The above is true. But I have seem special duty loads that are made with coils and magnets but no paper cone. They are nearly silent even at high power.
Their main use is recording guitar amps But also good for amp tweaking late night.
Another old inconclusive thread
The original question was - If one feeds sine wave into amplifier loaded with either a dummy load or real speaker, and takes oscilloscope traces from speaker output, how the two traces will compare?
The answers were:
1. Don't ever do it, or you will damage your speaker, hurt your ears, and annoy your neighbors.
2. Traces will be different.
3. A contraption made of resistors, coils, and capacitors can be designed that electrically simulates a speaker, and that would be a better dummy load.
Unfortunately, none of these answers gets to the point.
1. Hardships of this experiment can be easily overcome by using hearing protection and soundproofing. Anechoic chamber is a common tool in acoustics.
2. One could guess that they will be different, but in what way?
3. Such contraption lacks a very important property of a speaker - being electromechanical and mechanoelectrical transducer at the same time.
When excited, speaker cone stores mechanical energy, which is then released via decaying oscillations over a broad range of frequencies. At resonance nodes, decay times might be as long as tens of milliseconds. This is characterized by cumulative decay spectrum plot, or so-called waterfall.
Because speaker acts as a dynamic microphone, these decay oscillations are converted into electric signal, which then feeds into amplifier's input through feedback networks. This cannot be good - due to mechanical delay there will be no cancellation of decay noise, but feedback signal will rather add to distortion. My guess is that distortion measurement with actual speaker instead of dummy load will reveal it. I also suspect that this has been examined in the past, but I couldn't find any information.
Anyone with knowledge or opinion, please chime in.
The original question was - If one feeds sine wave into amplifier loaded with either a dummy load or real speaker, and takes oscilloscope traces from speaker output, how the two traces will compare?
The answers were:
1. Don't ever do it, or you will damage your speaker, hurt your ears, and annoy your neighbors.
2. Traces will be different.
3. A contraption made of resistors, coils, and capacitors can be designed that electrically simulates a speaker, and that would be a better dummy load.
Unfortunately, none of these answers gets to the point.
1. Hardships of this experiment can be easily overcome by using hearing protection and soundproofing. Anechoic chamber is a common tool in acoustics.
2. One could guess that they will be different, but in what way?
3. Such contraption lacks a very important property of a speaker - being electromechanical and mechanoelectrical transducer at the same time.
When excited, speaker cone stores mechanical energy, which is then released via decaying oscillations over a broad range of frequencies. At resonance nodes, decay times might be as long as tens of milliseconds. This is characterized by cumulative decay spectrum plot, or so-called waterfall.
Because speaker acts as a dynamic microphone, these decay oscillations are converted into electric signal, which then feeds into amplifier's input through feedback networks. This cannot be good - due to mechanical delay there will be no cancellation of decay noise, but feedback signal will rather add to distortion. My guess is that distortion measurement with actual speaker instead of dummy load will reveal it. I also suspect that this has been examined in the past, but I couldn't find any information.
Anyone with knowledge or opinion, please chime in.
Thank you sser2.
You summarized and listed much of what has been said.
A couple of more things to think about:
A. Testing an amplifier at full power at 3, 5 or 20kHz into a loudspeaker is probably going to burn out the tweeter.
But I like the idea for anyone who has a bursted signal source, plus gated spectrum analysis capability.
The same goes for mid and low frequencies. This is a system, a load resistor is not a system (unless you listen to the output transformer "sing", it is a very poor loudspeaker).
This is not an easy test. But not everything good is easy.
B. Most loudspeakers impedances are very complex. At very low frequencies, and as frequency increases, they can be resistive, then LR, then R, then RC, then L. I just described a closed box woofer, its DCR, lower side of resonance, resonance, upper side of resonance, and voice coil inductance in that order. And that is just the woofer without a crossover.
How about the crossover, midrange, crossover, tweeter?
C. If you test with a resistive load (I do), you get some reasonable results (and everyone gets to play on the same field conditions).
D. But, consider the set of curves of a triode, pentode, or beam power tube.
Draw a load line on a set of triode curves. If you know what to look for, you can see the 2nd harmonic distortion in the unequal spacing of curves along the load line (crowded at one end, and open spaced at the other end). You will see the same for the pentode and beam power tubes.
