max atmel speed

Go To Last Post
21 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello guys.
I'm wondering. The data sheet of the atmegas tell thatthe maximum speed is 16 mhz.
But at the baudrate section you see a table for 20 mhz.
And some sites sell development board with 20 mhz cristal.

So my question is does the atmega work stable on 20 mhz, any one try it.

And a silly question can the atmega run at 100mhz cooled with liquid nitro or something.

Thanks

Still Pink

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Pinkpanter wrote:
The data sheet of the atmegas tell thatthe maximum speed is 16 mhz.

There's not one data sheet for all AVR models. There is one data sheet for each AVR model. Some will run as fast as 16 MHz. Some will, IIRC, run at 20 MHz.

Tell us what data sheet You are reading and the discussion will be more meaningful.

As of January 15, 2018, Site fix-up work has begun! Now do your part and report any bugs or deficiencies here

No guarantees, but if we don't report problems they won't get much of  a chance to be fixed! Details/discussions at link given just above.

 

"Some questions have no answers."[C Baird] "There comes a point where the spoon-feeding has to stop and the independent thinking has to start." [C Lawson] "There are always ways to disagree, without being disagreeable."[E Weddington] "Words represent concepts. Use the wrong words, communicate the wrong concept." [J Morin] "Persistence only goes so far if you set yourself up for failure." [Kartman]

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
And a silly question can the atmega run at 100mhz cooled with liquid nitro or something.

I don't think that the speed is limited by temperature considerations, rather by FLASH access times, but maybe sombody knows better?

Four legs good, two legs bad, three legs stable.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It depends strongly on the model. Some of the low-voltage versions are, I think, spec'd at 8MHz. I seem to recall one or more that are actually spec'd at 24MHz though I think that is the highest for the family. 16 is pretty generic for 5V AVR operation.

Generally, as processes "improve" (that is, feature size is made smaller), the clock rate goes up and the power consumption goes down. Thus, you will generally see the newer ones spec'd at higher clock rates than the older ones.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The spec is a value that all of that model will operate at. Some individual chips within the model can operate higher, and can be overclocked. Use of EEProm used to be one of the limiting factors. You will only find out by trying.

Laurence Boyd II

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think that the specs give the safe oparation margins for the whole production family line.

I may push the hardware beyond the specs' limmits, but I will not be able to tell for sure when each of the periferals will follow the overdriving, or will stop responding, due to the manufacturing differences. No two chips are exactly alike.

I hope for nothing; I fear nothing; I am free. (Nikos Kazantzakis)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Atmel sent out a flyer a couple years ago announcing all the new megas at 24MHz. This was later revised to 20MHz. Evidently the flash is the bottleneck. Evidence: the usb with the avr core boots a flash rom into ram and runs at 48mhz out of ram. Why they dont offer this as a regular avr option who knows? You could bit bang ethernet, usb, do dsp stuff... holy cow what a screamer.

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Bob. I think you are right.
I still have in my hands one of the first AT90S1200s, rated at 16MHz while the S2313s were rated at 10MHz and the S8515 at 8MHz.
I treasure this chip, hopping than it will become very valuable at the near future!

I hope for nothing; I fear nothing; I am free. (Nikos Kazantzakis)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Giorgos_K wrote:
I treasure this chip, hopping than it will become very valuable at the near future!

I seriously doubt it, although I hope it does. I have one too.
And I also have a couple of the first 2313's, which were 12 MHz.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I spoke with some Atmel factory people about speed. They sed the process was good to 24 MHz, but their production testing was limited to 16 or 20 MHz. At least for newer i.e. mega chips, try 'em at 24 MHz and see!

Tim Ressel
Portland, OR
timr@earthling.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think what the OP was refering to was this:

On page one of the mega128 data sheet at the bottom is says "16Mhz"
On page 197 of the same data sheet it has a UART setting chart for 20MHz.

I've always wondered about this.

Go electric!
Happy electric car owner / builder

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Things like the UART setting chart are generic. They are written to be used with any of the AVRs at any speed. Thus, the fact that a setting for 20MHz clock on a 16MHz device should NOT be used to conclude that the part runs at 20MHz!

