Agilent 34461A Blog #2 - Histograms

RE: Agilent 34461A Blog #2 - Histograms

DAB — Wed, 20 Nov 2013 19:58:36 GMT

Great post.

This is one of those exercises that we should all do with our test equipment, but never find the time to do.

I agree that a comparison measurement of the same resistor would also be useful to see how repeatable the measurement app performs.

Overall, you did a good experiment, verified the device and provided an entertaining presentation.

Thanks

DAB

RE: Agilent 34461A Blog #2 - Histograms

michaelkellett — Wed, 20 Nov 2013 09:21:25 GMT

Nice to see some more playing with the 34461A. I see that you (and others) are posting stuff about these roadtest items as blogs rather than reviews - is there some special reason for that ? I ask because I was looking for activity in the Roadtest section and the blogs don't (always) show up there.

Reflecting on Morgaine's comments - it would be a nice addition to your experiment if you were to repeat it but using the same resistor many times - this would give a useful indication of the statistical capability of the measurements. For even more fun you could to it twice, once where you disconnect the probes each time and once when you don't. (If you have any life left after that I can think of other variations.)

I've just bought a 34416A so I might have time to try it myself.

RE: Agilent 34461A Blog #2 - Histograms

morgaine — Wed, 20 Nov 2013 08:50:39 GMT

Interesting post, Mark. It did leave me with an immediate question about how the measurements were triggered though, as you don't detail your procedure.

Am I right in thinking that the instrument is normally in free-sampling but non-recording mode, so that as you connect up a resistor you can immediately see an indication on the screen that good contact has been made --- without the instrument using that stream of varying data? And only then, after you decide that all is well, you manually trigger a single sample for the measurement record and histogram plot? If this is so then the manual trigger provides an arbitrary datum sampled from a continually varying input, since these probes are not surface-penetrating and therefore on a 6.5 digit instrument some visible variation would be expected from ambient effects like vibration and thermal creep.

Or, is the multimeter doing continual statistical processing on the input, so that when you trigger a sample for the record, you're actually obtaining a more dependable datum than a point reading could provide? (The simplest continual processing algorithm would of course be "hold lowest reading" in the case of measurements of passive resistance, since probe faults cannot decrease the value measured.)

I'd very much like to hear about such operational issues that make the most of powerful instrumentation.