Can output impedance measurement/plot be added to the amplifier test repertoire?

[Deleted member 48726]

It's 40 watts into a >200 oz magnet and 3 inch voice coil. If properly vented it could dissipate 300 watts. What temperature would everything be then?

Who knows? Not me, not you. Because we don't know the VC temp.

The issue is the heat transfer from the VC to it's surroundings. As air is a very good insulator the transfer is poor, hence you can not just guess at how hot the coil gets from its surrounding materials temperature.

 

[MAB]

Yeah, that's an extreme case. This is like I described though, 40 watts continuous into a magnet and voice coil sized for a concert... temps are low.

Overhung and underhung heat transfer is very different, too

Yes, agree.

If those speakers would survive running like that for 6 hours, the magnets would get much hotter. Those are also very likely vented - this hot air had nowhere to go

This is B&O showing that their DSP solution can limit thermal compression. But like you say, only for a limited time.

 

[mike7877]

Who knows? Not me, not you. Because we don't know the VC temp.

The issue is the heat transfer from the VC to it's surroundings. As air is a very good insulator the transfer is poor, hence you can not just guess at how hot the coil gets from its surrounding materials temperature.

When the gap is small, it's almost like there is no gap. That's physics!

Something to consider - nothing actually touches anything else in this realm - they just get infinitely closer together.


edit: for clarity- what you quoted, I was asking you to estimate - extrapolate from other reasonable knowns. Not to come to an exact answer

 

[Deleted member 48726]

When the gap is small, it's almost like there is no gap. That's physics!

Ehm. No.

 

[mike7877]

Sucks being wrong, I'll get over it lol

 

[MAB]

Yes. How much thermal resistance is there over a mm of air?

Again, not looking for an exact number.
How much does your hand burn at 1mm away from a heating element vs 10mm vs 25.

Neither is this (looking for an exact number)

FYI, the entire reason voice coils don't all burn up instantly in most speakers when more than 10-20 watts are going in is because most of the heat is transferred to the top plate and magnet.

I think you are oversimplifying a bit. Perhaps read that Klippel paper I suggested. The heat is dissipated from the Voice Coil through convection and conduction through the leads, and to the pole-peice and magnet. This is spelled out explicitly in the Klippel paper. But, most models assume the time constant for the voice coil is much shorter than the pole-piece or magnet, i.e. they are totally not the reason the VC don't instantly burn up...

I'll leave the rest alone.

 

[mike7877]

I think you are oversimplifying a bit. Perhaps read that Klippel paper I suggested. The heat is dissipated from the Voice Coil through convection and conduction through the leads, and to the pole-peice and magnet. This is spelled out explicitly in the Klippel paper. But, most models assume the time constant for the voice coil is much shorter than the pole-piece or magnet, i.e. they are totally not the reason the VC don't instantly burn up...

I'll leave the rest alone.

Good to know. It's been a few years since I read parts of that - it'd be interesting to go through in more detail.

 

[Deleted member 48726]

Sucks being wrong, I'll get over it lol

There's a thread about break in of speaker drivers where I remember the link from Klippel mab posted and also measurements of VC temperature from B&O I think.

 

[KSTR]

VC temp rise is real and its effects are real as well.
It actually happened to me that a speaker I did got voiced wrongly because I had used too loud too long sweeps (in an attempt to get cleanest, noise-free data) when measuring the individual drivers. It occured to me only when I happend to try the sweeps backwards out of curiosity, now going high to low, and baam! more than a dB of difference at the passband edges of the drivers.

 

[MAB]

VC temp rise is real and its effects are real as well.
It actually happened to me that a speaker I did got voiced wrongly because I had used too loud too long sweeps (in an attempt to get cleanest, noise-free data) when measuring the individual drivers. It occured to me only when I happend to try the sweeps backwards out of curiosity, now going high to low, and baam! more than a dB of difference at the passband edges of the drivers.

Yes, in another thread I posted a series of measurements over time of a speaker including after it had been operated at moderately high power for 30 mins.

