CeramicMan said:
I see what you're saying. Do you know of anywhere that has some actual measurements on this? It would be interesting to see exactly how much rise you get over a very small time window. This would happen regardless of the XO being active or passive, so I'm not sure what we are supposed to do with this info. Using only PA drivers may be teh kneejerk response but I'm not sure that any given PA driver is better than a "audiophile" driver.
I'd have to agree with ScottG and DSP-geek. I mean the smaller window of time we're talking about (it is transient after all) the less important I think it becomes... I mean who is going to notice a 2dB drop in SPL when it only lasts a ms?
CeramicMan said:
Take a lot at this example of Dynaudio's toneburst tests for their d52af midrange. This clearly demonstrates no compression of a 3kHz toneburst at 120dB! This is a relatively small (54mm diameter, 7mm high) voicecoil. They have similar plots for their tweeters and woofers. Does this not disprove the existence of transient compression due to voice-coil heating?
An externally hosted image should be here but it was not working when we last tested it.
http://www.gattiweb.com/images/dynaudio/d52af_data2.pdf
http://www.gattiweb.com/dynaudio
A four cycle tone burst is hardly comprehensive. All we can see is a fuzzy squiggle and it's hard to tell anything about it.
If I wanted to measure "dynamic compression",
1) I'd use a continuous tone that's long enough to clearly show the time constant and the steady-state response. A 1.33 millisecond tone burst just isn't long enough for that.
2) I'd then window the steady-state data and do an FFT to see the distortion level.
If I wanted to measure "dynamic compression",
1) I'd use a continuous tone that's long enough to clearly show the time constant and the steady-state response. A 1.33 millisecond tone burst just isn't long enough for that.
2) I'd then window the steady-state data and do an FFT to see the distortion level.
augerpro said:
I'm not sure. I think that normally it would just be lumped together with overall THD and IM measurements for everything including motor and spider non-linearities.
If there's a 2dB drop in as little as 1ms, the driver is probably in serious trouble. Eg: a "weak link" in the coil starts undergoing thermal run-away, whereby its resistance goes up so that it hogs more and more of the available power.
Hi,
Good reading :
http://stereophile.com/reference/1106hot/
Maybe I've not read carefully all the posts, but data about copper or alumium thermal resistivity changes with temperature and about voice coil thermal time constants could help. The heating process can then be modelled.
To study the thermal behaviour in real conditions, you can use a dual voice-coil driver, one coil used as the standard input and the other one as a thermal sensor, it has a surprising fast response.
Good reading :
http://stereophile.com/reference/1106hot/
Maybe I've not read carefully all the posts, but data about copper or alumium thermal resistivity changes with temperature and about voice coil thermal time constants could help. The heating process can then be modelled.
To study the thermal behaviour in real conditions, you can use a dual voice-coil driver, one coil used as the standard input and the other one as a thermal sensor, it has a surprising fast response.
"This clearly demonstrates no compression of a 3kHz toneburst at 120dB! "
Hmmmm......did you notice the start of the toneburst output ?
I think that it shows the converse of this. At that start the voice coil is heating , IIRC stereophile paper also have sudden jump of voice coil resistance (look at the woofer impedance over time )at the start of the test.
I also have the same opinion with ceramicman, the transient heating is ignored and people see the data smoothed over time.
that DynaAudio paper also said :
"[This]compression effect is under rated or ignored very often"
Hartono
Hmmmm......did you notice the start of the toneburst output ?
I think that it shows the converse of this. At that start the voice coil is heating , IIRC stereophile paper also have sudden jump of voice coil resistance (look at the woofer impedance over time )at the start of the test.
I also have the same opinion with ceramicman, the transient heating is ignored and people see the data smoothed over time.
that DynaAudio paper also said :
"[This]compression effect is under rated or ignored very often"
Hartono
Hi Hartono,
---did you notice the start of the toneburst output ?
I think that it shows the converse of this. At that start the voice coil is heating , IIRC stereophile paper also have sudden jump of voice coil resistance (look at the woofer impedance over time )at the start of the test.---
Howard's text explains that the glitch (the sudden jump of the voice coil resistance) is due to an artefact of the measurement process. It does not correspond to a burst of the voice coil temperature which will otherwise be very different from widely observed thermal phenomenons.
---the cooling process is not that linear, the voice coil moves variably and it's not heatsink.---
Here is an interesting text about voice coil heating,
http://www.sfa.asso.fr/fr/gea/Lesage.pdf
However, it is in french and I do not know if it has been translated in english. Towards the end of the document (two pages before the last one), there is a diagram showing the heating of the voice coil. Is saw nothing on it which is not purely conventionnal from what is already known about the temperature effects on an object.
Voice coil temperature can be modelled as if it was static and this can be checked by putting the voice coil outside of its usual magnetic environment so it does not move. When the same voice coil is in the magnet gap and is moving, its heat dissipation can only been enhanced, due to air flow even if it is not a linear process, and, on the long term, is assisted by the metallic mass of the magnet which behaves as a heatsink.
---did you notice the start of the toneburst output ?
I think that it shows the converse of this. At that start the voice coil is heating , IIRC stereophile paper also have sudden jump of voice coil resistance (look at the woofer impedance over time )at the start of the test.---
Howard's text explains that the glitch (the sudden jump of the voice coil resistance) is due to an artefact of the measurement process. It does not correspond to a burst of the voice coil temperature which will otherwise be very different from widely observed thermal phenomenons.
---the cooling process is not that linear, the voice coil moves variably and it's not heatsink.---
Here is an interesting text about voice coil heating,
http://www.sfa.asso.fr/fr/gea/Lesage.pdf
However, it is in french and I do not know if it has been translated in english. Towards the end of the document (two pages before the last one), there is a diagram showing the heating of the voice coil. Is saw nothing on it which is not purely conventionnal from what is already known about the temperature effects on an object.
Voice coil temperature can be modelled as if it was static and this can be checked by putting the voice coil outside of its usual magnetic environment so it does not move. When the same voice coil is in the magnet gap and is moving, its heat dissipation can only been enhanced, due to air flow even if it is not a linear process, and, on the long term, is assisted by the metallic mass of the magnet which behaves as a heatsink.
Hi Forr
Thanks for correcting me on that point
, I must've misinterpret it,
however the next sentence of the Stereophile article says :
This glitch hides the initial exponential rise of voice-coil resistance, which, just before the end of the test track, at around 340 seconds, reaches about 4.2 ohms, an increase of 8%
still 8% is significant
But I'm still confused by their test setup, how can a transient glitch happen with 1 hz low pass filter ? maybe it isn't a glitch. and how come the measurement on the tweeter didn't have it ? I assume that the initial heating/cooling in the woofer to be more significant than the tweeter considering the tweeter receive smaller power throughout the testing.
"When the same voice coil is in the magnet gap and is moving, its heat dissipation can only been enhanced due to air flow even if it is not a linear process"
Yes, true, the movement will enhance the dissipation, but it's nonlinear , so there's no point in modelling the linear heating if the cooling is non-linear , do you agree ?
I think the real problem is that the non-linear cooling cause resistance modulation of the voice coil resistance, would you buy or build an amp that modulate it's output by 0.3 dB ??
Best regards,
Hartono
Thanks for correcting me on that point
, I must've misinterpret it, however the next sentence of the Stereophile article says :
This glitch hides the initial exponential rise of voice-coil resistance, which, just before the end of the test track, at around 340 seconds, reaches about 4.2 ohms, an increase of 8%
still 8% is significant
But I'm still confused by their test setup, how can a transient glitch happen with 1 hz low pass filter ? maybe it isn't a glitch. and how come the measurement on the tweeter didn't have it ? I assume that the initial heating/cooling in the woofer to be more significant than the tweeter considering the tweeter receive smaller power throughout the testing.
"When the same voice coil is in the magnet gap and is moving, its heat dissipation can only been enhanced due to air flow even if it is not a linear process"
Yes, true, the movement will enhance the dissipation, but it's nonlinear , so there's no point in modelling the linear heating if the cooling is non-linear , do you agree ?
I think the real problem is that the non-linear cooling cause resistance modulation of the voice coil resistance, would you buy or build an amp that modulate it's output by 0.3 dB ??
Best regards,
Hartono
Hi Svante,
I had a dual voice coil coming from a broken driver so I did some experience with it and I was suprised how using one of the voice coil as a thermal sensor had a fast response. So I think it could be used to monitor temperature when investigating voice coil heating in real time, in real life conditions, on real signals and all this without a calculator.
Hi Hartono,
Neville Thiele says that all his BR calculations imply that the Qt of the driver will not vary more than 10%. I think a variation of DC resistance of 8% is within this rule.
I did not anlayse Howard's setup, it seems complicated. Using an experimental dual voice-coil speaker would have been more simple. However you can't have a glitch like that shown in a heating process.
---Yes, true, the movement will enhance the dissipation, but it's nonlinear , so there's no point in modelling the linear heating if the cooling is non-linear , do you agree ?---
Not linear, but fairly exponential, I think. We do not need to precisely quantify the heating but looking for an evaluation to appreciate in which circumstances it could become a factor altering the sound quality, this seems to happen in professionnal applications, but not so at home.
Thermal resistance and thermal time constant are very rarely given by drivers makers. Here some collected from old Kef :
T33A SP1074 (25 mm dome tweeter) --> TTC = 3.5 s
B110 (5' bass medium driver used in the BBC LS3/5A)
B110 SP1057 --> TR = 7 °C/W, TTC = 4s
BD139 (elliptical bass driver famous around 1970)
BD139 SP1037 --> R = 4.5 °C/W, TTC = 16 s
B200 (8' bass medium driver)
B200 SP1014 --> TR = 7° C/W, TTC = 4 s
B200 SP1022 --> TR = 7° C/W, TTC = 4 s
B200 SP1039 --> TR = 7 °C/W, TTC = 4.5 s
B200 SP1054 --> TR = 6 °C/W, TTC = 12.5 s
B200 SP1063 --> TR = 6.2 °C/W TTC = 11 s
B200G SP1075 --> TR = 6.2 °C/W TTC = 11 s
B300 (12' bass driver)
B300 SP1071 --> TTC = 15 s
---I think the real problem is that the non-linear cooling cause resistance modulation of the voice coil resistance, would you buy or build an amp that modulate it's output by 0.3 dB ?---
Due to the involved time constants, this is not modulation but compression, and a compression is not a distortion. 0.3 dB is the threshold for a noticeable difference of level in sinewaves, it's about 0.5 to 1.0 dB on music.
So I think that Howard's conclusion is right. There many other issues to concentrate on than thermal compression in voice coil in standard domestic conditions.
I had a dual voice coil coming from a broken driver so I did some experience with it and I was suprised how using one of the voice coil as a thermal sensor had a fast response. So I think it could be used to monitor temperature when investigating voice coil heating in real time, in real life conditions, on real signals and all this without a calculator.
Hi Hartono,
Neville Thiele says that all his BR calculations imply that the Qt of the driver will not vary more than 10%. I think a variation of DC resistance of 8% is within this rule.
I did not anlayse Howard's setup, it seems complicated. Using an experimental dual voice-coil speaker would have been more simple. However you can't have a glitch like that shown in a heating process.
---Yes, true, the movement will enhance the dissipation, but it's nonlinear , so there's no point in modelling the linear heating if the cooling is non-linear , do you agree ?---
Not linear, but fairly exponential, I think. We do not need to precisely quantify the heating but looking for an evaluation to appreciate in which circumstances it could become a factor altering the sound quality, this seems to happen in professionnal applications, but not so at home.
