High-Speed Digitizers

cancel
Showing results for 
Search instead for 
Did you mean: 

Digitizer PXI 5922 : antialiasing step response

Hi everybody,
I post a new thread because my problem is different than the previous.
 
On the digitizer 5922, there is an antialiasing filter with a bad step response for my system ! So, i can't use the 5922 card !
See FIR step response.
 
The system :
-> clk in : 10 Mhz (min : 4 mv / max : 680 mv for 50 Ohms)
-> channel 0 : 0,5 Mhz (min : -6 mv / max : 212 mv for 50 Ohms)
 
Beginning of the first measure :
See Signal w CLK 10 mhz and CH0 0,5 mhz FIR 48 tap std.jpg
 
You can see the ten first points at the first high level (delta = 20 mv).
The input signal have a 0,5 mhz frequency, so normally, there are no differences on these points (but delta = 20 mv).
I have to take care, antialiasing filter is too slow for my system.
After this problem, i tried to configure this filter : 8 tap hanning have a better step response than others.
Beginning of the second measure :
See Signal w CLK 10 mhz and CH0 0,5 mhz FIR 8 tap hanning.jpg
 
Finally , i have viewed that i could not having good points before 400 ns (same time in the specification).
My system must have at least 50 ns or 100 ns step response for having some points at about 100, 200 and 300 ns.
 
What can i do for being faster ?
 
Bye,
 

 
 


Message Edité par laulaurent le 03-19-2008 12:04 PM

Message Edité par laulaurent le 03-19-2008 12:06 PM

Message Edité par laulaurent le 03-19-2008 12:13 PM
----------------------------------------------------------------
Gendre Laurent
Stagiaire au service Détection à Thalès Alenia Space
E-mail : gendre@polytech.unice.fr
----------------------------------------------------------------
0 Kudos
Message 1 of 4
(6,938 Views)
Here is the FIR step response curve.
----------------------------------------------------------------
Gendre Laurent
Stagiaire au service Détection à Thalès Alenia Space
E-mail : gendre@polytech.unice.fr
----------------------------------------------------------------
0 Kudos
Message 2 of 4
(6,931 Views)
You might be at the limitation of the hardware. Because of the oversampling architecture of the board, there must be a low-pass filter to eliminate the quantization noise. The 8-tap Hann filter has the fastest step response of the filters for this board. The only thing you could do to speed things up would be to increase your sample rate from 10 MS/s to 15 MS/s, which would decrease the rise time by 33%. I don't suppose that would help you very much.

Is there a particular reason you're using this board for your project? I use the 5922 all the time, but mostly for frequency-domain applications, where I don't care (much) about transient response. The 5922 excels in the frequency domain but is mediocre in the time domain.

In any case, there may be a board more suited to your project. Can you share some of the details of what you're trying to do?

Cheers,
Ed L.

0 Kudos
Message 3 of 4
(6,906 Views)
HI,
Thanks for your help but 15MS/s don't help me !
See Pixel Measure.jpg and you will understand : i just need to pick up some points at 100:400 ns and 500:900 ns.

Now i'm trying to use a 5122 card because in this one there is no FIR (and 14 bits is the minimum for my application).
Maybe i will have a memory problem and i will optimize my development because of some 8 Millions points acquisitions.

Bye,




----------------------------------------------------------------
Gendre Laurent
Stagiaire au service Détection à Thalès Alenia Space
E-mail : gendre@polytech.unice.fr
----------------------------------------------------------------
0 Kudos
Message 4 of 4
(6,898 Views)