Multimeter Bandwidth

[jhufford]

Hi, I'm in the market for a new multimeter and I'm curious as to how much true rms bandwidth one should have. I actually can't remember ever even using the AC voltage setting on my current multimeter at home other than to merely see if an outlet is on or not. At work I've used them to measure overall power consumption of a rack of equipment for sizing up power supplies, so all just 60Hz stuff.

From what I can tell, here's the bandwidths of 3 nice $100 class meters
BK Precision BK2709B 500Hz
Amprobe AM270 20KHz
Brymen BM257 400Hz (per Lightages correction, thanks)

and for reference, the venerable Fluke 87-V is listed as 20KHz

So clearly the amprobe sticks out of the $100 bunch, but what is that good for?  My guess is that the trms is often used with crest mode to get the RMS reading of the ac wave along with the max value, to give you an idea of the level of distortion present. I could see where this could be very useful in an industrial setting with large inductive loads like motors. So 50/60Hz is obviously important (and distortion introduces higher harmonics, so you'd want several multiples of that). But is past a couple of kilohertz needed often?

So what do you use AC voltage mode for and how much bandwidth is necessary?

[Lightages]

Actually the BM257s is rated from 50Hz to 400Hz at 1% accuracy on AC voltage and AC current.

There was another discussion at one point as to what was really a useful bandwidth for a multimeter. It was basically the consensus that there was not much use, except for audio work, to need beyond 1kHz if even that much.

For the crest function on my Brymen BM869s, it functions like you said, I think. If I measure my mains at 220VAC, I see a max of 304.6 and a minimum of -304.6.

[retiredcaps]

In the $100 USD range, you can buy an used Fluke 87 (original) series I and it has a 20kHz as well with the usual qualifications as per

https://physics.ucsd.edu/neurophysics/Manuals/Fluke/87______umeng0800

[Wytnucls]

Error in RMS measurements arise when there is significant input signal energy at frequencies above the multimeter’s bandwidth, like in a series of pulses of short duration.

http://rfmw.em.keysight.com/spdhelpfiles/truevolt/webhelp/US/Content/__F_Tutorial/Tutorial%2004%20AC.htm

[dom0]

Well, for most wideband signals in excess of a few millivolts the shiny smart thinkin' DSOs everyone has on their desks are just as good as any RMS voltmeter.

[TimFox]

When doing precision noise measurements, I found the Marconi 2610 wide-band (10 MHz) true rms digital voltmeter very useful, since one can change the averaging time ("m" in "rms") by front panel control.  Of course, this is not a multimeter. 

[EEVblog]

There was another discussion at one point as to what was really a useful bandwidth for a multimeter. It was basically the consensus that there was not much use, except for audio work, to need beyond 1kHz if even that much.

Yes, over in the supporters section.
And yes, the consensus seemed to be that bandwidth isn't as important as what is used to be.
It's a bit of a throw-back to the old days of the first digital multimeters when testing audio circuits was a big use for a meter, so at least 20KHz bandwidth was important.

[G0HZU]

A meter with a flat response beyond maybe 50kHz can be useful for doing basic (S+N)/N measurements for comms radios (eg ham/PMR/CB)especially if the meter has a dB and dB relative function.

Part of the appeal of the Keithley 2015 THD for me was the ability to go a step further and do SINAD measurements using its onboard DSP. I found this meter to be remarkably accurate for this when testing at typical comms S/N ratios.

But usually I reach for my old Racal true rms meter for general purpose true rms measurements. It has a BW of 20MHz and can give useful measurements down to 30uV. It's a trusted workhorse and I've used this type of meter all my working life

[Wytnucls]

20MHz? Which Racal is that?
Most have a bandwidth of 300 kHz with a crest factor of 7.

[TopLoser]

20MHz? Which Racal is that?
Most have a bandwidth of 300 kHz with a crest factor of 7.

RACAL 9300 is my guess.

[G0HZU]

Yes, it's an old Racal 9300. Sorry for any confusion, I should have mentioned that it isn't a multimeter as such.  It just measure Vrms.
My Keithley 2015 bench multimeter has a true rms detector BW up to several hundred kHz but the accuracy suffers at high frequency, especially at small signal levels. I think the DSP has a BW up to 20kHz with up to 50kHz for THD measurements.

[macboy]

A nice flat passband up to 20 kHz and preferably up to 50 or 100 kHz can be used as part of a system to measure the frequency response of an audio amplifier.

Another use is measuring ripple (and noise) in SMPS supplies. This is usually down to a few or few tens of mV AC, and maybe 25 kHz to 200 kHz depending on the design. Most meters will require at least a couple mV at the input to overcome the noise of the RMS converter. Meters with a very limited passband (500 Hz, or a few kHz) often use sampling and calculation of RMS so they have a lower noise floor than the type that uses a wideband RMS-to-DC converter IC.