Now, draw an Ellipse around the load line. You will see the upper portion of the ellipse that is above the load line. It is in the region where the rp is lower and more in control of the load (more linear). You will see the lower portion of the ellipse that is below the load line. It is in the region where the rp is higher and less in control of the load (less linear).
That is what the loudspeaker impedance causes at so many frequencies (at Oh, so many frequencies).
This is what you get in Single Ended, driving real world loudspeakers.
Don't get me wrong, I listen to SE amps, and enjoy them. I do reduce the effect of the ellipse by designing the amp to have a higher damping factor. You can put the 8 Ohm loudspeaker (it's not 8 Ohms at most frequencies) on the 4 Ohm tap, and get similar results (but you may get less power).
Now, take 2 sets of triode curves, with load lines, and twist one over the other (180 Degrees). That is push pull.
As the ellipse of one direction is less controlled by one tube's rp, it is more controlled by the other tube's rp.
I do enjoy listening to Push Pull amps too.
As you can easily see on the sets of curves, testing amplifiers with resistive loads (load lines), and loudspeakers with elliptical loads will not necessarily give the same results.
You summarized and listed much of what has been said.
A couple of more things to think about:
A. Testing an amplifier at full power at 3, 5 or 20kHz into a loudspeaker is probably going to burn out the tweeter.
But I like the idea for anyone who has a bursted signal source, plus gated spectrum analysis capability.
The same goes for mid and low frequencies. This is a system, a load resistor is not a system (unless you listen to the output transformer "sing", it is a very poor loudspeaker).
This is not an easy test. But not everything good is easy.
B. Most loudspeakers impedances are very complex. At very low frequencies, and as frequency increases, they can be resistive, then LR, then R, then RC, then L. I just described a closed box woofer, its DCR, lower side of resonance, resonance, upper side of resonance, and voice coil inductance in that order. And that is just the woofer without a crossover.
How about the crossover, midrange, crossover, tweeter?
C. If you test with a resistive load (I do), you get some reasonable results (and everyone gets to play on the same field conditions).
D. But, consider the set of curves of a triode, pentode, or beam power tube.
Draw a load line on a set of triode curves. If you know what to look for, you can see the 2nd harmonic distortion in the unequal spacing of curves along the load line (crowded at one end, and open spaced at the other end). You will see the same for the pentode and beam power tubes.
Now, draw an Ellipse around the load line. You will see the upper portion of the ellipse that is above the load line. It is in the region where the rp is lower and more in control of the load (more linear). You will see the lower portion of the ellipse that is below the load line. It is in the region where the rp is higher and less in control of the load (less linear).
That is what the loudspeaker impedance causes at so many frequencies (at Oh, so many frequencies).
This is what you get in Single Ended, driving real world loudspeakers.
Don't get me wrong, I listen to SE amps, and enjoy them. I do reduce the effect of the ellipse by designing the amp to have a higher damping factor. You can put the 8 Ohm loudspeaker (it's not 8 Ohms at most frequencies) on the 4 Ohm tap, and get similar results (but you may get less power).
Now, take 2 sets of triode curves, with load lines, and twist one over the other (180 Degrees). That is push pull.
As the ellipse of one direction is less controlled by one tube's rp, it is more controlled by the other tube's rp.
I do enjoy listening to Push Pull amps too.
As you can easily see on the sets of curves, testing amplifiers with resistive loads (load lines), and loudspeakers with elliptical loads will not necessarily give the same results.
Re. burning tweeter at full power, if you feed 20 W of 20 kHz sine wave into a tweeter rated 20 W, and the tweeter burns, it means that tweeter rating was bogus. And, there is no need for such sacrifices. The test can be done at the output level that is safe for speaker.
The point about speaker presenting a complex load to amplifier is well taken. But that's not the only concern in evaluating behavior of the amplifier-speaker system. I am more concerned with speaker sending garbage signal into amplifier's input through NFB loops.
The point about speaker presenting a complex load to amplifier is well taken. But that's not the only concern in evaluating behavior of the amplifier-speaker system. I am more concerned with speaker sending garbage signal into amplifier's input through NFB loops.
The speaker WILL send its garbage back into the negative feedback port of the amp, which is why the end result is a very clean signal, since it essentially happens in real time. What's interesting is what Bob Carver mentioned, where the listening room acoustics will color this garbage, and thereby potentially create some short term low level reverb effects... Not likely much, but it's interesting.
