When looking at just the thermal power handling of a subwoofer voice coil, what variables are used to determine a rough power handling number?
Say I am building an underhung subwoofer with a 4.25" diameter coil using 22awg wire, .5" winding height with 4 layers, 72 turns, and a DCR of 1.3 ohms. There is 80 feet of wire in this coil.
If you are comparing it to another underhung coil that is 2.5" diameter, using 24awg wire, .53" winding height with 8 layers, 208 turns,, and a DCR of 3.7 ohms. There is 145 feet of wire in this coil.
The 2.5" voice coil exists in an actual production subwoofer with a power rating of 600w RMS. Since the 4" voice coil has a little over half of the length of wire as the 2.5", does its power handling scale to a little over half which would be around 400w?? Does the gauge of wire change things?
I know the proper way to test is just to do an AES-1984 test on it, but this question is from a design standpoint as if you were designing your subwoofer from the ground up.
Say I am building an underhung subwoofer with a 4.25" diameter coil using 22awg wire, .5" winding height with 4 layers, 72 turns, and a DCR of 1.3 ohms. There is 80 feet of wire in this coil.
If you are comparing it to another underhung coil that is 2.5" diameter, using 24awg wire, .53" winding height with 8 layers, 208 turns,, and a DCR of 3.7 ohms. There is 145 feet of wire in this coil.
The 2.5" voice coil exists in an actual production subwoofer with a power rating of 600w RMS. Since the 4" voice coil has a little over half of the length of wire as the 2.5", does its power handling scale to a little over half which would be around 400w?? Does the gauge of wire change things?
I know the proper way to test is just to do an AES-1984 test on it, but this question is from a design standpoint as if you were designing your subwoofer from the ground up.
I can see three "layers" of thermal power handling:
AES-1984 requires a two-hour test in free-air. ie, so long as the voice coil can sink the heat to the motor okay, the air exchange ability of the motor will be the limiting factor. The "layers" above are why it's possible to hit speakers with relatively large power inputs for short durations, so long as the "fuse" rating isn't exceeded. I've hit drivers with 5x their RMS (long-term) power rating and got away with it just fine, because music is peaky and they won't be held at full power for more than a fraction of a second.
Without further information, I would suggest that your subwoofer, with a voicecoil with 1.6x the radiating area of the 2.5" diameter VC, would handle proportionally more power in the long-term. ie, a smidge under 1KW.
Wire length won't factor in much: most of the extra length has been put into layers which can't radiate heat away.
Chris
- Instantaneous current capacity - ie, above which the VC becomes a fuse.
- Thermal mass of the voice coil - of the order of seconds for subwoofers. This is where the glue melts and the coil unwinds.
- Thermal saturation of the magnetic structure - minutes to hours. If the motor gets too hot overall, the voicecoil can't sink heat into it.
AES-1984 requires a two-hour test in free-air. ie, so long as the voice coil can sink the heat to the motor okay, the air exchange ability of the motor will be the limiting factor. The "layers" above are why it's possible to hit speakers with relatively large power inputs for short durations, so long as the "fuse" rating isn't exceeded. I've hit drivers with 5x their RMS (long-term) power rating and got away with it just fine, because music is peaky and they won't be held at full power for more than a fraction of a second.
Without further information, I would suggest that your subwoofer, with a voicecoil with 1.6x the radiating area of the 2.5" diameter VC, would handle proportionally more power in the long-term. ie, a smidge under 1KW.
Wire length won't factor in much: most of the extra length has been put into layers which can't radiate heat away.
Chris
Thank you for the reply, it brought some variables to the table that I hadn't thought about such as radiating area.
I had considered the fuse aspect and my original prototype used 150c rated magnet wire on the voice coil hanging over N42SH neo magnets with the theory that the voice coil would be the fuse before overheating the magnets, but the magnets barely got hot so I wound some new voice coils with 200c magnet wire.
The motor is a cup motor with axial air vents on the bottom and a .06" thick aluminum shorting ring pressed into the OD of the gap. The coil former is Titanium to block heat transfer, so the coil transfers heat directly to the aluminum, which extends up to contact the aluminum basket.
Not sure if there's anything that hasn't already been covered but this B & C video gets into the nitty gritty of the different limits:
Power Handling Deep Dive - Bennett Prescott - B&C Speakers
There are big differences between thermal and mechanical power handling. The other one is compression and output loss, but 99% of people out there assume we talk about reaching the point of VC winding failure. If that's what you want to go by, just assume half the rated input voltage of VC failure as a safe input range. It also depends on what type of enclosure the driver is working in ie. ported, sealed, rear or front horn loaded, etc.
