EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: mawyatt on January 26, 2022, 07:43:46 pm
-
Our ongoing project is demanding additional hi res DMMs (long story), so we got a new KS34465A to support our old KS34465A and newer DMM6500 (may need to get another one of these for the graphical display features).
Anyway, we decided to do some tests to see how well these 3 DMMs behaved. First off were some new to us, but very old Caddock 0.01% low TC Resistors, and some new Dale Vishay 1% somewhat low TC WW Power types. Good agreement, and was somewhat amazed how well the new and almost 2 year old KS34465A agree!!
Then decided to measure some voltages from out home-brew references which are based upon the well aged LM399, LTZ1000 and ADR4550. These have been powered for a couple years almost continuously. Again very good agreement between the DMM6500 and two KS34465A. So decided to run some longer term graphical displays and tried to set the 3 DMMs up as close as possible, with the DMM6500 set to Line Synch, 15 cycles and 100 term moving average. The two KS were set to 100 cycles and 100 term moving average, and all were using the MX+B scaling for the extra resolution. Note the setup is just some long twisted pair copper wire connecting the references to the DMMs, with a ~1M cable from the reference to the new KS, then a 500mm twisted pair to the DM65500 from the KS, and another 1M long pair to the older KS. So lots of potential "antennas".
We noted that the graphic displays look quite different, one reason is the time scales are different (DMM is 100s/div whereas the KS are 30min span which is effectively 180sec/div). Note the drop out in the DMM6500 that does not show up in either KS34465A? All DMMS have the vertical scale set to 10uV span.
What appears to be happening is the DMM6500 is more sensitive to EMI likely from my iPhone (our our WiFi) than the KS34465A. The setup isn't good from an EMI standpoint as the 3 DMM are connected with copper twisted pair cable without shielding. I know the first response would be the LM399 is being effected, which would make perfect sense except the 2 KS34465A don't show the large variation as the DM6500, in fact the older KS34465A has about an extra ~500mm of twisted pair.
Has anyone seen a similar EMI (if this is the cause??) sensitivity and compared various DMMs? BTW not interested in doing any detailed testing at this time, too many other important things on our plate at the moment, so just a curious notion here ;)
Best,
-
All 3 metes us a LM399 reference as well. The measurend voltage is a combination of the external reference voltage and the DMM voltage references. Both can vary to about a similar degree. A LM399 may very well show popcorn type sudden jumps of some 4 or 5 µV and the curve shown for the 6500 could to a large part be explained by such jumps. For the KS meter, the graphics is poor and hard to judge - not sure about the sacle and also very small.
It would really make sense to log data to a PC and than show the data togeter in the same diagram. This way one gets a good idea which part is correlated between the 2 meters (and thus likely from the external reference) and what is not and thus more like a meter effect.
-
Yes know they all use the LM399, one reason they were selected :-+
Also think the 15 cycles vs 100 cycles line synch with the sliding average may have some influence besides just the popcorn noise.
May hook up the SDM3065X and see what it looks like, although it doesn't have any sliding filters, but does have a LM399 reference.
Thanks for comments,
Edit: Just got back from dinner and noted these results. DMMs were reading ~2.6, 2.7 and 2.7uV SD for the KS old and new and the DMM6500 respectively, yet the DMM6500 shows slightly more apparent "randomness" in the trend display when all were "set" to 100 moving average on an aperture of 0.25s for all DMMs. This leads to the conclusion the DMM6500 is slightly more noisy, maybe more EMI susceptible and may help explain the displayed differences in the trend charts when using the different averaging routines and my presence (lack of) with the Smart Phone as a potential EM Interferer.
Let this run overnight and it seems the 15 cycle and 100 cycle line synch was causing the difference between the DMM and two KS34465A trend displays. Using the same 0.25s aperture, all the displays look similar and show similar features, although the DMM is slightly more "noisy". So it seems that the additional aperture time (100 vs. 15 cycles) coupled with the long smoothing filter (100) caused the KS to smooth over the perturbations, whereas the DMM6500 tended to follow those perturbations due to the smaller aperture window.
Also used the SDM3065X setup as close to the aperture as possible using 10 cycle line synch, it has no smoothing filter and thus the trend display isn't smoothed and useless for comparisons. It has it's own issues with random jumps of many uV likely due to internal LM399 popcorn noise, so we decided to keep it powered on 24/7 awhile back to see if these random jumps subside, so far they haven't! The SDM3065X isn't in the same league as the KS and DMM, more of an entry level hi res than a true professional level, but also lower cost.
Anyway, all's good now as think we have a plausible explanation for what we were seeing in the trend graphs!!
Best,