Just wondering:
Has anyone tried the method of mitigating the speaker cable resistance by enclosing it in the feedback loop? How has it worked? I'm considering doing it in a bridged configuration (so that both the "outgoing" and "incoming" speaker cables can be integrated in the feedback loop).
==> see attached
Has anyone tried the method of mitigating the speaker cable resistance by enclosing it in the feedback loop? How has it worked? I'm considering doing it in a bridged configuration (so that both the "outgoing" and "incoming" speaker cables can be integrated in the feedback loop).
==> see attached
Attachments
Good idea.
I think the main challenge you'll face is the impedance of the speaker cable in series with your feedback loop reducing the phase margin. Speak cables are very inductive. In other words, your closed-loop bandwidth may be limited.
This technique is used with some dc power supplies. A "sense" wire is run from the circuit back to the psu so that any voltage drops across the power leads are compensated for. This works very well but the frequency response is limited.
I think the main challenge you'll face is the impedance of the speaker cable in series with your feedback loop reducing the phase margin. Speak cables are very inductive. In other words, your closed-loop bandwidth may be limited.
This technique is used with some dc power supplies. A "sense" wire is run from the circuit back to the psu so that any voltage drops across the power leads are compensated for. This works very well but the frequency response is limited.
this has been done a few times i can think of the laye 70 when i attendedd a semina held by trio at the royal horse guards hotel and heard such a system the tio method was a 4 wire system thjat monitored both spk cables the result was outstanding
i also at that time came across a toshiba system i think that was 3 wire
yes you do open the feed back system to rf and noise etc but the main disadvantage is if you lose a feed back cable the amplifier may go and put on a large dc offset
the way that i remember trio doing it was to use a high value resistor 20 k or so inside the amp between the feed and sensor wire so that the out put did not float high or low away from 0
go on give it a try and see what you think
i also at that time came across a toshiba system i think that was 3 wire
yes you do open the feed back system to rf and noise etc but the main disadvantage is if you lose a feed back cable the amplifier may go and put on a large dc offset
the way that i remember trio doing it was to use a high value resistor 20 k or so inside the amp between the feed and sensor wire so that the out put did not float high or low away from 0
go on give it a try and see what you think
Yes, it can be done. As for the RF issue; that's already there - we already connect an aerial to our feedback input, so no real change there. Transistor amplifiers always include a series inductor anyway, so I don't see that a bit of loudspeaker cable will make much difference to amplifier stability but I'm prepared for the smoke to prove me wrong.
As I see it, the greatest danger is in losing the feedback or drive signal through a poor external connection. If that happened, you could rely on the amplifier to swing (probably fairly briefly) from rail to rail, settling at one rail when it had destroyed a transistor or two. The loudspeaker would follow shortly thereafter. I think if you're going to pull this trick you probably want soldered joints all the way. No 5-way binding posts, no crimp connectors, no solder tags, and definitely no 4mm banana plugs.
As I see it, the greatest danger is in losing the feedback or drive signal through a poor external connection. If that happened, you could rely on the amplifier to swing (probably fairly briefly) from rail to rail, settling at one rail when it had destroyed a transistor or two. The loudspeaker would follow shortly thereafter. I think if you're going to pull this trick you probably want soldered joints all the way. No 5-way binding posts, no crimp connectors, no solder tags, and definitely no 4mm banana plugs.
This was done commercially by Kenwood about 25 years ago (called something like LO-50M, a medium-powered monoblock). I don't know if they had a safety resistor or something to prevent disaster if the feedback wire came loose or if they just derived part of the feedback signal from the remote sensing. The amps were mildly popular, but the concept really didn't go anywhere in the marketplace.
You would need to consider your cable construction most carefully. In addition to the points already metioned; you have the potential for a large magnetic loop that can feed any field passing by (through) into the input of your amp. You would need 4 conductors total... all tightly twisted.
The inductance in your speaker cables is small in comaparison to inductance of the drivers... by a long shot. This leaves resistance, which should (cable looneyosity aside) have a proportional or linear effect on the signal; this can be compensated by turning the volume knob 0.001 degree.
I do sensing (KELVIN) loops like this all time... none are ever easy... you would think they should be. It seems that without a real purpose for doing so (absolute rather than relative accuracy), the risks, especially an open feedback loop (smoked driver), outweigh the benefits.
The inductance in your speaker cables is small in comaparison to inductance of the drivers... by a long shot. This leaves resistance, which should (cable looneyosity aside) have a proportional or linear effect on the signal; this can be compensated by turning the volume knob 0.001 degree.
I do sensing (KELVIN) loops like this all time... none are ever easy... you would think they should be. It seems that without a real purpose for doing so (absolute rather than relative accuracy), the risks, especially an open feedback loop (smoked driver), outweigh the benefits.
Yes, the speaker wires inside the negative feedback loop is a great idea. This is done all the time for large low voltage power supplies to lower cable voltage drop and regulation issues.
For audio it would be easier to implement for bridged amps because grounding is not used to drive the speaker. For non-bridge amplifier applications the ground sense error signal would have to be lifted off the high current ground. This could be done fairly easily with proper lay-out and grounding analysis. For safety reasons the sense loop would be connected near the speaker output terminals with maybe 100 ohm resistors. The sense cabling would require shielded twisted pair for RFI noise rejection.
