FPGA Central

Hi all,
I have an application where my data refreshes every 10
seconds...It takes so long as I am doing some sort of an averaging
over a few million samples and then taking the statistics once in
those few million samples.My system clock is of 40 MHz and it's fed
from outside through a SMA connector in Virtex2P board. Hence to get
16 meaningful samples I should have to wait around 160 seconds.
These are the methods I tried....
option 1: I used a clock for Chipscope pro which has a period of 10
seconds....But in this case it tkes a huge time to get the
data...around 600 sec for getting 16 samples...even though the values
make sence
Option 2: I am using the system clock iteslf for chipscope clock, and
once in 1 seconds I generate a high pulse of width of 1 period of sys.
clock and trigger data with that...but it takes all the data points
after the trigger condition and hence I effectively get 1 data
point.It's fast but doesn't help much

Is there any method by which I can take the data values only at the
high values of the pulse and hence all data point will be different.

Thanks in advance,
Pratap

"Pratap" <[email protected]> wrote in message
news:[email protected]...
> These are the methods I tried....
> option 1: I used a clock for Chipscope pro which has a period of 10
> seconds....But in this case it tkes a huge time to get the
> data...around 600 sec for getting 16 samples...even though the values
> make sence
> Option 2: I am using the system clock iteslf for chipscope clock, and
> once in 1 seconds I generate a high pulse of width of 1 period of sys.
> clock and trigger data with that...but it takes all the data points
> after the trigger condition and hence I effectively get 1 data
> point.It's fast but doesn't help much

The ILA works like a logic analyzer; there is a "trigger" and then on each
clock the data is sampled until the buffer is full. My guess is that in
constructing the ILA core, 16 samples is the minimum buffer length (I've
generally used many more). I think you would need to use some form of
register I/O off the board to instead sample the data yourself to do any
better.

Good Luck,
Marty

Pratap wrote:
> Hi all,
[snip]

> Is there any method by which I can take the data values only at the
> high values of the pulse and hence all data point will be different.
I don't have ChipScope in front of me, so this is from memory. The
trigger settings allow you to specify a number of 'windows' (I think
that's what they're called) and the number of samples per window. If
your buffer size is 1024 and you choose 512 windows, then each window
would have 2 samples. You would set the trigger to the pulse signal, and
then Chipscope would show you 512 consecutive triggers.


---
Joe Samson
Pixel Velocity

On Sat, 10 May 2008 13:32:07 -0700 (PDT), Pratap <[email protected]>
wrote:

>Hi all,
> I have an application where my data refreshes every 10
>seconds...

>Option 2: I am using the system clock iteslf for chipscope clock, and
>once in 1 seconds I generate a high pulse of width of 1 period of sys.
>clock and trigger data with that...but it takes all the data points
>after the trigger condition and hence I effectively get 1 data
>point.It's fast but doesn't help much
>
>Is there any method by which I can take the data values only at the
>high values of the pulse and hence all data point will be different.
>

You can add that pulse to the data, and use it as a qualifier, so that
only data samples matching that qualifier are stored.

- Brian

On May 10, 6:20 pm, Joseph Samson <u...@not.my.company> wrote:
> Pratap wrote:
> > Hi all,
>
> [snip]
>
> > Is there any method by which I can take the data values only at the
> > high values of the pulse and hence all data point will be different.
>
> I don't have ChipScope in front of me, so this is from memory. The
> trigger settings allow you to specify a number of 'windows' (I think
> that's what they're called) and the number of samples per window. If
> your buffer size is 1024 and you choose 512 windows, then each window
> would have 2 samples. You would set the trigger to the pulse signal, and
> then Chipscope would show you 512 consecutive triggers.
>
> ---
> Joe Samson
> Pixel Velocity

Yes,
The windows method worked....Thanks for the help...
Now on each window I am taking 1 sample and I am taking 512
windows...which solves my purpose.
Thanks again...