Gentlemen this is about the Smooth Feature on the DMM's

[Kiriakos-GR]

People I need lots of enlightenment about the  Smooth Feature - Function. ( what ever you called it .)

In theory :
When the input signal changes rapidly, “smoothing”
provides a steadier reading on the display.

Or..

Smooth is used to smoothen the refresh rate of the readings
in order to reduce the impact of unexpected noise and to
help you achieve a stable reading

Ok, so far so good, the point is how much is enough ?

The advanced versions of the Fluke DMM, they had an fixed refresh about the Smooth Feature.
And the latest Agilent U1272A offers the potential of manually adjust. 

And the key question is: How slow it called as very slow ?
Fluke does not say, how many milliseconds delay added, when their  “smoothing” gets enabled.

From the other hand, the Agilent U1272A it is so extremely fast,
that definitely needs active “smoothing” in some types of measurements.
In their User Manual they speak about adjustment from (0001 to 9999) but they do not say what those are ?
Are they milliseconds ?

9999 Milliseconds = 9.999 Seconds

[EEVblog]

In their User Manual they speak about adjustment from (0001 to 9999) but they do not say what those are ?
Are they milliseconds ?

Averaging is almost always quoted in number of samples, most likely a running average in this case of a smoothing feature on multimeters.

Dave.

[Kiriakos-GR]

ok I am trying hard here so to get the point or to translate it in something understandable.

Questions : With out “smoothing” , the DMM operates at the Max sample rate that could be 9999 ?
And so if I set less samples per second, I am slowing it down ?

[alm]

It's likely using a windowed moving average, which calculates the average over the last n points, but resets the sample population (restart from scratch) if the new value is outside the window (eg. avg +/- 10%). This is designed to smooth small fluctuations, but leave transient response to major changes in tact. Averaging over a larger number of samples will slow down response to gradual changes.

[EEVblog]

ok I am trying hard here so to get the point or to translate it in something understandable.

Questions : With out “smoothing” , the DMM operates at the Max sample rate that could be 9999 ?
And so if I set less samples per second, I am slowing it down ?

The ADC in the DMM always operates at the max sample rate, and the display update speed (say 4 times/sec) likely remains the same too.
The difference is what value is displayed.
Without smoothing, the ADC just send the value directly to the display each time.
But with smoothing set to say 4, the processor calculates the average of the last 4 readings and then displays that. Perhaps with "windowed" method like ALM explained which is able to reset the average count.

Dave.

[Kiriakos-GR]

Ok let me present an practical problem.

I have one AC/DC clamp (Hall type),
it has an output of 1mV /A (0.4-400A) & 10mV per Ampere ( 0.2 - 40A )

At the 10mV Per Ampere range, I have an continues play-up of 0.3mV no matter what.
This small play-up forces the  three last digits from the five, to dance ...
 
I did experiments about 4 hours, and I found, that I had to add at the Agilent one amount of samples as smoothing,
that has an analogy with the play-up voltage.

In order to find that, I had to use first the Min Max Average,  and to extract the Min from the Max,
so to find the value of the actual play-up voltage.

I think that I succeeded my goal, by finding the method of how much smoothing to add at the unstable source,
but I need to verify that and with another unstable source, but I do not have any ..   
 

[alm]

Smoothing settings likely depend on the frequency of the signal you want to reject, not the amplitude. If it does a windowed average, a large amplitude may effectively disable the smoothing, unless you increase the window size.

[Kiriakos-GR]

In my case, I did not needed to smooth the reading of a measurement.
I just needed to improve the cooperation of the DMM with the AC/DC clamp.

My measurements at my tests, was with regulated DC.
About AC, the clamp is so professionally made, that it will handle anything up to 1Khz or even up to 10KHz,
In a hardware level.
The output would still be in mV DC. 

Today I did some long lasting tests,  I left my large APC 1000XL UPS unit with external battery package,
to totally discharge, and charge.
I have about two hours on battery life, and I did monitored the all process ( took about 4 hours) by monitoring the Amperes of the external battery pack.
 
Those APC are extremely clever made,  when I stopped the Mains,
the UPS become active and started to use exclusively the external battery pack,
and so I measured about 6.5A at 24V as consumption.
Later on, it started to move the load from the external package to the internal one.
And many minutes later, when the internal package started to run out of charge,
the UPS combined the strength of both battery packs.
When both battery packs (27V) fall down to 22V =  11V  per cell ( 13.8V), it shut down.

I use to run battery calibration on the UPS once per year,
but this time I had a complete view of what was going on.
I had the Agilent measuring the Amperes with the clamp.
And the Fluke 28II was measuring the voltage. 

[Kiriakos-GR]

I came to update this thread with the answer of Agilent, about the Smooth function on their latest DMM.

Basically, there are three samples rate when SMOOTH MODE is engaged.
Namely:
SLOW = 15 averaging/sec
NORMAL = 10 averaging/sec
FAST = 5 averaging/sec.



And I will say thanks Agilent, but if you give a choice to enter  four digits in that menu, you making it confusing.
Just try to add this info too, in the next update of the users manual.