So if I understand correctly:
a) Indications are that voice coil temperature may be significantly modulated at much lower power levels than those where thermal compression normally becomes a factor;
b) This modulation is not rapid enough to single out transient peaks for compression, but it is rapid enough to compress the local signal envelope, which is probably even more audible; and
c) In general, that which minimizes thermal compression (high efficiency and large-diameter voice coils) also minimzes thermal modulation.
Please correct me if I've got it wrong.
And a question, if I may: At low signal levels, are there suspension system issues that could have a significant effect on dynamic contrast?
Thanks!
Duke
Last edited:
This is what I was saying except for a difference in opinion if the initial rise depends ONLY on the thermal capacity or the thermal capacity times the thermal resistance. Either way the rest of my discussion is correct.
I will have to check on the temperature rise because the thermal capacity is in degrees Joules per kelvin which means that the temperature change depends only on the heat injected into the material. Perhaps the "heat" here is the heat difference between the input heat and the output heat (that would bring the thermal resistance into the situation). Either way the instantaneous change in temperature is a function of the thermal capacity which is strongly related to the efficiency. Higher efficiency speakers have dramatically lower instantaneous temperature changes for a given SPL.
The static or long term temperature depends only on the thermal resistance. We both agree with that.
Hi Duke
Mostly right except:
a) The modulation would be linear in the power, but not the signal, so there would still me more modulation at higher SPLs than lower ones.
b) audibility of any of this is just a guess, but it does have the potential to explain what is often called "dynamics". people keep using a woofer as an example, but I don't see a woofer being the main culprit. I see it being the mid/tweeter, drivers acting above about 700 Hz. To me the ubiquitous 1" tweeter just sounds strained - no dynamics. What is that? Its certainly not frequency response.
c) in general that which makes for high efficiency makes for low thermal modulation. It would also tend to lower thermal compression but that is not guaranteed.
Mostly right except:
a) The modulation would be linear in the power, but not the signal, so there would still me more modulation at higher SPLs than lower ones.
b) audibility of any of this is just a guess, but it does have the potential to explain what is often called "dynamics". people keep using a woofer as an example, but I don't see a woofer being the main culprit. I see it being the mid/tweeter, drivers acting above about 700 Hz. To me the ubiquitous 1" tweeter just sounds strained - no dynamics. What is that? Its certainly not frequency response.
c) in general that which makes for high efficiency makes for low thermal modulation. It would also tend to lower thermal compression but that is not guaranteed.
Nice writeup Earl
Having done about everything under the sun in process control can attest that this modulation exists. Riding on the back of the thermal constant.
As to "large sound" I don't think it has anything todo with sub bass level or depth. low midbass I think is where we enter this large fullbodied domain spectrum. Perhaps has something todo with balance of bass wavelength distribution e.g. larger woofer sound better than smaller so to speak. But if we limit the bass to a point depending on lets say driver size/room, small can sound large within it's limits. Exceed those limits and it falls back on it's face to small strained sound. Similar to tuning a box with a Q 0.707 vs a low one that overly extends it. It's like without knowingly doing this the electrical wave (E) in a waveguide (aka RF Transmission line) is akin to what is typical of speakers normally built, but the missing element of the large sound is the magnetic wave (H) component.
Anywho, this is something I plan on testing late next week sometime (earliest). The final volley of drivers are on the way and a dozen should be enough to run a few experiments to disprove my curiosity 😉
HES do sound clearer, more dynamic for the above reasons but also because of the reduced noise floor innate to this type of driver. Lynn Olson was talking about this over in Beyond the Ariel a bit back and his thoughts are mine on the subject.
Having done about everything under the sun in process control can attest that this modulation exists. Riding on the back of the thermal constant.
As to "large sound" I don't think it has anything todo with sub bass level or depth. low midbass I think is where we enter this large fullbodied domain spectrum. Perhaps has something todo with balance of bass wavelength distribution e.g. larger woofer sound better than smaller so to speak. But if we limit the bass to a point depending on lets say driver size/room, small can sound large within it's limits. Exceed those limits and it falls back on it's face to small strained sound. Similar to tuning a box with a Q 0.707 vs a low one that overly extends it. It's like without knowingly doing this the electrical wave (E) in a waveguide (aka RF Transmission line) is akin to what is typical of speakers normally built, but the missing element of the large sound is the magnetic wave (H) component.