I would still want to see some data proving that such real time cancellation occurs. Theoretically it would require an ideal speaker with an absolutely rigid and weightless cone/voice coil. But real speakers are not like that. In a real speaker, there is a difference (or error) between waveforms of incoming signal and cone movement, which error cannot be corrected by electric feedback.
I would suggest that any testing in this area of interest has to utilise such modern measurement techniques.
It would be a poor test process to use continuous steady-state measurements, as any tests are aimed at comparing resistive load with reference speakers in a reference room (ie. your own listening speakers). Such a test would really need to focus on turn-on, steady state, and turn-off durations, with just enough cycles of operation where steady-state conditions have been set up for the speaker and room loading conditions to stabilise, before turning off the signal and watching decay. For a woofer range of frequencies that may require a few seconds of applied signal for room modes to stabilise, but for a tweeter the test period would be way less than a second.
The timing for a speaker load test should then suit the timing for a resistive load test, where disturbances of internal power supply rails need to be appreciated, as they would likely colour both test situations.
Then there is likely to be no need for resorting to pseudo electrical-only speaker loads, or some other deviation (such as mic boxing the speaker to reduce audible annoyance when testing).
Last edited:
a speaker can burn out with just a few watts of sine wave input, not a good thing to do...i do not as a matter of practice listen to sine waves.....
Few points;
As 6A3sUMMER said (post #43).
Regarding loads for initial testing, I simply dunk a 5 - 10W resistor in a container of water. (The bypass resistance of water is high enough not to influence the ohmic value of said resistor at loudspeaker values.)
Depending on what kind of loudspeaker used, yes - there does occur extra damped h.f. oscillation but mostly outside the audio spectrum. Those are however indicative of tendency to go unstable.
The effect is also dependant on the amplifier output resistance. (I hesitate to use the term "damping factor", because its simplistic definition has little meaning in the real world except for amplifiers of low D.F. ) For the effects of real loudspeaker impedance, one does not have to measure at full power. With due rerspect, any half-good loudspeaker should not exhibit serious impedance changes with power up to where it is overloaded. I have never found serious response differences close to the audio band.
Regarding fed-back room effects because of loudspeaker action as a microphone, again this should not significantly affect the frequency response. One must keep in mind that such a signal is quite attenuated; easy to check with a 'passive' driver as mic. close to the driven loudspeaker. Then, what is the output impedance of the amplifier? With a D.F. of 50 that impedance will be 0,16 ohm. The driver then sees 0,16/(0,16 + Rl), where Rl is at least the loudspeaker voice coil resistance. (For an 8 ohm loudspeaker this attenuation will be roughly a factor of 36 or more.) I have never tried measuring it but intuitively feel that this is not going to have significant influence.
(Caveat: There may be a problem here with very low D.F.s as in SET topology; not my favourite topology.)
As 6A3sUMMER said (post #43).
Regarding loads for initial testing, I simply dunk a 5 - 10W resistor in a container of water. (The bypass resistance of water is high enough not to influence the ohmic value of said resistor at loudspeaker values.)
Depending on what kind of loudspeaker used, yes - there does occur extra damped h.f. oscillation but mostly outside the audio spectrum. Those are however indicative of tendency to go unstable.
The effect is also dependant on the amplifier output resistance. (I hesitate to use the term "damping factor", because its simplistic definition has little meaning in the real world except for amplifiers of low D.F. ) For the effects of real loudspeaker impedance, one does not have to measure at full power. With due rerspect, any half-good loudspeaker should not exhibit serious impedance changes with power up to where it is overloaded. I have never found serious response differences close to the audio band.
Regarding fed-back room effects because of loudspeaker action as a microphone, again this should not significantly affect the frequency response. One must keep in mind that such a signal is quite attenuated; easy to check with a 'passive' driver as mic. close to the driven loudspeaker. Then, what is the output impedance of the amplifier? With a D.F. of 50 that impedance will be 0,16 ohm. The driver then sees 0,16/(0,16 + Rl), where Rl is at least the loudspeaker voice coil resistance. (For an 8 ohm loudspeaker this attenuation will be roughly a factor of 36 or more.) I have never tried measuring it but intuitively feel that this is not going to have significant influence.
(Caveat: There may be a problem here with very low D.F.s as in SET topology; not my favourite topology.)
