New Posts New Posts RSS Feed - Measuring Distortion
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Welcome to the Graham Slee Audio Products Owners Forum

 

Open to all owners plus those contemplating the purchase of a Graham Slee HiFi System Components audio product and wishing to use this forum's loaner program: join here (Rules on posting can be found here)

This website along with trade marks Graham Slee and HiFi System Components are owned by Cadman Enterprises Ltd


Measuring Distortion

 Post Reply Post Reply
Author
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Topic: Measuring Distortion
    Posted: 03 Feb 2011 at 11:24pm
As is my usual way I'm doing this article/thread little by little. I thought I'd begin with something amusing...

The clever or the dumb can obtain fantastic total harmonic distortion figures using the Audio Precision ATS-1 audio test set:



Take a look at the above reading from a Solo SRGII - fantastic isn't it? Only 0.0014% distortion!

Take another look and you'll see I measured it at 20kHz - wow! Everybody knows distortion gets really bad the higher in frequency you go, so the Solo SRGII must have incredibly low distortion at the customary 1kHz!

It's only when you understand that total harmonic distortion is the measure of harmonics and the first one appears at 40kHz (twice the 20kHz frequency and therefore often referred to as second or even order harmonic), then you take a look at the bottom line of the readout and realise where I cheated - yes I cheated. I stuck the high frequency filter in at 22kHz!

Like I said, the clever or the dumb can obtain fantastic distortion figures Wink

Edited by Graham Slee - 04 Feb 2011 at 12:27pm
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 04 Feb 2011 at 11:03am
Introducing N

OK, I started with a bit of fun above, but it does have a purpose.

If the 0.0014% isn't a measure of distortion then surely the meter should have read absolute zero instead of 0.0014%?

So what is that 0.0014%? Firstly please understand that nothing is perfect here on planet earth - well certainly nothing man makes!

Let's assume the 22kHz high frequency filter (sometimes called a low-pass filter or LPF) IS perfect, so the 0.0014% consists of no harmonics and no residual signal, therefore what's left must be noise.

The abbreviation of noise is the capital letter N. Therefore the 0.0014% must be N. But noise is usually represented in decibels (minus decibels). I know the noise figure of the SRGII is -84dB which is 1/15,848 or in percentage terms 0.0063%, so is this a miracle SRGII?

Let's take a look at the meters bottom line again. There's a lower filter point of 400Hz, but when we measure noise we measure from 20 or 22Hz to 22kHz, so we have the lower end of the noise missing, and that's why the reading is only 0.0014% instead of 0.0063%. For those who like to know about these things, low frequency noise is more dominant and here we've filtered it out, so we're only reading the noise above 400Hz.

So, what have we learned? We have learned that when we measure distortion (THD) we also measure noise (N), so the correct convention for measuring distortion is to state it as being THD+N.

Hope you all grasped the above? Don't worry if you didn't, this is not an exam.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
mrarroyo View Drop Down
Moderator Group
Moderator Group
Avatar

Joined: 28 Jul 2008
Location: Miami Beach, FL
Status: Offline
Points: 1401
Post Options Post Options   Thanks (0) Thanks(0)   Quote mrarroyo Quote  Post ReplyReply Direct Link To This Post Posted: 04 Feb 2011 at 11:40am
Actually I understood, and I am not very savy of things electrical.
Miguel
Back to Top
tg [RIP] View Drop Down
Moderator Group
Moderator Group
Avatar

Joined: 19 Jan 2008
Location: Sydney
Status: Offline
Points: 1866
Post Options Post Options   Thanks (0) Thanks(0)   Quote tg [RIP] Quote  Post ReplyReply Direct Link To This Post Posted: 04 Feb 2011 at 12:34pm

So we start by distorting the measurements do we ?

Thought that was for the spin doctors in marketing after the design has been sonically compromised by the accounting department.

OK, back on topic, I am reading this and so far I understand (I think), that this sort of measurement is vital in the design prototype and production QC chain as a quick and repeatable method of ensuring that nothing is grossly misbehaving (or even mildly misbehaving).  

