11-26-2011 04:13 PM
Thanks for the additional test data. I think I see what is happening.
1) the shape of your stimilus remains a square pulse. This is essentially a high frequency and is likely stimulating the resonance frequency of your transducer, producing the impulse like response in the signal. Can you provide a sinusoid stimilus?
2) the AC coupling circuit of the 9233 removes DC or steady state signals. This is why the green trace measured by the 9233 returns to zero during the "flat" portion of your pressure curve. the 9233 measures dynamics and the AC coupling circuit filters out anything below 0.5Hz in frequency.
I think you will find a much more realistic response if you use a sinusoid stimilus.
11-27-2011 05:06 AM
I tend to have an opposite opinion. In sensor specification they give resonant frequency 250 kHz. I open the ball valve manually to discharge air - I doubt that I can reach even a few percent of the resonant frequency by manual opening of the valve.
The final sensitivity of the couple: sensor/amplifier is around 17 mV/bar I noticed that as slower is the air discharge the farer is the resulting voltage from the expectation. Now I try to deduct how fast should I discharge the air to reach the maker's sensitivity.
My real signals from the engine combustion has much higher frequencies than 0.5 Hz and is distorted as well.
11-27-2011 10:45 AM
there could be a bad channel on the 9233, have you tried other input channels on the 9233?
I think it would still be nice to see say a 100hz or 1kHz sinusoidal stimulus to see of the sinusoidal pattern appears in the measured response of the 9233.
11-27-2011 11:56 AM
I use 6 sets of sensors for 6 cylinders on the engine. All behave in the same way, so there is no failure to one channel. Especially that I use 2 modules and tested all channels.
It is very difficult to generate sinusoidal signal of high pressure. especially with such high frequency as you suggest. But I have peak-like signals with frequency of 10-15 Hz. That was generated inside engine cylinder.
I attached a record from the engine in TDMS file. there is digital and analog part of the measurement. In analog part channels 0-5 represent signal read form 6 cylinders pressures.
The shape of pressure course is almost correct (however not exactly), but if I use factory sensitivity to recalculate voltage into pressures the results are to low. I was expecting max pressure 130 bar +/- 2% on each cylinder, and I got pressures varying from 55 to 83 bar, that is much to low. the shape of the pressure cours is also distorted especially when the pressure decreases.
11-27-2011 06:43 PM
the shapes look promising. I see 0.2 to 1.2 V peak to peak on some of the sensors. There is some noise.
Do you have the TDMS file, instead of the XLS? I can easily load it into DIAdem.
11-28-2011 12:48 AM
For the 4705M5 in-line charge amplifier, sensitivity = 1 mV/pC. How do we get to 20 mV/psi?
From the data sets provided, the AC coupling appears to be largely distorting the test signals and the engine signals. You will need a data acquisition device that can power IEPE and effectively acquire the sensor output signal including the significant output bias voltage of the sensor/charge amplifier.
11-28-2011 05:54 AM
Reply for Preston:
I have TDMS file, I attached it zipped as this forum do not accept files with TDMS extension.
Reply for dbs@NI:
In my initial post I made mistake giving the amplifier sensitivity, I gave it as I remembered it but that was wrong. The true sensitivity is 1mV/pC as in datasheet.
I do not understand your second part of post, can you give me some more details what do you mean, it seems to be important.
11-28-2011 09:40 AM
AC coupling distorts a pulse train due to the significant DC component (mean). The input signal has peak value ~1, but the output signal has peak-to-peak level ~1 but shifted such that mean ~0.
11-28-2011 09:52 AM
Probably I start understanding. Because in long term my signal has also a constant (DC) component and the 9233 can operate only with AC signals, there is distortion of my signal. My module 9233 is not capable to process correctly my signal.
Can you suggest any solution? I checked available NI modules and found module 9234. there is software selectable AC/DC coupling, will that be helpful to utilize this module with DC coupling selection?
11-28-2011 10:16 AM
With the 9234, with the IEPE on, it will use AC coupling. Even if you could use DC coupling with IEPE on, you would then have a 8V DC offset from the IEPE bias voltage.
The best thing to do is to re-shift the waveform. Take the minimum value over a period of time and add that back to shift your waveform back to 0 based.
The DIAdem trace looks pretty good (see page two of the pdf). You could add 0.1 to the waveform values to shift the signal back to zero based.