So yeah, operating the woofer dramatically changes parameters. Even the room temperature impacts the measurements significantly.

 

[mike7877]

Yes, in another thread I posted a series of measurements over time of a speaker including after it had been operated at moderately high power for 30 mins.

So yeah, operating the woofer dramatically changes parameters. Even the room temperature impacts the measurements significantly.

I wonder if anyone's ever designed a commercial speaker's crossover for VC resistance at its intended operating temperature.
It'd have to be a higher end speaker for it to matter. Diy not necessarily obviously, but for a corporation to be using 1% parts, I'd expect the drivers and enclosures to be pretty high end - ATC for example.

Most of the time I've got about 1w RMS going through my speakers. Loud is like 10 - 10w RMS is like 150 peak for rock. 2 to the tweeter, 8 to the woofer. Those coils, especially tweeter, are so light! Fraction of a gram, it's insane

 

[MAB]

I wonder if anyone's ever designed a commercial speaker's crossover for VC resistance at its intended operating temperature.
It'd have to be a higher end speaker for it to matter. Diy not necessarily obviously, but for a corporation to be using 1% parts, I'd expect the drivers and enclosures to be pretty high end - ATC for example.

Most of the time I've got about 1w RMS going through my speakers. Loud is like 10 - 10w RMS is like 150 peak for rock. 2 to the tweeter, 8 to the woofer. Those coils, especially tweeter, are so light! Fraction of a gram, it's insane

I believe many do.

The B&O DSP approach I mentioned above takes the thermal model of their drivers to not only limit the driver's maximum temperature, but also reduce the non-linearity of the driver when operating in the high thermal regime. They have a bunch of papers on it. I have no idea if they make their speakers sound better or if it allows them to use drivers with modest thermal capabilites and efficiency.

JBL, Meyer Sound, Genelec, etc. have detailed physical models of the thermal circuits of their drivers. Aside from drivers with high efficiency and massive thermal capabilities:

It does seem that most of these serious speaker designs are being done with absolute understanding of how the driver performance changes at high power and high temperature. But I am sure not all... A long time ago I had a pair of Vandersteen 1B that I bought from a shop I worked at in the '80s. I liked them in the store, and sold many. I didn't like them at home, even at moderate volume they 'ran out of gas'. 1st order crossover slopes were a big Vandersteen thing. I thought it was why we had quite a few back because of blown tweeters, and why I perceived the tweeter faded out as I turned the volume up.

I can't imagine listening at full blast for more even a single song on the gear I now have, so I am pretty sure I am not heating things up to the point of audible compression and distortion. And there is a ton of affordable gear available now with impressive dynamic range that sound great...

Which I guess is the point... if power and thermal compression is the priority because you enjoy slightly higher SPL than the average for listening, it is probably best to get speakers with at least moderately high efficiency. The idea is that under moderately loud listening conditions you don't have to suffer the large changes in driver parameters. The B&O data shows there is a limit, as you continue to dump heat into the magnet and chassis they slowly heat up which incrementally compromises the VC's dissipation. This is PA/House Party territory, then you want even higher efficiency. Reading Meyer Sound or L-Acoustics' literature for example, it seems they are totally aware of the intricacies of power and thermal compression, and are absolutely designing with high temperature fidelity in mind.

 

[Jukka]

Heat transfer is 7th grade physics, maybe some people here skipped that year

Copper melts at >1000 C, so that's most likely not an issue, but the glue melts already at ~250 C, which is the obvious issue. Melted glue can make the VC short or become loose.

As for crossover drift, has anyone simulated how much it actually drifts for increase in output impedance or heating up of speaker components? A typical xo for 2-way is at 2500 Hz, a hundred more or less is not much and I'd be surprised if the drift was that much. Yes, it's out of spec and problems may rise at crossover region, but the life of the voice coil is a bigger in my opinion so far. Crossover components are often specced with 5 % tolerances, which could be more than what you get from heat build up.

I favor active speakers, which are free from these issues.

