transfer function vi

[Luca8]

I am enquiring as to whether there have been any bugs reported with the transfer function.vi in the control toolset.

I have processed a signal using this function but the frequency response of the transfer function when implemented with this vi is completely different from what it should be. I have verified the theoretical frequency response of this transfer fucntion by using the bode plotting routine in two different software packages.

The transfer function ought to have a low pass frequency response characteristic. Instead the vi processes my signal with a netework with a passband characteristic.

I have implemented a number of other transfer functions using this vi and they respond as they should to the stimulus.

I would greatly appre
ciate any help with this issue.

[Barp]

Thank you for your interested in the PID Toolset.

The problem you are experiencing can be related to two problems:
1) The sampling time that you are specifying is too big (low frequency) for the dynamics of the system. This would generate alias and would give an unexpected output simular to band pass. Verify what is the maximum frequency of your transfer function and what is the maximum frequency of your sampling time. As a rule of thomb, this should be at least 2 times the maximum frequency but 10 times is always recommended;
2)the coefficients that you are trying to create use are very sensitive and this change could modify the transfer function and, consequently, the response of the VI;
3) There is a problem with this specific set of data/coefficients that ca
n't be handled by the VI. So, I would suggest to convert the polynomial coefficients to pole/zero/gain representation or State-Space representation and try to substitute the Transfer Function. If you have difficult to do this, you could look into this Toolkit: http://sine.ni.com/apps/we/nioc.vp?cid=13854〈=US

If nothing works, I would be happy to look into this issue in more details. However, I would need a VI that can simulate the problem with the data used to apply to the Transfer Function and the coefficients used in simulate.

[Luca8]

Hi,

Thank you for the prompt response.

With regards to your first suggestion, my input signal consists of data that has been sampled at 500 Hz and subsequentlt I have been setting dt to 0.002seconds. I have played around with the dt setting and have found that if I reduce the value of dt the characteristics of teh frequency response resembles a lowpass but the rolloff of the lowpass characteristic is much higher and the cutoff frequency is much lower than it ought to be. On top of this if I gradually increase the value of dt a large 'bump' appears in the frequency respone at the lower end of teh response. As I increase dt this 'bump' shifts to the right until I have this 'bandpass' characteristic.

With regards to the second suggestion, I have tried
altering the value of the coefficients in order to try and recreate what LabVIEW was producing but although changing teh coefficnets did change teh frequency response to some degree the shape of the frequency response is still of a lowpass characteristic and does not in any way resemble what LabVIEW produces.

I have converted the polynomials to a pole/zero/gain representation and used the appropriate function in the control toolset to simulate this transfer function, unfortunately teh results were exactly the same as I was obtaining with transfer function.vi. I have not tried the state space representation though.

Luca

[Barp]

Hi Luca8,

So, if I understand your reply, you still with the same problem you posted before? If that is case, can you send the data and coefficients to reproduce the problem?

Thank you!