The lower amp is amplifying the error of the upper one (and contributing a teensy bit of its own distortion). Then the net output signal is the difference of the two opamp outputs. The output signal of the lower amp is small.
Of course the device(s) that do the latter differencing are critical, so the overall system performance needs to be evaluated from that perspective.
First schematic is a balanced bridge arrangement. Yes, distortion sums in Rload.
Trim R values as needed to get null --- will progressively cancel (sum of + and - error) for lower and lower levels of thd.
Error take-off for thd reduction does not depend on gnfb and its falling open loop gain issues (stability, gain/phase, reduced nfb at high freq, increasing thd vs freq etal).
I'd like to see more alternate ideas tried and optimized.... See what it can do... pro and cons.... 'cause its getting boring with just gnfb to talk about. Good place to use SIM ?
Thx-RNMarsh
Trim R values as needed to get null --- will progressively cancel (sum of + and - error) for lower and lower levels of thd.
Error take-off for thd reduction does not depend on gnfb and its falling open loop gain issues (stability, gain/phase, reduced nfb at high freq, increasing thd vs freq etal).
I'd like to see more alternate ideas tried and optimized.... See what it can do... pro and cons.... 'cause its getting boring with just gnfb to talk about. Good place to use SIM ?
Thx-RNMarsh
Last edited:
Thank you Richard
The second attachment is from this article:
http://www.keith-snook.info/Wireless-World-Stuff/Wireless-World-1974/Reducing%20Amplifier%20Distortion%20-%20Sandman.pdf
The first attachment reminded me of this:
http://www.ti.com/lit/an/sboa031/sboa031.pdf
Although is a totally different circuit and has nothing to do with Richard’s distortion reduction one, I post the link due to the discussion of output buffering
(I have built the circuit with LF353N that I had available . It drives my 600 Ohm AKG K240 headphones).
Regarding distortion reduction on bridge circuits, there is one EE participant here who has done a lot of work on this. But he always keeps a low profile
George
Sandman's article has a Fig. 2 which the old guys might recognize as the
basis for Mike Wright's ill-fated XEC1000 power amplifier. The idea was to
use a smaller amplifier correcting the larger one. The problem of course is
that the smaller one had to deal with the large load current....
😎
basis for Mike Wright's ill-fated XEC1000 power amplifier. The idea was to
use a smaller amplifier correcting the larger one. The problem of course is
that the smaller one had to deal with the large load current....
😎
Attachments
Yeah, Deane (I'm not misspelling his name by the way) did promote a flat group delay, or what he termed "Deviation from Linear Phase" or DLP for short. He wrote about it in a 1986 paper he presented at the 81st AES Convention, titled High Frequency Phase Response Specifications: Useful of Misleading?
But as far as I'm aware, he just started out with very well engineered transformers and applied external RC networks to achieve the high frequency response he was looking for.
Basically, for a given transformer, there will be a particular resistive load that will result in a second order Bessel response. But that load resistance won't necessarily be what you want reflected back to the primary in the audio band. So you use an RC network that will provide damping at the resonant frequency, which is typically in the vicinity of 100kHz.
My information came from the late Tom Reichenbach, son of Ed Reichenbach. He founded CineMag Transformers and had all of Ed's engineering, including the engineering he'd done for Deane.
se
It looks stupid to me. It takes the distorded signal from the out of the first amp to add current. Why do not // them, more simply.
Last edited:
With the capability to measure in everyones hands - we ought to start measuring this more often in respect to tube pre/amps (transformers) as well as termination of speaker cable lines at a given loudspeaker and the like... Microphones, too. Sware and HWare that run by PC can do this easily now.
