2. tech-talk
  3. 1366579878

Electronics 101: Use of Negative Feedback; a Q

The topic of negative feedback (NF) has been beaten to death in other posts. Nevertheless, I have a question that touches on another aspect of NF that I do not recall having been explained. The Q starts in the middle of this OP.

By way of summary, many of our technical members explained why NF is used in amplifier circuits, e.g., extends bandwidth, helps to prevent oscillation, reduces distortion, reduces output impedance and therefore increases DF, and so forth. However, there is a cost.

Many of our technical members also explained that NF creates TIM (temporal intermodular distortion) because of the nano-second of time it takes for the NF circuit to tell the input circuit to make adjustments. TIM distortion results in distortion in odd-ordered harmonics (particularly the 5th, 7th and 9th) which our brains interpret as loudness, and in turn acoustic harshness.

Hopefully, I got the basics down. I'm sure my summary will be "clarified" by our techies.

But here's the Q (or 2). I read that NF also enables the amplifier (SS or tube) to regulate its output power to match a speaker's changing impedance stats as frequency changes. Could the tech members please clarify my understanding about this point.

It's kind of important because even though the subject of "tube friendly" and "tube unfriendly" speakers has also been killed, it would appear that electrical matching concerns between an amp and speakers having fluctuating impedance stats as a function of frequency may be mitigated in whole or part by using NF. Of course, I not suggesting that NF will compensate for a short or an open circuit, but somewhere in between there will be harmony. ;>')

Perhaps the technical members can put some "flesh on the bones" with respect to this Q.

I suppose the Q raises a related Q about amps (SS or tube) that do not use NF because of concerns pertaining to TIM distortion as mentioned. Great . . . the TIM distortion problem is solved if no NF is used. But how does the amp regulate its output power to compensate for fluctuating impedance curves?

Thanks.
bifwynne
bifwynne OP
2,206 posts
 
Thank you Kirkus for your cogent and thoughtful response. Hopefully, we will read more posts that address the very narrow issue about using NF to compensate for "varying impedance-vs-frequency characteristic[s]" of speakers.

You also said that "[i]f the amplifier's output impedance differs significantly from that which the speaker designer used for evaluation, then the response of the speaker will be different from what the designer intended."

I assume from a lay person's perspective, in plain English, that means if a speaker was voiced to be driven by a "Voltage Paradigm" amplifer (as described in Ralph Karsten's White Paper, a SS amp), then using a "Power Paradigm" amp (usually a tube amp) to drive the speakers may affect the sonic presentation. This assumes of course, that the tube amp has a relatively high putput impedance if little or no NF is used.

Therefore, the amount of acoustic deviation between the actual versus intended presentation presumably will be affected by how flat (or "tube friendly") the speaker's impedance curve is, and how much NF is being using to reduce the output impedance of the amp, be it SS or tube. I surmise that if a tube amp uses NF to reduce its output impedance, it will behave somewhat like a "Voltage Paradigm" amp. In other words, the acoustic deviation between what the designer intended and actual performance may be mitigated (i.e., reduced). Is this correct?

So Kirkus, the follow up to my Q was how do so called zero NF amps manage to compensated for changing speaker impedances?? Is the bottom line objective simply to reduce the amp's output impedance by whatever means is possible, i.e., through NF or otherwise?

I believe that Ayre SS amps do not use NF in the circuit design, yet have relatively high DFs, thus suggested low output impedances. I don't know how Ayre achieves these results without NF, but they do. Is that all that is needed, low output impedances and the amp will be a champ??

Thanks again.

BIF
bifwynne OP
2,206 posts
 
Quick follow-up to my last post. As I recall, some time back, one or more of our EE audio tech members commented that a DF of more than 20 DF doesn't add much acoustic benefit. I think a DF of 20 correlates to an output impedance of .4 ohms.

In late 2011 or sometime in 2012, ARC released a new Referance tube amp, the Ref 150. This baby has a 1000 joule power supply, which I assume is a lot. In addition, the ARCDB web site says the Ref 150 uses 14 db of NF and has a DF of 17, the later stat getting close to the DF factor of 20 mentioned above.

Synthesizing all of this at a high level, what I understand is that lowering output impedance much beyond .4 ohms may yield little benefit. However, at .4 ohms or lower, the amp (be it tube or SS) will behave like a Voltage Paradigm amp and be better able to compensate for the changing impedances values of a speaker in a way the designer intended. Hence, using the Ref 150 as an example, can one expect that it might do a fair job handling a speaker that was designed to be driven by a tube amp, even if the speaker does not have tube friendly impedance curves and negative phase angles??

