Nelson Pass said:
Where there is a standing wave there is a change of impedance. That's not the same thing as resonance. Resonance is caused by the reactive elements of the circuit. The change of impedance due to standing waves is simply due to the constructive and destructive addition of the reflections.
How can RF pickup be unrelated? The electrical wavelengths of audio band signals are far too long to cause any issues relating to reflections and standing waves. So where else are you going to find frequencies so high that their electrical wavelengths are short relative to the length of the cable?
Even if you assume a very slow propagation velocity of say 50% C, at 20kHz, you're talking about an electrical wavelength on the order of five miles. How is a five mile long wavelength going to create any sort of reflection issues in a line just a few methers in length?
Incidental? At frequencies where reflections are any sort of concern, the cable's characteristic impedance is the square root of L/C. So how do you figure that inductance is merely incidental to the cable's characteristic impedance?
It's incidental at audio frequencies, but that's because there really is no characteristic impedance at audio frequencies because the cable's impedance is a function of frequency.
se
mirlo said:
You end up with the same result.
For that 80 ohm speaker, you need to add about 10 times as much voice coil wire, which results in a voice coil with 10 times the mass which will drastically reduce the efficiency of the speaker, which means you need to increase the amount of power you feed it to get the same output.
So where's the advantage?
se
To clarify:
By 80 Ohm speaker, I meant use 10 8 Ohm drivers in series, and a huge enclosure, a-la Genesis 1.1 or Infinity IRS (which probably wire their multiple drivers in series-parallel), not a single driver with a heacy, high resistance voice coil made of lots of fine gauge wire.
I'm not really sure though what was in mind at the beginning of this thread.
Is it _really_ easier to build a good amplifier that has 10 times the voltage output capability and 1/10 the current output capability? Unless you are talking about tubes, high voltage output devices aren't that great. So I think you'd trade problems for other problems...
By 80 Ohm speaker, I meant use 10 8 Ohm drivers in series, and a huge enclosure, a-la Genesis 1.1 or Infinity IRS (which probably wire their multiple drivers in series-parallel), not a single driver with a heacy, high resistance voice coil made of lots of fine gauge wire.
I'm not really sure though what was in mind at the beginning of this thread.
Is it _really_ easier to build a good amplifier that has 10 times the voltage output capability and 1/10 the current output capability? Unless you are talking about tubes, high voltage output devices aren't that great. So I think you'd trade problems for other problems...
Re: To clarify:
Ah, ok. Sorry for the misunderstanding.
But you're ultimately getting the same output for the same power. 10 times the voltage and 1/10th the current.
Getting 10 times the voltage output requires 10 times the voltage gain which in my experience is more problematic than current gain. The current gain of a typical amplifier is handled by what amount to simple emitter followers and getting the current you want is a relatively simple matter of having enough output devices, enough heatsinking and enough power supply current.
se
mirlo said:
Ah, ok. Sorry for the misunderstanding.
But you're ultimately getting the same output for the same power. 10 times the voltage and 1/10th the current.
Getting 10 times the voltage output requires 10 times the voltage gain which in my experience is more problematic than current gain. The current gain of a typical amplifier is handled by what amount to simple emitter followers and getting the current you want is a relatively simple matter of having enough output devices, enough heatsinking and enough power supply current.
se
HifiZen,
Whether the output impedance is reduced by loop feedback or by localised feedback (emitter followers driven by a low Z) may not be so important. Both are feedback systems.
I'd have thought a good system, in theory, would be to have a transconductance amp driving a 0.1 ohm resistive load, then connect the speaker across the 0.1 ohm resistor. Of course, power wastage would be immense and so this isn't practical.
Instabilities due to phase shifts in the MHz region where the loop gain is >1 may be problematical. But consider this: what business has an audio amp in having an open loop gain above 1MHz? What's the point.
BAM
Whether the output impedance is reduced by loop feedback or by localised feedback (emitter followers driven by a low Z) may not be so important. Both are feedback systems.
I'd have thought a good system, in theory, would be to have a transconductance amp driving a 0.1 ohm resistive load, then connect the speaker across the 0.1 ohm resistor. Of course, power wastage would be immense and so this isn't practical.
Instabilities due to phase shifts in the MHz region where the loop gain is >1 may be problematical. But consider this: what business has an audio amp in having an open loop gain above 1MHz? What's the point.
BAM
Steve Eddy said:
Making an amp with 80 ohm output impedance might be a little overkill.
I’m really trying to make an amp which is immune to reflections coming from the (speaker / speaker wire circuit).
If I were to make an amp at double voltage with an 8 ohm series output, designed the amp, before the 8 Ohm series, to drive without distortion loads ranging from 4 Ohm to 1 KOhm. Can I be sure that with an inductance free 8 ohm series resistance will virtually make the amp’s internal gain/feedback circuit immune from the (speaker / speaker wire circuit)’s reflections & picked up transmition noise?
