I have done quite a bit of tuning with a pot in the R2 position. The amp takes a while to get to full temp and is thermally stable at 1.2A. 2nd harmonic is dominant until the frequency gets above 10kHz is so. My main concern has been keeping 2nd dominant and distortion as low as I can get it. I think all that's pretty good. My recent attempts with square wave was to see if the amp is otherwise performing properly. Maybe I'm over working it. Thanks for the great inputs. Maybe I need to spend more time listening and less time measuring!
Member
Joined 2009
Paid Member
Yes I read that long time ago, and seems I got I the wrong way, thought majority preferred 2nd just like instruments tend to produce 2nd, he can play with bias and find what's his best setting.
Thanks for the reminder bigun
I'm no expert, just remembered this snippet as it was also counter to my thinking, I was also thinking like you.
The key is that distortion should be low and avoid higher orders (above 3rd).
We read that humans are not sensitive to high 2nd order, we also read that providing the distortion of higher orders falls off monotonically after the 2nd we are also not sensitive to it and so everyone proclaims single ended triode amplifiers are the bees knees.
I found something different. i found that testing amplifiers with such distortion on simple music (female singer not much noise) at low volume they sound very nice indeed. But then, when you have more complex music and if you turn up the volume it all goes sideways. The bass is never punchy but is mushy and the mids turn to mud. You got yourself a guitar amplifier !
My ears were damaged by my own abusive listening habits and I have hyper accusis - certain treble frequencies irritate me. Some of these frequencies seem to line up with known resonances of our ears, like the ear canal - all of which is well known to the headphone people. So I seek out amps that don't irritate me (it is sometimes the speaker, often the source material rather than the amp) and I have found that I usually prefer amps with low distortion. This was a big disappointment to me as I was really into building amps with a bit of distortion, mostly with the idea that it would be a 'secret ingredient' and the resulting amplifier would let me have something I could smugly claim as superior to all those produced by the engineering types who only measure things
Last edited:
Do you have some FFT plots you could share? I have been looking at 1kHz, dual tone & multi-tone, I posted some earlier in this thread while doing the same tuning as you. I'm a bit curious about the performance you get with your setup.
I assume you're aware that the distortion will change depending on the load resistor you use, so depending on the load (impedance curve of your speakers), there is a compromise to be made on the Iq and balance of the resistors.
I'm no expert, just remembered this snippet as it was also counter to my thinking, I was also thinking like you.
The key is that distortion should be low and avoid higher orders (above 3rd).
We read that humans are not sensitive to high 2nd order, we also read that providing the distortion of higher orders falls off monotonically after the 2nd we are also not sensitive to it and so everyone proclaims single ended triode amplifiers are the bees knees.
I found something different. i found that testing amplifiers with such distortion on simple music (female singer not much noise) at low volume they sound very nice indeed. But then, when you have more complex music and if you turn up the volume it all goes sideways. The bass is never punchy but is mushy and the mids turn to mud. You got yourself a guitar amplifier !
My ears were damaged by my own abusive listening habits and I have hyper accusis - certain treble frequencies irritate me. Some of these frequencies seem to line up with known resonances of our ears, like the ear canal - all of which is well known to the headphone people. So I seek out amps that don't irritate me (it is sometimes the speaker, often the source material rather than the amp) and I have found that I usually prefer amps with low distortion. This was a big disappointment to me as I was really into building amps with a bit of distortion, mostly with the idea that it would be a 'secret ingredient' and the resulting amplifier would let me have something I could smugly claim as superior to all those produced by the engineering types who only measure things
I can relate to a lot of this
Hyper accusis: -Now I know the name for my 'easily irritated ears' too! However I still have some doubts about best measuring being most pleasing to my ears. I have built and measured some LJM class B amps that measure really nice, but still prefer the JLH that measures slightly worse. The distortion is so low in level that I find it hard to accept that the presence of the higher harmonics (around -100dB or buried in noise) would explain it.. With the JLH it seems I prefer the settings that gives lowest overall distortion. Muddy mids with higher distortion on JLH, or harsh mids & treble in some cases class (A)B.
Last edited:
Hi, Here's a 3V rms 1kHz tone with a 40dB notch over. The notch has negligible loss at 2kHz and helps to measure harmonics as I can use my Picoscope on a more sensitive range.
I have played with the balance of R1/R2 a bit but not with enough method to understand it yet. At the moment I have 100/433.
I don't know picoscope, I like using ARTA FFT for this together with a decent (but not very expensive) sound card. You have a lot of different output signals to chose from to see what happens with IM etc, no notches needed. You can jump between different input signals and watch the FFT spectrum while tuning the trimmers.
A lot of people here seem to use REW for distortion measurements, but I have only used it for room measurements a few times. I guess I'm reluctant to learn a new program
Would your graph indicate that you are in the -70dB range with the harmonics? I think 3V is a relevant amplitude to do measurements, not too low to me buried in noise, and a relevant listening level. I rarely listen at high levels any more.
