
|
Posted: Aug 05, 2012 - 12:33 PM |
|
Joined: Feb 12, 2005
Posts: 16305
Location: Wormshill, England
|
|
It should take about 10 seconds per chip. i.e. most of your time will be spent inserting the chip into a ZIF socket.
You can always buy the chips ready programmed if you don't want to do them yourself.
Yes. You can have 10 ZIF sockets if you want but you need to have custom software to burn each ZIF.
Likewise, you can devise test suites that do not involve clicking IDEs.
David. |
|
|
| |
|
|
|
|
|
|
Posted: Aug 05, 2012 - 05:18 PM |
|
Joined: Nov 17, 2004
Posts: 6137
Location: Great Smokey Mountains.
|
|
Google 'Arduino Gang Programmer"
First hit:
http://www.sparkfun.com/tutorials/233
Followed by other interesting possibilities.
Although is you pay someone $50 a day in wages you get your chips programmed for 50 cents each. I'm not sure how much a gang programmer would cost to build and test or what the total number of Arduino's you'd need to program would be to hit a break even point for building one, but I'd guess it would be above 5,000.
Smiley |
_________________
FREE TUTORIAL: 'Quick Start Guide for Using the WinAVR C Compiler with ATMEL's AVR Butterfly' AVAILABLE AT: http://www.smileymicros.com
|
| |
|
|
|
|
|
|
Posted: Aug 05, 2012 - 06:38 PM |
|
Joined: Jun 08, 2011
Posts: 301
Location: Maryland, USA
|
|
You cannot simply daisy chain or parallel a bunch of AVRs on one ISP connection, as you've discovered. In a daisy chain, you'll just feed junk data to the 2nd through last chip and back to the programmer, and in the parallel setup you'll have bus contention on the MISO line that could damage the MISO pins on the AVRs.
What you could do is move to a device with multiple UARTs that you can use in SPI mode, and modify the ISP firmware to write out the same data to each SPI connection, although you then have to figure out how to handle the verification data coming back from all of those chips. Or you could use some buffers to route one ISP connection to multiple target chips--you'd still be programing them sequentially, but it would still probably save you some time in the simple handling of the chips. |
|
|
| |
|
|
|
|
|
|
Posted: Aug 05, 2012 - 09:10 PM |
|
Joined: Feb 12, 2005
Posts: 16305
Location: Wormshill, England
|
|
I am sure that if I thought hard enough about it, you could daisy-chain the SPI. However I doubt if it would gain you much advantage. You would lose time with propagating NULL commands.
Just how fast do you imagine that you can load ZIF sockets or pogo-pin jigs? The theoretical minimum time to connect, set fuse bits, erase the chip, burn and verify a 4kB bootloader, and set lock bits is about 20ms + 15ms + 25ms + 194ms + 5ms = 259ms or one quarter of one second.
There was a similar question a few weeks ago about programming a ATmega2560. I did not have a 2560 but did have a ATmega128. I seem to remember the theoretical time for 128kB was 5.5 seconds. I managed 6.2 seconds, so the theoretical and actual are not too far apart.
The OP of that thread said that a AVRISP-2 was taking 12 seconds for 250kB of the ATmega2560. In other words, the regular Atmel tools are pretty efficient. (I guess that this was without a verify pass)
David. |
|
|
| |
|
|
|
|
|
|
Posted: Aug 05, 2012 - 10:29 PM |
|
Joined: Jul 02, 2003
Posts: 1029
Location: Tricky Dicky City
|
|
Maybe just make stand alone programmers that use the arduino running a sketch to do the bootloader and fuses (no need for the IDE). Could also make it once done to load the blinky program. See this site for
some good ideas. http://www.gammon.com.au/forum/?bbtopic_id=123 |
|
|
| |
|
|
|
|
|
|
Posted: Aug 05, 2012 - 10:42 PM |
|
Joined: Jun 19, 2002
Posts: 955
Location: SF Bay area
|
|
|
|
|
|
|
|
Posted: Aug 06, 2012 - 02:58 AM |
|
Joined: Dec 30, 2004
Posts: 8767
Location: Melbourne,Australia
|
|
What about parallel programming mode?
Why do you bother testing by loading a sketch? Have your default image load the bootloader and the blinky app. If the led blinks, there's a very good chance the micro is programmed correctly. |
|
|
| |
|
|
|
|
|
FAQ
Search
Register
Log in to check your private messages
Log in
Maximize
RSS
