Agilent 34970a Voltage Measurement

[Grant_C]

Hi all, i am using an Agilent 34970A datalogger with a 16 channel multiplexer inside to measure 4 input voltage pulses.  Individually, they occur once per minute.  I believe that the instrument scans from a set list each channel in turn.  My problem is that the 4 input voltage pulses could, worst case scenario, arrive at the same time.  Is there a way round this scan list problem?

Regards.

[Broken_Arrow]

Sounds like you'll need a hardware "black box" before the meter to solve that issue.

Since it's likely the signals will not arrive at the same time down to the micro-second, you could consider a FIFO. Alternately, four different analog delay lines, separated by ~ 0.1 seconds, might also work, and that method would work even if they DO arrive at the exact same time.

[johnsold]

How long do the pulses last? What characteristic of the pulses are you trying to measure? Amplitude? Width (duration)? Start time? Do all the pulses have the period or are they only approximately the same? What accuracy and resolution do the measurements require?

How fast can your instrument make a single measurment? How fast can it switch between channels?

Lots of questions, but the solution to your problem lies in the answers.

Lynn

[Grant_C]

The pulses are from a power detect circuit and produce a pulse 520ms long with amplitude of about 2V.  All the pulses are the same and are from seperate circuits that is why the pulses could arrive at the same time.  The pulses are created from 4 different devices that burst every minute.  I would somehow like to measure the pulse in the centre of the peak.  Normally i would use a scope and set the cursors for gating and then measure the mean.  Dont think the datalogger can do this and not sure exactly how long the datalogger samples each measurement for.  I have to be carefull not to include rising and falling edges when measuring the pulse.  Accuracy of 2 decimal places would suffice.  The 16 channel multiplexer has a switching speed of 250channels per second.

Ideally i would like the datalogger to be waiting for each signal and trigger on the rising edge then somehow measure the peak of the pulse only avoiding edges.  This result would then be read into my PC.

[johnsold]

At 250 channels per second it samples a channel every 4 ms. If you only scan 4 channels, each would be sampled every 16 ms. Since the pulses are 520 ms long, you will get about 32 samples of each pulse. Watch for the rising edge of the pulse: One sample near zero and the next near 2 volts. (Be aware that you might capture the transition and get an intermediate value, so you might need to look at 3 successive samples to be sure). Then take the 15th or 16th sample after the rising edge as the middle of the pulse. You do need to keep the data from each channel separate of course. But with those numbers overlapping pulses should not be a problem.

Lynn

[SachaE]

Hi,

  an alternative would be to replace the hardware entirely. National Instruments do have simultaneous sampling cards  (so no MUX involved) to fit inside a PC or on USB which would give your timing results to a greater resolution than 4ms, but of course there's purchasing cost involved in that.

Here's a good starting point : http://sine.ni.com/nifn/cds/view/main/p/sn/n1:7689,n18:Y/lang/en/nid/1036/ap/daq
and here http://sine.ni.com/nips/cds/view/p/lang/en/nid/201880

The main difference between one of the simultaneous sampling solutions and your existing Agilent 34970A is of course the digits resolution. You have 6 1/2 digits voltage resolution (about 21/22 bits), versus the 16 bit products I've pointed at above.

Other alternatives are to go with a much higher sample rate and not simultaneous sampling which would lower the cost quite a bit.

Not pushing this as "the" solution, just something you could bear in mind for the future. If 16ms accuracy is all you need from the system, then as Lynn says, overlapping pulses aren't an issue. (16mS is the time sampled for all the channels to go through if smapling 4 channels so that's the samples / second / channel numbers.)

Thanks

Sacha Emery
National Instruments (UK)