NI 6250 for pulse analyzing

[Rafi2003]

Hi,

I'll be using the NI 6250 for my application.

I have 4 channels which output 500 us wide pulses at 4 Hz.

Questions:

1.  I would like to sample each channel for 4 seconds.  With a sampling rate of 1Mhz it will require a 4M samples.  Is it possible to acquire so many samples?

2.  Will the card be able to acquire at least 2 channels simultaneously?

3.  Will the card be able to acquire 4 channels simultaneously?

Thanks

RK

[Rafi2003]

HI,

I'm adding 2 waveform diagrams for better understanding the waveforms.

1.  Waveform to input 1

2.  Waveforms to inputs 1 and 2.

3.  It's important to sample inputs 1&2 simultaneously in order to measure the timing and voltage between them.

4.  The task is to measure:  signal voltage, pulse width, frequency and shaping (few width measurements along the pulse).

5.  I understand that in order to sample 2 channels, the maximum rate of each one is 1M/2  samples per second.

           Is this correct?

           Will DAQmx  give me 2 synchronized array as a result of this?

6. I'm basically interested in the 4 high level signals.  I'm looking for a way to sample just them (in order to minimize the array size).

7. The attached waveform is just one sample.  There are others with different voltages and different frequencies.

Any advice is highly appreciated

Thanks

RK

[Jim_S]

Hello RK,

The NI-6250 is not a simultaneously sampled card where it has its on ADC for each channel like our S-Series boards or our some of our X-Series boards. Instead the board multiplexes the inputs thus it can sample 1 channel at a max of 1.25MHz and 1 MHz for all the channels. This can be seen in the Specifications manual for the card. For more information on how the M-Series converts the data with one ADC please consult this KnowledgeBase and the M-Series User manual on page 53.

[Rafi2003]

Hi Jim and thanks for your reply.

I have read the material you've suggested but still quite confused.

Please try to answer the following questions:

1.  Do I always have to choose the Sample Clock and Convert clock?  or is it set automatically by my card setup. (see the following examples)

2.  I have 4 inputs with the waveform seen in the attached file.  I only sample one channel at a time.  Is the Sample Rate be 1M/4 S/ch?  What will be the convert clock?  Why do I care about the convert clock?

3.  I set up just one channel, make the measurement.  Then, setup the second channel, make the measurement and so forth.  What will be the Sample and Convert clock in this case?

4.  I want read "simultaneously"  (multiplexed) reading of 2 channels.  Assuming I setup 4 inputs channels, what is the Sample and Convert clock?

5.  Looking at the waveform I provided (500us pulse), what is the minimum requirements of Sample/Convert clocks in order to take a width measurement of max +/- 1%?

Thanks

RK

[Jim_S]

Hello RK,

When looking over your first posts, you wanted to simultaneously acquire from 2 channels, meaning that each channel has its own dedicated ADC to convert each channel. This means that if you use the same sample clock (the frequency to take each point) for all the channels on a simultaneous board, the digitizing will occur all at the same time. It seemed from you first posts that this was a major concern for the measurements needed.

With your card or M-Series boards, it uses a multiplexer where for each pulse of the sample clock it will need to create separate pulses to change the state of the multiplexer. For example, you want to sample 2 channels from your board at 500kHz. The sample clock pulse will occur and then the board will create 2 pulses to change the mux to acquire both signals. How the card creates these signals has to do with the settling rate of the card. It states in the specification how much settling time is required in order to get a valid measurement. For this senario, the minimum settling rate is 1us for the + or - 10V range. You also can monitor this by using the timing property node calle AIConvertClock.Rate

Ultimately, I introduce this concept because of the need to understand the difference between simultaneous and multiplexed. Finally, the 6250 card can't sample several channels at 1MHz but instead it can sample for all channels at 1MHz. Referring back to my example, each channel is being sampled at 500kHz or totally 1MHz. Ultimately, this card doesn't seem like it would meet you application but an S-Series card would be a better fit.

[Rafi2003]

Hi Jim and thanks again,

The specifications requirement came up after I have bought the card, so, now I'm stuck with it and I'm trying to achieve the best I can.

My approach will be to sample 2 channels simultaneously, hopping (still to be found)  that the signal will not change too much between the 2 samples (ch1 and ch2).

I think that the 500KHz sampling and the 1us setting time between samples is not the only issue....

As I dig into this material, I've read that one should also take into account the voltage variations between 2 samples.  Am I right?

In my case, one channel is at 9 Volts and the other is close to 0. 

Digging more into the specifications of the NI 6250 I found the graph "Setting time errors for different input impedance".

I ask you to verify my thought here....  If my impedance is 10K, and I can allow for 1K ppm of error, it means that the required setting time is approximately 12 us. From here I deduct that my sampling rate is   1/12us= 83 KHz.  Is this correct?

Does it mean, now, that the sample of ch2 will occur 6 us after the sample of ch 1?

How will this timing change if I multiplex 4 channels?  ( I don't need it....just for understanding purpose)

Thank you very much

RK

[Jim_S]

Hello RK,

I am not completely sure what you mean with the following sentence below. Could you explain what you mean below 1K ppm? Is this mean .1%?

I ask you to verify my thought here....  If my impedance is 10K, and I can allow for 1K ppm of error, it means that the required setting time is approximately 12 us. From here I deduct that my sampling rate is   1/12us= 83 KHz.  Is this correct?

The settling time has to do with the RC circuit created by your transducer and the capacitance inside the board. In your NI-6250, its capacitance is 100pF so it is ideal to have your device's impedance to be under 1kohm to prevent lengthening the settling time. In order to reduce the settling time, it is recommend to use a unity gain buffer which reduce the impedance of it greatly. Please consult this KnowledgeBase about ghosting due to settling time and this Developerzone to learn more about unity gain buffers.

[Rafi2003]

Hi Jim,

Apparently you didn't follow my thoughts...

Please look at the attached graph.  It is taken from the NI6250 Specifications.

This graph show the signal error for different input impedance.  My impedance is 10 K to ground and (1.1 K between ch1 and ch2 if it makes any difference).

If you follow the graph you will see the point where 10K graph meets the 1k ppm error, the setting time is 12usec.

My question is if this line of thought is correct, and whether I can conclude that sampling at 1/12us=83Khz will give me a max error of 1kppm (0.1%) ?

Thanks again

RK

[Jim_S · Accepted solution]

Hello RK,

I understand now and yes you are correct in your assumption of the max sampling rate for 1 channel. If you increase the number of channels, you will need to reduce the sampling rate. Here is the formula for that:

Sampling Frequency =1/(number of channels * 12us)

In order to reduce this impedance though, you can imploy the unity gain buffer that I mentioned in my previous post.