Distortion of the closed-loop amplifier is determined basically by two things: the distortion of the open-loop amplifier and the amount of feedback. Because of stability concerns, the amount of feedback must decrease as frequency increases. Thus distortion of a feedback amplifier will typically increase as frequency increases. Of course, as frequency increases, the phase error between input and output will increase too. But this is not a root cause, it's more of a side effect of decreasing feedback with increasing frequency. So what you're seeing is correlation of this phase error with increasing distortion, but it is not the cause per se.
It doesn't actually work this way, and that is why it hasn't been done. The concept of "accuracy of the NFB" may be appealing at an intuitive level, but it doesn't actually lead to the root cause of distortion issues.
Exactly. Your post is what made me think of the article, and I had the exact same thoughts as you regarding the exclusion of alternative 2.
That will be hard unless you're doing a class A output stage or deliberately adding H2 somehow.
Last edited:
simultions matters..it s a good tool to make a fast preview.
proceding otherwise, building and testing is too time consuming
and is more painfull without simulation tests before, as we can
eradicate mosts of the drawbacks before even making a pcb...
if the simulator say 0.01% distorsion and that real world is 0.03%,
that s a good accuracy..at least, enough for me..
Member
Joined 2009
Paid Member
My sims are showing a phase shift as frequency increases not because I have limited the feedback on stability grounds. On the other hand, the amount of feedback falls off - I assume because the gain is limited by the topology / output devices.
I am doing Class A at this point - having never done it before so I want to go down this route for the fun of it. I have some 'junk' to use up that provides an artificial constraint on what I can build. Starting with 4 transformers, each one has a single secondary at 25AC 4.5A and I want to use them to build 4 separate single rail stereo amplifiers with some differences between each one of them. I'm looking at only 6W into an 8 Ohm load. The point of the exercise is the learning and the fun rather than some kind of professional product. After that, maybe I'll want something different.
Member
Joined 2009
Paid Member
Okay. This sounds like a topology that's stable without explicitly adding frequency compensation. The topology itself is reducing feedback enough at high frequencies to make it stable even though you may not have explicitly designed it with that goal in mind. What I'm saying is it's that decrease in feedback with increasing frequency that's causing distortion to increase with increasing frequency - even if you didn't explicitly put some compensation components in there. Care is required with such topologies though. What can happen in a BJT design is that an inverting stage is compensated by its own collector-base capacitance. If the SPICE model is way off in its collector-base capacitance parameter, the actual circuit could end up being unstable. Explicit frequency compensation is not as evil as it might seem at first, because an added capacitor, if it's of good quality, will have a capacitance value independent of the voltage across it. OTOH, the internal collector-base capacitance of a BJT varies a lot with the instantaneous collector-base voltage. This can cause stability to vary strongly with signal level if the circuit is self-compensating, in addition to increasing distortion.
That makes sense. Especially single-ended class A could well have distortion dominated by 2nd order.
btw, lumba, what kind of amp do you use ?
i remember having already asked this question,
but with no answer in sight...
tubes ?..mosfets?. a mix of boths?..
let us know..
Telstar,
The annoying noise floor, raised by the very same multiplying mechanism, is made up of various sub-harmonic and intermodulation products, modulating the signal components, causing the notorious grainy highs (the more feedback the grainier highs), clearly not masked.
Also, quiet background despite high static noise level.
The ear generates distortion in a certain pattern and has a complex masking mechanism, but you cannot refer to it so loosely.
The annoying noise floor, raised by the very same multiplying mechanism, is made up of various sub-harmonic and intermodulation products, modulating the signal components, causing the notorious grainy highs (the more feedback the grainier highs), clearly not masked.
How would you describe phase distortion?
I strongly doubt it. How do you know which types of distortion are actually present?
Hearing has been subject to continuous intensive research, the self-distortion is very well-understood, as a matter of fact, far more intricate details on hearing are.
Tubes are vastly superior chiefly due to high electron mobility. The most linear types (not many) offer a harmonic distribution pretty closely covering the aural distortion envelope: wonderful transparency, richness of detail and spatial presentation. Global feedback would ruin all that.
Also, quiet background despite high static noise level.
Since you mentioned Science...
Observation is the friend of science, assumption is its enemy. You seem to be accumulating assumptions like a magnet. My advice is that you keep a log of observations and proofs separate from your log of conjectures.
Don't trust any theory you come across no matter who supports it unless it is provable (mathematically or by simulation) to your satisfaction or you have made an observation under controlled circumstances to prove it. Be suspicious of people who offer qualitative explanations and evade mathematical justification when asked.
Also, be careful that you note the conditions of the proof/observation.
Don't believe a theory just because it is understandable, easy to measure or popular. Your default stance should be to treat everything as conjecture.
Just some free advice. Take it or leave it.
