Comparison of distortion components in IC, triode and pentode amp

[rotation]

Hi,

yesterday i had a short conversation with my teacher, i was asking questions related to exam. Beside things directly related to it (digital control), we also talked about distortion and overload characteristics of preamps.
He was very interested in my modification to 47 preamp output stage where i wanted to get different harmonic distortion spread/sound, so i used ECC99 tube and tried various amounts of local feedback.
I got interested in this modification by reading EMI paper about 47 preamp, it says that the output stage produces more 3rd than 2nd harmonic distortion etc. There was also post from Winstion O'Boogie where he is confirming this claim:
http://www.groupdiy.com/index.php?topic=7292.80
and a little more here:
http://www.groupdiy.com/index.php?topic=17847.20

Instead of "normal" exam i will probably present my modification to the output stage, measurements of distortion spread before and after, and also write a little about overload/distortion characteristics of IC, pentode and triode amplifier.

The only reading i could find about this is "Tubes vs Transistors: Is there an audible difference?" by Russel O. Hamm:
http://www.dwfearn.com/tubes_vs_transistors.html
It is interesting that the article in fig.6 shows more 3rd than 2nd distortion for pentode, while EMI says their EF86 input stage has more 2nd than 3rd. Well, R.Hamm wrote that different circuits show different characteristics which is understandable.

My plan now is to build simple IC preamp, 47 type with two output stages and measure what i can. I hope i get a chance to use university's laboratory because i don't have distortion analyzer and other lab gear. I'm all excited about it and this is for sure more interesting than digital control, at least for me ;D

As i'm going to write few pages about this thematic, i would like to find more information if possible.
Can anyone please point me to further reading about this things?
And did anyone measure distortion spread of standard EF86 input stage used by Mullard, Quad, EMI and many others?
I would also like to hear any comments about what i'm going to do.

I hope i get a chance to use university's laboratory to measure this things myself. I'm all excited about it and this is for sure more interesting than digital control, at least for me ;D
Miha

 

[Samuel Groner]

The differences between various ICs regarding distortion and overload characteristics is so gross that it makes IMO no sense trying to compare them to other technologies.

Samuel

 

[rotation]

I thought it is so. Then i will focus on 47 output stage and write a little more in general about the rest.

 

[pstamler]

What Samuel says is correct; he's done the definitive work on the harmonic distortion spectra of ICs. SG, could you post the link here?

It's also the case that you can obtain distortion spectra from tubed circuits which vary widely; although the general rule that triodes have more 2nd harmonic while pentodes have more 3rd harmonic applies, there are plenty of exceptions. Also note that many circuits which use EF86 tubes on the input use them in triode mode, not pentode mode.

Peace,
Paul

 

[rotation]

What Samuel says is correct; he's done the definitive work on the harmonic distortion spectra of ICs. SG, could you post the link here?

It's also the case that you can obtain distortion spectra from tubed circuits which vary widely; although the general rule that triodes have more 2nd harmonic while pentodes have more 3rd harmonic applies, there are plenty of exceptions. Also note that many circuits which use EF86 tubes on the input use them in triode mode, not pentode mode.

Peace,
Paul

It is here, lots of good reading:

Work on IC opamp distortion measurements:
http://www.sg-acoustics.ch/analogue_audio/ic_opamps
/pdf/opamp_distortion.pdf

And website:
http://www.sg-acoustics.ch/analogue_audio/ic_opamps/pdf/opamp_distortion.pdf

 

[Jonte Knif]

Hi Rotation,

Do you know how to read load lines from tube data sheets after looking at a schematic? If not, then I suggest that you learn. It is essential. It is fairly easy to determine which distortion component will be prominent in a given stage. Then you can proceed to measure those which look interesting. And modifying your own circuits to taste will get much easier. Endless trial and error not needed after that.

For small signal amplification with pentodes the load line usually produces _much_ more 2nd harmonic than third. I would even dare to say that small signal pentode stage with a lot of 3rd is bad design. OTOH in power amps the load line is chosen to maximize power output. This leads to a lot of 3rd harmonic.

I happened to build a simple SE output stage with E81L, switchable between pentode, ultralinear and triode modes and with adjustable FB.  In this particular case the most "beautiful" odd harmonic spectrum happened to be in ultra linear mode, while pentode had much more 2nd harmonic until clipping.

You don't necessarily need a spectrum analyzer. Any sound card with for example RMA audio analyzer works great. And I would always start first with oscilloscope to get the big picture.

 

[rotation]

Hi Rotation,

Do you know how to read load lines from tube data sheets after looking at a schematic? If not, then I suggest that you learn. It is essential. It is fairly easy to determine which distortion component will be prominent in a given stage. Then you can proceed to measure those which look interesting. And modifying your own circuits to taste will get much easier. Endless trial and error not needed after that.

For small signal amplification with pentodes the load line usually produces _much_ more 2nd harmonic than third. I would even dare to say that small signal pentode stage with a lot of 3rd is bad design. OTOH in power amps the load line is chosen to maximize power output. This leads to a lot of 3rd harmonic.

