precision sine wave reference

[mdanh2002]

Hi,

Are there any ICs that generate a precise sine wave reference? I want to test the true RMS performance of my multimeter.

So far to test resistance measurement, I use 0.001% 1K and 10K precisions resistors purchased from eBay and to test DC voltage, I use MAX6161AESA+ for 1.25V 0.02% reference, which work just fine. For frequency I use a Japan-made 26MHz 1.8V 1.5ppm TCXO which I purchased from eBay to verify the performance of the multimeter and my frequency counter.

Thanks.

 

[NorthGuy]

True RMS depends on voltage measurements and calculations. You already got measurements with 0.02% precision. So, the problem is calculations. You do not need a reference signal to verify your calculations.

True sine wave is the most favorable shape where interpolation errors cancel each perfectly over the course of the cycle. Various deviation will only increase rms errors. Therefore, rms tests on pure sine wave may create an impression of false accuracy.

 

[mdanh2002]

True RMS depends on voltage measurements and calculations. You already got measurements with 0.02% precision. So, the problem is calculations. You do not need a reference signal to verify your calculations.

True sine wave is the most favorable shape where interpolation errors cancel each perfectly over the course of the cycle. Various deviation will only increase rms errors. Therefore, rms tests on pure sine wave may create an impression of false accuracy.

Thanks. I understand now.

Is there any way to verify if a multimeter indeed performs true RMS measurement instead of just performing normal averaging? A lot of cheap multimeters on eBay (extech, uni-T, mastech, etc.) claim to be true RMS meters.

 

[audioguru]

How can you trust the precision of junk bought on eBay??

 

[mdanh2002]

How can you trust the precision of junk bought on eBay??

Hi there,

It is definitely not true that all stuffs from eBay are junk. Maybe some from Chinese sellers, in particular rechargeable batteries and fake stuff (Chinese iPhone), are indeed junk. But as far as electronics components are concerned, I have never had any major issues. For these precisions resistors and oscillators which I bought, I verify them with a lab-calibrated multimeter and another frequency counter at my company, and they are indeed within their claimed accuracies.

 

[NorthGuy]

A lot of cheap multimeters on eBay (extech, uni-T, mastech, etc.) claim to be true RMS meters.

They're unlikely to have a good accuracy. Even my Fluke-289 only measures RMS voltage with 0.5% accuracy (compare to 0.025% on DC voltage), and I'm not sure there are more precise hand-held meters out there. Perhaps someone at nearby University has a calibrated benchtop multimeter and they will let you use it to verify/calibrate your device.

 

[dougy83]

Is there any way to verify if a multimeter indeed performs true RMS measurement instead of just performing normal averaging? A lot of cheap multimeters on eBay (extech, uni-T, mastech, etc.) claim to be true RMS meters.

The averaging method is used (and is valid) when it's assumed that you're measuring a sinusoidal signal. You should therefore measure a non-sinusoidal signal to see if it's measuring RMS. e.g. a squarewave could be used (in which case the averaging method will be quite wrong).

 

[mdanh2002]

The averaging method is used (and is valid) when it's assumed that you're measuring a sinusoidal signal. You should therefore measure a non-sinusoidal signal to see if it's measuring RMS. e.g. a squarewave could be used (in which case the averaging method will be quite wrong).

Thanks! Let me try with square wave and see how my cheap true RMS meter performs.

 

[crutschow]

Yes, a square-wave is a good measure of the accuracy of a true RMS meter since it's relatively easy to generate a square-wave of known amplitude. For example, using a CMOS toggle FF operating from a known DC source to give an exact 50% duty-cycle with amplitude equal to the supply voltage which gives an RMS value of 50% of the supply voltage. An average responding meter would display a value of .707/.636 = 55.5% of the supply voltage.

 

[MrAl]

Hi,

I think you meant to say that the average power of a square wave is 50 percent, as the RMS value is 1/sqrt(2) which is about 0.7071, and unfortunately that's the same as a sine wave so if the meter uses a peak measurement then the square wave will read the same as the sine wave. If it does use an average measurement then it will read around 55 percent while the sine will read around 70.71 of the peak.

 

[NorthGuy]

If you make it into AC square wave (by measuring it relative to V/2), all - peak, average, and rms - will be the same and equal to V/2. So, you'll know right away which of the three the meter uses.

 

[crutschow]

...............................
I think you meant to say that the average power of a square wave is 50 percent, as the RMS value is 1/sqrt(2) which is about 0.7071, and unfortunately that's the same as a sine wave so if the meter uses a peak measurement then the square wave will read the same as the sine wave. If it does use an average measurement then it will read around 55 percent while the sine will read around 70.71 of the peak.

Yes, I was off on my math. Your calculation for a peak reading meter is if it's DC coupled. If it is AC coupled the it would read 1/2 of the peak-peak value and would thus display .707 * 0.5 = .344 of the PP value. An AC coupled average responding meter would display .707/.636 * 0.5 = .55 as you noted.

That brings up an interesting point. To measure true RMS of an asymmetrical waveform (average value not 0V) the meter must be DC coupled since a 0 to 1V square-wave has a different RMS value than a +0.5v to -0.5V square-wave, for example.

 

[NorthGuy]

That brings up an interesting point. To measure true RMS of an asymmetrical waveform (average value not 0V) the meter must be DC coupled since a 0 to 1V square-wave has a different RMS value than a +0.5v to -0.5V square-wave, for example.