You can measure it and hopefully stop before it becomes destructive.
Your thermal limit is .... duh! .... temperature .... and you can measure it in real time.
Long ago (think late 60´s early 70´s) Fane published in Wireless World a very clever method. (why am I not surprised? 😉 )
Since copper resistance thermal coefficient is well known, if you can measure DCR with precision you will very accurately calculate temperature, with less than 1 degree error, go figure.
Problem is doing it real time, you can´t simply apply Audio, switch it OFF and use your trusty old multimeter to measure DCR, VC has such a low thermal mass that when you switch to measure, even within seconds, actual value will have drifted away significantly, making measurement useless.
So these clever guys made a Wheatstone/impedance bridge instead.
If you know 3 of the impedances and balance the bridge (zero current through the meter or zero voltage between right and left nodes) you can easily find the fourth.
Say speaker is Z1, you use a small resistor as Z2 , say 1 ohm , pick a convenient value for Z4, say 100 ohm and a variable resistor for Z3, Z3/Z4 ratio is exact same as Z1/Z2 ratio.
The beauty of this method is that when you turn amp OFF for measuring, Z3 will keep value needed for balancing so keep the temperature relevant parameter for your convenience.
Only "problem" is that speakers are reactive, so can´t be strictly compared to plain resistors.
FANE trick around that was that although indeed speakers are reactive over most of the range, there is a certain frequency, typically in the midrange, often between 200Hz and 400Hz, where current and voltage are in phase , so it "behaves" resistive 😱
You can measure phase, of course, but the simple alternative is to find the minimum impedance point and test there, it´s close enough.
Knowing VC temperature you can deduce thermal destruction point, if you know safe temperatures for your adhesives, wire enamel and former, and stop just before that.
Your thermal limit is .... duh! .... temperature .... and you can measure it in real time.
Long ago (think late 60´s early 70´s) Fane published in Wireless World a very clever method. (why am I not surprised? 😉 )
Since copper resistance thermal coefficient is well known, if you can measure DCR with precision you will very accurately calculate temperature, with less than 1 degree error, go figure.
Problem is doing it real time, you can´t simply apply Audio, switch it OFF and use your trusty old multimeter to measure DCR, VC has such a low thermal mass that when you switch to measure, even within seconds, actual value will have drifted away significantly, making measurement useless.
So these clever guys made a Wheatstone/impedance bridge instead.
If you know 3 of the impedances and balance the bridge (zero current through the meter or zero voltage between right and left nodes) you can easily find the fourth.
Say speaker is Z1, you use a small resistor as Z2 , say 1 ohm , pick a convenient value for Z4, say 100 ohm and a variable resistor for Z3, Z3/Z4 ratio is exact same as Z1/Z2 ratio.
The beauty of this method is that when you turn amp OFF for measuring, Z3 will keep value needed for balancing so keep the temperature relevant parameter for your convenience.
Only "problem" is that speakers are reactive, so can´t be strictly compared to plain resistors.
FANE trick around that was that although indeed speakers are reactive over most of the range, there is a certain frequency, typically in the midrange, often between 200Hz and 400Hz, where current and voltage are in phase , so it "behaves" resistive 😱
You can measure phase, of course, but the simple alternative is to find the minimum impedance point and test there, it´s close enough.
Knowing VC temperature you can deduce thermal destruction point, if you know safe temperatures for your adhesives, wire enamel and former, and stop just before that.
Yes, obviously there is the impedance dip where the motor will behave somewhat resistive ie. In phase I and V where the most power can be sunk into the driver. Fane got that right and did a great job at showing where to expect the most current across the VC.
A good sense of smell can be helpful here as a crude means of knowing whether the adhesives used in the windings and former are starting to "gas out". Expect a little of that being ok when the driver is new, but once the adhesives are set, there should be little tell of any odor, plus many manufacturers will "set" the windings with a quick blast of current across the VC, solidifying the winding as a solid component. Other than this scenario, any more odor usually signifies the VC is under great thermal stress and failure of the adhesives is not far away.
The good thing about a high reactance motor is the impedance will protect the VC at a broader range of frequencies, given the mechanical limits are still not being reached. Thats why many tweeters will survive an oscillating amplifier outout stage as long as the rise in Le continues higher up in frequency and there is no Zobel used to compensate. Many amps have Zobels at there outputs and I've seen the resistors go up in smoke from over ambitious techs bench racing amplifiers with high amplitude square wave signals. Hence you shouldn't try bench testing an amplifier at higher power levels with HF heavy waveforms.
There is also a certain factor of power derating necessary to compensate for different materials used in the windings and VC former material itself, which can make a huge difference. A paper former from a Fostex or Lowther FR isn't going to handle nearly as much input as a Kapton or Nomex former on a big PA driver. Large VC excursion capabilities also help with cooling to a large extent if the motor is vented properly.
A good sense of smell can be helpful here as a crude means of knowing whether the adhesives used in the windings and former are starting to "gas out". Expect a little of that being ok when the driver is new, but once the adhesives are set, there should be little tell of any odor, plus many manufacturers will "set" the windings with a quick blast of current across the VC, solidifying the winding as a solid component. Other than this scenario, any more odor usually signifies the VC is under great thermal stress and failure of the adhesives is not far away.
The good thing about a high reactance motor is the impedance will protect the VC at a broader range of frequencies, given the mechanical limits are still not being reached. Thats why many tweeters will survive an oscillating amplifier outout stage as long as the rise in Le continues higher up in frequency and there is no Zobel used to compensate. Many amps have Zobels at there outputs and I've seen the resistors go up in smoke from over ambitious techs bench racing amplifiers with high amplitude square wave signals. Hence you shouldn't try bench testing an amplifier at higher power levels with HF heavy waveforms.
There is also a certain factor of power derating necessary to compensate for different materials used in the windings and VC former material itself, which can make a huge difference. A paper former from a Fostex or Lowther FR isn't going to handle nearly as much input as a Kapton or Nomex former on a big PA driver. Large VC excursion capabilities also help with cooling to a large extent if the motor is vented properly.
Last edited:
@JMFahey - My apprehension in getting a real world, meaningful measurement of an actual VC temp limit is not really knowing whether the actual limits of the former, adhesives and winding materials have been reached or exceeded. It makes a huge difference in power handling depending which materials are chosen and how well the driver is designed to shed heat from the motor while its in heavy continuous use.
An alnico driver will have much lower limits due to the magnetic material and its sensitivity to heat induced fluz loss. The temp limits of the magnet/motor are very important to know, but in case of ferrite and neo, this is much less of a concern than with those older altec and JBL alnico drivers we all love. Ive seen some VC failures where I just have to scratch my head how they were accomplished and how it would require such a ape fisted hand on the volume knob to get those results.
An alnico driver will have much lower limits due to the magnetic material and its sensitivity to heat induced fluz loss. The temp limits of the magnet/motor are very important to know, but in case of ferrite and neo, this is much less of a concern than with those older altec and JBL alnico drivers we all love. Ive seen some VC failures where I just have to scratch my head how they were accomplished and how it would require such a ape fisted hand on the volume knob to get those results.
For modern subwoofers the high power handling is obtained through high temp adhesives and forced air through the motor so its very difficult to say a 4" or 2.5" coil has a certian power handling as its dependent on the adhesives and the airflow design of the motor. Removed from the subwoofer in static air driven from a power supply the power handling will only be in the 10's of W.
I'm sure everyone has seen the barbarians on YouTube who plug car audio subs directly into the wall and see how long they last with cones flapping in the breeze to the tune of 60hz. Funny thing is if you are lucky to land Fs close to 60hz impedance peak, the driver can last a very long time due to the voltage dropping across a high reactive resistance. Its just such a stupid thing to do if you have good drivers and more money than common sense in your pockets to "burn" away.
The guy from B&C made a video about doing that to one of their woofers! He hung it from some bungee cords and let it play, and the tinsels are what failed.
I didn't expect this thread to continue getting replies, but the information is good. My original query was from a design standpoint on whether I would need to upsize my voice coil diameter from 3" to 4" to get more power handling to better fit its usecase. I feel like an 18" subwoofer with +/- 18mm excursion and a bl^2/re motor force of 160 would not be able to reach its potential with just a 3" voice coil; I am trying to see if bumping it up to a 4" would give me enough power handling to fit its usecase. If I can get a 1000w rms rating it would be perfect.
I use Hernon Voice Coil Bonder 360 and 22 awg 200c magnet wire for the coils, 4 layer wet wound with bonder under each layer, on a titanium former. So the coil itself should take some abuse. I am thinking about just bonding a k-type thermocouple onto the top wind of the voice coil and running a lead out to measure the actual voice coil temp in real time.
I didn't expect this thread to continue getting replies, but the information is good. My original query was from a design standpoint on whether I would need to upsize my voice coil diameter from 3" to 4" to get more power handling to better fit its usecase. I feel like an 18" subwoofer with +/- 18mm excursion and a bl^2/re motor force of 160 would not be able to reach its potential with just a 3" voice coil; I am trying to see if bumping it up to a 4" would give me enough power handling to fit its usecase. If I can get a 1000w rms rating it would be perfect.
