ATXmega16A4U runs happily at 64 Mhz

Although I don't believe in shipping products
with parts pushed past their datasheet specs,
I'm not above playing around.  After a day of
figuring out the clock and pll on the above
chip and getting the pll to run at 64 Mhz, I
thought "what the hell" and changed the cpu
divider from 2 to 1.  The part runs just fine
and a scope verified that it is indeed running
at 64 Mhz.  Not bad for a $1.50 part.

Thought someone might like to know.

"Jim Stewart" <jstewart@jkmicro.com> wrote in message 
news:ke98j4$rba$1@dont-email.me...
> Although I don't believe in shipping products
> with parts pushed past their datasheet specs,
> I'm not above playing around.  After a day of
> figuring out the clock and pll on the above
> chip and getting the pll to run at 64 Mhz, I
> thought "what the hell" and changed the cpu
> divider from 2 to 1.  The part runs just fine
> and a scope verified that it is indeed running
> at 64 Mhz.  Not bad for a $1.50 part.

Quite often, modern small MCUs could be far overclocked.  I've run 25Mhz 
rated 68HC12 at 70 MHz; no problem.

Vladimir Vassilevsky
DSP and Mixed Signal Consultant
www.abvolt.com

On 29/01/13 20:40, Jim Stewart wrote:
> Although I don't believe in shipping products
> with parts pushed past their datasheet specs,
> I'm not above playing around.  After a day of
> figuring out the clock and pll on the above
> chip and getting the pll to run at 64 Mhz, I
> thought "what the hell" and changed the cpu
> divider from 2 to 1.  The part runs just fine
> and a scope verified that it is indeed running
> at 64 Mhz.  Not bad for a $1.50 part.
> 
> Thought someone might like to know.

I've often used overclocking as part of reliability testing - if your
board works at a higher speed, then you know your timing margins are good.

But I don't think I've overclocked /that/ much - aren't these chips
spec'ed to just 20 MHz?  It doesn't surprise me that the cpu core is
happy at that speed, or the other digital parts (timers, uarts, etc.),
but reading from the flash could easily be a problem.

David Brown <david@westcontrol.removethisbit.com> writes:

> On 29/01/13 20:40, Jim Stewart wrote:
>> Although I don't believe in shipping products
>> with parts pushed past their datasheet specs,
>> I'm not above playing around.  After a day of
>> figuring out the clock and pll on the above
>> chip and getting the pll to run at 64 Mhz, I
>> thought "what the hell" and changed the cpu
>> divider from 2 to 1.  The part runs just fine
>> and a scope verified that it is indeed running
>> at 64 Mhz.  Not bad for a $1.50 part.
>> 
>> Thought someone might like to know.
>
> I've often used overclocking as part of reliability testing - if your
> board works at a higher speed, then you know your timing margins are good.
>
> But I don't think I've overclocked /that/ much - aren't these chips
> spec'ed to just 20 MHz?  It doesn't surprise me that the cpu core is
> happy at that speed, or the other digital parts (timers, uarts, etc.),
> but reading from the flash could easily be a problem.

Ha ha, this is fun. I just overclocked my STM32F4-Discovery to 250MHz
(it is already a screaming fast little micro at 168MHz).

Now it can toggle a port pin at 125MHz...

-- 

John Devereux

David Brown <david@westcontrol.removethisbit.com> writes:
> But I don't think I've overclocked /that/ much - aren't these chips
> spec'ed to just 20 MHz?  

32 MHz.  See: http://www.atmel.com/devices/ATXMEGA16A4U.aspx

David Brown wrote:
> On 29/01/13 20:40, Jim Stewart wrote:
>> Although I don't believe in shipping products
>> with parts pushed past their datasheet specs,
>> I'm not above playing around.  After a day of
>> figuring out the clock and pll on the above
>> chip and getting the pll to run at 64 Mhz, I
>> thought "what the hell" and changed the cpu
>> divider from 2 to 1.  The part runs just fine
>> and a scope verified that it is indeed running
>> at 64 Mhz.  Not bad for a $1.50 part.
>>
>> Thought someone might like to know.
>
> I've often used overclocking as part of reliability testing - if your
> board works at a higher speed, then you know your timing margins are good.
>
> But I don't think I've overclocked /that/ much - aren't these chips
> spec'ed to just 20 MHz?  It doesn't surprise me that the cpu core is
> happy at that speed, or the other digital parts (timers, uarts, etc.),
> but reading from the flash could easily be a problem.

32 Mhz part.  It has a nifty PLL clock
generator that lets you use a reasonable
16 Mhz ceramic resonator and multiply it
up to as high as 128 Mhz for the counter/
timer section and divide it down for the
CPU.  This tells me they are using a process
good for 128 Mhz so I went back and ran
the CPU on 128 Mhz and it worked fine there
too.   I suspect the CPU clock rate is more
a function of test expense and product
positioning than that of operational capability.

Jim Stewart <jstewart@jkmicro.com> wrote:

> 32 Mhz part.  It has a nifty PLL clock
> generator that lets you use a reasonable
> 16 Mhz ceramic resonator and multiply it
> up to as high as 128 Mhz for the counter/
> timer section and divide it down for the
> CPU.  This tells me they are using a process
> good for 128 Mhz so I went back and ran
> the CPU on 128 Mhz and it worked fine there
> too.

Did you try testing outside normal room temperatures?

-a

Anders.Montonen@kapsi.spam.stop.fi.invalid wrote:
> Jim Stewart<jstewart@jkmicro.com>  wrote:
>
>> 32 Mhz part.  It has a nifty PLL clock
>> generator that lets you use a reasonable
>> 16 Mhz ceramic resonator and multiply it
>> up to as high as 128 Mhz for the counter/
>> timer section and divide it down for the
>> CPU.  This tells me they are using a process
>> good for 128 Mhz so I went back and ran
>> the CPU on 128 Mhz and it worked fine there
>> too.
>
> Did you try testing outside normal room temperatures?

I don't have the time right now.  It's on
my "playtime" list.