Audio Asylum Thread Printer

Get a view of an entire thread on one page

FAQ / News / Events

Digital Drive

RE: Feedback - Part 4

72.211.237.239

13th Duke of Wymbourne

In Reply to: RE: Feedback - Part 4 posted by Story on February 15, 2024 at 06:09:53

The type of amplifying device - BJT, J-FET, MOSFFET or GaNFET - doesn't really make a difference to the discussion. They all thave non-linear Voltage-In to Current-Out characteristics hence they distort and, hence, feedback is needed to correct that.

The GaN advantage is their bandwidth - maybe doesn't bring too much to the linear amplifier table but it is advantageous for switching amplifiers in that they can switch more quickly and push the switching noise further from the audio band. They are the device du jour for switching amplifiers. Going back to Bruno Putzey, when asked about re-designing his Purifi module to use GaN instead of MOSFETs he said that there would only be a marginal improvement because he has applied a lot of feedback so errors are well suppressed. He thinks that maths is more important (i.e. being able to design an architecture with a lot of feedback across the audio band) than sexy FETs. If designers can't do the maths, and have lower feedback designs, then they will see more benefit from GaN.

Follow Ups:

RE: Feedback - Part 4 - morricab 07:24:38 02/16/24 (3)
Yes and triode amps have the best linearity of any amplification device ever made. ARe they completely linear? No, but they are linear enough to get away with any feedback beyond degeneration (sometimes even without that). The only other device that can kind of do this is the FET if run in Class A and without feedback beyond degenration.

The thing not mentioned is what feedback does to the distortion characteristics beyond just the level of THD and IMD. It has a clear impact on the pattern of those harmonics away from a low harmonic order dominant to a high harmonic order dominant pattern.
- Sort of-ish - 13th Duke of Wymbourne 14:52:23 02/16/24 (2)
  Triodes are square law devices, like FETS. This is not linear. You could operate them only over a narrow range of their characteristic where they will approximate a linear characteristic but that is the same with any non-linear device - if you operate it only over a tiny portion of its range it will approximate to linear (whether that range is large enough to do anything practical is TBD). Degeneration and global feedback work in the same way, so if you think global feedback is bad you should not look at degeneration as good (a difference is that degeneration applies to one device so stability is not an issue, but then global feedback stability around many stages is not an issue if it is done right).
  The topic of feedback translating low order harmonics to high order harmonics is an interesting one. It was raised decades ago in Wireless world magazine with an example of a single stage amplifier that had a large amount of purely second harmonic. After applying feedback the second harmonic was reduced considerably but also low levels of higher harmonics appeared. The F-Word article that I linked to in a previous post covers this case in the Section 'Storyline 2: Re-entrant distortion'. At worst, this is a case of going from no high order harmonics to a small amount of high order harmonics but never 'high order dominant'. And, there are no truly square-law devices, they all have some imperfections that make them not pure square law so they create high order harmonics on their own - I have never seen a single-ended triode power amp spectrum that did not already contain a spray of 3rd, 4th, 5th etc harmonics.
  
  Edits: 02/16/24 02/16/24
  - RE: Sort of-ish - morricab 13:14:23 02/18/24 (1)
    3/2 law actually...not square law (or quadratic if you prefer)
    
    It is actually pretty simple. A single ended triode will yield all harmonics but with a monotonic decay where each subsequent harmonic is significantly lower than the one before it...with 2nd of course being the most abundant.
    
    Perfect FETs actually behave in some cases better, like a Class A push/pull output stage... the problem is that you will never have perfect FETs and so you will get plenty of distortion in the real world.
    
    A good simulation paper is the following:
    
    http://www.its.caltech.edu/~musiclab/feedback-paper-acrobat.pdf
    - RE: Sort of-ish - 13th Duke of Wymbourne 00:01:37 02/19/24 (0)
      Thanks. I have a Beethoven recording by James Boyk of which I am fond. I will read through in detail and see whether Boyk is a better EE or pianist :)

Post a Followup:

FAQ

Post a Message!

Forgot Password?

Moniker (Username):
Password (Optional):
	Remember my Moniker & Password (What's this?) Eat Me
E-Mail (Optional):
Subject:

Message:	(Posts are subject to Content Rules)

Optional Link URL:
Optional Link Title:
Optional Image URL:
Upload Image:
E-mail Replies:	Automagically notify you when someone responds.

To view your new posting or follow-up, click on the RELOAD or REFRESH button on your browser.

This post is made possible by the generous support of people like you and our sponsors: