# THD measurements

#### gerrittube

Hi,

I’ve done some THD measurements on my breadboard 6SN7 - 6SN7 - EL34PP project. THD measures 0,39% with 36 Watt output. Now my signal generator has 0,10% THD for 0,32 Volt PP input. My USB audio input also has approximately 0,10% THD.

What THD figure is “real” for the amplifier part only? Should I calculate 0,39 - 0,10 - 0,10 = 0,19%?

Regards, Gerrit

#### rayma

The distortions could add or subtract in various ways, depending on the phase of the harmonics.
Usually you want the test equipment to be 5-10 times better than the DUT for accurate measurements.

#### jhstewart9

D% TOTAL is calculated using ROOT of SUM of SQUARES.
So D total is (d1^2 + d2^2)^(1/2)
Or Total Distortion is root of d1 squared plus d2 squared.
To get your real distortion you can do the math on a pocket calculator
Or with pencil, paper & slide rule like us fossils used to do!

#### Attachments

• Pickett & Eckel Slide Rule 10C 8W.jpg
39.5 KB · Views: 615

#### lcsaszar

My USB audio input also has approximately 0,10% THD.
I would invest in a better sound card, capable of low distortion output and input. You can check it with a loop test.

#### jhstewart9

Hi,

I’ve done some THD measurements on my breadboard 6SN7 - 6SN7 - EL34PP project. THD measures 0,39% with 36 Watt output. Now my signal generator has 0,10% THD for 0,32 Volt PP input. My USB audio input also has approximately 0,10% THD.

What THD figure is “real” for the amplifier part only? Should I calculate 0,39 - 0,10 - 0,10 = 0,19%?

Regards, Gerrit

When you know the source generator D%, square the indicated 0.39, then square the known generator D, 0.10, subtract & take the square root.
So the bottom line is 0.377%, a good number.
Some of the reading on the meter could be hum & other noise, a measurement often called SINAD. Signal + Noise + Distortion, a term often used in the Communication Business.
That is the rdg a common THD Distortion Meter such as the HP 330 Series will indicate.The fundamental is removed & the remainder measured by an average responding meter.
Later THD meters would use an RMS responding meter. But the difference in the indication is small up to about 20%.

#### pblix

I would say "better than 1/2%" and call it good.

#### 6A3sUMMER

Are you going to use the USB input?
Or are you going to be using the analog audio input that the generator is connected to?
You will only be using one input at a time, right?
One input distortion does not add to the other input distortion.

How did you measure the harmonic distortion of the USB input?
Or is it just a specification from the manufacturer?

Measure the 2nd Harmonic, 3rd Harmonic, 4th Harmonic, 5th Harmonic of the signal generator. Write it down.

Measure the 2nd Harmonic, 3rd Harmonic, 4th Harmonic, 5th Harmonic of the signal generator plus amplifier (signal generator connected to the amplifier). Write it down.

Probably, the 2nd and 3rd harmonics are the only ones with any significant percentage.
Higer order, 4, 5, 6, 7 etc. should be far lower in level.

Unless you are near clipping; or unless the open loop amplifier distortion is not very good
and so you are using lots of negative feedback to correct for that, you should not be getting any significant level of higher order harmonic distortion.

Negative feedback pushes down lower order harmonics, and from that, creates upper order harmonics (interesting fact, Huh?).
Refer to documents by Mr. Norman H. Crowhurst.

The 2nd harmonic of the generator and the 2nd harmonic of the amplifier is the most likely harmonic that either partially adds to, or that partially subtracts from each other.

There was some good advice in the posts about using the Root Sum Square method.
For example, an amplifier with 0.5% 2nd, 0.5% 3rd, and no other significant distortion,
the THD is 0.7%, not 1.0%.

If you are going to manufacture and sell amplifiers, you may want to invest in a better signal generator.

Otherwise, if this is for your personal home system, then if you are real worried about the amplifier distortion, you may also want to tool up with some expensive and sophisticated measurement equipment to check the distortion of your loudspeakers at moderate and loud drive levels.
I suspect that for most speaker systems, the speaker distortion is the issue, not your amplifier.

You can do a partial cancellation of the 2nd harmonic distortion of your amplifier plus/minus the 2nd harmonic distortion of your loudspeaker (particularly of the woofer).
Just use your signal generator into the amplifier, and a microphone in front of the speaker, with the mic output to your distortion measuring device.
Measure with the speakers connected one way (red to +, white to -), and then reverse the connection (red to -, white to +). One will be lower distortion.
Just be sure when you get done to connect both speakers the same way, or you will have out-of-phase stereo, and bass cancellation.

If you use Vinyl, you may be surprised by the rated % levels of the phono cartridge distortion.
But take heart, I love my Vinyl recordings. They sound great!
One thing about Vinyl is the distortion is almost all 2nd and 3rd.
And, there are no quantization levels in vinyl.
And there are no fold-back of signals that are out of band, like there can be with CD players.

Then there are the Red Book CDs.
And I love my CDs and CD player too.
They sound great too!

I find that my FM tuners have more distortion, noise, fold-back signals, and birdies than
using the streaming 'signal' of the local classic station.

Just take advice from the publication that is named: "Enjoy the Music".
Relax, and take it in.

