Gibbs phenomenon issue in square wave from labview builtin function generator

[johnsold]

Apparently Mike watches too much television.

Bob's suggestion of a load is probably the right direction. However, the neurobiologist needs to take a few more electronics courses.  (Bob: I have nothing against neurobiologists. I have worked with some  for many years.) The ideal situation has the signal source and the signal destination having matched impedances and being connected by a transmission line with that same impedance. The NI 9263 has a source impedance of 0.1 ohms with a voltage range to just above 10 V and a current limit of 1 mA. So, it cannot drive a matched load (0.1 ohms) because 10 V into 0.1 ohms requires 100 A! The NI 9239 has an input impedance of 1E6 ohms with the capacitance not specified.

The two graphs you posted show different amplitudes: Signal out = +/-5 while the meausred signal is +/-500, nominal. What are the units? How does the amplitude change if the devices are only connected by wires? How long are the wires?

What is the final application? If we know what you will actually be driving and how you will be using the signal, we can better advise on a possible solution.

Lynn

I need to say that I am an engineer (I'm not a neurobiologist).

[ritch_by_nature]

Thanks Lynn, to answer your question, the amplitude differenvy in the plots is because I preset it with a configuration file that has different gains for each channels of data. I did that so the waveform doesnt all have the same amplitude when I'm trouble shooting. That way I can see how they behave. I multiply each channel of data by a diff gain. The wires are very shorts probably 4 to 5 in. Long

[johnsold]

The first thing I would try is loading the signal with about 10000 ohms at the measurement end.  This will not exceed the rating of the generation device for any voltage up to 10 V.

Lynn

[Bob_Schor]

OK, so my explanation sucks, but we agree, you need to include a load (I'm happy with a 10K, rather than 1M, load -- being a neurobiologist, I tend to "experiment", i.e. try a few different resistors varying by order of magnitude.

BS

[ritch_by_nature]

Ok. I will try to do that is see what happens. Thanks guys.

[ritch_by_nature]

Negative, I tried using a 10k on the generating end still the same issue. I tried it on the acquiring endstill the the same. nothing changed

[Bob_Schor]

Draw us a circuit diagram to show us the wiring.  We should see two ports of the A/D, two ports of the D/A, possibly a ground, resistors (show values, please) and some wires.  This so much sounds like a wiring problem ...

BS

[johnsold]

I agree. We are missing something.

Lynn

[BarryGRAEllc]

That is not ringing, it is most likely caused by a brickwall filter eliminating some of the freq components required to make the tops and bottoms flat.

Notice how it starts "ringing" before the transition? An electrical circuit will NOT do that.

Put an RC roll-off before it OR use some LabView filter function to minimize the apparent shape change.

Barry Galvin / GRAE LLC / LLNL

[ritch_by_nature]

Thanks fo particpating,

  which labview one would you suggest?