08-03-2010 06:07 PM
I'm having trouble when trying to make two measurements at the same time with my PCI 6010 DAQ board. I use differential measurements because my voltage signals are rather small - and I use bias resistors because my transducers are floating and tend to charge to one of the registers otherwise. I have generally measured one channel at a time with little difficulty (some irritating persistent noise, but that's it). My problems arose when I tried to measure two transducers at the same time (at a high sampling rate, 100kHz, which I can't lower by much) - in particular, one of them produces a large DC offset, while the other has none. The max input range of my DAQ board is +/- 5V, and the DC offset of the transducer is 4 - 4.2 V. The signal I use is an AC one on top of this offset - I've always filtered the offset out after the fact.
When I initially tried measuring the two (diff.) channels there was a large amount of ghosting present, causing the offset of both channels to be affected. I built four unity gain buffers (following the 'Using a Unity Gain Buffer with a DAQ Device' tutorial, Fig. 5) to lower the source impedance of my transducers, but that did not get rid of the ghosting. I realized that the major difficulty came from the DC offset, so I tried putting a high pass filter in front of the unity gain buffer on the high side of my transducer output. I already had a bias resistor to ground across the non-inverting input of the op amp (as per Fig. 5 of the tutorial), so I just chose an appropriate capacitor for the cutoff frequency I wanted, and added this before the buffer. In testing this alone, I found that it did a decent job of eliminating the offset from the transducer signal. I had been concerned that the DC current from my RC circuit was going into the AIGND of my board, but the resistance is so high that it should be at most about .1mA - I can't imagine that that would be a problem.
After this addition, I tested it by measuring both channels (I use Signal Express) with a signal (a few mV, roughly 15kHz) hooked up to only one of the channels - but in the output there was a sinusoidal signal on both channels, of similar magnitudes. Apparently I've resolved the DC issues, but this would seem to show that there is still a problem with ghosting between my channels...
Does anyone have an idea for why I would still be seeing ghosting? I've checked my buffer circuits multiple times, and they seem to be wired correctly. Could it possibly be a faulty op amp? Any ideas for how I would test for that? Would it be better to use an active high pass filter in place of the RC circuit/unity gain buffer combination?
Sorry for the super long post - but I wanted to give as much of the information that I have as possible.
08-04-2010 12:42 AM
I wanted to add that I have now tested each of the unity gain buffer circuits individually, and they seem to be working as expected. I used a function generator to create a square wave signal, and used the high side of this signal as the input for the buffer. I put the ground from the signal into AISENSE, and then measured each channel (by itself) using a NRSE measurement. Each of the channels accurately reproduced the input, and they all measured the same amplitude. What else could the problem be? Any suggestions would be greatly appreciated.
08-04-2010 09:31 AM
Here is a document that talks about potential remedies for ghosting. There are a couple more suggestions you can try, such as increasing interchannel delay or placing more channels between your signals.
Hopefully this information helps. Best of luck with your application!
08-04-2010 01:29 PM
Thanks - I've spent a lot of time with that tutorial! I'm limited by how much I can reduce the sampling rate of the two channels... The maximum sampling rate of my card is 200kS/s, and I initially divided that between the two channels as 100kS/s each. I can go a little bit lower, but I'm looking at signals that are around 25kHz, so I certainly can't go below 50kHz. Yesterday I tried measuring at this rate and I still saw appreciable ghosting. My understanding is that adding an additional channel or increasing the interchannel delay is tantamount to reducing the sampling rate, right? That is why I tried to solve my problem by lowering the source impedance and bleeding off the DC component.
I really don't understand why my set up doesn't work. Since the unity gain buffers seem to be working properly, my effective source impedance should now be much less than 1k ohm, and the RC circuit I set up is removing the DC offset so there should not be large voltage swings required when switching between the two channels. I would think that this should now be a pretty straightforward measurement - my two signals are the same frequency, in phase, and of similar magnitude - I don't understand why there would still be ghosting in my setup. Any ideas?
08-05-2010 10:05 AM
Adding and additional channel or increasing the interchannel delay is not necessarily tantamount to reducing sample rate. Your card multiplexes through all channels at the fastest sampling rate of the card + 10 us, regardless of what your chosen sampling rate is. When you decrease your sampling rate, you just increase the amount of time between each multiplex action, not the interchannel delay. Adding a channel between the signals is somewhat tantamount because you are technically adding time between when the ADC digitizes each signal. This will allow any stray voltage on the multiplexer line to bleed off when it connects to the open channel before it connects to the next signal. You can even try placing several channels between the signals and place a resistor from one of the open channels to ground to bleed off your voltages.
I would follow the instructions in the KnowledgeBase entitled How Do I Increase Interchannel Delay Using NI-DAQmx or Traditional NI-DAQ (Legacy)?. Let us know how that goes.
08-05-2010 03:56 PM