 

[mike7877]

I think you are oversimplifying a bit. Perhaps read that Klippel paper I suggested. The heat is dissipated from the Voice Coil through convection and conduction through the leads, and to the pole-peice and magnet. This is spelled out explicitly in the Klippel paper. But, most models assume the time constant for the voice coil is much shorter than the pole-piece or magnet, i.e. they are totally not the reason the VC don't instantly burn up...

I'll leave the rest alone.

Hold on... I got thinking some more about this:

If it was so hard for a driver's voice coil, say a tweeter, to dissipate heat, it wouldn't take long at all to reach thousands of degrees with just a watt or two. If literally all of it wasn't transferred and radiated away rather quickly, tweeters (and speakers of all types) just wouldn't survive. There isn't a lot of movement in a tweeter, and very often the design behind the dome is entirely sealed. This means that, other than through the dome, nearly all of the heat (so nearly all of the power, as at 90dB/W, most power applied isn't converted to movement, but heat) is being conducted to the magnet assembly and then radiated to the environment. Since I've pointed my IR thermometer at many a dome's dome, I can say that their temperature is maybe 1 deg C above ambient at quieter listening. Domes in medium sized rooms never seem to rise to more than 10 deg C over ambient at comfortable listening levels (not quiet, just not painful).
Considering the above, most conduction has to be to the magnet assembly, and most of it has to be pretty quick, because also, considering the temperature rise of 0.5g copper with 1W applied, if the thermal resistance of its surroundings was high it wouldn't take long to get red hot and melt. Yes there needs to be an increase in temperature first (so there's enough of a gradient for efficient/effective transfer to take place), but once this is reached, I firmly believe that most heat is dissipated through the magnet.

Also, using the time constant alone as evidence of dissipation or absorption, is only looking at one piece of the puzzle - without knowing the weight, it can't be judged how well heat transfer is taking place. And using weight + time constant is also probably a pretty convoluted way to determine the temperature of the radiator - I think you'd first need the contact area of the radiator plus the power being absorbed by the surrounding material to estimate the temperature of the radiator. Regarding contact area, if it's not direct contact, the distance between the two materials is probably required as well. IR radiation flows freely until absorbed. I believe its power dissipates like any microwave radiation. A millimeter of oxygen and nitrogen isn't much of a barrier, and the air present all has nowhere to go because behind the dome is sealed.

There's also this observation I've made:
I've seen a subwoofer, in the middle of normal loud operation, get hit with a transient that knocks its coil out of the magnetic gap. In seconds the coil was smoking. By 5, the coil was done. The only thing that can be concluded from this is that transfer to free air is much lower than to metal with a 1-2mm buffer of air between. Much lower.

Yes, I didn't spend an inordinate amount of time finding the applicable equations to make an example with notes describing each step of the processes I described, and my last observation is completely anecdotal, but
What do you think after reading this?

edit: about woofer leads being an avenue for voice coil cooling - I've taken a spool of 50 feet of 20ga wire and put a few amps through it. The ends coming off the coil were a few degrees over ambient, while the coil itself rose by over 30 degrees. I didn't have extremely long leads, but from 2-3 inches from the coil to the ~1.5 feet to the leads (I should say foot as it was only one side - the other was connected to the inside of the roll) was the same, lower temperature.

 

[Jukka]

Hold on... I got thinking some more about this:

If it was so hard for a driver's voice coil, say a tweeter, to dissipate heat, it wouldn't take long at all to reach thousands of degrees with just a watt or two. If literally all of it wasn't transferred and radiated away rather quickly, tweeters (and speakers of all types) just wouldn't survive. There isn't a lot of movement in a tweeter, and very often the design behind the dome is entirely sealed. This means that, other than through the dome, nearly all of the heat (so nearly all of the power, as at 90dB/W, most power applied isn't converted to movement, but heat) is being conducted to the magnet assembly and then radiated to the environment. Since I've pointed my IR thermometer at many a dome's dome, I can say that their temperature is maybe 1 deg C above ambient at quieter listening. Domes in medium sized rooms never seem to rise to more than 10 deg C over ambient at comfortable listening levels (not quiet, just not painful).
Considering the above, most conduction has to be to the magnet assembly, and most of it has to be pretty quick, because also, considering the temperature rise of 0.5g copper with 1W applied, if the thermal resistance of its surroundings was high it wouldn't take long to get red hot and melt. Yes there needs to be an increase in temperature first (so there's enough of a gradient for efficient/effective transfer to take place), but once this is reached, I firmly believe that most heat is dissipated through the magnet.

