Why don't we care about TIM (transient intermodulation distortion) anymore?

[PaulK]

Well that certainly looks impressive.

The 36 dB rule I refer to comes from my textbook and is based on 35 years of professional experience
designing analogue systems for the largest record company in the world.
That company is now part of UMG.

 

[solderdude]

Rules are there to be broken.

 

[PaulK]

I fully agree!

 

[MaxwellsEq]

Well that certainly looks impressive.

The 36 dB rule I refer to comes from my textbook and is based on 35 years of professional experience
designing analogue systems for the largest record company in the world.
That company is now part of UMG.

Can you share an image of the relevant page of your textbook? Which one is it?

I don't recall encountering a 36 dB feedback limit, and I don't understand how a limit could be established, so I want to understand what I've missed.

 

[JeremyFife]

Rules are there to be broken.

Apart from that rule ...

 

[antcollinet]

That graph is exactly what I hear every time I play a track through a class D amp, the cymbals sound like crap and no amount of EQ can make it sound natural.

Cymbals are not a square wave - nor a step - not even close.

Rimshot!

Nor is a rimshot.

 

[solderdude]

I don't understand how a limit could be established

By using a circuit that has less than 36dB open-loop gain ... so no op-amp.

I assume that local feedback is 'allowed' ?

 

[PaulK]

I am sorry I can’t oblige to answer all questions.
It concerns an American textbook that I found in a local second hand bookshop.
It dates back to the late fifties. It is a true treasure of information about designing tube amplifiers.
I still have the book but it is located in my house in Switzerland. I am living in Sweden now.

To make feedback possible in a power amp without disturbing stability the author suggested a wide band driving stage and phase inverter for the output tubes. To achieve that he suggested the use of penthode tubes. Video amplifiers used penthode tubes in those days. A frequency range of 500 kHz was minimum. I chose E80L tubes from Philips SQ series. I designed that amplifier in 1976.

During listening tests in the studio where I worked it made clear the mixing console we used suffered from TIM. The standard studio amplifier did not reveal that TIM.
The custom built console used early opamps used for telecommunication. They had an open loop gain of 100 dB which meant feedback of 50 to 80 dB. Too much for TIM free operation.

I still use that power amp. Never cared to look for anything else with one exception. The Danish Lyngdorf amplifiers that offer digital correction of room acoustics. I heard that system at an audio fair twenty years ago. An amazing system.

 

[Sokel]

Cymbals are not a square wave - nor a step - not even close.

About normal, recorded stuff we listen to, clapping is, it's like a small explosion starting low and going way higher than 20kHz, probably the reason it's so hard to sound life-like.
But not very close to square.

 

[PaulK]

Clapping is often manipulated during recording with Keepex and similar gear.

I wonder how the sound of cymbals can be distorted during replay.
The waveform is not spectacular. Could be a problem for class D amplifiers.

 

[MaxwellsEq]

Could be a problem for class D amplifiers.

It's not a problem for the Purifi and latest Nord. Also, pobably not a problem for previous generations of discrete Class D.

In a blind test, level matched and operating within design parameters you would not detect a quality Class D amplifier relative to another class.

 

[RexrothPigeon]

Could be a problem for class D amplifiers.

Lol it is not

 

[Cbdb2]

I question the last two recommendations in the list.
Application of massive amounts of negative feedback indicates the basic design of the amp is not sound.
Large amounts of feedback will induce stability problems especially for power amps with a complex load.

The more feedback the better it sounds provided it is never less than 30 dB at any audio frequency is definitely not something I would advise.

I would like to point to the importance of Nyquist diagrams when negative feedback is applied in power amplifiers.

Negative feedback is not a magic potion that cures basic design flaws.

Have you studied control theory? Do you understand the math in "The F-word"? Your constant referance to Nyquist diagrams shows you know "just enough to be dangerous". Your questioning however just shows your ignorance.

 

[Cbdb2]

By using a circuit that has less than 36dB open-loop gain ... so no op-amp.

I assume that local feedback is 'allowed' ?

"Nested feedback is functionaly identical to global feedback." B Putzeys

 

[Cbdb2]

The custom built console used early opamps used for telecommunication. They had an open loop gain of 100 dB which meant feedback of 50 to 80 dB. Too much for TIM free operation.