I suspect I'm probably mixing and matching concepts here, but hopefully you can untangle what I just wrote in this post.

Thanks again.
gregm
3,405 posts
 
I believe that Ayre SS amps do not use NF in the circuit design, (...)
I find that hard to believe, Ayre is a very serious constructor... Maybe what they are saying is there is no global FB, only local?
Is that all that is needed, low output impedances and the amp will be a champ??
Usually, a lowish impedance does yield good results if/ when the speakers are designed with that in mind. Many speakers are...
If not (take the Lowther based designs for example), then output impedance in the ohm range (rather than milli-ohm) would be better!
kirkus
502 posts
 
I believe that Ayre SS amps do not use NF in the circuit design, yet have relatively high DFs, thus suggested low output impedances. I don't know how Ayre achieves these results without NF, but they do.
No. They use feedback, even if they say they don't, or come up with a reason why their feedback is "different". This is akin to all the "LED Televisions" on the market - virtually all of them are in fact LCD televisions with LEDs used as a backlight. But they needed a new acronym to differentiate them from the previous models, and marketing departments aren't exactly known for their terminological precision. How many times have you seen "solid-state" on the outside of a television who's principal component was a picture tube? "Zero-feedback" is a similar label.

In reality, virtually every audio power amplifier without an output transformer (tube or transistor) uses some sort of unity-gain voltage follower circuit as the final stage to lower its output impedance. This type of circuit has 100% local negative feedback, hence the unity gain.
I assume from a lay person's perspective, in plain English, that means if a speaker was voiced to be driven by a "Voltage Paradigm" amplifer (as described in Ralph Karsten's White Paper, a SS amp), then using a "Power Paradigm" amp (usually a tube amp) to drive the speakers may affect the sonic presentation. This assumes of course, that the tube amp has a relatively high putput impedance if little or no NF is used.
Ralph really seems to favor a binary viewpoint of the subject with these two clearly-defined camps of amplifiers and speakers . . . but it's indeed correct that certain some loudspeakers produce far more variation in response with a high output impedance than others. This is an important subject to him because his circuit design preferences place practical limits on how low his amplifiers' output impedances can be.

As for the subject of how low an amplifier's output impedance *needs* to be, I actually think John Atkinson approaches this subject in a rational and practical manner when he measures an amplifier's response into an IHF speaker load, and the corresponding impedance/phase plots for the loudspeakers he measures. It's also possible that one may prefer a loudspeaker's response to be somewhat different that its designer, and that a higher-output-impedance amplifier may yield a pleasing result.

Personally, I can't say I've found such a modification to a speaker's frequency response to be a pleasing one, except for a few very rare occasions. But your mileage may vary . . .
bifwynne OP
2,206 posts
 
Thanks again Kirkus. Admittedly, I am not familiar with Atkinson's test involving an IHF speaker load. Care to explain what that test is all about?

So, going back to the use of NF to reduce an amp's output impedance, in the case of the ARC Ref 150 which uses 14 db of NF and probably has an output impedance between .4 and 1 ohm (probably closer to .4 ohm because its "rated" and reported DF is 17), can one assume that the Ref 150 will behave "SS-like" if presented with a speaker load that was voiced to be driven by a high current SS amp?

Isn't that the bottom line consideration here? Short of shlepping a 75 pound tube amp to a dealer to properly audition speakers, it's helpful to know (or at least be able to reasonably predict) if a tube amp can "switch-hit" and function in a "SS-like" way if presented with a speaker load which was voiced to be driven by a low-impedance, high DF, high-current SS amp. If the foregoing proposition is technically accurate, is it also fair to consider a tube amp's rated output impedance and correlatively its DF, when matching it to a speaker load? Any back of the hand guidelines?

Also, is output impedance less critical in the higher frequencies? Many 3-way speakers have an impedance bump at the midrange-tweeter cross-over point. I believe Ralph's White predicts that a "pure" Power Paradigm amp will deliver more power (watts) than a "pure" Voltage Paradigm amp at higher impedance levels. If so, it may matter.

Thanks again.

P.S. check the Ayre web site re the zero NF point. And I do have a very high-level understanding of the difference between local versus global feedback.
More to discover