Thermal resistance and thermal time constant are very rarely given by drivers makers. Here some collected from old Kef :
T33A SP1074 (25 mm dome tweeter) --> TTC = 3.5 s
B110 (5' bass medium driver used in the BBC LS3/5A)
B110 SP1057 --> TR = 7 °C/W, TTC = 4s
BD139 (elliptical bass driver famous around 1970)
BD139 SP1037 --> R = 4.5 °C/W, TTC = 16 s
B200 (8' bass medium driver)
B200 SP1014 --> TR = 7° C/W, TTC = 4 s
B200 SP1022 --> TR = 7° C/W, TTC = 4 s
B200 SP1039 --> TR = 7 °C/W, TTC = 4.5 s
B200 SP1054 --> TR = 6 °C/W, TTC = 12.5 s
B200 SP1063 --> TR = 6.2 °C/W TTC = 11 s
B200G SP1075 --> TR = 6.2 °C/W TTC = 11 s
B300 (12' bass driver)
B300 SP1071 --> TTC = 15 s
---I think the real problem is that the non-linear cooling cause resistance modulation of the voice coil resistance, would you buy or build an amp that modulate it's output by 0.3 dB ?---
Due to the involved time constants, this is not modulation but compression, and a compression is not a distortion. 0.3 dB is the threshold for a noticeable difference of level in sinewaves, it's about 0.5 to 1.0 dB on music.
So I think that Howard's conclusion is right. There many other issues to concentrate on than thermal compression in voice coil in standard domestic conditions.
Re: Voice coil heating and the performance of passive XO, plus my funny experience at HTG
Yup, no personal attacks in those quotes. You're just a poor misunderstood lad seeking the truth.
Jay, I really tried to be civil and answer your questions with graphs that showed what was going on under realistic conditions. You, on the other hand presented no proof of anything, just a quote from a web page taken out of context. When I suggested you do some experiments yourself to see the effects, you replied that you couldn't be bothered to waste your time on that. You presented no scientific rebuttal at all to my points but claimed that somehow the graphs I presented weren't quote up to your standards. That's when I'd had enough and decided you are a troll. Given the tone of your opening salvo here, that opinion hasn't changed.
Dennis H
Jay_WJ said:
Yup, no personal attacks in those quotes. You're just a poor misunderstood lad seeking the truth.
Jay, I really tried to be civil and answer your questions with graphs that showed what was going on under realistic conditions. You, on the other hand presented no proof of anything, just a quote from a web page taken out of context. When I suggested you do some experiments yourself to see the effects, you replied that you couldn't be bothered to waste your time on that. You presented no scientific rebuttal at all to my points but claimed that somehow the graphs I presented weren't quote up to your standards. That's when I'd had enough and decided you are a troll. Given the tone of your opening salvo here, that opinion hasn't changed.
Dennis H
forr said:
If you are interested in the time constants specifically, I'd say it isactually better to use a single-coil driver. If you use the secondary coil, there will be a time constant between the coils that is hard to tell how large it is. Using the copper in the wire that the current runs through ensures that it is the actual temperature of the coil that is measured. It might be that the difference is not that big, but this way you know.
OT: I have heard a level difference of 0.3 dB in a blind test (99%), with a music signal. I have read reports of less than 0.1 dB too, which is something to think of when designing blind tests.
Here is the effect on the frequency response of heating the voice coil to 100 °C:
An externally hosted image should be here but it was not working when we last tested it.
BTW, Jon's last post in the HTG thread answers all your technical questions very specifically. I think you can see that there's a reason why many people respect his opinion. If you want to call that "adoration" that's your option and tells us more about you than about him or us. If you'll recall, this whole thing started because, in another thread, you quite vocally claimed your design is superior to Jon's because it uses fewer components, even though you have never built nor listened to yours, just modelled it in software. Lots of people have built Jon's and like the sound. 'Nuff said.