"Improving loop gain improves TIM" B Putzeys. Maybe those old opamps just didnt have a high enough slew rate?
You keep bringing up examples from 60 years ago as if nothing has been learned since than.

 

[bmc0]

It concerns an American textbook that I found in a local second hand bookshop.
It dates back to the late fifties. It is a true treasure of information about designing tube amplifiers.

Was that specifically in reference to tube power amps? It's difficult/impossible to use gobs of negative feedback due to the bandwidth limitations of output transformers. 36dB or so sounds about right as a practical limit in that particular context.

 

[solderdude]

"Nested feedback is functionally identical to global feedback." B Putzeys

Yes, but this is using amps with open-loop gains well over 35dB (< 1kHz) in those nested designs with more than one high open-loop gain amplifier sections where each of them has local and global feedback.
This results in high BW and very low distortion.

For tube amplification stages such high gains and having a > 10kHz BW is possible. The problems occur when high gain and transformers were used in the feedback loop.

Indeed many things have changed in the last 60 years... TIM isn't a problem, also not with >100dB open loop gain op-amps.

 

[Cbdb2]

Yes, but this is using amps with open-loop gains well over 35dB (< 1kHz) in those nested designs with more than one high open-loop gain amplifier sections where each of them has local and global feedback.
This results in high BW and very low distortion.

Dosnt this mean you can have very little global feedback as long as your total local feedback amount is the same as your global feedback?

"From the viewpoint of distortion, there is no differ-
ence whatsoever between one macho feedback loop and local feedback with a simpler global loop
around it. Designers who propose to use “mostly local feedback and only a little global feedback”
are labouring under an illusion. It makes no difference. Whether you choose to use a nested loop
or global feedback depends on other practicalities but has no bearing at all on actual audio performance."

Are you saying that the total feedback, local plus global has to be greater than 35db for this to be true? Are there simplifications in the equations that assume the gains are >>>1? Ild have to look at his math again.

 

[solderdude]

You'll have more open-loop gain available in nested designs and thus can apply more feedback resulting in lower distortion.

Of course there is no difference between nested and just one global feedback loop when the open-loop gain is the same.

Are you saying that the total feedback, local plus global has to be greater than 35db for this to be true?

Not saying that....
The more linear the open loop amplification is the less feedback is needed to reach high BW & low distortion.
With a higher BW open-loop amplification the less feedback is needed to reach a flat FR response.
With transformer outputs the limitation and non-linearity that is caused by the OP transformer can only be corrected up to a point.
There is little point to throwing in 'more gain' to correct for those limitations, I guess that's where the 35dB 'limit' came from all those decades ago.

It is not applicable to OTL SS, for OTL tube amps the max. output current capabilities of tubes in low speaker impedance is the bottle neck.

This 'problematic' output transformer is not present in SS designs (aside from some purposely designed amps out there) so the open-loop limitations are not there (in the audible band).
Of course, the higher the open-loop gain is there will be other technical limitations that pop up and throw dirt in the design.

That said .... well designed amps using output transformers and carefully tuned feedback loops are more than sufficient for audio applications.
The 'well designed' part determines the signal fidelity side of it.

That spoken off '35dB limit' of past (tube amp) days is not applicable any more but may have made designing tube amps back in those days a bit easier to do properly.

 

[PaulK]

@ Cbdb2,



It was extremely tempting to neglect your post. There are two good reasons not to react.


First of all you do not honor the established forum etiquette “avoid ad hominem“


Secondly my remarks are specifically related to the design of the tube power amplifiers.


Somehow you missed that information.





The insulting tone in your post did not offend me, it takes more to do that.


You gave me an excellent opportunity to question your authority on this subject.





Your comment on my reference to Nyquist was kicking an open door.


For amplifier design Nyquist diagrams are often paired with Bode plots.


Together they give information about the stability margin in amplifiers that use NFB.


A pity you did not notice the omission to mention Bode.





You honor me by calling me dangerous.


It is the first time I am called dangerous in a debate about design principles.





After 35 years work as balance engineer at a major record company what included testing


and design of profession alequipment I do not feel obliged prove my credentials.





https://hifisonix.com/wp-content/uploads/2017/10/The-Theory-of-TIM-Matti-Otala.pdf





The link gives information about TIM.


Please note Matti Otala opens with the thesis to limit NFB to 40 dB





I decided to refrain from reading and posting at ASR. This is my last post here.