I happened to build a simple SE output stage with E81L, switchable between pentode, ultralinear and triode modes and with adjustable FB.  In this particular case the most "beautiful" odd harmonic spectrum happened to be in ultra linear mode, while pentode had much more 2nd harmonic until clipping.

You don't necessarily need a spectrum analyzer. Any sound card with for example RMA audio analyzer works great. And I would always start first with oscilloscope to get the big picture.

Yes, i understand loadlines but never draw one from given circuit.
I just found nice article with explanation of load lines and equations for harmonic distortion:
http://www.freewebs.com/valvewizard1/Common_Gain_Stage.pdf
It says that it 2nd harmonic distortion is not the only one produced, but it might be enough for the start, i can go into more details later.
This one is nice too, i learned a lot from it:
http://www.electronicstheory.com/html/e101-37.htm
During the weekend i will try to calculate load lines and distortions for the circuits i know best.

Btw, when you say RMA audio analyzer, you probably mean this one:
http://rightmark-audio-analyzer.software.informer.com/
If so, you probably use some kind of attenuator before the sound card, no?
If professor accepts my proposition, which i think he will, i will have access to good laboratory at University of Ljubljana.  They have some very good gear there, can't wait to see and try it.

Thanks for replies so far. I will try to write down some basics of my presentation and see what professor thinks.
Btw, i've seen some similar presentations from other students and they call them innovations. I don't know, if i extend my writing to several pages, make some measurement etc. (until it is more or less what others do), this is still not innovation, at least not for me. Hmm

Miha

 

[Jonte Knif]

Nice papers. I have to admit that it is sometimes a bit difficult to see the operating point of a pentode, but fortunately most schematics have at least some voltages printed.
RMA yes, that one. I guess older versions are free? And yes, pads needed. This can be quite cumbersome, because the pad has to sometimes have very high input impedance to not change the load line. But since you have the access to pro equipment you don't have to worry about this.

Btw, i've seen some similar presentations from other students and they call them innovations. I don't know, if i extend my writing to several pages, make some measurement etc. (until it is more or less what others do), this is still not innovation, at least not for me. Hmm

Well, I guess so. But it is sometimes mind boggling to see what kind of papers are worth publishing.

 

[playboss]

there was an AES paper about this a while ago.

read this one first tell me if you r done
http://www.q-audio.com/johncurl.pdf

 

[Speedskater]

Dr. Earl Geddes has some recent papers on Sound Qualilty Perception and Distortion Perception:

http://www.gedlee.com/distortion_perception.htm

This is a very different view-point than the traditional almost urban legend ideas about distortion.

 

[playboss]

what do you mean?

•  The masking effect of the human ear will
tend to make higher order nonlinearities
more audible than lower order ones.
•  Nonlinear by-products that increase with
level can be completely masked if the order
of the nonlinearity is low.
•  Nonlinearities that occur at low signal levels
will be more audible than those that occur at
higher signal levels.
Again these may seem intuitively obvious.  

hows that different???!  
I follow Earl's posts over there but not closely. He found himself a chipamp that has no high order distortion below noise floor and says thats good for him because the matching inside the IC is much better. OK ,be it.

 

[Jonte Knif]

I read the arrogant "paper" by Curl i.e. collection of posts with 90% redundancy, full of hostility towards other people, devoid of humor etc. But I just had to, because there were interesting pieces of information and anecdotes in it, and of course it warms my Finnish heart to see how important Matti Otala has been to his thinking.

One thing that really irritated me was his rebuttal of ABX tests. Not that they wouldn't have problems, but what is he raving about the impossibility to arrange a technically relevant test? Is he scared?

I don't understand where he got the idea that  people don't know about the importance of high order harmonics. If it really is like that I feel very sorry for audiophiles.

Also Earl Geddes' papers didn't bring much new info, but I found them very interesting and important because the quantitative method seemed to work pretty well. I'm very pessimistic about a scenario in which manufacturers would adopt such parameter in their spec sheets. The completely useless THD figures will reign. I would definitively want to see Geddes to go forward to include more variables in his model. If I remember frequency dependence wasn't yet there, but it is extremely important and should be included and poses no problems to his math.

So, thanks for those links even if it remains a bit mystery to me why their findings concerning distortion would have been novel.

This is a very different view-point than the traditional almost urban legend ideas about distortion.

Never heard these urban legends. What are you referring to?

Best,

Jonte

 

[Speedskater]

Never heard these urban legends. What are you referring to?

The three common legends I was thinking of are:

a) Musical instruments only produce even order harmonics.

b) A little bit of odd order harmonic distortion sounds worse than lots of even order distortion.

c) A little bit of high order harmonic distortion sounds worse than lots of low order distortion.

 

[Jonte Knif]

I see.

"a" is really weird. I didn't "know" it. Funny.
"b" and "c" are of course related to what we mean by "little" and "lots" and "high" and "low". Anyhow "b" is obviously quite misleading, yes.

"c" does make sense if we talking about 1% 2nd and 0,1% 11th.