Anywho, this is something I plan on testing late next week sometime (earliest). The final volley of drivers are on the way and a dozen should be enough to run a few experiments to disprove my curiosity 😉
HES do sound clearer, more dynamic for the above reasons but also because of the reduced noise floor innate to this type of driver. Lynn Olson was talking about this over in Beyond the Ariel a bit back and his thoughts are mine on the subject.
Thanks for your reply, Earl.
I think it depends on the system, in particular the thermal compression or modulation behavior of the woofer vs that of the tweeter.
I have heard a lot of small speakers that sounded dull at low levels; at their best at medium levels; and bright at high levels. This would be consistent with the woofer experiencing more thermal compression than the tweeter, and the designer "voicing" the speaker to sound correctly balanced at medium levels. BUT obviously I do not know that this is what is really happening - only that it seems to me like a plausible explanation. And of course the sample size of my listening panel is on the small side.
Say we have a 6.5" two-way speaker with a 90 dB, 60 watt woofer crossed over at 2.5 kHz to a 93 dB, 100 watt dome tweeter. Power handling rated per IEC 268-5. This is not so far fetched - both drivers are manufactured by the same company, SB Acoustics (they don't make a 6.5" woofer rated at more than 60 watts, and only one dome tweeter rated at less than 100 watts).
If we divide the spectrum at 2.5 kHz, approximately 20% of the power goes to the high frequency section, according to Eminence:
High Frequency Driver Power Handling | Eminence Speaker
Our system power handling will be 75 watts, and if we pad back our tweeter by 3 dB, at 75 watts our tweeter will be seeing about 7.5 watts (take 20% of 75 watts, then pad it back by 3 dB). Assume we hit it with 75 watts for long enough for the temperatures to stabilize (just so we know we'll be seeing textbook thermal compression). So that's 60 watts into the woofer, 7.5 watts into the tweeter, and 7.5 watts into the tweeter's resistor. I don't think it would be unreasonable to expect the woofer to exhibit 3 dB or more of thermal compression, while the tweeter is exhibiting less than 1 dB of thermal compression. If we increase the padding on the tweeter so that it's not so screaming bright at high SPLs, we end up with the situation I claim to have observed: Dull at low levels, at its best at medium levels, and somewhat bright at high levels.
If anything, I think the thermal modulation of the envelope that you described would make the tonal balance vary accordingly over much smaller time intervals.
Of course the cause for the change in tonal balace with level that I observed may be something else entirely, or my observation may be biased or otherwise incorrect.
I think it depends on the system, in particular the thermal compression or modulation behavior of the woofer vs that of the tweeter.
I have heard a lot of small speakers that sounded dull at low levels; at their best at medium levels; and bright at high levels. This would be consistent with the woofer experiencing more thermal compression than the tweeter, and the designer "voicing" the speaker to sound correctly balanced at medium levels. BUT obviously I do not know that this is what is really happening - only that it seems to me like a plausible explanation. And of course the sample size of my listening panel is on the small side.
Say we have a 6.5" two-way speaker with a 90 dB, 60 watt woofer crossed over at 2.5 kHz to a 93 dB, 100 watt dome tweeter. Power handling rated per IEC 268-5. This is not so far fetched - both drivers are manufactured by the same company, SB Acoustics (they don't make a 6.5" woofer rated at more than 60 watts, and only one dome tweeter rated at less than 100 watts).
If we divide the spectrum at 2.5 kHz, approximately 20% of the power goes to the high frequency section, according to Eminence:
High Frequency Driver Power Handling | Eminence Speaker
Our system power handling will be 75 watts, and if we pad back our tweeter by 3 dB, at 75 watts our tweeter will be seeing about 7.5 watts (take 20% of 75 watts, then pad it back by 3 dB). Assume we hit it with 75 watts for long enough for the temperatures to stabilize (just so we know we'll be seeing textbook thermal compression). So that's 60 watts into the woofer, 7.5 watts into the tweeter, and 7.5 watts into the tweeter's resistor. I don't think it would be unreasonable to expect the woofer to exhibit 3 dB or more of thermal compression, while the tweeter is exhibiting less than 1 dB of thermal compression. If we increase the padding on the tweeter so that it's not so screaming bright at high SPLs, we end up with the situation I claim to have observed: Dull at low levels, at its best at medium levels, and somewhat bright at high levels.