I'm not sure why it is not done for more high end audio equipment. Possibly is deemed to complex for general public, being that it would require more external hook ups. Possibly exotic cable OEM's would feel threatened.
For audio it would be easier to implement for bridged amps because grounding is not used to drive the speaker. For non-bridge amplifier applications the ground sense error signal would have to be lifted off the high current ground. This could be done fairly easily with proper lay-out and grounding analysis. For safety reasons the sense loop would be connected near the speaker output terminals with maybe 100 ohm resistors. The sense cabling would require shielded twisted pair for RFI noise rejection.
I'm not sure why it is not done for more high end audio equipment. Possibly is deemed to complex for general public, being that it would require more external hook ups. Possibly exotic cable OEM's would feel threatened.
Easy to do. (force conductors are the current carrying ones, sense conductors are the voltage feedback lines).
The most important issues are safety of feedback connection. As infinia suggested, use 100 ohm safety resistors at the amp, should one or both of the sense lines disconnect.
If you use parallel wires for both force and sense, you will intercept the field of the force conductors to varying degrees with the sense ones. This is not important as a DC consideration, but will be one for time varying signals, like music. Should complete coupling occur, the inductance of the force run will not be seen nor corrected by the amp.
Wire considerations:
1. Two independent twisted pairs can be used, but they MUST have different pitches..two identical pitches will couple darn near like parallel ones.
2. quad twist. The force pair being opposing conductors, the sense pair the other two...like an analog clock, force at 12 and 6, sense at 3 and 9. Polarities of each do not matter with respect to coupling. The quad twist setup this way force the two pair to be orthogonal, meaning that they do not couple magnetically.
3. Quadaxial..I made a 25 foot length of quadaxial expressly for this purpose, but the person who evaluated the sound of the cable never returned my wire, so I could not offer it to an amp designer to try this method.
4. Parallel force, coax sense: a ground loop will form between, and would require differential sensing...yuk.
5. Twisted/coax force, coax sense: ok..
Cheers, John
The most important issues are safety of feedback connection. As infinia suggested, use 100 ohm safety resistors at the amp, should one or both of the sense lines disconnect.
If you use parallel wires for both force and sense, you will intercept the field of the force conductors to varying degrees with the sense ones. This is not important as a DC consideration, but will be one for time varying signals, like music. Should complete coupling occur, the inductance of the force run will not be seen nor corrected by the amp.
Wire considerations:
1. Two independent twisted pairs can be used, but they MUST have different pitches..two identical pitches will couple darn near like parallel ones.
2. quad twist. The force pair being opposing conductors, the sense pair the other two...like an analog clock, force at 12 and 6, sense at 3 and 9. Polarities of each do not matter with respect to coupling. The quad twist setup this way force the two pair to be orthogonal, meaning that they do not couple magnetically.
3. Quadaxial..I made a 25 foot length of quadaxial expressly for this purpose, but the person who evaluated the sound of the cable never returned my wire, so I could not offer it to an amp designer to try this method.
4. Parallel force, coax sense: a ground loop will form between, and would require differential sensing...yuk.
5. Twisted/coax force, coax sense: ok..
Cheers, John
poobah said:All good points... but what is the audible advantage?
Audible advantage??? Who knows. Well, at least it gets rid of the arguments and flames..that's quieter..
It does eliminate the speaker wire, though..if done correctly..
No speaker wire, no speaker wire flame war..
But it does create the possibility of a very easy A/B test scenario, now doesn't it?
Click...wires there...click, wires gone..Hear a diff? and no terminals with connection resistance, nothing moved...no piece of garbage A/B speaker connection box with lousy switches..
Cheers, John
poobah said:your feedback wires should go to accelerometers mounted on the cones... that where the REAL trouble lies.
For me the real trouble lies with ohms law..
Now, you want double integrations of a double differential system???
Sheesh, base jumping without a parachute is easier..
Cheers, John
I didn't say it would be easy... but that's where the biggest problem lies (drivers). And actually it might not be that hard if you have an good amp... phase correct up to 200 kHz.
And the A/b test is fine... as long as the volume shift is accounted for and nulled out... louder is ALWAYS better i.e. Stradivarious.
And the A/b test is fine... as long as the volume shift is accounted for and nulled out... louder is ALWAYS better i.e. Stradivarious.
Yes, it's been done.
Even the speaker itself has been put into the loop before using various sensor methods to measure speaker position.
(most commonly done and called 'motional feedback') The idea is almost as old as electronic audio.
Another option is to make the amp output impedence negative which can 'correct' for cable resistance or even speaker voicecoil resistance.
Even the speaker itself has been put into the loop before using various sensor methods to measure speaker position.
(most commonly done and called 'motional feedback') The idea is almost as old as electronic audio.
Another option is to make the amp output impedence negative which can 'correct' for cable resistance or even speaker voicecoil resistance.
But isn't the series inductor outside of the feedback loop? It is normally there to isolate the feedback loop from the impedance of cable and speaker.
A few metres of speaker cable will have several uH of inductance and will have some capacitance. The speaker impedance will vary quite a lot under different signal conditions...sometimes looking inductive and sometime capacitive, as well as resistive. There is a good chance a standard amp will get upset.
This sort of system can obviously be made to work but I think it requires some modification of a normal feedback amp to avoid problems. Just a heads up.
I remember reading about the Philips motion-feedback speaker. I never read any info about how they sounded. If memory serves the feedback was fairly low frequency and taken off the woofer only.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.