AnsweredAssumed Answered

Path delay

Question asked by H.ELGALA on Mar 3, 2006
Latest reply on Mar 3, 2006 by H.ELGALA

Like • Show 0 Likes0
Comment • 5

I would like to determine the path delay between modulator and demodulator in order to set the correct Phase reference in the demodulator using the delay_path() function. In the documentation it says that I can use two TimedSink components to specify the start and end path, but when I do that I get the error:

Function "delay_path" : This function is only valid for 1-Dimensional Data.

But T1 and T2 time sink components are by definition Voltage vs Time (2-dimensional). Probably the documentation is correct, but I don't understand what I am doing wrong. I would appreciate your help again!

No one else has this question

Outcomes

5 Replies

kkladopoulos Mar 3, 2006 3:19 AM
Mark Correct
Correct Answer
I have never used the delay_path() function so I do not know how it works.

Have you tried the MeasuringDelay design under the BER_Validation_prj example project?
Like • Show 0 Likes0
Actions
kkladopoulos Mar 3, 2006 4:27 AM
Mark Correct
Correct Answer
I also tried the path_delay() function and it worked for me for data generated from a simulation that does not have any sweeps.

When I added a sweep in my simulation I saw the same error message you mentioned
Function "delay_path" : This function is only valid for 1-Dimensional Data.

The error message is a little misleading. When it mentions 1-Dimensional data it really means data with only 1 independent variable. A sweep creates additional independent variables.

For swept data you can use the following two equations to fix the value of the independent variable the sweep creates and so reduce the dimensionality of the data you pass to path_delay(0 by 1:

sweep_idx = 0
x = delay_path(T2[sweep_idx,::], T1[sweep_idx,::])

Plot x[0] and x[1] in a table.

Then manually change sweep_idx to 1, 2, 3 to see the values of x for other values of your swept variable.
Like • Show 0 Likes0
Actions
H.ELGALA Mar 3, 2006 5:17 AM
Mark Correct
Correct Answer
I understand now! I measured the delay with cross correlation and delay_path() and I get very different results. But I am pretty sure that delay_path() measures something else than what I want. The correlation result makes more sense.
Thank you again for the reply.
Like • Show 0 Likes0
Actions
kkladopoulos Mar 3, 2006 5:21 AM
Mark Correct
Correct Answer
If you used the Ptolemy CrossCorr model (used in the MeasuringDelay example design) then it will give you the estimated delay in number of samples, so the actual Delay is (number of samples) * TStep. The delay_path() function returns the delay in sec.
Like • Show 0 Likes0
Actions
H.ELGALA Mar 3, 2006 6:15 AM
Mark Correct
Correct Answer
..I knew that and I even checked on a different design to see whether I'm doing something wrong.
I measured the delay between QPSK_MOD and QPSK_DEM with the cross correlation and with the delay_path() in the QPSK BER Validation design.
With cross correlation I get the exact number of time steps delay: 2 TStep (interpolator) + 5 TStep (ChannelDelay) = 7 TStep
TStep = 1.0E-7
With path_delay(), I get a delay of 1.5E-6.
I guess that it measures some other kind of time delay.
Like • Show 0 Likes0
Actions

Path delay

Outcomes

Related Content

Recommended Content