How does terminating a cable at the speaker terminals improve Group delay? What would it take? Thx-RNMarsh
Thank you Richard
The second attachment is from this article:
http://www.keith-snook.info/Wireless-World-Stuff/Wireless-World-1974/Reducing%20Amplifier%20Distortion%20-%20Sandman.pdf
The first attachment reminded me of this:
http://www.ti.com/lit/an/sboa031/sboa031.pdf
Although is a totally different circuit and has nothing to do with Richard’s distortion reduction one, I post the link due to the discussion of output buffering
(I have built the circuit with LF353N that I had available . It drives my 600 Ohm AKG K240 headphones).
Regarding distortion reduction on bridge circuits, there is one EE participant here who has done a lot of work on this. But he always keeps a low profile
George
Boy you are good! That was fast, too. yes, Sandman in Wireless World discloses this error-take off approach and gives some back ground to Black (Mr neg feedback). It is Sandmans own original idea. Wireless World 1974.
I think it has great potential. Maybe with power amps? They still need a lot of help. Or low thd oscillators... the concept can be used somewhere to advantage. It seems to have been under-developed.... Is this what Walker/Quad amps were inspired by? Thx-RNMarsh
Last edited:
I think that's called Hawksford error correction, but there's *always* priors in electronics. Usually Blumlein or Baxandall!
Thanks,
Chris
No, that is a comparitor between I/O, etc. Sandman shows something different in functionality. Inverting the error signal and feeding it into the output (inverted) to cancel thd. Thx-RNMarsh
Last edited:
Chris, Richard is right. It is not Hawksford.
Yes there are priors, which AKSA highlighted after I originally posted it up a few months ago. However, as far as I can tell, it's never hit the mainstream.
Yes there are priors, which AKSA highlighted after I originally posted it up a few months ago. However, as far as I can tell, it's never hit the mainstream.
My comment wasn't about the Sandman; it was about "This approach measures the difference between the input (the reference) and the main amplifier output which also includes the error signal. The output of the measuring amplifier represents the error difference and is applied to the main amplifier feedback node as an error correction signal." No biggie either which way.
Thanks,
Chris
http://hifisonix.com/wordpress/wp-content/uploads/2012/12/AFEC-V1.0.pdf
Thanks,
Chris
http://hifisonix.com/wordpress/wp-content/uploads/2012/12/AFEC-V1.0.pdf
I understand. However, we here have a collective capability to make things happen.
There are guys involved in SIM who are wizards at it but just rehash the usual topology until it is a Taj Mahal. Bonsai -- we just need to keep pushing and show it to people until it gathers enough head of steam. Maybe others want some one else to start and show the way then they will take it and run with it. Even though John's forum isnt the right place... I am getting it exposed to as many people as possible to explore it further. I thought of something like this in concept only as a means to lower my oscillator thd on another forum. Then I rediscovered Sandman's invention. Thx-RNM
There are guys involved in SIM who are wizards at it but just rehash the usual topology until it is a Taj Mahal. Bonsai -- we just need to keep pushing and show it to people until it gathers enough head of steam. Maybe others want some one else to start and show the way then they will take it and run with it. Even though John's forum isnt the right place... I am getting it exposed to as many people as possible to explore it further. I thought of something like this in concept only as a means to lower my oscillator thd on another forum. Then I rediscovered Sandman's invention. Thx-RNM
Last edited:
I had played a lot with error correction since the 70s. Everything playing with feedback is too acrobatic, on my opinion, as it deal with comparisons ( phase problems due to delays). The method used with some low HD OPAs (including Scott's baby) seems more healthy: Low open loop distortion, huge gain, huge feedback.
Error cancellation (not the same thing) looks more promising, but need good paired parts, so not industrial neither.
Richard, i tried, in the same spirit to start a collaborative project with my protection circuit (same idea).
It works wonderfull i my own amp since 20years. Can be used to add a correction error to any existing amp too.
http://www.diyaudio.com/forums/solid-state/221737-ultimate-amp-protection-circuit.html#post3254403
Collaboration ? Nada, peanuts, nothing !!!!
Thread is dead and i'm too discouraged to do everything by myself, BOM, printed CI etc.
Last edited:
- Status
- Not open for further replies.