Dennis H
Dennis H
Hi Forr,
"I was suprised how using one of the voice coil as a thermal sensor had a fast response"
I think this says it all, I mean it's pretty fast to heat up a coil, and it doesn't need to wait a few seconds, mind you that the insulation coating of the coil still hinder this measurement.
"I think a variation of DC resistance of 8% is within this rule."
Which is significant.
"Not linear, but fairly exponential, I think"
Exponential maybe, I don't know, make that nonlinear exponential cooling,
yes cooling not heating, actually both.
"Due to the involved time constants, this is not modulation but compression, and a compression is not a distortion."
If the cooling rate is not exactly the same with heating(which is impossible), then it will modulate the resistance.
If you think there's no instantaneous heating in metal try touching arc welder tip. I thought you already try heating a voice coil and know it for yourself, it doesn't take a few seconds as you might have believed to heat up a coil.
If you looked carefully at the stereophile graph, the resistance is looks to be modulated by the music.
"So I think that Howard's conclusion is right. There many other issues to concentrate on than thermal compression in voice coil in standard domestic conditions"
Thank you for your suggestion, but I do want to reduce it.
Do you want to use resistor made from copper ?, it will only compress when it heats up, not distort, right? and hopefully it will do it slowly, with perfectly exponential cooling, at the same rate as the heating.
Hartono
"I was suprised how using one of the voice coil as a thermal sensor had a fast response"
I think this says it all, I mean it's pretty fast to heat up a coil, and it doesn't need to wait a few seconds, mind you that the insulation coating of the coil still hinder this measurement.
"I think a variation of DC resistance of 8% is within this rule."
Which is significant.
"Not linear, but fairly exponential, I think"
Exponential maybe, I don't know, make that nonlinear exponential cooling,
yes cooling not heating, actually both.
"Due to the involved time constants, this is not modulation but compression, and a compression is not a distortion."
If the cooling rate is not exactly the same with heating(which is impossible), then it will modulate the resistance.
If you think there's no instantaneous heating in metal try touching arc welder tip. I thought you already try heating a voice coil and know it for yourself, it doesn't take a few seconds as you might have believed to heat up a coil.
If you looked carefully at the stereophile graph, the resistance is looks to be modulated by the music.
"So I think that Howard's conclusion is right. There many other issues to concentrate on than thermal compression in voice coil in standard domestic conditions"
Thank you for your suggestion, but I do want to reduce it.
Do you want to use resistor made from copper ?, it will only compress when it heats up, not distort, right? and hopefully it will do it slowly, with perfectly exponential cooling, at the same rate as the heating.
Hartono
Thanks, guys, for good discussion. I'm learning a lot!
Dennis:
How could I do an experiment when I have no measurement setup? Go back to the thread and read carefully. I didn't particularly pointed you when I gave the "... not up to my standard" comment. I just assessed my overall knowledge I gained from the discussion at that time. I needed to know more. Look at discussion here and also read Jon's explanation. The matter is not so simple. I just wanted to have more details and consider other factors. That's all. I'm not an expert at all on these things. As I said, my intention was not to debate. So no wonder that I couldn't give any scientific, knowledgeable details there. I don't understand why you took my statement personal and overreacted to it. And I did NOT assert that my design IS superior to Jon's. I just raised a hypothetical question to see people's responses and learn from there. Please, don't add irrelevant comments any more here. Let's have people discuss the subject.
Thanks,
Jay
P.S. I have no problem with Jon. A dictator always says that he didn't do anything to force people to adore him, but they, especially a few around him, just did. And please, please, be nice. "lad?" "troll?" I think I'm old and civilized enough.
