| Author |
Message |
|
|
 Posted: May 04, 2012 - 11:59 PM |
|
Joined: Dec 29, 2011
Posts: 81
Location: Pittsburgh, PA
|
|
I've seen several sites that claim their boards run at 3.3v rather than 5. Doing this would make some downstream designs easier, but I was wondering what risk there is.
Is it based on temperature?, endurance? or are do some items just not work (read digital returns wrong results)?
This has to do with the atmega1284 chip.
I will likely use 5 volts and the level shifter, I'm just not real clear why. |
|
|
| |
|
|
|
|
|
|
Posted: May 05, 2012 - 12:55 AM |
|
Joined: Sep 15, 2011
Posts: 131
Location: Israel
|
|
Check this graph, according to it the max frequency you can get at 3.3v would be approx 12MHz or so.
So at 3.3v and 16MHz it might not run at all, or not run in a stable manner. |
|
|
| |
|
|
|
|
|
|
Posted: May 05, 2012 - 01:05 AM |
|
Joined: Nov 22, 2002
Posts: 12036
Location: Tangent, OR, USA
|
|
All depends if it remains in specs with the combination of 3.3V and 16MHz. Some AVRs will. Some won't.
If its hobby stuff, you can probably get away with it out of specs. Commercial? I would not do it.
Jim |
_________________
Jim Wagner
Oregon Research Electronics, Consulting Div.
Tangent, OR, USA
"The only thing standing between us and victory is defeat" P.G.Wodhouse in Wooster & Jeeves series
|
| |
|
|
|
|
|
|
Posted: May 05, 2012 - 05:19 PM |
|
Joined: Dec 29, 2011
Posts: 81
Location: Pittsburgh, PA
|
|
| I agree that I should stick with the specs, but was wondering how would you know what failed? would there be intermittent problems or just not work, or wrong computations or all of the above. |
|
|
| |
|
|
|
|
|
|
Posted: May 06, 2012 - 08:53 AM |
|
Joined: Oct 05, 2006
Posts: 2242
Location: Poland
|
|
|
Quote:
but was wondering how would you know what failed?
Search for "overclocking".
The datasheet will not tell you what happens when you violate specification. |
_________________
No RSTDISBL, no fun!
|
| |
|
|
|
|
|
|
Posted: May 06, 2012 - 10:07 PM |
|
Joined: Sep 04, 2002
Posts: 21256
Location: Orlando Florida
|
|
Looks like if its an xmega, the elpm instruction takes a poop at about 2x the speced max clock. We should collaborate on a test program that would light one of the stk500 leds (or leave it lit if the avr hung up)
if failure was detected in eeprom, ram, flash, ext mem bus, timers, etc. I have no idea how to write these modules but if they ran round robin, you could stick a model X avr into the stk and keep cranking up the clk till the lights came on and publish the results. Like the seti project. |
_________________
Imagecraft compiler user
|
| |
|
|
|
|
|
|
Posted: May 07, 2012 - 03:39 PM |
|
Joined: Sep 12, 2009
Posts: 2398
Location: Sacramento, CA
|
|
| Propagation delay and transition times increase at lower voltage. Look at e.g. a 74HC04 datasheet and you'll see these times @2V are 2-5 times longer than the times @4.5V. So running too fast for a given voltage (given propagation/transition times) means you're running on the ragged edge of whether all the internal setup times are met, which will eventually mean random errors, temperature and voltage sensitivity, etc. |
|
|
| |
|
|
|
|
|
|
Posted: May 07, 2012 - 04:28 PM |
|
Joined: Nov 22, 2002
Posts: 12036
Location: Tangent, OR, USA
|
|
There may not be any one consistent failure.
With over-clocking, one common reported failure is EEPROM write at temperature extremes.
But, whether or not that is what you might see is anyone's guess.
Jim |
_________________
Jim Wagner
Oregon Research Electronics, Consulting Div.
Tangent, OR, USA
"The only thing standing between us and victory is defeat" P.G.Wodhouse in Wooster & Jeeves series
|
| |
|
|
|
|
|
FAQ
Search
Register
Log in to check your private messages
Log in
Maximize
RSS
