Fluke 732a vs 3458a

[nikonoid]

Specifications for 732a and 3458a are within a ballpark of each other at 1yr. If you have several 8.5 digit meters with similar specs to check each other and have 10v standard that is stable to 0.2ppm for 10 minutes, is there a point in owning 732a too?

What would you be gaining?

Thanks.



Sent from my iPhone using Tapatalk

[Moon Winx]

Traceability. And plus a good chunk of the year after a 732A is cal'd it has a known value to < 1 ppm.

[MisterDiodes]

A few 732a/b + KVD + null meter is going to get you into the range of low ppm measurement uncertainty of absolute measures, say within ~~2ppm @ 95% confidence.  The 732 is an analog ovenized reference and doesn't generate switching noise on it's own - and just about the only thing mortals might own (short of a JJA) that might be a bit more stable than a single LTZ1000a.  With an older 732 with good drift records you can estimate a closer estimate of it's output voltage on any day between calibration cycles.  The linearity of a KVD is known, self calibrating and measurable at any time, and the input impedance of a null meter goes basically to infinity theoretically at null point.    Not quite that in real life with real fets but typically the impedance jumps to some 10's of GOhm at null, typically higher than any DVM.  The other advantage is that with this system the only required calibration is on the 732a/b itself - everything else is self-calibrating.  If you have two or three 732's circulating back and forth on their calibration schedule, you really do have a good verifiable voltage reference with ~2ppm uncertainty compared to NIST.  You're also looking at sensitivity down to 10 or few 10's nV on a 1uV full scale range - and some null meters do better than that.  Don't worry about the fact that an analog meter might only be spec'd at 2% accuracy - remember that's on a 1uV or better full scale, and most of the time you're looking for the null point - which is also going to be "zero" signal anyway, and FS meter accuracy has minimal effect.  Compare that to a 3458a trying to resolve 50nV.

An analog null meter also trims off noise higher than 10Hz for you, which is usually what you're after anyway.

A standard issue 3458a is an excellent meter, but for -absolute- measure approaching it's next cal date is going to be more in the area of 8~10 ppm uncertainty.  A high stability version will get you in the 4 ppm uncertainty range - and even over a 24hr period an older 732 will be more stable as we know from long experience.  Again:  sensitivity of DVM is typically not as good as what you see with a null meter method.  The other problem is the meter's own noise it brings to the table, how long since AutoCal, etc.

Of course the 3458a gets you easier data logging, more measures per hour, resistance, Ac measures, etc.  After all that though it is not considered a true voltage transfer standard, and doesn't promise that.

A well equipped lab for low ppm measures will always have several 3458as (or similar) PLUS 732's, KVD's and null meters.  You get the best quality measures by using a wide array of tools and techniques to increase confidence and lower uncertainty.

 If ALL you're going to do was measure measure DCV, then a 732 + KVD + null meter will get you lowest measurement uncertainty per buck, but having lots of tools available is always your best confidence booster.

Here's an example of how to calculate DCV measurement uncertainty on 3458a - the longer you've gone since calibration "the more you're sailing on the ocean of unknown absolute value":

https://www.manualslib.com/manual/529213/Hp-E1328a.html?page=57

Here's a link to an older 1988 paper that explains that troublesome grey area of when you measure a 3458a against a standard, and did it pass or fail the test - Where do Test Limits Come From:

https://community.keysight.com/thread/18420