waveform analysis

[BartvStC]

I have to measure the V NULL (see attachment).

What is the correct way to do that ? 

[DFGray]

Given that plot, the easiest way I know is to make a histogram of the waveform Y values. There should be three peaks, corresponding to the positive and negative peaks and the V NULL (the middle one). You can make the histogram as fine grained as you wish. You can also use the V NULL peak items to find points to average or as a starting point to do something like a linear fit to a region. Let us know if you need more info.

[BartvStC]

Thanks for the advise for V-NULL, it works well.
But now another problem.
The scope measures two signals A and B, each on a separate channel.
Waveform B is the opposite of waveform A, as displayed in the attached .jpeg file.
Waveform B is negated and build with waveform A.
The composed waveform is analysed for bit time (period and pulse duration).
There is a problem, however.
The unit under test sends more than one ARINC-429 label(how many depends on the unit)
Each label is sent with a different transmit rate. Edge triggering is on signal A.
If a unit is sending a lot of labels, all with different transmit rates, you get a
B signal of a different label than of signal A, and therefor not opposite of signal A.
The period and pulse duration of the composed waveform is then irregular and unreliable.
Is there a way to solve this problem ? 
 

[DFGray]

It is unclear from your message exactly what the problem is. I think it is one of the following:

1. Since the B waveform is made from the A, it appears some time after the A, making it difficult to create a composite waveform.
2. The A and B waveforms are taken at different times, making it difficult to determine what part of A goes with what part of B.

If the problem is the first one, you can find the B part which corresponds to your A part by inverting the A part and running a correlation function. The peak will occur when the two functions are aligned (with sample period accuracy - if you need more, ask, it can be done).

If you problem is the second one, use simultaneous acquistion on your scope. They will be automatically aligned. If you need further tweaking, see the first method.

I feel like I have not answered your question. Please post more details. Some real data would be nice.

[BartvStC]

Okay, I will try to explain more in detail. Look at attachment "A429analyse.doc".
The green waveform is signal A, the red waveform is signal B.
Signal B is negated and added to signal A, which gives the green waveform in the
lower graph indicator. From this waveform I have to measure the bit time and the
high time, which you can be calculate from period and pulse duration.
This sample was taken from an ideal test unit, which has the B signal always opposite
from A signal. Other units have sometime signal B momentarily (random) in phase with
signal A. The waveform in the lower graph then gets sometimes zeroed out, and therefor
has a very irregular and false period and pulse duration time. The question is :
How do I prevent this ? 

[DFGray]

If you simply want to detect the condition and do an add instead of invert-add, you can run a correlation function (e.g. Correlation Coefficient) on the A and B waveforms. If it comes up positive, just add them. If it is negative, do the invert-add.

From the data and your description, it sounds like you have a hardware issue. Since you are using an NI-SCOPE device, the inputs are simultaneously acquired, which rules out a synchronization issue with the scope. The problem could be with anything in your test chain - device under test, cables, amplifiers, or scope. If you can, run the test using multiple scope cards on the same signal (you can probably use a DAQ card as well, just for quick diagnostics). Try to use scopes from the failing and not failing tests. If the scopes all show the issue, then work down your test chain until you find the problem. If the scopes are failing, send them back to us for repairs.

Good luck.

[BartvStC]

Okay thanks, the correlation Coefficient works well.
The only thing now is to manipulate the A+B waveform, so that it consists only from,
for example, positive pulses so that the period time is consistent. 

[DFGray]

Since you already have your low, mid, and high levels from the histogram, this should be fairly easy. Subtract the mid level from everything, then take the absolute value of the waveform. This will reflect the negative pulses across the X axis, giving you only positive pulses. You will get some slight distortion around the X axis, depending on your noise level and the accuracy of your mid level, but it should not prevent you from getting fairly accurate pulse width measurements.

[BartvStC]

Okay, all good so far.

I have to measure rise and fall times of the A+B signal.
First I did this with Transition Measurements .vi. The low, mid, and
high ref levels for this were 5, 25, 45 for the negative and 55, 75, 95
for the positive pulse. This, however, is not allowed. If the signal
is very asymmetric the rise and fall times can be false (VNULL is compensated for).
I am trying the Waveform Scalar Limit Comparison with zero as limit,
to split the waveform in a positive pulse train, and a negative pulse train.
This gives an error massage at the Transition Measurements.vi.(see attached file).
 

[DFGray]

Since we are trying to use the transient measurement VIs in a way they were not supposed to be used, it is not surprising they are failing. Let's try something else.

You have the top, mid, and bottom values for your waveforms. From these, you can easily calculate the 10% and 90% voltage levels for the rise/fall times. You can find the X and Y values of the curves at these levels by subtracting the level value from the curve and using one of the root finding or zero crossing VIs. Sort your data by X value to enable you to easily pair the 10% and 90% points, then calculate the rise and fall times.

P.S. Please post screenshots in PNG format so we can view it with our browsers. Word takes a long time to start. The PNG image may also take less space, if you use compression level 9.