• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

New Feedback Path

mike7877

Addicted to Fun and Learning
Joined
Aug 5, 2021
Messages
938
Likes
210
So I was thinking and the thought occurred to me:

You know the last stage of feedback in most amplifiers, the one that goes from the final output to the (usually negative) input of the second last stage?

Well, what if, instead of the signal for feedback being taken from, essentially, the output terminals: a second wire was run alll the way to the speaker's input terminals for it?

Would this negate any/all of the [unwanted] effects from the properties of the wire? If not perfectly, mostly? (say >90% in most cases?)

I'm wondering because, for a good 100 watt RMS passive speaker system to perform properly, you need an 8-10 gauge speaker wire. If, instead, some lowly 16-18 gauge oxidized lamp cord could be used with a single, equally pathetic (but much less corroded), 28 gauge conductor (think phone cable) that'd be great!

I'm not thinking this should be a feature standard with all new amplifiers, just (if it's viable) a possible feature that some high fidelity amp manufacturers could include on various models of their amps, maybe especially ones to appeal to some of their, more DIY-type, customers. Enable it with a dip switch on the back, which once slid from "internal" to "sense", the trace to the input of the second last stage (for feedback) that usually pulled from the part of the PCB feeding the 5-way binding post (output), it would be directed an RJ11 jack on the rear. Included with the amp, then, would be a custom cable with male RJ11 on one side, and banana connector on the other. The banana connector would be to clamp on the far end of the speaker wire with an alligator clip. Pic:
1737443551380.png



Right now, to perform really well, passive speakers' crossovers need to be made with 1% parts, and the drivers' properties need to be really consistent + exactly what the crossover was designed for. When you add 0.1 ohms series resistance between the amplifier and that extremely well-tuned system, the crossover slopes no longer align (both slope and cutoff frequency are affected, the higher the order, the more severe the effect). The most obvious and detrimental result of this is the change of the speaker's frequency response - the drivers no longer work constructively through the transition band.

To keep this poast short, speaker wire does matter, and it matters immensely when high quality, consistently manufactured drivers are used with crossovers that are actually designed to match their optimal operating frequency ranges, and those crossovers use quality parts like low-gauge air-core inductors and audio-grade capacitors. And when those quality parts are actually the exact values for the design and not the next closest thing off the shelf. Of course, the amplifier needs to have a low output impedance as well. More matters, but not to this conversation (it's already getting long...). In a lot of cases, whether you use a 16/2 extension cord or 10 gauge speaker wire won't make an audible difference. But also, in a lot of cases when you're dealing with quality equipment designed for high-fidelity sound reproduction, the difference between those two speaker wires is real, and profound. And present at all playback levels.


Thoughts on this feedback? What about method (does the cable need to be shielded or anything else?)
 
Last edited:
 
Would this negate any/all of the [unwanted] effects from the properties of the wire? If not perfectly, mostly? (say >90% in most cases?)
It would but now you are exposing the internal signal path of the amp to the outside. Improper usage could cause oscillations and other ways to kill the amp. If you were building a closed system of amp and speaker, you could do this.

FYI, my Klippel NFS speaker measurement system uses such a "sense" feature to eliminate frequency response errors of the amp and cabling.
 
Back in 2010.. someone else was thinking too :)
The whole article is here: https://www.tubecad.com/2010/03/blog0184.htm

Thus, the faulty logic goes like this, since the loudspeaker's negative input terminal attaches to the amplifier's ground, nothing can ever go wrong at this point. How could it? It's ground after all! If only that were true, life would be so much easier. Here is an example of how three wires are not enough.
 
So I was thinking and the thought occurred to me:

You know the last stage of feedback in most amplifiers, the one that goes from the final output to the (usually negative) input of the second last stage?

Well, what if, instead of the signal for feedback being taken from, essentially, the output terminals: a second wire was run alll the way to the speaker's input terminals for it?

Would this negate any/all of the [unwanted] effects from the properties of the wire? If not perfectly, mostly? (say >90% in most cases?)

I'm wondering because, for a good 100 watt RMS passive speaker system to perform properly, you need an 8-10 gauge speaker wire. If, instead, some lowly 16-18 gauge oxidized lamp cord could be used with a single, equally pathetic (but much less corroded), 28 gauge conductor (think phone cable) that'd be great!

I'm not thinking this should be a feature standard with all new amplifiers, just (if it's viable) a possible feature that some high fidelity amp manufacturers could include on various models of their amps, maybe especially ones to appeal to some of their, more DIY-type, customers. Enable it with a dip switch on the back, which once slid from "internal" to "sense", the trace to the input of the second last stage (for feedback) that usually pulled from the part of the PCB feeding the 5-way binding post (output), it would be directed an RJ11 jack on the rear. Included with the amp, then, would be a custom cable with male RJ11 on one side, and banana connector on the other. The banana connector would be to clamp on the far end of the speaker wire with an alligator clip. Pic:
View attachment 422732


Right now, to perform really well, passive speakers' crossovers need to be made with 1% parts, and the drivers' properties need to be really consistent + exactly what the crossover was designed for. When you add 0.1 ohms series resistance between the amplifier and that extremely well-tuned system, the crossover slopes no longer align (both slope and cutoff frequency are affected, the higher the order, the more severe the effect). The most obvious and detrimental result of this is the change of the speaker's frequency response - the drivers no longer work constructively through the transition band.