Your proposal to build 4 class A amps and use these to compare circuits is excellent. Takes me back about 30 years.
lumba, i agree that tubes have a better THD profile in matter
of monotonic deacrease than most class AB amps, though the total THD
is hugely higher which make the amp tube so called superior sonics
somewhat suspect, not to mention that this monotonic deacrease is
ruined as soon as the tubes are used in push pull mode..
indeed , tubes are vastly superior as natural exciters and
distorsion adders..
Member
Joined 2009
Paid Member
I can't see how this is true - can you explain further ?
Is it instead not due to their transfer curves being derived from vacuum space charge physics, large physical size allowing good geometry without too many parasitics, large voltage swings that sit well above the noise floor etc.? - and a nice warm glow
Since you mentioned Science...
Observation is the friend of science, assumption is its enemy. You seem to be accumulating assumptions like a magnet. My advice is that you keep a log of observations and proofs separate from your log of conjectures.
Don't trust any theory you come across no matter who supports it unless it is provable (mathematically or by simulation) to your satisfaction or you have made an observation under controlled circumstances to prove it. Be suspicious of people who offer qualitative explanations and evade mathematical justification when asked.
Also, be careful that you note the conditions of the proof/observation.
Don't believe a theory just because it is understandable, easy to measure or popular. Your default stance should be to treat everything as conjecture.
Just some free advice. Take it or leave it.
Your proposal to build 4 class A amps and use these to compare circuits is excellent. Takes me back about 30 years.
Many of your guys were building audio "30 yrs ago" and I feel that I may have missed out on some of the simple pleasures such as the classic JLH amplifier topology etc.
I do like to collect assumptions, I like to compare them, consolidate, resolve discrepancies and generally build a mental model of what might be going on. I'm the consummate armchair LTSpice adict. For me it's an enjoyable part of the hobby. I may not have started into this hobby (nearly 1 year ago now) had I not enjoyed reading all sorts of weird theories about audio, the magic and critical secrets that people feel that they have discovered along the way, the claims of wonderful things from tubes, many of which seem to require descriptions that would challenge a poet !!
The proper science is unforgiving, it deals only in the hard truth, leaves little room for some romance sometimes. It doesn't allow you the luxury of making assumptions and cutting corners. After many years as a scientist (e.g. I once built a vertical hot electron transistor where the 'grid' was a 2d electron gas in heterostructure of GaAs and AlGaAs - a Static Induction Transistor if you like) - now I like the freedom to wander off track.
These theories and discussions also shape the direction that my eventual construction plans will take - I don't have time to do a lot of building and recently I've changed my plans from 4 very similar amplifiers with NFB around an OLG of 90dB+ to looking at 4 different designs that include something with zero global nfb. I'm quite looking forward to it
Last edited:
This is a fun hobby and is very intriguing if you have a science background. Hearing is believing. It is a really good idea to build a simple, low power amp, without or with low NFB, class A, single-ended and listen to it. Then, say, go and listen to a Conrad Johnson tube amp driven by an extremely expensive CD player or vinyl. Listen to live music. I try to avoid technology prejudice and focus on how convincing the sound is. Then I work out the physics of why so many designs kill the music.Bigun said:These theories and discussions also shape the direction that my eventual construction plans will take - I don't have time to do a lot of building and recently I've changed my plans from 4 very similar amplifiers with NFB around an OLG of 90dB+ to looking at 4 different designs that include something with zero global nfb. I'm quite looking forward to it
Brian
(PS: I have never build a JLH)
Or it may just be that the perception of "richness of detail" is because the ear/brain is to able to resolve more musical details due to less distortion.So you like to hear two distortions, the added by tubes is not in music. Added distortions ("richness of detail ") are very musical
Surely NOT the majority of consumer amplifiers. Imho at least. If really the most of amplifiers were really at this level of excellence, none would have to regrets about their perfomances.
Regards
Piercarlo
Or it may just be that the perception of "richness of detail" is because the ear/brain is to able to resolve more musical details due to less distortion.
I guess it depends on how much you believe the old Carver CJ challenge. If it's true that Bob was able to tweak his BJT, low-bias, amp by adding distortion, to make it indistinguishable to the golder ears from a Conrad Johnson Premier, that pretty much drives a spike through the theory that tubes sound better because they're in some unmeasurable way more accurate.
Personally, I believe it's, at least mostly, true. With modern DSP's, it should be possible to build a box that would map the distortion products and general transfer function from CJ's, Marantz D-9's, SETs, and all sorts of other highly regarded "musical" amps, to add euphonic color to modern, accurate, solid state amps.
Member
Joined 2009
Paid Member
A lot happens before the music ever reaches the power amp. I've found a HUGE difference between different CDs in my collection when played on my YBA source. If I listen critically, there are very few CDs that can even kneel at the feet of my Chesky disks. There seems to be a critical dependency here on recording quality. The 'niceness' control is something I've thought much about and it seems to me that we would first need very good recordings, preferably unmixed tracks so we can apply our own mixing - but this sounds like another hobby altogether