I use Hernon Voice Coil Bonder 360 and 22 awg 200c magnet wire for the coils, 4 layer wet wound with bonder under each layer, on a titanium former. So the coil itself should take some abuse. I am thinking about just bonding a k-type thermocouple onto the top wind of the voice coil and running a lead out to measure the actual voice coil temp in real time.
To be very exact and honest, you CAN'T compare power handling from different datasheets very well at all unfortunately.
Some will refer to some AES standard, but even that one is pretty vague to begin with, and very open te interpretation.
By default it's tested with a relative wide band frequency range, which acts very differently compared to low frequencies.
In sense of heat, high output power with low cone excursion is absolutely killing for example.
Even a bit of cone excursion will give so much more cooling, depending on the motor design obviously.
Also the temperature of the environment is extremely important (which is rarely being considered)
It also says very little about dynamic power handling, which is far more important for PA systems (well, sound-reinforcement)
It's also a bad measurement for (very) long term power handling (or heat rather).
The type of system (closed, BR, PR, horn, dipole) also really really matters.
What I would do for designing a subwoofer? I assume this is for home-hifi application.
That totally depends on the context, sound pressure levels, environment and especially the type of music!
Usually I design in such a way that the xmax would be reached before the given power handling.
If you want to do it really well, take about half of the Pa_max at the peak of the VA (power) graph.
That will give plenty of safety margin.
But to be honest, for home hifi applications, this is very rarely a problem.
Unless you want permanent hearing damage or just push tiny little woofers non-stop at max.
So to be answer your question, what is the exact application, driver, system and cabinet volume?
Some will refer to some AES standard, but even that one is pretty vague to begin with, and very open te interpretation.
By default it's tested with a relative wide band frequency range, which acts very differently compared to low frequencies.
In sense of heat, high output power with low cone excursion is absolutely killing for example.
Even a bit of cone excursion will give so much more cooling, depending on the motor design obviously.
Also the temperature of the environment is extremely important (which is rarely being considered)
It also says very little about dynamic power handling, which is far more important for PA systems (well, sound-reinforcement)
It's also a bad measurement for (very) long term power handling (or heat rather).
The type of system (closed, BR, PR, horn, dipole) also really really matters.
What I would do for designing a subwoofer? I assume this is for home-hifi application.
That totally depends on the context, sound pressure levels, environment and especially the type of music!
Usually I design in such a way that the xmax would be reached before the given power handling.
If you want to do it really well, take about half of the Pa_max at the peak of the VA (power) graph.
That will give plenty of safety margin.
But to be honest, for home hifi applications, this is very rarely a problem.
Unless you want permanent hearing damage or just push tiny little woofers non-stop at max.
So to be answer your question, what is the exact application, driver, system and cabinet volume?
Yes, this is how I am designing it as well.
Usage is sealed for either hifi or car audio. The sub is one I'm designing from the ground up. Nothing higher than around 100hz, so power with no excursion isn't a problem. I need around 800-1000 watts of power to reach design xmax. My first iteration is complete and uses a 3" coil specified in the first post, but I know based on other underhung subwoofers from the past that use a similar coil size that my 3" coil is limited to around 600-700 watts rms. I am just trying to see if there is a method that real world subwoofer designers use to calculate thermal power handling based on coil dimensions. If the jump up to a 4" coil won't get the power needed, then I will have to go with a larger diameter coil. I'm trying to see if there is a way to calculate or estimate it instead of investing in the coil winding and motor tooling to do a 4" sample and have it not meet the power handling goal.
@hurrication
I am a bit confused, you're making your own driver?
Like I said before, the majority of power numbers that can be found in datasheets are down with wide band noise.
So you can't really go off of that.
But like you already mentioned, it's quite a tricky set of compromises.
As far as I now, in the end it's all about testing and doing quite some FEM/BEM simulations.
Reason being is that heat dissipation all depends on movement of the air and construction of the motor design.
Both things can't be estimated anymore by some simple calculations.
We are waaaay into complex thermodynamics at that point.
The best practical approach is just see what other are doing.
I am a bit confused, you're making your own driver?
Like I said before, the majority of power numbers that can be found in datasheets are down with wide band noise.
So you can't really go off of that.
But like you already mentioned, it's quite a tricky set of compromises.
As far as I now, in the end it's all about testing and doing quite some FEM/BEM simulations.
Reason being is that heat dissipation all depends on movement of the air and construction of the motor design.
Both things can't be estimated anymore by some simple calculations.
We are waaaay into complex thermodynamics at that point.
The best practical approach is just see what other are doing.
