06-27-2011 04:29 PM - edited 06-27-2011 04:30 PM
We are using all 64 AI channels on PCI 6071E for real-time fluorescence detection via photodiodes array. The operating mode is following:
External digital start trigger
Internal clock
Finite sample mode: ~2000 samples per trigger
Sample rate: 1MS/s
Input range: (1) 0~1V or (2) 0~100mV
Platform: WIN-XP/ CVI
For our experiment the signal size is switched between range (1) and (2), and we want to use full 12bits resolution on the card for the two situations in order to maintain a decent S/N ratio. We could potentially do the following three things, but I am writing here in case anyone has a better sense of this device so that we can work out the best solution.
A. Dynamically change the 6071E device settings and switch between range (1) and (2)
How fast can one dynamically change the input gain settings on this device? If it is fast enough, A would be most cost-effective solution .
B. T- off all 64 signals and use an additional 6071E card which is pre-set with a different input range.
C. Dynamically change the signals (build additional switching electronics) such that they are with the same input range.
Thanks!
Solved! Go to Solution.
06-28-2011 11:30 AM
Hi,
How would a typical acquisition experiment go?
Would it be as follows?
1. Arm trigger.
2. Acquire 2000 samples at 1 Ms/s upon triggering at gain (1)
3. Change gain setting
3. Arm trigger.
3. Acquire 2000 samples at 1 Ms/s upon triggering at gain (2)
What amount of time is there between triggers?
Or do you need to change the gain on a per channel basis?
Thanks,
Jon S
06-28-2011 11:44 AM
Hi Jon,
Thanks fro your reply. Yes, it goes in the way you mentioned:
1. Arm trigger.
2. Acquire 2000 samples at 1 Ms/s upon triggering at gain (1)
3. Change gain setting
4. Arm trigger.
5. Acquire 2000 samples at 1 Ms/s upon triggering at gain (2)
6. Go to 1 and repeat
At a given time, all 64 channels have the same gain (1) or (2).
Between cycles (one cycle= step 1 to step 5), time delay is around 2 seconds;
Bteween step1 and step4, time delay may vary from 20ms to 3 seconds. If step3 can be excuted in less than 10ms and the device can be ready for step4 and step5, it will certainly do it.
Thanks,
Kunyan
06-29-2011 11:45 AM
Thanks for the clarification. How quickly you can make this change will depend on the computer you are running it on. I've attached a VI that can help benchmark this. When I ran it on my machine, I was able to change the setting in around 3 ms, so this would definitely meet your spec.
Give this a shot, and it should help you feel out whether or not it will work.
Jon
06-29-2011 11:59 AM
That's great, Jon!
We have checked our system more carefully and it seems that we will operate and change the gain settings on a time scale longer than 100ms, in whcih case this won't be a problem. Thanks,
Kunyan
06-29-2011 04:49 PM
Hi Kunyan,
You mentioned you need a 1 MS/s sample rate but you are acquiring on 64 AI channels. The 6071E is multiplexed, so I'll assume that you are acquiring an aggregate sample rate of 1 MHz, which is really only 15.625 kHz on each of the 64 channels. Or, perhaps you are just acquiring on a single channel at a time?
Anyway, I just wanted to point out that you can multiplex the same channel multiple times at different gain settings by calling multiple instances of DAQmx Create Channel:
As shown above, the rate has to be reduced since you have to effectively sample the channel twice for each sample clock edge. If you want to sample the channels sequentially at the full rate (e.g. 2000 samples on ai0 at 0-1V @1 MHz, then 2000 samples at 0-100 mV @ 1 MHz) it gets kind of ugly but you can do it by calling DAQmx Create Channel for each sample that you wish to read (see here).
If reconfiguring the task every time is suitable for your application there's no harm in doing that, I just wanted to point out this option as well.
Best Regards,
06-29-2011 05:15 PM
Thanks John for pinting this out. We misunderstood the device functions: We were thinking the device can read 64 AI channels in parallel, obviously sequential read out at 1.25MS/s /64 ~ 19.5KS/s is little bit slow for us. Are there any devices that read AI in parallel or much faster serially? Best, Kunyan
06-29-2011 05:38 PM - edited 06-29-2011 05:39 PM
As an additional note:
we prefer to have parallel analog read in with >= 100KS/s/ch
06-30-2011 11:48 AM - edited 06-30-2011 11:55 AM
Hi Kunyan,
A cDAQ chassis full of 9215s would give you 32 simultaneous channels @ 100 kHz (see here). Two of such chassis would give 64 channels.
In the PCI form-factor you could look at the 6143, but it only has 8 channels per board so you'd need 8 of them for all 64 channels. You might consider PXI instead to allow for easier expandability (the 6143 is also available in PXI form-factor).
One large downside with the above two options is that these products both have fixed ranges. The 9215 is fixed at ±10V, whereas the 6143 is ±5 V.
NI has other options with variable ranges but the cost starts to go up. If you're intersted in these, I'd look at a PXIe chassis with either 4300s or 6358s. You might also consider 6123s instead of 6143s in the PCI/PXI form-factor.
For the best dynamic performance you might look at the Dynamic Signal Acquisition products.
Best Regards,
06-30-2011 12:01 PM - edited 06-30-2011 12:01 PM
The 9215s are available with Screw Terminal or BNC connectors FYI.