The combined RMS is V_RMS = Square_Root(V_DC^2 + V_AC_RMS^2). Many true RMS meters can measure both V_DC and V_AC_RMS at the same time. It's easy to calculate. V_DC is the average voltage. Then V_AC_RMS = Square_Root(V_RMS^2 - V_DC^2) - sort of standard deviation. In the case of 0V to 1V 50% square wave, V_DC = 0.5V and V_AC_RMS = 0.5V, so V_RMS = Square_Root(0.5^2 + 0.5^2) = 0.707V.

Power loss through a resistor can also be calculated as a sum of losses caused by V_DC and V_AC_RMS components.

 

[mdanh2002]

I just experimented again today on using a GW-INSTEK GFG-8019G function generator to produce ~50Hz square and sine wave. I measure the waveforms using a HP 54600B 100MHz oscilloscope to know roughly the expected measurement values and measure them using 4 different multimeters: Victor VC921, Radio-shack 22-812, Fluke 17B and Appa 207. Except for the Appa 207 which is true RMS, the rest just uses (I presume) averaging method when measuring AC.

I know that's a lot of equipments for a normal electronics hobbyist, but I can't resist buying cheap test gears from eBay from time to time

Below is the result:

Sine wave which measures 3.844V peak-to-peak and 1.357V RMS on oscilloscope reads 1.329, 1.340 and 1.346V on the averaging meters and 1.3527V on the true rms multimeter.
Square wave which measures 3.906V peak-to-peak and 1.922V RMS on oscilloscope reads 2.108V, 2.129V and 2.141V on the averaging meters and and 1.939V on the true rms multimeter.

So with a sine wave, all meters show approximately the correct value. With a square wave, the averaging meters show higher than the expected rms value, and only the Appa 207 (true RMS) shows the correct value. So the Appa 207 is indeed true RMS.

One thing I notice is that the AC voltage measurement in the averaging meters is good to about 1kHz only - any frequency higher than that will read lower value. The Appa 207 multimeter can still measure the RMS value of frequency in the range of 50kHz. Did not try higher frequency though.

Attached are some of the photos taken during the experiment.

 

[MrAl]

Hi,

The rms and avg values of a square wave centered at 0.000 volts. are the same.
For the case of 3.906v peak to peak, that gives us 1.953v rms and 1.953v avg and 1.953v peak.
If the other three meters read avg, then they will multiply it by 1.11 and that gives us 2.168v which they would try to pass off as rms. If the other three read peak they will do the same thing and display the same result.

Another way is to use a triangle wave with a DC offset. The avg and rms values will then be very different.

Only the better meters can usually read well when measuring a signal over say 100Hz, so 50Hz should be good.

 

[NorthGuy]

These non-rms meters do not pretend to be true-rms.

 

[MrAl]

Hi,

QUOTE MrAL: "they try to pass off as rms"
QUOTE NorthGuy: "These non-rms meters do not pretend to be true-rms"

Nobody said "true rms". For myself, i said simply "rms", not "true rms".
What they do do (ha ha) is they try to show the rms value via built in (and constant) scaling, but it only works for sine waves. So for sine waves they show the rms value, and if they get lucky it works for some other waveshapes. It certainly does not work for all wave shapes.

 

[NorthGuy]

Hi,

QUOTE MrAL: "they try to pass off as rms"
QUOTE NorthGuy: "These non-rms meters do not pretend to be true-rms"

Nobody said "true rms". For myself, i said simply "rms", not "true rms".
What they do do (ha ha) is they try to show the rms value via built in (and constant) scaling, but it only works for sine waves. So for sine waves they show the rms value, and if they get lucky it works for some other waveshapes. It certainly does not work for all wave shapes.

MrAl,

I wasn't responding to your post. The whole true-rms thing started from discussing Chinese DMMs who pretend to be true-rms while they're not. But the DMMs tested didn't have any "true-rms" markings (nor were they Chinese for that matter (perhaps one)), so we still don't know if Chinese "true-rms" DMMs are true rms or not

 

[mdanh2002]

MrAl,

I wasn't responding to your post. The whole true-rms thing started from discussing Chinese DMMs who pretend to be true-rms while they're not. But the DMMs tested didn't have any "true-rms" markings (nor were they Chinese for that matter (perhaps one)), so we still don't know if Chinese "true-rms" DMMs are true rms or not

Hi MrAl,

Thanks for your calculations and suggestions. As always, I learned a lot by experimenting and posting the findings on this forum

The Appa 207 is made by Appatech (http://www.appatech.com/) - a Taiwanese company. Not sure if this is "considered" Chinese-made. Anyway it claims to be true RMS (see specs at **broken link removed**) and seems to be during my testing.

As far as my electronics hobby goes, I am prepared to purchase another Uni-T 61E or Uni-T 139C, which claimed to be true-RMS and see how good the true RMS performance is.

 

[MrAl]

Hi,

NorthGuy:
Ok, thanks for clearing that up. I also just wanted to bring out the distinction between rms meters and true rms meters. I believe if they state "true rms" then they should be able to handle several waveform types without having to manually apply some correction factor.

mdanh2002:
Yes it does look decent, and maybe you can try some measurements on a triangle wave, both centered at zero and also with some large DC offset. That would be interesting too.