There is no way to keep 5-6 digits precision at so fast measurement.
Obviously, but people are suggesting it drops down to less then 2 digits.
My understanding of the accuracy statement is that the crest function is capable of measuring changes in a measurement parameter that last for longer than .8ms, which is why they only assign a tolerance to values greater than this threshold. Any change in the parameter less than .8 ms does not get an assigned tolerance because the meter is not capable of registering it. I could be wrong, but this is how I interpreted it.
OK, so a change that lasts for at least .8ms.
That makes sense.
+/- 100 digit accuracy
But this makes no sense. My Brymen drops down to 4 digits in crest mode so if I put it in the 50V range and tap a 9V battery with the probes, you're suggesting it can read anywhere between 8 and 11 volts?
It's an easy test to do...
I've got a 9V battery here and if I set it to the 50V range and measure the voltage I get "09.465V".
I pressed the crest button and tapped the battery terminal with the probe and I got "09.59V" on screen.
I did it again and got "09.62V".
Third time gave "09.60V".
Fourth time: "09.60V"..
Fifth time: "09.62V".
OK, it's a tiny bit high (1.4%) but the error is consistent with "10 displayed digits", ie. 0.1V.
The exact phrase use in the datasheet is: "Specified accuracy +/1 100 digits for changes > 0.8ms in duration"
I think that "specified accuracy" refers to the meter's accuracy in 50,000 count mode, not the "visible accuracy" of the number of digits displayed in crest mode.
It's consistent with my quick test and much more believable than the meter only having +/-1V accuracy (which would be completely useless!)
I agree with you that this accuracy level would be practically useless for my measurement, so I hope someone can point out where I have gone wrong in my understanding.
I am not suggesting that every meter will necessarily read to the limits of the tolerance range when taking a measurement, but any random meter could read at the max. or min. published tolerance and still be considered to be within spec. I assume that if these meters are tested for quality at some point during the manufacturing process, then one that reads say 10 volts in crest mode would pass and be sold to a customer assuming my interpretation of this accuracy statement is correct. Every meter is going to read a little differently due to manufacturing variances.
I'm not sure using a 9 volt battery like you are describing would adequately test the function of crest mode, because you are simply taking multiple measurements of a constant 9 volt source that doesn't see any rapid changes once you have the meter connected to the battery. I think a more accurate test would be to simulate a short duration voltage spike on a constant voltage source at varying time intervals.
The exact statement for the BM857s is "Specified accuracy +/- 100 digits for changes > 0.8ms in duration", but this is the only meter with a greater than 4 digit display in Brymen's lineup that doesn't also include an additional caveat statement "6000 count mode only", or "Resolution 5000 counts" when describing crest mode. I would be surprised if this meter is capable of .001 resolution in crest mode when the flagship BM869s is not.
My understanding is that the number of additional counts has a different effect on accuracy based on the resolution of the range you are measuring in. The resolution is .001 in 6.000V range, and the resolution is .01 in 60.00V range, so the effect on accuracy of +/- 100 counts is the difference between .1 and 1 in those two ranges. Since your BM857 meter switches to a resolution of .01 when you enter crest mode measuring 9V (09.00V), then +/- 100 digits should result in +/- 1V of tolerance, regardless of the normal screen resolution.
As an example calculation, if we assume that we are limited to the 6000 count mode for the crest function as is stated in the BM-839 manual, then the only ranges available in DCV are 6.000, 60.00, 600.0, 1000. To measure 9 V, the meter would have to be in the 60.00 V range, otherwise it should show an overload in 6.000 V range. The "specified accuracy" in the 60.00 V range for the BM-839 is +/-(.045% + 1D), so a reading of 9 V would give +/- .014 V (.00045*9 + .01*1), then because we are also in crest mode we need to add the additional +/-100 digits to this (.01*100) = 1 V. The total tolerance per spec for reading for 9 volts in crest mode using the 6000 count range would be +/- 1.014 V.
If we were reading say 5 V in the same 6000 count crest mode, then we would have a resolution of .001 (5.000V) because we are in the 6.000 V range, and the +/- 100 counts would only amount to +/-.1 V (100*.001). The specified accuracy would also be less at +/-.005V (.00045*9 + .001*1), resulting in a total tolerance of +/- .105 V.