EEVblog 1379 - What's all this NPLC Stuff Anyhow?

[EEVblog]

Why does a handheld multimeter read zero volts and a bench meter reads several volts?
It's all about the NPLC's man...

[pplaninsky]

What does ‘buela’ means? Is that some Aussie thing? Dave uses it in the same context as one would use ‘wtf’. So, what does it mean? Where does it come from?

[EEVblog]

What does ‘buela’ means? Is that some Aussie thing? Dave uses it in the same context as one would use ‘wtf’. So, what does it mean? Where does it come from?

[TheSteve]

Too many switching supplies spewing garbage in your lab Dave. I have no where near that much noise in my lab.

[pplaninsky]

Thanks! Now, I got to watch that movie

[EEVblog]

Too many switching supplies spewing garbage in your lab Dave. I have no where near that much noise in my lab.

It's 50Hz, not switching noise.

[TheSteve]

Too many switching supplies spewing garbage in your lab Dave. I have no where near that much noise in my lab.

It's 50Hz, not switching noise.

Fair enough - although it sure seemed like you had a lot of cables running around in the ceiling etc connected to the switching supplies for the lights. What is causing so much 50 Hz to be spewing everywhere? Is this a bigger issue with 50 Hz vs 60 Hz?

[ejeffrey]

If you put the handheld meter on ACV does it show 0.5 Vrms?  NPLC averaging won't help you on AC volts, but I certainly don't see big AC signals on my handheld meter -- maybe your lab is a lot worse than mine but I doubt the difference is that great.  I would guess that part of the issue is that the battery powered handheld meter has better common mode rejection because it isn't mains referenced.

[dcac]

If you put the handheld meter on ACV does it show 0.5 Vrms?  NPLC averaging won't help you on AC volts, but I certainly don't see big AC signals on my handheld meter -- maybe your lab is a lot worse than mine but I doubt the difference is that great.  I would guess that part of the issue is that the battery powered handheld meter has better common mode rejection because it isn't mains referenced.

On my BM869 with two standard 1m test leads depending on how they are laying on my work bench I can easily get something like 0.3 to 0.8 Vrms. If I i.e. let one lead hang towards the floor I can get 2 to 4 Vrms. This is without really having any mains powered equipment on - though this is an electrically heated house and probably has more mains wires running in the walls than normal.

If I change to the DC position I get just a few mV but I can see the bargraph flickering madly so the 50Hz mains is still being picked up but gets filtered out almost completely by the averaging sinc filter. The BM869 number display has a fixed output rate of 5 SPS or 200mS per sample which correspond to 10 cycles of 50Hz.

[Kleinstein]

The mains isolation helps the handheld DMM a littel, but not very much. Especially the higher end bend DMMs have quite good separation from the mains supply - though still a few 100 pF (in part intentional) to ground.

If one uses the right number of samples and has the sampling rate fixed and right, one can to post measurement filtering and average to also get 50 Hz / 60 Hz suppresion by post processing.  It is not as simple as averaging 5 readings at 0.2 PLC to get again 20 ms total integration time. It is more about getting the samples evenly spaced over the mains phase. So this may work with some 4 or 9 or 14 readings at 0.2 PLC depending on the time lost between samples. Post processing/averaging) in some aspects is not as good (especially with sharp main synchronous spikes), but can also have some adjantage, if the mains frequency is not perfect and the toltal aperture time is long.

Not all bench DMMs use an integrating ADC. Some older Fluke (and AFAIK also Phillips) meters like the 8840 actual use a more classical (kind of successive approximation) fast sampling ADC and than post processing (averaging) to get mains suppresion.

With the Dual slope ADC in especially older meters the integration time is only part of the conversion. So more like 100 ms integration and some 200 ms for the rest of the conversion to get some 3 readings per second. Some new handheld DMMs start using sigma delta ADCs and these can get near 100% aperture, so 200 ms integration and 5 SPS.

The auto zero mode with the  bench meters also adds additional time, often doubling the time needed for a conversions. So even there 10 PLC integration does not mean 5 readings per second, but maybe only 2.5 SPS.

[Labrat101]

Hi ,
I also don't have that much Noise in my lab ether ..
It's properly your inverter AC's  they are notorious for line noise ..
Also they are wired to the same phase as your Lab  ..