I think he may be customizing his own driver to suit the needs of his project. Ive taken a few older 18" EV drivers with their highly limited 2.5" VCs to the point of failure with misc VC assemblies sourced from various places. I would run out of xmax in most cases before I ran out of thermal power limits, but what I've discovered in the process was the edgewound Cu coils on nomex formers would hold up to 800W at a 50% duty cycle over 1 hr with the driver loaded in a FLH without any visible signs of distress. The weak link was always the cone and spider adhesive junction. It needed to be secured with a 200C UV cured epoxy, so the heat and mech stress (horn loaded application with small back chamber) wouldn't be an issue. Failure was usually seen in the cone itself.
A Ti former helps, but its prone to warping at higher temps given the adhesives and winding bond were up to it. The other thing is a smaller ID coil is stiffer than a larger one and you can concentrate the flux of the motor across a tighter area. A larger VC and motor is never a dumb idea and the only way to build in a margin of safety if you want reliability other than using exotic materials which may not be enough to hold up over time. Of course a ptc/ntc sensor in the winding is a great way to know exactly whats going on, but in most cases its more than sufficient to build a mathematical model in the software used to control power limiting based on known materials used and box tuning parameters. Most diy people who buy a raw frame driver dont care about this and will push the driver beyond xmax in a ported cab way before the electrical limit is reached. The exception is horn loading the driver in a small back chamber, which can be thermally brutal on any driver.
A Ti former helps, but its prone to warping at higher temps given the adhesives and winding bond were up to it. The other thing is a smaller ID coil is stiffer than a larger one and you can concentrate the flux of the motor across a tighter area. A larger VC and motor is never a dumb idea and the only way to build in a margin of safety if you want reliability other than using exotic materials which may not be enough to hold up over time. Of course a ptc/ntc sensor in the winding is a great way to know exactly whats going on, but in most cases its more than sufficient to build a mathematical model in the software used to control power limiting based on known materials used and box tuning parameters. Most diy people who buy a raw frame driver dont care about this and will push the driver beyond xmax in a ported cab way before the electrical limit is reached. The exception is horn loading the driver in a small back chamber, which can be thermally brutal on any driver.
Yes. I know, hard to believe. But I'm already on the 3rd revision of working prototypes. I am winding my own voice coils in house. All of the FEA and design work was done last year, I'm trying to optimize things right now for my 3rd revision.
That's the problem, nobody is doing what I am. I haven't found any subwoofer on the market that is comparable in design. Low distortion (.16 Re/Le ratio), underhung (flat bl), light mms, compliant, neo with high force factor, high excursion subwoofer. Some of the big neo prosound woofers come close, but they are all either overhung or split coil and do not support the amount of excursion mine does. But that's not my market.
It's my own design from the ground up, from 5" diameter solid chunk of round 1018 to a subwoofer all in house. I have a cnc voice coil winder and make my own coils.
One of the main reasons I went with titanium is for its low thermal conductivity: I need it to block heat instead of transfer it. The voice coil windings are air gapped straight to aluminum heatsink in the motor, so all of the heat transfer is done directly to the windings instead of through the former. Titanium also has a working temp of up to 400c, which far exceeds both the wire temp and bonder temp. Here is the tech data for the bonder, here is the data sheet for the wire.
The only reason I made this post to see if anybody knew of a way to estimate the increase of thermal power handling from a known amount of a given size coil to that of a larger diameter.
As per your last sentence, here is an example Selenium uses for determining VC temp delta and power handling. You need to determine your VC temp coefficient and a few other preliminary things, as you can see in the formula used (driver specs are for a 18WS600 so you can see what specs are required to calculate this).
Attachments
Riiiight, now finally the context becomes clear to me 🙂
I can only speak from experience, but like every pro company either does just tons of practical testing or tons of BEM/FEM research but MOSTLY a combination of the two when it comes down to thermals = max power.
At that point you're so much in a rabbit hole of difficult thermodynamics, omg.
I have even seen situations that when a specific spot on a voice coil isn't cooled down enough, it becomes a weak point.
We are talking at least an order of magnitude (or more) from the "standard" given max power handling (from the datasheet)
I know quite a bit of thermodynamics, but this requires so much of hands-on experience, omg.
None of the "simple" linear equations are valid anymore, not even close.
A (extremely simple) explanation, is that heat dissipation (or rather removal), is just not a simple linear expression anymore.
So all the ideas above (very well ment, so thumbs up) are not valid anymore.
Without any fancy simulations methods, my best advice would be just to try on stuff.
But that also involves a good air flow in the motor structure.
At the same time I wonder why you need so much power???
That is even quite a lot for sound-reinforcement applications!