CPU usage too high, probably from reading 15 N4351 channels, solution?

[Ariel]

I've used the profiler, the AI Buffer Read is far and away the big (huge) hog of time. The hardware TBX-68T on the N4351 is slow and doesn't (to my knowledge) support occurances. I'm trying to read each of 15 channels each second, in addition to reading strain gauges from a SCXI card (which is real fast, uses occurances, and in an entirely seperate while loop).

My CPU usage spikes up to 100% from ~51% just under every second or so and, oddly, increases and decreases in spiking from ~70% usage every ~4 seconds. (like a signal in an envelope).

all my loops have delays, no array manipulation where there shouldn't be, etc etc. i feel confident that it's the reading of the TBX-68T (at 60 hz not
ch frequncy) that's causing this. any suggestions for a different/faster way to read it?

[PdB]

First of all, I think the measurement hardware-section is a better forum for this question.
I don't think the TBX-68T is the problem. That are just some electronics parts to get rid of the AC-hum.

I urge you to take a look at the specifications of the hi-res DVM. If I interpret the data, it says that for a 60 Hz rejection, you have a maximum of 9.7 readings per second.
That means that you have new samples every 15/9.7 = 1.5 seconds, so you cannot get the 1 sec. speed you need.
Why is that?
Sigma-delta ADC's (the ADC of the 4351) are capable of hi resolution, but they use digital filtering-techniques inside the silicon to acchieve that.
And as you know, every filter (analog or digital) takes some time to settle to a stable read-out.
And exactly
that is the reason why a single-channel read-out is so much faster than a multi-channel read-out in sigma-delta designs. It doesn't have to wait until the output is settled.
So what happens, you try to read-out the DVM, but internally it waits for an end-of-conversion of the ADC. Your vi just waits until it is ready.

Note: I have not used the 4351 in practice but my company designed a compact digitiser with sigma-delta convertor under LabVIEW control.
Unfortuately it never reached the market, but it was nice to see temperature vary in the 0,001K range.

Hope this helps,

Patrick de Boevere
Serenity systems design

[Ariel]

Thanks, i spent some time with it last night, and came to the same conclusion...