If anything, I think the thermal modulation of the envelope that you described would make the tonal balance vary accordingly over much smaller time intervals.
Of course the cause for the change in tonal balace with level that I observed may be something else entirely, or my observation may be biased or otherwise incorrect.
Last edited:
Thanks Rob,
That is an interesting graph and it illustrates my descriptions of how VC temperature varies. The upper curves appear to be two blasts of noise or music input. The lower curve shows the exponential rise and fall of voice coil temperature in response. There is clearly a time constant associated with temperature change and it has the look of a heavily low pass filtered square wave, much as a slow charging RC filter would show. The time constant could be directly pulled from the graph by looking at the rate of change of temperature. Since the music or noise was on off on, there are several cycles of temperature swing and an upwards drift as the magnet structure slowly warms (at least 2 time constants are involved).
The practical effect of all this is that the DCR of the driver is changing in proportion to VC temp (0.38% / Degree C rise for copper wire). As VCR changes sensitivity drops exactly as if we added a few ohms of parasitic resistance in series with woofer and tweeter. We used to measure the resultant "power compression" curves at JBL (1, 10, and 100 Watts were a popular choice) although they were misleading as the sweep rate had much to do with the compression seen. If power compression looked like too much for advertising purposes, the we just ran the curves a little faster.
Is it audible? Again, it is a slowly varying frequency response error. It requires a steady application of power input (the thermal masses provide an integrating element) and you have to be running the voice coils damn hot to get a few dB of loss. Much as Wayne described, I first noticed the phenomenon when doing pink noise testing on tweeters. If you switch on pink noise you see output SPL jump to a level and then drift slowly downwards a few dB.
Probably a real enough factor in PA applications but I wouldn't lose sleep over it at home.
David
Last edited:
As to the modulation of the music envelope, take another look at the woofer curve. From the initial rise time it looks to me like a time constant in the 10 to 20 second range. It is taking 100s of seconds to approach a stable temperature when the 50 Watt blast is applied.
Anybody that thinks music is being quickly modulated is ignoring the data.
Anybody that thinks music is being quickly modulated is ignoring the data.
Dave
As I said look at a tweeter, not a woofer, the time constants are much shorter. At any rate it is the initial slope that determines the modulation NOT the shape of the curve. I never disagreed that the temperature would have an exponential asymptotic behavior over a long time period. My argument depends only on the initial slope of the curve.
I have measured tweeter thermal changes in less than a second. This is very much in the range of musical dynamics.
(5:20 am - Dave! Really! Insomnia much?)
As I said look at a tweeter, not a woofer, the time constants are much shorter. At any rate it is the initial slope that determines the modulation NOT the shape of the curve. I never disagreed that the temperature would have an exponential asymptotic behavior over a long time period. My argument depends only on the initial slope of the curve.
I have measured tweeter thermal changes in less than a second. This is very much in the range of musical dynamics.
(5:20 am - Dave! Really! Insomnia much?)
Last edited:
HI Duke
a 2.5 kHz crossover is rather high, but certainly does exist. I think that the issue would become more pronounced as one tried to lower the crossover. And remember than the thermal capacity of a tweeter is far far lower than the woofer is. Far more that the 5:1 in the power that you show. So I still think that the tweeter is a more likely culprit, but this, of course would be design specific.
a 2.5 kHz crossover is rather high, but certainly does exist. I think that the issue would become more pronounced as one tried to lower the crossover. And remember than the thermal capacity of a tweeter is far far lower than the woofer is. Far more that the 5:1 in the power that you show. So I still think that the tweeter is a more likely culprit, but this, of course would be design specific.
Combination of the heat and a new puppy in the house had me awake early. What better to do than to write ranting letters to audio forums!
Granted that tweeters respond quicker, but they still have time constants in multiple seconds.
Before you hang your hat on this new distortion discovery you really should do some measurements to see the magnitude of the effect. The injected current approach that I mentioned is quite easy. Capacitor couple the driver of interest and then connect a current source via an inductor. You want to filter signal out of the voltage that you measure but not filter to the point where the measuring time constant is approaching the thermal time constant.