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

For Pinkpanter (and anyone else who cares) here's my experience on overclocking AVRs.

ATmega16/32 (L versions too): 18.432MHz rock solid, 20MHz often works.
ATmega128 (L version too): 18.432MHz rock solid, 20MHz problematic.
ATmega168 (didnt try L vers): 22.11MHz rock solid, 24MHz works great as far as I can tell.

BTW, be sure to turn on the CLKOPT fuse bit, or have a solid noise-free external clock source.

I expected problems to show up when running non-L-version processors at 3.3V AND overclocking them. But amazingly, no problem at all. Example:

ATmega128 (not L ver) at 3.3V and 18.432MHz works fine.

Flash access is indeed the rate-limiting step. Atmel's IP division advertises their AVR "core" at beyond 100MHz! Makes me wish they'd put in a flash accelerator aka wider access bus (just like on the 60MHz Flash-based Philips LPC21xx ARM processors).

Cheers,
-Pascal

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Atmel's IP division advertises their AVR "core" at beyond 100MHz!

So, if one used external high speed ROM for code, I wonder what the speed limitation would be?

You can avoid reality, for a while.  But you can't avoid the consequences of reality! - C.W. Livingston

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ya know, there have been repeated threads on the "quest for speed". But we have seen very, very few mentions, much less detailed app discussions, of the AT94 family.

For all the apps I've done so far, I've yet to run out of gas at 8MHz. But I'm just a simple bit jockey, drawn to the AVR for the near-single-chip microcontroller solutions that can be developed.

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'd like to see SRAM versions myself. I'd take the hit of pageing in a kb of flash into SRAM and then executing it from ram. But I don't use very large chunks of code either. (Hobbyist ASM'er, I haven't even run across a need from SRAM yet) I just think it would be interesting from a digital signal perspective, as the sampling rate in and out could be so dramatically higher.

-Curiosity may have killed the cat
-But that's why they have nine lives

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

bobgardner wrote:
Evidence: the usb with the avr core boots a flash rom into ram and runs at 48mhz out of ram. Why they dont offer this as a regular avr option who knows? .

Note that that can be a fairly expensive proposition, compared to a normal AVR.. They could easily be selling these at a mild loss, simply to have a competing USB product of this type out there. While the option would be nice, it might not make much sense on a corporate level.

Heck it could even take a bite out of the low end ARM category for speed, which certainly wouldn't make sense for them..

Alan

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What did Scenix know about flash that allowed the sx (pic clone w/pipeline) to run at 100 MHz (10 ns instruction cycle)? (Dropped because of heat problems, now at 75 MHz.) It can be done.

Rick

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

[busily pulling out Atmel SAM7S datasheet ...]

Quote:
–Fast access time, 30 MHz single-cycle access in Worst Case conditions

So Atmel >>does<< know how to make it at least somewhat faster than current AVR models.

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@theusch and RickB

To get to higher speeds, many manufacturers simply make their internal FLASH wider, as in 32, 64, or even 128 bit wide data bus. For example, the LPC21xx ARM processors fetch instructions from FLASH in single 128bit reads. Sure, each fetch may take 50us (20MHz), but each fetch "contains" four 32-bit ARM instructions. So the processor can run at 60MIPS+ out of FLASH (with full 32-bit ARM instructions) even though the FLASH could not support this speed if it were a 32-bit device.

Of course, there's a price. A 128-bit access bus is costly in die space.

@theusch
Good point about the 30MHz Atmel FLASH access speed. Maybe we'll see some 30MHz AVRs in the future.

Of course, this isn't far from the 24MHz that many AVRs achieve today. Any of the ATmegaXX8 that I've tried seem to run well at 24MHz, although in my products, I run them at only 22.11MHz.

-Pascal

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

@theusch
Good point about the 30MHz Atmel FLASH access speed. Maybe we'll see some 30MHz AVRs in the future.

But your point about wider flash is well taken, too. I might have read something that the "no wait states" at full speed in Thum mode is due to something like you said--maybe 32-bit accesses & double-buffered. So it could be really the same flash with the same speed capabilities as we have in AVRs.

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.