Problem with oversampling with the Pico3424 units

Forum for discussing PicoScope (version 5)
Post Reply
dmalan
User
User
Posts: 3
Joined: Tue Nov 21, 2006 9:21 am

Problem with oversampling with the Pico3424 units

Post by dmalan »

Hi,
I am busy writing an application for the Pico3424 in VB.NET 2005. All is going well but I ran into a problem trying to use the oversampling feature.
If I understand the manual correctly, the unit is supposed to take more than 1 sample per sample period (2 for an oversampling factor of 2) and then calculate and return the average of these samples.
This implies that for a particular measurement cycle, the total amount of samples taken for various oversampling factors should remain the same and therefore also the total time for the cycle event.
When I run my code using an oversampling factor of 2 for example, the amount of data points remain the same (as for oversampling at a factor of 1), but the total event time doubles. Does this mean that the effective sampling rate has halved? ( or doubled????)

I would also like to know what the noise spec is for the PICO3424 units. (How many of the 12 bits are "jumping" with the inputs to the channels shorted?)


As a matter of interest:
I have read many posts on this forum from people asking about using .NET as a programming tool for the PICO products. I had the same questions before I started with my project. I can safely say now that it is definately possible to call all of the dll. functions successfully from VB.NET. The secret is just to make very sure that you define the parameters that you pass to the dll to be exactly the same (the same amount of bytes) as that in the prototypes documented in the user manual. The main thing to remember is that a "long" type becomes and "integer" type in VB.NET 2005. Also, a "String" type should be used for "char" and be passed "byval", not "byref".
I hope this settles some queries.

Regards,
Niel.

Michael
Advanced User
Advanced User
Posts: 656
Joined: Thu Jul 07, 2005 12:41 pm
Location: St Neots, Cambridgeshire

Post by Michael »

Hello,

Thank you for your post.

Oversampling must be understood to implement it correctly. Your oversample factor of two will half the sampling rate and buffer memory and increase the effective resolution by 0.5 bits. Oversample factor 4 will quarter the sampling rate and buffer memory and increase the effective resolution by 1 bit. Noise can be attributed to the LSB.

Good luck with your .Net wrappers.

Best regards,
Michael
Michael - Tech Support
Pico Technology
Web Support Forum

dmalan
User
User
Posts: 3
Joined: Tue Nov 21, 2006 9:21 am

Oversampling theory

Post by dmalan »

Hi,

Thanks for the reply.
I think I still may have a problem understanding the theory :oops: :
What is the difference between sampling at say 1000ns per sample taking 1000 sample points with an oversampling factor of 2, and sampling at 500ns per sample taking 500 samples with an oversampling factor of 1?
If there is no difference, why bother with oversampling?

How about the noise specs: I would realy like to know what is typical for the average Pico3424 unit. (How many of the lsb's toggle when the inputs are shorted?)

Regards,
Niel.

Post Reply