Last edited:
Hi, 2nd is -75 and 3rd is -87dB. Thanks for the feedback on software. I'll check it out.
I agree on listening levels. I started working on my JLH at 10W and then found it wasn't right at levels I could listen to!
I know I need to do more work on the noise floor but the distortion seems OK. Am I right or should I work on it?
Cheers
Jon
I think you're ok in the distortion, but could probably improve a little bit more. I think I have been between -80 to -90dB in most of my better JLH experiments using modern output transistors.
The way I did it was to first try to find the Iq current range where the output transistors are in their most linear range (vary Iq and watch distortion) and after that I balanced the resistors to try to lower the distortion further. Balancing the resistors can improve the current sharing between the output transistors, and it seems improved current sharing mostly reduces the even harmonics, so you can tune this to taste. Good current sharing also improves the efficiency of the amp. For some reason I also found that the DC offset can have some effect on distortion, sometimes giving lower distortion when it's slightly 'off'. You should also check clipping to see that it's fairly symmetrical.
The way I did it was to first try to find the Iq current range where the output transistors are in their most linear range (vary Iq and watch distortion) and after that I balanced the resistors to try to lower the distortion further. Balancing the resistors can improve the current sharing between the output transistors, and it seems improved current sharing mostly reduces the even harmonics, so you can tune this to taste. Good current sharing also improves the efficiency of the amp. For some reason I also found that the DC offset can have some effect on distortion, sometimes giving lower distortion when it's slightly 'off'. You should also check clipping to see that it's fairly symmetrical.
Last edited:
Hi. Great to have a target so I'll work on reducing the distortion a bit more. I think I'll also install a temporary trimmer to adjust R1.
There are obviously high level specs in the original article but I find it useful to have other performance 'markers' to aim at. This is my first amp build and I want it to be the best it can be. Many thanks for your input.
Cheers
Jon
I'm no expert, just remembered this snippet as it was also counter to my thinking, I was also thinking like you.
The key is that distortion should be low and avoid higher orders (above 3rd).
We read that humans are not sensitive to high 2nd order, we also read that providing the distortion of higher orders falls off monotonically after the 2nd we are also not sensitive to it and so everyone proclaims single ended triode amplifiers are the bees knees.
I found something different. i found that testing amplifiers with such distortion on simple music (female singer not much noise) at low volume they sound very nice indeed. But then, when you have more complex music and if you turn up the volume it all goes sideways. The bass is never punchy but is mushy and the mids turn to mud. You got yourself a guitar amplifier !
What you describe there is a typical cheaply built valve amplifier, when complicated music turns to mush its the power supply that's modulated by the signal, sort that and a SET will play anything and sound just as good as with the simple female vocalist.
Many stereo amps from the 50s have the same issues.
Its distortion but not as you know it..
My sets which get aired in public a couple of times a year, can be played loud, and sound just the same at all volumes. Sort of rules out 2nd harmonic as the thing that makes them sound good in my view.
Sorry for the diversion..
Most people who have a view on this think a nice progression from 2nd 3rd 4th and 5th will have a distortion very like the ear itself. I helped Oxford University measure this. THD can be 30% in the ear. The brain knows this and hears a clear sound. The arguement these people put forward was sounds that distort in the same way as the ear pass as perfect unless a better version is shown. A tiny amount of 5th ( 0.05% perhaps ) and not much else as some feedback amplifiers can have will sound worse than 1% THD with the suspension bridge distortion curve. Someone ( me perhaps ) said 1% THD to this recipe or 0% will sound good, everything else might not. 0.1% to the same recipe might just be best of all. This is sometimes said to be our preference for 2nd harmonic. Not really. Like food things need balance. Also. A real speker with real music may not be the same as the snapshot THD result. It would be surprising if it was telling us everything. Telling us enough perhaps.
When the Quad ESL57 came along it was possible to test this. 1% was thought to be OK. The ESL being 20dB better than that. 0.1% was thought to be a safe hi fi standard. Due to loudspeaker science not really improving 0.1% is as good today as every it was. 1957 had mostly class A amplifiers. Speakers became rather dull when transistor amplifiers arrived to combat class B distortion that mostly was unknown before. The speakers gradually came back to a more neutral balance as AB types improved. Sometimes I think reducing THD to the point where it sounds less good is a number science that isn't science at all. A bit like pure water for drinking. H20 made from gases isn't the nicest and might be slightly harmful.
When the Quad ESL57 came along it was possible to test this. 1% was thought to be OK. The ESL being 20dB better than that. 0.1% was thought to be a safe hi fi standard. Due to loudspeaker science not really improving 0.1% is as good today as every it was. 1957 had mostly class A amplifiers. Speakers became rather dull when transistor amplifiers arrived to combat class B distortion that mostly was unknown before. The speakers gradually came back to a more neutral balance as AB types improved. Sometimes I think reducing THD to the point where it sounds less good is a number science that isn't science at all. A bit like pure water for drinking. H20 made from gases isn't the nicest and might be slightly harmful.
See Crowhurst, N. H., 1957 "Some Defects in Amplifier Performance Not Covered by Standard Specifications" JAES 5(4). It's a bit of an easier read than his earlier work which does the hard math (1953 May IIRC)
Bruno Putzeys had a reasonable ripost in 2011 (Linear Audio Volume 1, "The F-word or, why there is no such thing as too much feedback") but still relies on nigh-infinite headroom when we're building circuits that rely on small scale models.
Particularly in the almost-ubiquitous differential amplifier, which is outside the feedback loop by definition.
Which brings us back to this topic, where (a few hundred pages ago) various builders opined that original JLH sounded better than the one with a differential front end.
How similar 99% of amplifiers are (and therefore the limits of differentiating them) is only made clear when you look at something completely different, like Susan Parker's amp or Tim de Paravincini's Musical Fidelity A1 (or half a dozen other designs I can't find now since Google's recent lobotomy)
It's not easy to follow what you're saying there but I take it to mean that people don't really like euphonic distortion, they just find it better than the alternative of measures, such as negative feedback, which are supposed to reduce it.
I don't see the logic of any argument there. SET amplification is renowned for high levels of euphonic distortion (typically 1-3%) but devotees that I know, think that's the bee's knees of audio reproduction for just that reason. Of course, one can't call it distortion as that seems ugly. Measuring it would be anathema too, instead we say "rich in harmonics" like H2,H3,H4 etc.
Take a considered look at Jean Hiraga's magazine articles on the perception of distortion - always interesting to read how he developed his theory and designed some other fascinating, hot little hot little beasties like "le monstre" that brought the theory into practice. You can easily google and find his articles discussed here in the solid state forum and elsewhere on the web. The articles were also translated from the French by Jan Didden in AudioExpress, for example.
As best I know fifth and seventh if that's all that's there is usually disliked. In real music somethings sound to have second harmonic distortion. Triangles sound very like modern amplifiers. Played together I have to remember these are live sounds and not distortion. Beethoven should sound like 78s and Mahler like FM radio. Same venue, same orchestra. FM radio is still a high grade source despite the world moving on. I wonder if Beethoven set his sound from memory when already not able to appreciate higher harmonics. I believe many who are loosing their hearing detest HF distortion.
Recently the locals celebrated local heroes of the pandemic. The music was clear and spread for 400 metres. A line array and 3 watts!!!!
Recently the locals celebrated local heroes of the pandemic. The music was clear and spread for 400 metres. A line array and 3 watts!!!!
Read Crowhurst
If you can't be bothered, here is the readers digest version
A dose of global feedback
a) takes the existing 2nd and 3rd harmonic distortion and changes it into 5th and higher orders, which is much more audible
b) pushes the amplifier closer to instability
c) makes it much more likely other stages are likely to clip when the output clips and makes it clip harder, which also sounds awful
d) causes early stage clipping which usually upsets bias settings, moving the operating point away from the design point
e) usually implies a reduced open loop gain (and NFB) at higher frequencies (see (b) above)
f) pushes the non-feedback compensated differential amp away from it's linear region
g) encourages the designer to fiddle to "tidy up" the square wave response, by adding non linear compensation
h) encourages the designer to forgo linear design principles as "feedback fixes everything".
i) makes it much more sensitive to speaker load, speaker energy storage and RFI (see b) and e) )
Been there, done that.
Tell me again, which of the above is a positive step?
p.s. I keep threatening to try to build the worst sounding (but empirically "perfect" measuring) amplifier I can but the thought just sucks the life out of me
And I built enough signal-modulated-oscillators early in my career as it is
Last edited:
As best I know re entrant harmonic distortion is an abbreviation of a old fashioned test gear. If you very carefully add negative feedback from nowhere the higher harmonics emerge. What you are doing is fitting the magnification of the device being measured. The distortion was always there but likely at -80dB. Add 30 dB feedback and it's at -50fB. That's nasty.
However you might as well have made this high order distortion as nothing is there to balance it off. So despite the technical arguments it is unpleasant.
I was reading a PYE book on audio. 1973. One Audio critic was patronising people know didn't like typical sterile sound of advanced hi fi. He was trying himself in knots. These reviewers couldn't write a book like this to save there lives. Donald Aldous and John Borwick being the exceptions. John Borwick was very easy going and likable. My hero Percy Wilson also. I lived near him but never knowing met him.
However you might as well have made this high order distortion as nothing is there to balance it off. So despite the technical arguments it is unpleasant.
I was reading a PYE book on audio. 1973. One Audio critic was patronising people know didn't like typical sterile sound of advanced hi fi. He was trying himself in knots. These reviewers couldn't write a book like this to save there lives. Donald Aldous and John Borwick being the exceptions. John Borwick was very easy going and likable. My hero Percy Wilson also. I lived near him but never knowing met him.