Put a DMM ohm meter across the driver and inductor (you want to filter out audio) and watch the voltage vary as the coil heats. As I mentioned before, most of the time the signal average will be fairly low and the coil resistance will stay low. If you really crank it then it will start to correspond to input (this is only a comment on the logarithmic relation between power and perceived loudness).
Remember those light bulb protection devices sometimes put in series with drivers? This is really just a milder version of that.
Note that we had an interesting observation with the KM1 tweeter. The coil and its adhesives were good to a higher temperature than ferrofluid was rated for. In other words, if we used ferrofluid we would have to set the protection to a lower temperature than if we didn't use it. On the other hand, since the thermal conductivity was that much greater with ferrofluid, we actually found a near match between the power required to get the bare coil to the danger point and the power to get the ferrofluid to its danger point. We ended up using the ferrofluid because the temps would stay down and compression would be reduced, plus we figured the ferrofluid limit could be exceeded occaisionally, but the bare VC limit must not be exceeded!
Best Regards,
David
Last edited:
HI Dave
Air conditioning!? Or don't they have that in Canada?🙂
As I have said many times, audibility of this issue is unknown, so I am certainly not "hanging my hat" on anything other than to say this is an option.
And I have done some studies, again, as I said, and found that this effect is detectable. But subjectively relating it to what we hear is virtually impossible with real device because there are so many confounding variables. In other words unless one can control the effect it is impossible to study it effectively.
I created a math model of it and the intent was to simulate the effect and do a study with Lidia at her school. This is the aspect that has not happened yet. Lots of things get in the way.
It is not necessary to inject anything into the signal path, one can determine all that you need by simply monitoring of the voltage and current. The DCR of the coil is easily determined from this data (even in real time if that is desired) and from that one can get the temperature if desired. I believe that this technique is far better than any other since I can get the DCR on very short intervals, almost continuously without any need to low pass filter any thing.
So, in conclusion, theoretically thermal modulation has to exist and it is either audible or it is not at "typical levels". Neither result would surprise me. But if it is not audible then it would be back to the lab to determine what it is that so many people detect as "dynamics". Why do some speakers have it while others don't? It isn't nonlinear distortion because we have ruled that one out (for the most part. There could be some effects of low orders at high levels on "dynamics" at higher power, but the data to support that is scant to nonexistent.)
Air conditioning!? Or don't they have that in Canada?🙂
As I have said many times, audibility of this issue is unknown, so I am certainly not "hanging my hat" on anything other than to say this is an option.
And I have done some studies, again, as I said, and found that this effect is detectable. But subjectively relating it to what we hear is virtually impossible with real device because there are so many confounding variables. In other words unless one can control the effect it is impossible to study it effectively.
I created a math model of it and the intent was to simulate the effect and do a study with Lidia at her school. This is the aspect that has not happened yet. Lots of things get in the way.
It is not necessary to inject anything into the signal path, one can determine all that you need by simply monitoring of the voltage and current. The DCR of the coil is easily determined from this data (even in real time if that is desired) and from that one can get the temperature if desired. I believe that this technique is far better than any other since I can get the DCR on very short intervals, almost continuously without any need to low pass filter any thing.
So, in conclusion, theoretically thermal modulation has to exist and it is either audible or it is not at "typical levels". Neither result would surprise me. But if it is not audible then it would be back to the lab to determine what it is that so many people detect as "dynamics". Why do some speakers have it while others don't? It isn't nonlinear distortion because we have ruled that one out (for the most part. There could be some effects of low orders at high levels on "dynamics" at higher power, but the data to support that is scant to nonexistent.)
Last edited:
I hate to interject a discussion between two engineers 🙂
Tom D. had talked about the modulation transfer function (MTF), which can be measured through ARTA as a window into dynamics. His claim is that it depends on the directional properties of the source, and how the reverberant envelope interacts with the direct sound. If we are able to clearly follow the direct sound, it is perceived as more dynamic. Example given by Keyser was drums being struck in a bare room versus a treated studio. In the bare room, the drums sound loud and harsh. In the treated room, they sound dynamic.
What about excursion and the non-linear properties of the suspension and cone? Less excursion always sound more dynamic. Bigger woofers, compression drivers. The coil stays in the gap more and the non-linearities of the suspension are not exposed as much?