I guess I was misunderstood. Room reflections picked up by speaker are indeed minor factor. What isn't minor though is speaker distortion signal directly picked up by voice coil and fed into amplifier's input.
But I guess as a matter of practice you test amplifier's performance using sine waves.
There's no need to blast full power into the speaker, and most speakers worth their price will not bust at few watts of sine signal. Also, there is no heed to listen to sine signal - place a speaker in soundproof enclosure, and put on ear protection.
Are you saying that any non-constant non-resistive behaviour of the voltage and current signal at the amplifier output is a distortion?
No doubt that time domain measurements would be the the most revealing.
It would be also interesting to compare distortion spectra with dummy load and real speaker.
No, it is not about reactive behavior of speaker. It is about distortion arising during conversion of electric signal into acoustic signal and back from acoustic into electric.
As an example, in a gedanken experiment an electric pulse is sent through amplifier into speaker. Voice coil is excited and sends radial acoustic wave through the cone, which is reflected back to voice coil by speaker's frame. Voice coil then picks up the reflected signal and sends it to amplifier input through NFB network. On a trace there will be two spikes separated by time of acoustic wave propagation. The second spike is distortion in question.
Would you group any non-instant signal as distortion, such as a speaker cone moving mass that generates a signal after the drive signal is stepped to zero?
What about room loading on speaker, such as a woofer building up a modal soundwave field over a couple of seconds, and then that pressure field decaying when signal to the woofer is removed.
What about a speaker box with vibrating walls.
What about room loading on speaker, such as a woofer building up a modal soundwave field over a couple of seconds, and then that pressure field decaying when signal to the woofer is removed.
What about a speaker box with vibrating walls.
Damping factor....how effectively the amplifier "shorts out" the stored energy in the speaker during the speaker's reverse cycle (cone moving back into the magnet). Tube amps depended on the low impedance output transformer secondary winding to do this. Solid state amplifiers depend on the return path through the low impedance output stage.
goes without saying....sine waves are convenient that is why it is alway used...
can you imaging a spec that says, 10 watts into an 8 ohm speaker when playing the Aida march, 8 watts when playing Norah Jones......
Although OT, you are right about speaker specs. They don't say "rated Norah Jones power at 1 W 1 m", but "rated music power" is the same BS.
An amplifier that is loaded with a loudspeaker's varying impedance may affect the amplifier's distortion, depending on the amplifier.
Consider a push pull amplifier with no negative feedback, and that uses very low rp triodes driving a moderate plate to plate primary impedance.
Now Consider a push pull amplifier that uses say, 14dB negative feedback, and that uses pentode-wired pentodes driving a moderate plate to plate primary impedance.
These 2 amplifiers may very well have the same damping factor number.
The damping factor of the triode amp is intrinsic; the damping factor of the pentode amp is almost completely due to negative feedback.
Norman Crowhurst was perhaps the first one to widely publish that negative feedback takes the low order distortion products and corrects them as best as possible, while instead creating smaller new higher order harmonics.
Remember, any signal that appears at the amp input, whether the original music signal from the RCA phone jack input to the tube grid, or negative feedback to the tube cathode gets distorted, whether it is the first, second, or third time around.
The loudspeaker elliptical load may have a different effect on the set of pentode curves than it does on the triode curves (the shapes of the curves are different). But it also has an effect on what the error signal is that has to be sent back as negative feedback to the pentode amp.
Consider a push pull amplifier with no negative feedback, and that uses very low rp triodes driving a moderate plate to plate primary impedance.
Now Consider a push pull amplifier that uses say, 14dB negative feedback, and that uses pentode-wired pentodes driving a moderate plate to plate primary impedance.
These 2 amplifiers may very well have the same damping factor number.
The damping factor of the triode amp is intrinsic; the damping factor of the pentode amp is almost completely due to negative feedback.
Norman Crowhurst was perhaps the first one to widely publish that negative feedback takes the low order distortion products and corrects them as best as possible, while instead creating smaller new higher order harmonics.
Remember, any signal that appears at the amp input, whether the original music signal from the RCA phone jack input to the tube grid, or negative feedback to the tube cathode gets distorted, whether it is the first, second, or third time around.
The loudspeaker elliptical load may have a different effect on the set of pentode curves than it does on the triode curves (the shapes of the curves are different). But it also has an effect on what the error signal is that has to be sent back as negative feedback to the pentode amp.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.