My Keithley 2001 are flat within 0.1 dB up to 6 MHz. They can also display the DCV, ACV, and frequency at the same time, very nice for measuring power supply ripple. They can also display peak, average, crest factor, etc.  It's an older design, but a very nice one. And that VFD... drool.

[G0HZU]

Another use is measuring ripple (and noise) in SMPS supplies. This is usually down to a few or few tens of mV AC, and maybe 25 kHz to 200 kHz depending on the design. Most meters will require at least a couple mV at the input to overcome the noise of the RMS converter.

That's where the old Racal meter really scores because it has 20MHz bandwidth and can measure reliably down to 30uV. eg measuring ripple and noise on a linear PSU across a very wide bandwidth down to << 1mV. It can also be used to check/calibrate an HF noise source down to very low signal levels.

[dom0]

Don't forget the old and trusty HP 3400A/B. The A is spec'd to ~10 MHz but has a typical bandwidth of ~20-25 MHz (-3 dB) on a 50 ? source. I don't know about the B, but it probably has a bandwidth in the 40+ MHz region

[G0HZU]

One other thing I've used the Vrms function on the Keithley 2015 for is measuring capacitors for capacitance and ESR.

It has enough BW to measure up to several hundred kHz using a function generator (sine) and a sense resistor.
I get good results with it even for small caps with low ESR.

[jpb]

I guess the lower ranges are designed for power supplies up to aircraft frequencies (400 Hz). The next step up is audio (20kHz).

I would have thought for audio work they wouldn't be much use for accurate work but they could be very useful for repair work to check the signal level at various points.
They might also allow the checking of audio transformers (I'm no expert - just surmising).

[DG5SAY]

Have a look with a DSO at the mains voltage today. It´s not a pure sine anymore, because there are a lot of SMPS and things like that. So you need a TrueTRMS DMM with at least up to 2 kHz to get the real RMS voltage. There is a reason that according to IEC 61000-3-2, the current of a DUT is measured up to the 40th harmonic (2 kHz @ 50 Hz mains).
I wish you always a clean mains power!

[dom0]

Have a look with a DSO at the mains voltage today. It´s not a pure sine anymore, because there are a lot of SMPS and things like that.

I don't know why people keep saying that, "heavy metal" (transformer + rectifier + filter caps) supplies load the mains very similarly to SMPS. SMPS with PFC are actually much better than aforementioned supplies.

... not even talking about universal motors ... which have horrifying waveforms...

[DG5SAY]

Have a look with a DSO at the mains voltage today. It´s not a pure sine anymore, because there are a lot of SMPS and things like that.

I don't know why people keep saying that, "heavy metal" (transformer + rectifier + filter caps) supplies load the mains very similarly to SMPS. SMPS with PFC are actually much better than aforementioned supplies.

... not even talking about universal motors ... which have horrifying waveforms...

... I guess because most of SMPS don´t have PFC

[dom0]

Most older ones don't have PFC - true, however, nearly every newer SMPS with more than a couple dozen Watts or so has PFC, either active or passive, simply because you can't match the regulations without one.

[eas]

A nice flat passband up to 20 kHz and preferably up to 50 or 100 kHz can be used as part of a system to measure the frequency response of an audio amplifier.

Another use is measuring ripple (and noise) in SMPS supplies. This is usually down to a few or few tens of mV AC, and maybe 25 kHz to 200 kHz depending on the design. Most meters will require at least a couple mV at the input to overcome the noise of the RMS converter. Meters with a very limited passband (500 Hz, or a few kHz) often use sampling and calculation of RMS so they have a lower noise floor than the type that uses a wideband RMS-to-DC converter IC.

My Keithley 2001 are flat within 0.1 dB up to 6 MHz. They can also display the DCV, ACV, and frequency at the same time, very nice for measuring power supply ripple. They can also display peak, average, crest factor, etc.  It's an older design, but a very nice one. And that VFD... drool.

That's good to know. I guess I need to keep my eye out for a deal on a 2001. I'd been thinking I might be happy with a 2010, but...

For what it is worth, Keithley gives states ACV and ACI bandwidth on the 2000, 2015 and 2700s goes up to 300KHz.  Accuracy ranges from ~0.4% to 0.65% below 100KHz and drops to ~4.5% from 100KHz to 300KHz.

[dom0]

A 3400 is much cheaper and has more bandwidth. And is less accurate... but who cares.

A 3403 has even more bandwidth (specified to 100 MHz or so)...