• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Measuring the power of an amp, with an oscilloscope

Bow_Wazoo

Addicted to Fun and Learning
Joined
Jan 3, 2021
Messages
758
Likes
661
Today I found this interesting video where the guy is doing a power measurement with an oscilloscope.


Is it actually possible to estimate from how much %THD, the sine wave on the oscilloscope appears to be cut off, or no longer 100% clean?
 

Doodski

Grand Contributor
Forum Donor
Joined
Dec 9, 2019
Messages
20,745
Likes
20,757
Location
Canada
Today I found this interesting video where the guy is doing a power measurement with an oscilloscope.


Is it actually possible to estimate from how much %THD, the sine wave on the oscilloscope appears to be cut off, or no longer 100% clean?
Oscilloscope power measurement is a @ clipping measurement only. %THD cannot be measured unless there is some special feature built into the O-scope. I have never seen a O-scope with %THD measurement. But then I have never had a digital O-scope. Always bought Tektronix analogue and analogue storage O-scopes. :D
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Today I found this interesting video where the guy is doing a power measurement with an oscilloscope.


Is it actually possible to estimate from how much %THD, the sine wave on the oscilloscope appears to be cut off, or no longer 100% clean?
Most DSOs include an FFT function so you could get the THD from that. Probably manually, using cursors, not sure they have a built-in THD function (but the 'scopes I use do not target audio measurements so maybe some do?)

Note most DSOs use 8-bit converters, though 12-bit DSOs are becoming more common (and some boards use 16-bit ADCs), and have fairly noisy front ends (relative to the uV levels needed for audio measurements) thus it is not a good tool for low-level distortion (or noise) measurements.

For power measurements on a DSO (when I do them, pretty rare) I tend to use a current probe plus a voltage probe to get RMS I/V values (letting the 'scope do the math for me), or a special sensor and amplifier to get the current (a current-to-voltage converter), then let it calculate the average power.
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia
Oscilloscope power measurement is a @ clipping measurement only. %THD cannot be measured unless there is some special feature built into the O-scope. I have never seen a O-scope with %THD measurement. But then I have never had a digital O-scope. Always bought Tektronix analogue and analogue storage O-scopes. :D

If we have the input signal and differenced that with the amp’s output (attenuated), then we could get THD+N?
Not sure how, but it seems possible??
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia
The O-scope is a low resolution measure. It's just not in the nature of the device for %THD measures.

We need a lot of dynamic range and resolution to get two numbers like 1.000000 and 0.999999 and difference them to get a meaningful number.
If we subtract one signal from the other in analogue, then we do not need a lot of resolution… because we have removed most of the offset.

I know O-Scopes have 2 inputs, but I have not used one in a long time.
I think we can do lissajou figures on a O-scope, so we could see it is linear to a point and then fold over?
 

Doodski

Grand Contributor
Forum Donor
Joined
Dec 9, 2019
Messages
20,745
Likes
20,757
Location
Canada
We need a lot of dynamic range and resolution to get two numbers like 1.000000 and 0.999999 and difference them to get a meaningful number.
If we subtract one signal from the other in analogue, then we do not need a lot of resolution… because we have removed most of the offset.

I know O-Scopes have 2 inputs, but I have not used one in a long time.
I think we can do lissajou figures on a O-scope, so we could see it is linear to a point and then fold over?
It's still a low resolution device and meant for low resolution measures. I have had differential/balanced O-scope resolution down to 5 micro volt per division and I never was able to measure %THD. A cassette tape head output at the max resolution yes it was viewable but anything else is pointless and not achievable. Maybe you know something I don't but I don't think it can be done.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
If we have the input signal and differenced that with the amp’s output (attenuated), then we could get THD+N?
Not sure how, but it seems possible??
You could try, having it calculate the difference and generate the RSS values to get the ratio to create a SINAD number (most DSOs have fairly good math functions), but as @Doodski said it would be low in resolution with a dynamic range of perhaps 40 dB or so. I have done it using one probe on the input and another on the output, then getting the difference matched as close as possible to create the error signal, but the results weren't all that great best I recall. It is a bit of a PITA. You could get more range with averaging, but fundamentally it's the wrong tool for the job IME/IMO. A spectrum analyzer or regular audio analyzer is a better tool for distortion measurements.