I will follow with interest, to see if there can be any correlation with what I might expect a given set of measurements to either do or not do, to the music I would wish to hear.

edited sometime later - 

Gee I can be a grouch when I'm off my feed, might need to up the meds Embarrassed

- meant to reassure Graham that his efforts to bring sanity and enlightenment were not going completely unnoticed - seems to have come out a bit backhanded.  Shocked Sorry 'bout that. Ouch




Edited by tg - 06 Feb 2011 at 3:18pm
Back to Top
iamalexis View Drop Down
Senior Member
Senior Member
Avatar

Joined: 07 Feb 2009
Location: London
Status: Offline
Points: 119
Post Options Post Options   Thanks (0) Thanks(0)   Quote iamalexis Quote  Post ReplyReply Direct Link To This Post Posted: 04 Feb 2011 at 4:30pm
thanks graham for the information and starting this thread. looking forward to the next instalment.
Back to Top
Graham Slee View Drop Down
Admin Group
Admin Group
Avatar
Retired

Joined: 11 Jan 2008
Location: South Yorkshire
Status: Offline
Points: 16298
Post Options Post Options   Thanks (0) Thanks(0)   Quote Graham Slee Quote  Post ReplyReply Direct Link To This Post Posted: 06 Feb 2011 at 2:47pm
OK, so we now know what THD+N means. But just to recap: THD means total harmonic distortion or the totals of all the harmonics added up, and N means the noise which is measured too simply because it's there.

But how does the test gear measure it? If you've seen distortion specifications you will also have probably seen them referenced to a particular frequency. That is the fundamental frequency - the signal frequency shoved into the input of the device under test (DUT) - the industry standard being 1kHz. The distortion meter (incorporated into the test set) filters out the fundamental frequency by a very effective notch filter. That's how there's only the harmonics the device under test produces plus the noise left.

Now let's examine how to measure THD+N...

If it's a power amp designed to drive speakers then there are two ways.

One way is to measure the distortion at full rated power (into its rated load) and if the fundamental frequency were to be viewed on an oscilloscope it would be adjusted upwards until the sine wave tops started to be flattened (clipping) and then adjusted back to just below clipping. The distortion would then be read off and the result would be referenced to "just before clipping".

That's OK for a disco or a group pa amplifier which is designed to be ran flat out, or near to it, but for home listening we hardly ever go that far (unless we don't have neighbours...). In such a case a manufacturer may choose the 12 o' clock volume control position. He would set the input signal level (the fundamental frequency) so that at full volume the sine wave is just below clipping and then wind the volume control to the 12 o' clock position, and then read off the distortion.

Some manufacturers quote both.

In a preamp, distortion is usually read at the rated output. The just below clipping level is probably never ever reached and so would be a meaningless exercise. On some preamps the noise (N) can get in the way of a true reading, and in which case it is acceptable to increase the signal level to somewhere between rated output and just below clipping.

When it comes to headphone amplifiers you have a cross situation between a preamp and a speaker amp. You could quote just below clipping but if you never got further than 11 o' clock on the volume control without deafening yourself it wouldn't be all that meaningful.

For our range of headphone amplifiers we devised a way of testing all functions at what we consider to be the worst operating conditions - where the volume control is electrically speaking half and half, and being logarithmic (audio taper) that point is roughly the 3 o' clock position. In this position the noise is at it's worst because of the volume control resistance, and so is the upper end of the frequency response due to the resistance coupled with the input filter. So we thought it only fair to measure distortion at this setting too.

The resultant THD+N distortion for the Solo SRGII (either power supply) is 0.0116% (but we publish 0.02% to be on the safe side because no two devices are exactly the same, on this planet anyway) - the amplifier driving a 32 Ohm load (the lowest of the two impedances we quote in the Power output spec.). The photo below shows the readout.



You will note the output level is approx. -4dB. This is the level the SRGII gives into a 32 Ohm load set at -6dB relative output (3 o' clock volume setting approx.) because of its current drive (IEC/DIN audio matching system, 120 Ohms) output. This level differs on the Solo Ultra-Linear and Novo headphone amplifiers because their outputs are not to IEC/DIN being 33 Ohms source impedance. They still give current-drive but more of it.

There has been a complaint from somewhere in deepest China regarding the non-correlation of our distortion figures with those of the challenger and I will answer that now. The challenger says it is 0.06%...

If you turn the Solo SRGII volume control to maximum and measure it as if it were a power amplifier into 32 Ohms, just below clipping the THD+N reads 0.0083% (far better than what we publish) and the evidence is provided in the scope and meter photos below. Actually this is a bit more than the stated 27mW into 32 Ohms - we always like to be on the safe side (2.21dBu = 0.9995435074mV, and that squared and divided by 32 = 31.22mW) .







However, if the input signal level is carefully tweaked to give just the required amount of clipping until the meter reads 0.06% you will obviously get 0.06% as I did below...







You will also see that the sine wave is actually just clipping and therefore is not just below clipping. The clever or the dumb can indeed also obtain any reading they like, but to achieve the above I must congratulate the challenger in being of the clever variety.

Now we've got that attack out of the way, I'll continue my next post with what use are distortion measurements.
That none should be able to buy or sell without a smartphone and the knowledge in how to use apps
Back to Top
 Post Reply Post Reply
  Share Topic   

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.01
Copyright ©2001-2018 Web Wiz Ltd.

This page was generated in 0.078 seconds.