My concern here is not deep base, or sub sonic, it’s for super refined voice & highs.
chasing windmills
Hi Brian,
You are in danger of doing a Don Quixote here.
Forget the reflections idea - it's not important. Its pursuit will cause you to make bad design decisions. It is another of those mis-interpretations of engineering principles that arise from time to time.
Better to concentrate on the principal issue: controlling a voltage across a reactive loudspeaker load. The impedance of a speaker varies widly with frequency and contains rapid phase shifts and resonances. It is typical for a speaker to dip as low as 1 or 2 ohms and exceed 10 or 15 ohms, and it is a mix of inductance and capacitance. Reflections are the least of your worries!
For most speakers I would say <1 ohms output resistance is adequate for a power amp, <0.5 ohms is good. At 8 ohms you'll hear the lack of control. But don't get tricked into pursuing much lower resistance than this - again it will lead you to making design choices that will corrupt the sound in other ways and you'll get frustrated. Everything in balance.
BAM
Hi Brian,
You are in danger of doing a Don Quixote here.
Forget the reflections idea - it's not important. Its pursuit will cause you to make bad design decisions. It is another of those mis-interpretations of engineering principles that arise from time to time.
Better to concentrate on the principal issue: controlling a voltage across a reactive loudspeaker load. The impedance of a speaker varies widly with frequency and contains rapid phase shifts and resonances. It is typical for a speaker to dip as low as 1 or 2 ohms and exceed 10 or 15 ohms, and it is a mix of inductance and capacitance. Reflections are the least of your worries!
For most speakers I would say <1 ohms output resistance is adequate for a power amp, <0.5 ohms is good. At 8 ohms you'll hear the lack of control. But don't get tricked into pursuing much lower resistance than this - again it will lead you to making design choices that will corrupt the sound in other ways and you'll get frustrated. Everything in balance.
BAM
Nelson Pass said:
Every amp is immune to cable reflections. Not every amp is immune to different reactive loads however. And it's reactance which leads to instability. Cable reflections are not reactances. They simply result in standing waves. The superposition of the direct and reflected waves.
se
Good vibrations?
Steve,
Nelson is saying, correctly, that the net effect of standing waves is to create resonances. Just like those in church organ pipes. An incorrectly terminated transmission line will start to resonate at certain frequencies - wavelength multiples of the transmission line length. Same effect as standing waves in radio transmission. When the line is terminated with a short circuit or an open circuit it will behave like a high Q resonance circuit: at certain wavelength multiples the cable will appear to be either a dead short or an open circuit . Obviously a dead short is problematic for a voltage amplifier.
Nelson's paper basically points out that if a feedback amplifier's bandwidth is high enough then a cable resonance may come close enough to it to change the phase margin and destabilize the amplifier. Nelson's solution is to terminate the cable at the speaker end (this assumes the speakers impedance at these frequencies is not influencial) with an RC load to help reduce reflections. Normally you would choose a pure R for this purpose but this would end up dissipating power at audio frequencies.
The alternative is to make sure the amp isn't pumping energy into the cable at wavelengths near to the cable length. It doesn't need to in an absolute sense since it really only needs to provide energy up to 20kHz (assuming a linear speaker load). Certain designs end up powering into the MHz region as a side effect, often to enable high levels of feedback in the audio band whilst trying to maintain closed loop phase margin. Alternatively, you can damp the amp output - put a damping resistor in series with the output and parallel it with an inductor to minimize the effect of the damper in the audio band - sounds familiar?
BAM
(still considering Morello's assertion that a transmission line does not resonate when one end is matched and one end isn't.
)
Steve,
Nelson is saying, correctly, that the net effect of standing waves is to create resonances. Just like those in church organ pipes. An incorrectly terminated transmission line will start to resonate at certain frequencies - wavelength multiples of the transmission line length. Same effect as standing waves in radio transmission. When the line is terminated with a short circuit or an open circuit it will behave like a high Q resonance circuit: at certain wavelength multiples the cable will appear to be either a dead short or an open circuit . Obviously a dead short is problematic for a voltage amplifier.
Nelson's paper basically points out that if a feedback amplifier's bandwidth is high enough then a cable resonance may come close enough to it to change the phase margin and destabilize the amplifier. Nelson's solution is to terminate the cable at the speaker end (this assumes the speakers impedance at these frequencies is not influencial) with an RC load to help reduce reflections. Normally you would choose a pure R for this purpose but this would end up dissipating power at audio frequencies.