To keep this poast short, speaker wire does matter, and it matters immensely when high quality, consistently manufactured drivers are used with crossovers that are actually designed to match their optimal operating frequency ranges, and those crossovers use quality parts like low-gauge air-core inductors and audio-grade capacitors. And when those quality parts are actually the exact values for the design and not the next closest thing off the shelf. Of course, the amplifier needs to have a low output impedance as well. More matters, but not to this conversation (it's already getting long...). In a lot of cases, whether you use a 16/2 extension cord or 10 gauge speaker wire won't make an audible difference. But also, in a lot of cases when you're dealing with quality equipment designed for high-fidelity sound reproduction, the difference between those two speaker wires is real, and profound. And present at all playback levels.


Thoughts on this feedback? What about method (does the cable need to be shielded or anything else?)
Google "Kelvin Sensing".

I had a Deltec power amplifier which did exactly what you describe. It worked well. You can read about the DPA50S and DPA100S here:

BUT, the mechanism is fragile. If you accidentally break the sensing pair, the amplifier can/will become dangerously unstable
 
I'm wondering because, for a good 100 watt RMS passive speaker system to perform properly, you need an 8-10 gauge speaker wire.
I'm not sure where you get that from. In typical 10ft lengths, 14 gauge is already quite significant overkill - even with 4ohm speakers.

(What is important is not the power, but the ratio of speaker wire impedance to speaker impedance)

12 gauge (as I use) is double overkill.

Given that as long as speaker wire is of sufficiently low impedance, then the effects of the wire is inaudible, it doesn't make much sense to reduce already inaudble effects with a complicated feedback method.
 
So I was thinking and the thought occurred to me:

Thoughts on this feedback? What about method (does the cable need to be shielded or anything else?)
Kenwood Sigma drive was basically that. (see post#5)
That method is not used, certainly for longer cable runs as it is easy to introduce timing errors due to cable length alone.
It is not a viable option but could be in active systems but not needed because of the short cable lengths.

MFB makes more sense (for low frequencies)
 
BUT, the mechanism is fragile. If you accidentally break the sensing pair, the amplifier can/will become dangerously unstable

Yeah, I hadn't thought of just how fragile it is. Maybe an entirely proprietary speaker cable would be required with the third conductor included and attached internally

edit: looking into your previous amp - interesting company!
 
Last edited:
Kenwood Sigma drive was basically that. (see post#5)
That method is not used, certainly for longer cable runs as it is easy to introduce timing errors due to cable length alone.
It is not a viable option but could be in active systems but not needed because of the short cable lengths.

MFB makes more sense (for low frequencies)

Makes sense. Are you sure the timing problem is severe enough to matter with signals under 20kHz for 10-20 foot runs?
 
Philips made speaker drivers with a feedback sensor years ago. Don’t remember what they called it, but I assume they also made amplifiers that could use the feedback.

Edit: here we are

 
Philips made speaker drivers with a feedback sensor years ago. Don’t remember what they called it, but I assume they also made amplifiers that could use the feedback.

Edit: here we are


This is different - it's a sensor on the woofer, not electrical feedback for the entire speaker. But the concept is similar
 
Makes sense. Are you sure the timing problem is severe enough to matter with signals under 20kHz for 10-20 foot runs?
Yep, simply because the amplifier itself has to be stable to much higher frequencies.
Part of this can be done with local feedback but there is going to be a transition band that must be well above the audible range when the desire is to compensate for not just the lower part of the audible range.
The more the circuit starts to lean on local feedback (can even be overall feedback within the amp) the less effective the overall (as in speaker wires + back EMF) feedback becomes.

Then there is a huge range of load impedances and cables. Also one would need 4-wire cables for this concept.
 
Yeah, I hadn't thought of just how fragile it is. Maybe an entirely proprietary speaker cable would be required with the third conductor included and attached internally

edit: looking into your previous amp - interesting company!
The amplifier shipped with its own proprietary cables using their own bundled single-core "slink". At the speaker end the 4 "wires" were bought together into 2 normal banana plugs and at the amplifier split out into a male XLR and a female XLR. All OK, but inconvenient if the cables were not long enough. After one house move, I had to make my own. I was VERY careful about the solder quality of the feedback cables.

After about 25 years of use I retired the amp last year. I'm not wealthy enough to keep swapping gear so I have to pick the right equipment (it measured well for the time) and look after it really well. But this amplifier did make me a bit nervous and I checked the cables and connections regularly.
 
As mentioned by @amirm , feedback point taken from behind the speaker cable, from speaker box terminals, is highly problematic and unpredictable and would result mostly in oscillations. Bad idea, very bad. And especially if not done by highly qualified specialist and tailored to just one exact setup.
 
As mentioned by @amirm , feedback point taken from behind the speaker cable, from speaker box terminals, is highly problematic and unpredictable and would result mostly in oscillations. Bad idea, very bad. And especially if not done by highly qualified specialist and tailored to just one exact setup.
I agree. I never had any issues in practice, but I was perpetually concerned about instability. I don't think it's a safe or suitable approach for domestic consumers.
 
  • Like
Reactions: pma
I agree. I never had any issues in practice, but I was perpetually concerned about instability. I don't think it's a safe or suitable approach for domestic consumers.

In the OP I did say more for the DIY crowd. But I get the point. It seems it already has been a niche product, so nice niche I hadn't heard of it and there are only a couple examples.
 
Last edited:
Back
Top Bottom