Unfortunately the equipment I have at work has high dynamic range and accuracy but cuts off above the audio band. The DSOs would work, since they can be set to sample slowly and have a lot of memory, but as I've said before I am queasy about signing out a DSO to bring home that costs more than my home. :)
 
OP
Bow_Wazoo

Bow_Wazoo

Addicted to Fun and Learning
Joined
Jan 3, 2021
Messages
758
Likes
661
Thanks guys, I understand.
I will still build a 50 Ohm dummy for fun and see what the FA-10 will show on the DOS.

Now I'm just wondering from what % THD the curve will no longer look clean.
2%, 10%?
Pure speculation, right?
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Thanks guys, I understand.
I will still build a 50 Ohm dummy for fun and see what the FA-10 will show on the DOS.
A 50-ohm load would be for RF; for an audio amplifier you'd probably want 8 ohms.

DSO? Digital Sampling Oscilloscope
Now I'm just wondering from what % THD the curve will no longer look clean.
2%, 10%?
Pure speculation, right?
I can see about 1% if I have a "pure" sine wave with which to compare. 10% should be fairly obvious. These threads might help:
 
OP
Bow_Wazoo

Bow_Wazoo

Addicted to Fun and Learning
Joined
Jan 3, 2021
Messages
758
Likes
661
Oh, I thought DSO is the abbreviation for digital oscilloscope....

8Ohm?!
The FA-10 is a headphone amplifier.
And my HE6SE has 50Ohm impedance.
 

EdW

Senior Member
Joined
Jul 5, 2020
Messages
329
Likes
413
Location
Cambridge, UK
If we have the input signal and differenced that with the amp’s output (attenuated), then we could get THD+N?
Not sure how, but it seems possible??
This was a technique used in the 1960’s by the likes of Peter Walker of Quad fame. He cancelled the audio oscillator signal with an attenuated version of the amplifier output and examined the residue signal. The oscillator signal feeding the cancellation signal was put through an RC filter to mimic the frequency response of the amplifier under test. The technique had the advantage that amplifiers with lower distortion than the oscillator could be evaluated.
 

HarmonicTHD

Major Contributor
Forum Donor
Joined
Mar 18, 2022
Messages
3,326
Likes
4,829
Today I found this interesting video where the guy is doing a power measurement with an oscilloscope.


Is it actually possible to estimate from how much %THD, the sine wave on the oscilloscope appears to be cut off, or no longer 100% clean?
Just as an alternative.
If you want to measure THD eg for DIY purposes. An audiointerface (eg Focusrite or Motu ca 150USD, ca 0.002% THDplusN) will work together with REW (free).

This is what Julian Krause (YouTube) uses asfik as an audio analyzer is quite expensive.
 

DVDdoug

Major Contributor
Joined
May 27, 2021
Messages
2,917
Likes
3,831
For the power measurement-calculation you shouldn't have to worry too much about distortion because solid state electronics tends to hard-clip and the peak voltage should be about the same at 1% or 10% distortion.

With bad clipping the voltage might even droop a bit giving you a more conservative calculation. ...That's more likely with a power amplifier than a headphone amp. The power supply voltage in a headphone amp should hold-up because the current requirements are minimal.

The true-power (related to the area under the curve) continues to go up with clipping but your calculations will be based on a sine wave.

And if you're not near clipping when listening to music it's unlikely that you have audible distortion (from the amplifier).
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Oh, I thought DSO is the abbreviation for digital oscilloscope....

8Ohm?!
The FA-10 is a headphone amplifier.
And my HE6SE has 50Ohm impedance.
Your post said "DOS", which to me is the old Disk Operating System.
I will still build a 50 Ohm dummy for fun and see what the FA-10 will show on the DOS.

I didn't go back to the top of the thread, sorry, just read about measuring power and jumped to speakers since that is usually the only place in the audio chain we care about power. I had no idea what an FA-10 is/was. 50 ohms would be appropriate if you want a simple load emulating your headphones.

Carry on - Don
 
OP
Bow_Wazoo

Bow_Wazoo

Addicted to Fun and Learning
Joined
Jan 3, 2021
Messages
758
Likes
661
I've made a small dummy.
However, only for one channel.
To what extent could the measured values be influenced by the fact that I only load one channel?
(balanced output)

20220624_075535.jpg
 
Top Bottom