PCIe 6738 control relays with digital output for analog signals in LabView

[tareyes]

Thank you for the insight. 

Some other sensors that are used are load cells, strain gauges, displacement, angular rate, pressure. All of these would be reproduced by a DAC and the external circuitry required.

While the 24bit capabilities of the 4463 would be optimal, the FIFO buffer size is far below what I would require as I need 300ms of data per channel output at 20kHz, so 6000 samples per channel optimally each time the device is ran. 

I'm starting to believe that this task may be more difficult to produce a working product than I previously thought.

I'm attaching a screen capture of the waveforms I'm currently working with produced in LabView just as a reference. I can manage a waveform that looks like this from my DAC currently, it just lacks the resolution needed.

At this point, I am just trying to get a working prototype with the 4 channels I have on the DT9834 device I have on hand, but different hardware would be needed for a tool that could be used in consistent testing. 

 

[santo_13]

---

@tareyes wrote:

While the 24bit capabilities of the 4463 would be optimal, the FIFO buffer size is far below what I would require as I need 300ms of data per channel output at 20kHz, so 6000 samples per channel optimally each time the device is ran. 

 

---

Nothing to worry about, that is where the DAQmx driver is very powerful, it will allocate DMA and stream it from memory. I have generated a few seconds of waveform at 51.2kS/s rate using 4463 at a massive 35 channels in parallel.

[tareyes]

So I could hypothetically purchase four PXI-4463 cards, a chassis, a communication card on the host PC, and I would be able to generate 8 waveforms to represent sensors without any external circuitry at all? Besides just a LabView program to run the samples? 

24 bits would be enough to not even need to consider relay controls into voltage dividers and introducing all that extra noise. Especially with the attenuation of 17dB applied to reduce the range to +/- 1.7V. And if it can produce 51.2kS/s per channel, that would be enough. 

[santo_13]

It is trickier than that. There is no one solution that fits all.

The speciality of the DSA cards is that they can produce very good signals composed of sines - these are sounds and vibration signals, like from accelerometers and microphones. This is achieved using the delta-sigma converters and a bunch of filters.

This DSA architecture does not suit well for other precision analog signals like a square, ramp or any other non-sine composed signals. You can generate a square wave using a DSA card, but the signal will look like the textbook equation for a Fourier transform of a square wave whereas the square wave generated by a DAQ card will have clean edges and more like a theoretical square wave.

In summary, either there is a compromise or choose the instrument best for each situation.

[tareyes]

Interesting information, so I could get a DSA card to reproduce the vibration sensor signals, but my original idea would have to be implemented for other types of sensors like load cells. 

Given the capabilities of a PXI chassis, this could all be implemented within one system. Using a DSA card in combination with a PXIe-6738 for example to reproduce the other sensors. But this just further complicates the system if there is no "one size fits all" solution.

Edit: A complete list of the sensors I am trying to replicate simultaneously: Accelerometers, Loadcells, IR, ARS, Potentiometers, Pressure

[tareyes]

Santo, with what you said about the DAQmx driver. I had an NI rep tell me that a cDAQ chassis with output modules like the 9262 would not be capable of my application due to the FIFO buffer size being very small. Are the cDAQ systems capable of allocating DMA and streaming it from memory? Those were much more desirable devices given their interface over USB and their compact size.

[rolfk]

---

@tareyes wrote:

Santo, with what you said about the DAQmx driver. I had an NI rep tell me that a cDAQ chassis with output modules like the 9262 would not be capable of my application due to the FIFO buffer size being very small. Are the cDAQ systems capable of allocating DMA and streaming it from memory? Those were much more desirable devices given their interface over USB and their compact size.

---

And those USB connections are exactly the reason why they can't reliably upheld continuous high speed data transfer. There is no DMA channel over USB unlike with PCI(e) and PXI(e).

[santo_13]

What do you mean by IR and ARS?

Loadcells, Potentiometer and Pressure can be simulated using regular DAQ AOs. Are these always 0-5V or some industrial standard voltage signals or 4-20mA loop signals?

[tareyes]

these are always recorded at +/- 5V, and the ADC typically applies a gain of around 79 to the received signal. I need to be able to reproduce the original signals, which can be very small values. My concern is that a 16bit AO doesn't have the resolution to produce such a signal, so I've concluded that some external circuitry would be required. 

ARS are Angular Rate Sensors and IR are infrared compression sensors.

As you can see below where I attached some more sample waveforms, I also included a max and min amplitude value in volts. these are very small values that a 16bit device cannot replicate accurately to my knowledge.