Phase error in scxi with TBX-1316

[Luiz Felipe]

Dear All,

We are facing a problem of phase error between signals with a SCXI system using TBX 1316 (HV attenuator). Here is a description of our system:

-SCXI 1600

-SCXI 1120/D to measure current signals;

-SCXI 1120 connected to a TBX 1316 through a SH-3232A cable, to measure 200 V signals

-LabView 8.2.

The problem is that, whatever are the signals we measure, a constant phase error of about 6 degrees (in 60 Hz) is introduced.

We did the following tests to find where and what the problem is:

1) Used a signal generator to input the same signal to both the 1120/D and 1120 modules WITHOUT the TBX and the cable connected. There was no phase error.

2) Used a power source to input a 120 V signal to the TBX WITHOUT the 1120 module connected to it. An oscilloscope was used to measure the input signal, the output signal of the TBX and the output signal in the open end of the SH-3232A cable.  There was no phase error.

3) Used a power source to input a 120 V signal to the TBX WITH the 1120 module connected to it.  An oscilloscope was used to measure the input signal and the output signal in the end of the SH-3232A cable connected to the 1120 module.  There was a phase error of about 6 degrees (at 60 Hz).

4) Used a power source to input a 120 V signal to the TBX WITH the 1120 module connected to it. A probe was used to connect the same 120 V signal to the 1120/D. The phase difference was both measured by software (with a vi) and with an oscilloscope.  There was a phase error of about 6 degrees (at 60 Hz).

Can anyone help us to find what is going wrong and eventually fix the problem?

Thank you!

Luiz Felipe Willcox

[Schneiater]

Hi Luiz Felipe,

First, I would like you to answer me some questions to clarify the scenario.

Do you have other terminal block? Or are you connecting the signals directly to the SCXI 1120/D?

Then, a test that I would like you to do, as you have a signal generator, is try to make some tests in different frequencies. Please try to measure signals at 400Hz and tell me if the phase difference between the signals change.

Thanks,

Regards,

[Luiz Felipe]

Hi Rodrigo Schneiater.

Thank you for answer me. I'll clarify the scenario for you.

Beyond the hardware that I had listed before, we have a scxi 1305 terminal block for current signal input.

We did some tests, in a frequency range between 60 up to 400Hz, like you sugested me. We used a current clamp, that have a constant phase error for all range of the test (comproved by oscilloscope measurements). This is the result for, discounted the error by the current clamp (3°):

60Hz - phase shift of 6.40° between voltage and current signal;

120Hz - phase shift of 11.15°;

180Hz - phase shift of 16.50°;

240Hz - phase shift of 21.66°;

300Hz - phase shift of 26.40°;

360Hz - phase shift of 30.70°;

400Hz - phase shift of 33.30°;

Thank you for help us.

[Henrik_Volkers]

Luiz,

if you calculate the group delay  you will find that one channel is about 250µs delayed. You didn't sample with 4kHz, did you ? 😉  

delay= P/(360*f)    (P=phase in deg, f =frequency)

I don't know how you measure the phase (I prefer SAM with a sinus), however the jump at 60Hz is normally due to line noise 😉 and the drift I assume is the current clamp.....

 

Message Edited by Henrik Volkers on 02-19-2010 12:53 PM

[Henrik_Volkers]

Output impedance of the TBX??

Input impedance of the scope ?

Input impedance of  1120?

Here is the clue.

Sorry, no time to investigate further , however I would measure the delay , proof that it's constant enough for my needs and compensate in software.

[Schneiater]

Hi Luiz Felipe,

The test you did confirms what I was thinking. The 1316 has a capacitive component and when using with a 1120 has a frequency response of 400Hz. But this frequency response is for gain (-3dB). Generally, when a circuit goes down to 3dB in gain its phase has already gone much lower and it is what we saw in your tests.

Also, something that can bring a phase difference is the fact that both 1120 are multiplexed to the same DAQ card, so you have a phase difference by this multiplexing delay.

All of this is the normal behavior of this hardware.

What I suggest is that you make a compensation by software of this phase difference. You can use this test you did to make a phase response that will be used to compensate the measurement.

I hope this has cleared the things.

If you need more help, feel free to ask.

Regards,

[Luiz Felipe]

Henrik and Rodrigo: thank you for your replies.

Both of you mentioned the group delay due to multiplexed operation of the SCXI-1600. My first message was incomplete. When we did test no. 1 (signal generator + voltage in SCXI 1120, no TBX 1316 + current in 1120/D), we said that there was no error. Actually, there was an error of 0,45 degrees due to multiplexed operation. We are sampling at 20 kHz per channel, what gives 0,15 degrees between a channel and its predecessor. Our current channel is number 0, while our voltage channel is number 4, what gives the 0,45 degrees.

Henrik mentioned output impedance of TBX 1316 and Rodrigo confirmed what we also suspected: it is a problem of impedance coupling between 1316 and 1120/D. We expected a plain frequency response up to 400 Hz for both gain and phase.

The application we are working on is for power measurement, so we need only to measure voltage signals of about 60 Hz. The correction of the voltage phase error will probably solve our problem.

In our application, we work in frequency domain and compensate the phase error by adding back the 6 degrees difference. However, we would prefer to do this software correction in time domain. Can you give me any suggestion on how to do it? Is there any function or vi in LabVIEW that will introduce the a time delay in a signal?

Regards,

Luiz Felipe

Cepel

Rio de Janeiro, Brasil

[Schneiater]

Luiz Felipe,

If you are using always the same configuration for the Data aquisition (number of samples and rate) the phase difference will be a time difference that can be thought as an array offset. Something that apears in the first sample in one channel will appear after some samples. This number of samples is your phase difference. You can make an array subset to this samples in one channel and an array subset in the end of the other channel. This way, you will have fewer samples but phased samples.

Did I make myself clear? 

If you have problems, make me know.

Regards,

[Luiz Felipe]

It is very clear, Rodrigo. In fact, that was our idea, but we just wanted to be sure that there was no simpler way of doing this. We will unbundle the signals from each channel from the array in the output of the N samples window acquired. We will delay the current channels and bundle again the signals in a single array. By doing this, we expect that no further changes will be necessary in the rest of the vi.

I'll inform you of the result of this implementation.

Once gain, thank you for your help.

Regards,

Luiz Felipe

Cepel

Rio de Janeiro, Brasil