below 10msec data acquisition : GPIB or NI-DAQ ?

[arienai]

Hello, 

I would like to measure continuously a voltage drop on a resistor, with a sampling time < 10msec for a 1sec, then sampling of 100msec for the next 10sec (this is to simplify). Then repeat this cycle of measurement until pressing STOP button. 

To go further, my resistance has a non negligeable TCR, so that when supplying a step voltage, it is self-heating. So I want to get the  "voltage" herafter setting 5V (heating), and also heafter setting 0.1V (cooling). Heating duration at 5V ranges from 1sec to 10sec, idem for cooling. So basically, I need fast acquisition only during the first 1sec of heating and cooling. 

My concern is how to get  <10msec sampling using GPIB. 

My setup is based on GPIB-USB-HS with VISA type communication using Labview 2012.

Voltage source (SMU) -  1 resistance (1kOhm) - GND. 

I put 3 DMM between the source and the GND to measure simultaneously the voltage drop on this resistance.

Don't make comments on that stupid setup, this is just to simplify my problem. 

The DMMs are Keithley 2182 (9 digits), Agilent 34410A (6 1/2 digits), HP 34401A (6 1/2 digits).  

Agilent E5263A provides the voltage source between 0V to 5V, in DC mode. As SMU, I can also read the voltage and the current using GPIB commands. (no comment like I don't need DMM because of SMU voltage measurement etc etc ..; thank you). 

The DMM are set to 4 1/2 digits in order to speed up the data acquisition using a minimal GPIB commands to each instrument. Rather than using existing sub-vi from the maker, I just send the necessary command, like READ?. 

I use a Model 9700 to measure the temperature on a holder using a TC (just for information). 

I measure the differential time between "before" sending the command and after reading the values. This is dt (msec). I expect to get the lowest dt, since I have no sampling time set on my while loop.

I also read the data, and display the indicator for each data, and also plot a waveform chart for each one. This is typically what I need. I agree that I can create a producer/consumer and enqueue the data then dequeue in a parallel while loop and display/graph. No problem. 

SMU E5263A : dt = 5msec  for reading Channel 1 

34401 only (NPLC = 0.2) : dt = 23msec

34410 only (NPLC = 0.2) : dt = 16msec

2182 (NPLC =0.2) :  dt = 33msec

Model 9700 : dt = 15 msec to read 1 temperature channel

But E5263 + 9700 + 34401 + 34410 + 2183 : dt = 45 msec

I can improve dt as I explained before (enqueue / dequeue and plot). 

However, I think that I reached the limits of using GPIB for what I want to do. 

I am wondering how fast could be the same data acquisition if I replace all the DMM by a DAQ-NI-USB-6251 and read the analog inputs (there are 8 channels available on the 6251 and 1.25MS/s) ? 

Should I use an oscilloscope with a trigger, and output the buffered data for each cycle ? 

Any other suggestion or example or link to similar topic ? 

Yours

[Kevin_Price]

Your 6251 can handle this easily with a buffered, hardware-timed task.

Start with the shipping example ("Help-->Find Examples...") for continuous voltage input.  It can easily keep up with a constant 100 Hz sample rate (10 msec intervals) and the timing intervals will be hardware-precise.

I would recommend you acquire both the control and the response signal (2 channels) as an easy way to correlate your data.

-Kevin P

[arienai]

Thank you Kevin, I will try it. However, is it possible to get a few 10microVolt resolution with this DAQ ? I am then wondering about the difference between DAQ and DMM for voltage measurement. 

In my set up, I will build a wheatstone bridge, so I need to measure voltage drop between some fixed resistances, and also the Vab (bridge). I guess I need to choose floating ground for that. 

Acquiring control and response signal : does it means that the voltage source will be also handled by the DAQ rather than the SMU ? 

Yours

Laurent. 

[bseguin]

You will need differential mode and 4 wire mode.

you might also prefer to be able to adjust your gain grammatically.

DMM is usually configurable as you wish. but not all of them allow you to acquire at high rate.

Most DAQ from NI have configurable gain.

Benoit

[Kevin_Price]

No, you won't get 10 microVolt resolution with the DAQ device.  16 bits spread over 5 volts will give you something nearer to 100 microVolt resolution.  But keep in mind that resolution is not the same as accuracy or repeatability.  You can help yourself on the accuracy front by acquiring both the reference and the test voltages -- the ratio between the two is much less sensitive to any systematic inaccuracy than either measurement alone.  The inaccuracy will tend in the direction of "canceling out" when you do the ratio.   (It won't *actually* cancel out, but the effect is generally reduced.)

Also, in the presence of fairly random noise, you can oversample and average.  As I recall, the extra *apparent* resolution is proportional to the square root of the oversampling factor, i.e., 100x oversampling yields up to a 10x improvement in *apparent* resolution.   (I emphasize *apparent* because real world noise & quantization effects are not as purely random as the tidy theory would like.)

In general, a DMM prioritizes resolution while a DAQ device prioritizes speed.  The right tool for the job will depend on how you trade off those priorities.  (*see footnote)

No, I wouldn't try to source the voltage with the DAQ device, it probably can't source enough current for your resistor network.  I would only try to capture both a reference voltage (probably the fixed resistor voltage drop) and your desired measurement voltage.

-Kevin P

*This brings to mind one of my favorite quotes from an engineering professor in college.  To emphasize that engineering is constantly about making trade-offs and setting priorities, he said something like this:

"The fundamental law of engineering is the Conservation of Misery.  You'll always be making trade-offs, and you'll always be stuck with compromises you wish you didn't have to make."

[arienai]

Thank you very much for your answer (and sorry for the delay in answering, I did not get a notice). 

I will continue without DAQ, following your advices.