I am trying to determine if I can get a 10 MHz signal into a gumstix
board, which uses the Intel PXA255 ARM CPU. http://gumstix.com/
Not easily. The timers in the PXA255 (and others in the family) are
clocked from the (various) CPU clocks, and are trying to be timers for the
OS, not general purpose.
The PXA255 is a pretty poor choice for timing apps, too - interrupt
latency is high, unless you sneak round the side with what remains of FIQ,
and even then, it's far from deterministic. Too many DMAs, too much
caching, and too complex a memory architecture.
If you're keen on the PXA architecture for this, how about Balloon3
instead, PXA270, but with an FPGA / CPLD to do the time critical things,
leaving the CPU to deal with the high level stuff. It's all open source,
so you can use / adapt / swipe what you like.
http://www.balloonboard.org/hardware/300/index.html#images
(disclaimer - Balloon's my baby...)
If hardware acceleration in programmable logic isn't acceptable, then
divinding your 10MHz by 3 or 4 and using it as the clock input might do
what you need, but I doubt that the PLL is exactly jitter free by the time
it gets up to 500MHz. Depends what you want to do, I guess. I'd definitely
be tempted with something smaller for the time-critical operations. For a
halfway house between a CPU and programmable logic, I really like Scenix
devices - PIC architecture (eww...) but clock at up to 75MHz, and retire
an instruction per clock.
http://www.parallax.com/detail.asp?product_id=SX20AC/SS-G
Completely deterministic, free and usable tools. I use them all over the
place.
Having de-lurked, I'd just like to say what an excellent place this is. I
care more about frequency than time, but obsessive engineering of any kind
makes me smile :)
I'm currently getting my fix from a TrueTime GPS XL-AK, and a new (to me)
PRS10.
The plan for this week is to run the 1PPS from the GPS to the PRS10 - the
"How do I know my GPS stabilized oscillator is working?" thread was well
timed. (sorry)
If anyone knows of any snags in my simple plan, feel free to point out the
obvious - this is all new to me.
<relurk>
Steve
On Mon, 14 Aug 2006 15:29:50 -0400, Glenn <glenn@net127.com> wrote:
>
> I am trying to determine if I can get a 10 MHz signal into a gumstix
> board, which uses the Intel PXA255 ARM CPU. http://gumstix.com/
Not easily. The timers in the PXA255 (and others in the family) are
clocked from the (various) CPU clocks, and are trying to be timers for the
OS, not general purpose.
The PXA255 is a pretty poor choice for timing apps, too - interrupt
latency is high, unless you sneak round the side with what remains of FIQ,
and even then, it's far from deterministic. Too many DMAs, too much
caching, and too complex a memory architecture.
If you're keen on the PXA architecture for this, how about Balloon3
instead, PXA270, but with an FPGA / CPLD to do the time critical things,
leaving the CPU to deal with the high level stuff. It's all open source,
so you can use / adapt / swipe what you like.
http://www.balloonboard.org/hardware/300/index.html#images
(disclaimer - Balloon's my baby...)
If hardware acceleration in programmable logic isn't acceptable, then
divinding your 10MHz by 3 or 4 and using it as the clock input might do
what you need, but I doubt that the PLL is exactly jitter free by the time
it gets up to 500MHz. Depends what you want to do, I guess. I'd definitely
be tempted with something smaller for the time-critical operations. For a
halfway house between a CPU and programmable logic, I really like Scenix
devices - PIC architecture (eww...) but clock at up to 75MHz, and retire
an instruction per clock.
http://www.parallax.com/detail.asp?product_id=SX20AC/SS-G
Completely deterministic, free and usable tools. I use them all over the
place.
Having de-lurked, I'd just like to say what an excellent place this is. I
care more about frequency than time, but obsessive engineering of any kind
makes me smile :)
I'm currently getting my fix from a TrueTime GPS XL-AK, and a new (to me)
PRS10.
The plan for this week is to run the 1PPS from the GPS to the PRS10 - the
"How do I know my GPS stabilized oscillator is working?" thread was well
timed. (sorry)
If anyone knows of any snags in my simple plan, feel free to point out the
obvious - this is all new to me.
<relurk>
Steve
