Multichannel Frequency Response Function Block?

[marc lawrence]

We are currently using the FRF block to evaluate our DUT with two channels (X and Y). However our DUT has 8 output channels. Ideally I would like to be able to wire up the existing FRF block with multi-channel support (ie 1x X Signal, n x Y Signal) and get the new block to correctly perform averaging. Currently if I wire up the FRF block in a loop it averages CH1, CH2, CH3 etc which obviously is the wrong result.

One solution was to have 8 FRF blocks - but this seemed a bit of an overkill. Is there a n Channel FRF block anywhere - or has someone modified it for n Channels? (if not, I suppose I will have to do the modifications!).

Many thanks,

Marc

[DSPGuy]

Hi Marc,

You asked if there was an N channel FRF block anywhere.  LabVIEW 8.0 added multichannel support for the FRF.  Modes include 1xN, Nx1, and NxM, where NxM can be interpreted as a pairwise FRF (first element of X with first element of Y, ...) or as all the combinations of X and Y array elements.

Barring an upgrade to LabVIEW 8.0, the simplest solution is probably your approach of placing N FRF VIs on a diagram.  This ensures that the averaging is the same as the 1x1 case with little programming effort. 

For the 1XN case it should be fairly simple to export the averaging state and manage that yourself.  If you use one of the other measurement VIs as a starting point (e.g. FFT Spectrum (mag-phase).vi) please note that the state cluster that is used is intended to cover the general case where each "channel" could have a different number of points, different sampling rate,  different timestamp...  This is the reason for the array-of-clusters-of-arrays approach.  If you make a simplifying assumption that each channel will have the same number of points, same timestamp, etc. then you could utilize a much simpler state cluster that contains some 2-D arrays to hold the multi-channel averaging information. 

Hope this helps.

-Jim

[marc lawrence]

Hi Jim,

Yes it does help - ie it means I have to write the block myself as we dont want to upgrade to LV 8 just yet... unless someone can send me the LV8 block and save it was a LV 7.1 version (I guess they did not use a CIN node and so it is just more block structure....)

Or I write it myself using a couple of averaging cycles from the FFT (Phase / Mag) blocks and do my own averaging... I just wanted a quick and easy solution ;-0)

[leocavadas]

I work in LV 7.1 and have same problem. I look for solution by structure case with 7 states (1x1,1x2,1x3,1x4,1x5,1x6,1x7 FRF functions) linked to number of channels in activity, but i dont like this solution...I try develop other solution at the moment.