Dennis:
How could I do an experiment when I have no measurement setup? Go back to the thread and read carefully. I didn't particularly pointed you when I gave the "... not up to my standard" comment. I just assessed my overall knowledge I gained from the discussion at that time. I needed to know more. Look at discussion here and also read Jon's explanation. The matter is not so simple. I just wanted to have more details and consider other factors. That's all. I'm not an expert at all on these things. As I said, my intention was not to debate. So no wonder that I couldn't give any scientific, knowledgeable details there. I don't understand why you took my statement personal and overreacted to it. And I did NOT assert that my design IS superior to Jon's. I just raised a hypothetical question to see people's responses and learn from there. Please, don't add irrelevant comments any more here. Let's have people discuss the subject.
Thanks,
Jay
P.S. I have no problem with Jon. A dictator always says that he didn't do anything to force people to adore him, but they, especially a few around him, just did. And please, please, be nice. "lad?" "troll?" I think I'm old and civilized enough.
Thanks for those, forr. I think those time constants show that it's average power, not peak power that determine VC temperature. Outside the pro concert arena, I think 20 C temp rise is pretty realistic for home speakers except for subwoofers that can take a real pounding. That's what stereophile measured in that test with loud, 'dense' music and most music has a much higher crest factor. Certainly none of us would pound our speakers as hard as Brandon did on his test (thanks for risking it Brandon), and even then the compression was less than a dB.
Hi Svante,
---If you use the secondary coil, there will be a time constant between the coils that is hard to tell how large it is.---
This could be calculated... However my "discover" was purely incidental, I was looking at the "diaphony" between coil, how both coils behave as a transformer, for servo applications. I retained the feature of the second voice coil as a fast reacting thermal sensor because it's an easy way to estimate temperature without much hardware.
0.35 dB was the just noticeable difference on sinewaves I arrived when I tested myself. This is in accordance with average people according to data I own. I am not aware of speakers having response deviations less than 0.35 dB.
Hi Hartono,
Taking data like RT = 7 °C/W, TTC = 12s, it is easy to calculate what the temperature of the speaker voice coil will be when 1 W will be applied for 12s, and then the corresponding DC resistance increase. Do you think the delivered power is often 1W for 12s at home ?
---If the cooling rate is not exactly the same with heating(which is impossible), then it will modulate the resistance.---
Modulating the resistance does not modulate the sound but compresses it. There are some AES papers on voice coil heating, they tell about sound compression, nowhere of sound modulation, by thermal effects.
Note I gave numbers and ways to estimate values of what we are speaking about. You didn't and just qualify as "fast" the thermal VC reactions, maybe you could provide more substantiated informations than mine.
---If you use the secondary coil, there will be a time constant between the coils that is hard to tell how large it is.---
This could be calculated... However my "discover" was purely incidental, I was looking at the "diaphony" between coil, how both coils behave as a transformer, for servo applications. I retained the feature of the second voice coil as a fast reacting thermal sensor because it's an easy way to estimate temperature without much hardware.
0.35 dB was the just noticeable difference on sinewaves I arrived when I tested myself. This is in accordance with average people according to data I own. I am not aware of speakers having response deviations less than 0.35 dB.
Hi Hartono,
Taking data like RT = 7 °C/W, TTC = 12s, it is easy to calculate what the temperature of the speaker voice coil will be when 1 W will be applied for 12s, and then the corresponding DC resistance increase. Do you think the delivered power is often 1W for 12s at home ?
---If the cooling rate is not exactly the same with heating(which is impossible), then it will modulate the resistance.---
Modulating the resistance does not modulate the sound but compresses it. There are some AES papers on voice coil heating, they tell about sound compression, nowhere of sound modulation, by thermal effects.
Note I gave numbers and ways to estimate values of what we are speaking about. You didn't and just qualify as "fast" the thermal VC reactions, maybe you could provide more substantiated informations than mine.
As brought up in the HTG thread, while VC temp rise may not be instantaneous, the Le rise with increased excursion is instantaneous and varies with the VC position in the gap. This is a far more serious effect than VC heating. It has a minor effect on the blend at the crossover frequency and a major effect on motor-induced distortion.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.