#### U130421

Test the distortion of your amplifier in open loop , you get a magnified value to be decreased by NFB .

#### 6A3sUMMER

THD meters do not tell the individual levels of the individual harmonics.

I had many spectrum analyzers at work; now am retired, so do not have access to that.

Most of my amplifiers do not have negative feedback.
I use a digital scope, and its FFT function to look at the individual harmonics.
The digital scope is a very useful tool.

I have a Function Generator. I use its square wave to test for rise time and "ringing".

I use a Denon Audio Test CD. It has extremely low distortion. It has many types of test signals.

Last edited:

#### Mark Tillotson

Negative feedback pushes down lower order harmonics, and from that, creates upper order harmonics (interesting fact, Huh?).
Refer to documents by Mr. Norman H. Crowhurst.

Refer to this and get a fuller story: feedback effects

#### kodabmx

Maybe I'm too simple, but I measure THD with my ears... If it sounds bitched, it probably is. As long as it's pleasing to the ear, Why bother with numbers?

#### U130421

Maybe I'm too simple, but I measure THD with my ears... If it sounds bitched, it probably is. As long as it's pleasing to the ear, Why bother with numbers?
I 100% agree with you , specially if you chose tube technology to enjoy extra harmonics.
You wouldn't buy mineral water and complain it is not pure H₂O.

#### 6A3sUMMER

Agreed, the ears plus brain are great tools for finding out that something about your system is wrong.

But fundamental measurements sometimes are helpful too.

If your amp has a problem, you might be better able to find the trouble with a scope.
An unusual shape to a sine wave
Slew
Clipping
Hum, what frequency, 1x line or 2x line?
An oscillation
etc.

Suppose you build a simple push pull pentode amp that has no negative feedback, but it does have low distortion (it is possible).
Then you connect that amp to your speakers and it sounds bad (the bass sounds un-dampened, and the mid ranges are too loud).
You are so surprised and so disappointed.
Did you forget to check the damping factor?
If you did not measure it, you should have predicted it (due to the un-fedback pentode circuit). But some newbe builders might not know to do either.

Now it is either time to measure the damping factor, or to do one of the things engineering is about . . . predicting the outcome.

Some tools are measurements, some tools are ears plus brain, and some tools are an engineer's brain.

Oh . . . and sometimes some engineers are wrong, for an example, look up the Tacoma Narrows Bridge that USED to exist in the state of Washington.

Last edited:

#### 6A3sUMMER

jhstewart9,

I still have my precision Plastic Laminated Bamboo Post VersaLog Slide Rule.

#### jan.didden

Paid Member
Refer to this and get a fuller story: feedback effects

... which clearly shows that IF you want to use feedback, use lots of it! About the best way to heavily modify your sound is using 6-12dB of feedback .
Unless you want to change the sound, of course.

Which you knew of course.

Jan

#### gerrittube

Hi,

I’m using an external signal generator (Velleman), and a USB audio card with RCA inputs (I get a better one this week). On my PC I have digital scope, spectrum analyzer, bode plotter, etc.. I know how to calculate THD manually by taking dB measurements from the spectrum analyzer. I even created a small Windows program to simplify the calculation by entering the measured dB values for the main frequency and it’s harmonics.

All values were measured, not taken from device spec sheets. I want to know the reality, so that’s why I take so many measurements. After all I have to listen to music, not to specs!

With my new USB audio card I will continue measuring each stage of the amplifier and see what more can be improved.

Regards, Gerrit

#### 6vheater

Did you forget to check the damping factor?
If you did not measure it, you should have predicted it (due to the un-fedback pentode circuit). But some newbe builders might not know to do either.

Now it is either time to measure the damping factor, or to do one of the things engineering is about . . . predicting the outcome.

Paranoia with damping factor is a common problem with "tubey guys".
There is almost no good theoretical reason to assume speakers behave the way they do.
The "damping factor" guff is almost entirely BS, partly because speaker impedances vary with frequency as do ESLs.
The impedance of a typical hi end woofer can go from a nominal 8-10ohms to up to 60 within a space of 1 octave.
How you gonna change that, add more voltage?
It's still not gonna generate any current, and to a NFB network looks like a giant hike in control voltage.

If you don't believe me, read about from the Finnish engineer who developed current feedback control of speakers.
He actually explains that most of the stuff talked about the importance of damping factors as being utter garbage.

Last edited:

#### tikiroo

I don't think it is "damping factor" as such that is the problem. It's that high output impedance causes the frequency response to noticeably follow the loudspeaker impedance, so that when impedance peaks at say 60 ohms, you will get increased output. Which was what 6A3sUMMER was talking about.

#### petertub

Paid Member
I don't think it is "damping factor" as such that is the problem. It's that high output impedance causes the frequency response to noticeably follow the loudspeaker impedance, so that when impedance peaks at say 60 ohms, you will get increased output. Which was what 6A3sUMMER was talking about.
If the impedance peaks at 60 ohm the output might be reduced as most sandamps
has a low output impedance, hence deliver less power into 60ohms.

The driving force of a speaker is dependent of the current flow. Not the voltage
applied across it!

#### kodabmx

But without the voltage, no current flows