Tom D. had talked about the modulation transfer function (MTF), which can be measured through ARTA as a window into dynamics. His claim is that it depends on the directional properties of the source, and how the reverberant envelope interacts with the direct sound. If we are able to clearly follow the direct sound, it is perceived as more dynamic. Example given by Keyser was drums being struck in a bare room versus a treated studio. In the bare room, the drums sound loud and harsh. In the treated room, they sound dynamic.
What about excursion and the non-linear properties of the suspension and cone? Less excursion always sound more dynamic. Bigger woofers, compression drivers. The coil stays in the gap more and the non-linearities of the suspension are not exposed as much?
If this were true then listening in an anechoic chamber would be the ultimate in "dynamics". I think that there is more to it than that.
My studies into nonlinear distortions indicted that loudspeakers would not have the most audible types of nonlinearities and that the types of distortion that they would have, almost exclusively low order, were not that audible. But the low orders do affect the amplitude of the signal and may be the source of "modulation" even if the changes to the harmonic structure were not audible. This has not been looked at as far as I know.
My studies into nonlinear distortions indicted that loudspeakers would not have the most audible types of nonlinearities and that the types of distortion that they would have, almost exclusively low order, were not that audible. But the low orders do affect the amplitude of the signal and may be the source of "modulation" even if the changes to the harmonic structure were not audible. This has not been looked at as far as I know.
I recorded a drum track once in a heavily damped radio studio. It was a large room with lots of absorbtion. The drums sounded really strange in there. Like the sound was coming just from the drum heads with no sense of "body" or "tone" from the wooden drum itself. I had to use headphones with the drum sound with added reverb just to be able to play the track.
Interesting. I guess it depends on what can be called a "treated" room. I can imagine the unpleasantness of a highly reverberant room. In fact, it's not comfortable being in a such a room. At the same time, anechoic chambers are not comfortable either, and your experience of a heavily absorbent room is similar.
I just like the idea of how Tom's theory explains the perception of dynamics.
I just like the idea of how Tom's theory explains the perception of dynamics.
Earlier I mentioned some tests I did that showed audible SPL change at about 1dB/second at high power levels. The total compression stopped at about 3dB, but it happened pretty fast, in just a few seconds. It is basically in the timeframe you are talking - in the 10 second range - but I think it still tends to show this effect occuring reasonably fast, especially since the woofer is the component with the largest thermal mass. A smaller driver would likely experience this faster, and with less power input.
Originally Posted by speaker dave
What phenomonon are you referring to? Is this just compression due to coil heating? Thats a slow moving effect tied to the time constants of the voice coils.
David S.
Exceeding Xmax or suspension limits results in an instantaneous reduction of the waveform magnitude.
Once a driver reaches Xmax, additional input won't result in a linear increase in output, regardless of the voice coil temperature.
Small tweeters in two way systems often have only a single capacitor for a crossover, high dynamic midrange can require more excursion than the suspension or the motor may be able to deliver.
Now, to bring this back to the question of uniform directivity importance, if the tweeter does reach a midrange dynamic limit at a lower SPL than the woofer, directivity would change with SPL.
Directivity changing with dynamics can't be good.
What phenomonon are you referring to? Is this just compression due to coil heating? Thats a slow moving effect tied to the time constants of the voice coils.
David S.
"Something else" being limited excursion in small drivers seems a more probable explanation of limited dynamics than modulation caused by "immediate" voice coil heating.
Exceeding Xmax or suspension limits results in an instantaneous reduction of the waveform magnitude.
Once a driver reaches Xmax, additional input won't result in a linear increase in output, regardless of the voice coil temperature.
Small tweeters in two way systems often have only a single capacitor for a crossover, high dynamic midrange can require more excursion than the suspension or the motor may be able to deliver.
Now, to bring this back to the question of uniform directivity importance, if the tweeter does reach a midrange dynamic limit at a lower SPL than the woofer, directivity would change with SPL.
Directivity changing with dynamics can't be good.
Thanks Earl, I had no educated "feel' for the difference in thermal capacity between the tweeter and woofer voice coils, and thought that the 5:1 would be in the ballpark.
Last edited:
- Status
- Not open for further replies.