The alternative is to make sure the amp isn't pumping energy into the cable at wavelengths near to the cable length. It doesn't need to in an absolute sense since it really only needs to provide energy up to 20kHz (assuming a linear speaker load). Certain designs end up powering into the MHz region as a side effect, often to enable high levels of feedback in the audio band whilst trying to maintain closed loop phase margin. Alternatively, you can damp the amp output - put a damping resistor in series with the output and parallel it with an inductor to minimize the effect of the damper in the audio band - sounds familiar?
BAM
(still considering Morello's assertion that a transmission line does not resonate when one end is matched and one end isn't.
)
idea
I suppose a really neat solution to protect over-zealous amps would be a self-damping speaker cable. Some dielectric material between the conductors that was lossy at high frequencies, creating a damping effect. The cable would exhibit resistive qualities at high frequencies thus lowering its Q at the resonant wavelengths - like a crummy capacitor or a leaky transformer.
Seems like something someone has probably already invented.
I suppose a really neat solution to protect over-zealous amps would be a self-damping speaker cable. Some dielectric material between the conductors that was lossy at high frequencies, creating a damping effect. The cable would exhibit resistive qualities at high frequencies thus lowering its Q at the resonant wavelengths - like a crummy capacitor or a leaky transformer.
Seems like something someone has probably already invented.
traderbam said:
A few people corrected you, but none explained why.
Nelson said matching impedances at the speaker end will stop all reflections, which is correct. Even if the line is non matched at the amplifier end, since it is a pure output, it won't matter. If you match the load end of a line, there will be no reflection hitting the source end, and it won't matter whether its matched or not.
Nelson Pass said:
You can look at standing waves as a type of resonance I suppose. But that's largely a semantics issue and detracts from the more fundamental issue. That being reactance.
Standing waves are not reactive and do not change the voltage/current phase relationships of the signal. In other words, it's not standing waves that are slipping a pole into works and causing the amplifier to go into a death spiral. Instead it's the capacitive and inductanve reactances of the load (which includes the cables).
se
Shiver me timbers!
Ok, let's get mathematical!
Reactance is defined as the part of impedance that isn't resistive (the "imaginary" part). So let's examine the impedance of a terminated transmission line.
Z = Zo x (Zload + j.Zo.tanW)/(Zo + j.Zload.tanW)
Where
Z is the impedance seen at the source end
Zo is the characteristic impedance of the transmission line
Zload is the termination impedance
W is the number of wavelengths between source and load x 2pi
When Zload = Zo the cable looks like Zload to the source. If not, the effect of the distributed inductance and capacitance of the cable causes it to act like a variable resonating circuit, the impedance swinging back and forth from inductive to capacitve along its length. If Zload is a total mismatch, ie a dead short OR an open circuit, there will be points along the length of the cable that have either zero or infinite impedance. If energy is injected into a point of zero impedance the cable will soak it up like a sponge and STORE it, just like a LC series circuit. In fact, with no resistive losses the alternating current and voltage in the cable will grow out of control until something fails catastrophically.
These considerations are crucial in all high frequency work. Even microwave ovens blow up if the load (the thing you want to cook) is severely mismatched; the reflected microwave energy damages the magnetron. In radio transmission mismatched coax cables and antennas can result in damaged transmitters and dangerous voltage potentials along the coax. It is possible to be electrocuted by a antenna feed cable even though the transmitter itself outputs only a small voltage. Yikes!
The reflected energy in mismatched cable literally affects the impedance of the cable and this includes its reactance.
Ok, let's get mathematical!
Reactance is defined as the part of impedance that isn't resistive (the "imaginary" part). So let's examine the impedance of a terminated transmission line.
Z = Zo x (Zload + j.Zo.tanW)/(Zo + j.Zload.tanW)
Where
Z is the impedance seen at the source end
Zo is the characteristic impedance of the transmission line
Zload is the termination impedance
W is the number of wavelengths between source and load x 2pi
When Zload = Zo the cable looks like Zload to the source. If not, the effect of the distributed inductance and capacitance of the cable causes it to act like a variable resonating circuit, the impedance swinging back and forth from inductive to capacitve along its length. If Zload is a total mismatch, ie a dead short OR an open circuit, there will be points along the length of the cable that have either zero or infinite impedance. If energy is injected into a point of zero impedance the cable will soak it up like a sponge and STORE it, just like a LC series circuit. In fact, with no resistive losses the alternating current and voltage in the cable will grow out of control until something fails catastrophically.
These considerations are crucial in all high frequency work. Even microwave ovens blow up if the load (the thing you want to cook) is severely mismatched; the reflected microwave energy damages the magnetron. In radio transmission mismatched coax cables and antennas can result in damaged transmitters and dangerous voltage potentials along the coax. It is possible to be electrocuted by a antenna feed cable even though the transmitter itself outputs only a small voltage. Yikes!
The reflected energy in mismatched cable literally affects the impedance of the cable and this includes its reactance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.