[Kleinstein]

Am inverter or PC system can add some high frequency baclground to the mains. This could cause some nasty spikes.  However it would not effect the general 50/60Hz background level. This depends more on the use of grounded / non grounded equipment and the position of cabels. A grounded ESD safe desk liner can also make a difference.

The principles do not change much, unless the background AC level is so high that some amplifier / ADC may reach saturation. The open input terminals is more like an unusual case, more for demonstration. One can also get some mains hum background in a low impedance circuit, from magentic coupling - it would normally not be as large, but for low level measurements one will usually have some hum.

A nice point about the mains hum suppression by integration is that it does not only suppless the 50 Hz quite good, but also the harmonics, e.g. 100 Hz after a simple rectifier.

[Labrat101]

Interesting point . but you missed the real problem .
When the probes are removed the Meter reads Zero.
 Most test leads are about 1 meter long approx .
So having 2 test leads connected they are working as an antenna or better still a dipole  .
any RF radiation from Long wires in the roofing walls will transmit RF at what ever harmonics .
 also the leads are a crazy dipole so your local Ham ,Police etc transmitting on or around 2mtr will be picked up.
as your meter has no tuner it will display random numbers . And all the other random RF harmonics that will tune to your
Test leads .
 Mains suppression will not stop it .

[Kleinstein]

RF influence is a different topic from mains hum. The open leads, but also a not very direct short will pic up some RF. For this, especially the modern meters include EMI filtering at the inputs: there are usually ferrites and capacitors to suppress much of the RF signal. For higheste accuracy one would still want to reduce EMI and keep the mobile phones and other strong emitters away from sensitive electronics. Unless in a really bad setup EMI usually causes small errors like a few steps of the last digit. There is still the chance to build a good (resonant) antenna for a strong local source by accident, but this is rare.
Dave should have the equipment so test how much effect some RF signal at the input could have. With modern meters they usually test this - in parts this is required by regulations. Especially some older meter can be quite bad with this - even analog meters are not totally immune to this. I have an old analog one that showed near full scale on all ranges (including 1200 V DC  ) from some 20 V 50 kHz (AFAIR) signal.

The large scattering that Dave showed in the video are essitially due to mains hum - in the graphing mode the 50Hz mains like waveform is clearnly visible.

[floobydust]

I tracked it down once, most noise on a bench is due to AC hum from the line cords. I monitor current flow on my ESD mat, and was surprised to see it all over the place, in a uA kind of way. Once a mains cord rests on an ESD mat, capacitive coupling (into the usual 1MEG resistor) adds several volts of hum to the worksurface.

For bench multimeters, some have Y-capacitors or at least the power transformer's winding capacitance, to earth-ground. Cheap ones have capacitance pri-sec which adds hum. Some (better quality) bench multimeter power xfmr have the electrostatic shield, others do not.
Point is you need to know how the particular bench multimeter floats, where the leakage is to mains or GND or something noisy in the Inguard portion.
Rarely you can find shielded IEC mains cords which do help a lot.

edit: it's also cell phones with RF TX bursts causing regular beeps and jumps as EMI.  A WiFi router, laptop with Wifi/Bluetooth, cordless phone - these are also sources of nearby RF.
People walk up to me (with phone in pocket) in the lab, and suddenly I'm getting crap on the scope. Grr.
LED workbench lighting is also a source of EMI, I've used grounded wire mesh on ceiling diffusers to stop that, fluorescent lamps are pretty bad as well.
At university, the EE building elevator VFD made so much noise it screwed up all the students lol their circuits acted up mysteriously every time it was running.
An electrically quiet lab is a bit of work.

[ejeffrey]

Interesting point . but you missed the real problem .
When the probes are removed the Meter reads Zero.
 Most test leads are about 1 meter long approx .
So having 2 test leads connected they are working as an antenna or better still a dipole  .

It's not antenna (far field) behavior at 50 Hz and harmonics. It's just capacitive coupling from nearby sources. Each lead has some picofarads of capacitance to the nearby environment.  That can pick up stray electric fields from ungrounded or unshielded equipment and generate a small current.  That current, if unbalanced, flows across the 10M input impedance and creates a signal voltage.