Also, using the time constant alone as evidence of dissipation or absorption, is only looking at one piece of the puzzle - without knowing the weight, it can't be judged how well heat transfer is taking place. And using weight + time constant is also probably a pretty convoluted way to determine the temperature of the radiator - I think you'd first need the contact area of the radiator plus the power being absorbed by the surrounding material to estimate the temperature of the radiator. Regarding contact area, if it's not direct contact, the distance between the two materials is probably required as well. IR radiation flows freely until absorbed. I believe its power dissipates like any microwave radiation. A millimeter of oxygen and nitrogen isn't much of a barrier, and the air present all has nowhere to go because behind the dome is sealed.

There's also this observation I've made:
I've seen a subwoofer, in the middle of normal loud operation, get hit with a transient that knocks its coil out of the magnetic gap. In seconds the coil was smoking. By 5, the coil was done. The only thing that can be concluded from this is that transfer to free air is much lower than to metal with a 1-2mm buffer of air between. Much lower.

Yes, I didn't spend an inordinate amount of time finding the applicable equations to make an example with notes describing each step of the processes I described, and my last observation is completely anecdotal, but
What do you think after reading this?

edit: about woofer leads being an avenue for voice coil cooling - I've taken a spool of 50 feet of 20ga wire and put a few amps through it. The ends coming off the coil were a few degrees over ambient, while the coil itself rose by over 30 degrees. I didn't have extremely long leads, but from 2-3 inches from the coil to the ~1.5 feet to the leads (I should say foot as it was only one side - the other was connected to the inside of the roll) was the same, lower temperature.

Well, about the only thing that generates heat in a speaker driver is the voice coil. The VC does not heat up solely because current runs through it, most of the heat comes from the current interacting with the magnet and it's forces. The VC conducts heat to the air around it (very inefficient, but it improves if the VC moves quickly, like in bass drivers) and through the leads (if copper, efficient). All else is heat radiation. Heat energy radiation speeds up as the temperature difference increases between the heated object and its surroundings. Voice coils are attached to voice coil formers (usually a plastic tube, very short and thin in tweeters, almost like a ring), that conducts heat poorly. The dome is attached to the former and the former to attached to the basket or the frame of the driver with the suspension. The VC touches physically only the voice coil former and air around it. Dome can get heat from VC via the former and air (very inefficient). Tweeters are air tight and often isolated from the rest of the air inside the speaker. The VC gets heat only when current is passing through, at all other times it's losing heat.

Where does the heat go? The magnet is big and heavy and so it can soak up a good deal of heat radiation before heating up itself. The magnet conducts heat to air and the basket. Some tweeters have a "ring radiator" or other type of metallic plate that's exposed on the outside of the speaker. This plate and the dome can conduct heat into air and radiate heat to surroundings. The leads (driver leads and cabinet internal leads and crossover parts) can soak up some heat as well, depending on the gauge. When soldering thick wires with lead-free solder and high temperatures, one can notice the lead in fact does conduct heat quite well (the other end may burn your fingers). Copper is one of the best materials for heat conduction. If the heat gets though the wire terminals into the speaker wires between speaker and amp, it also contributes small amount in heat radiation. All parts inside the cabinet heats the air inside it, which spreads it around. That is a major factor in heat dynamics.

All these together can dissipate heat for 1 W or more, which is often good enough for tweeters. Remember that radiation speeds up with higher heat difference.

What about a bass driver that has VC popping out from the gap? Well, I'm not entirely sure, but when looking at ways that heat can dissipate, my only idea is that the VC stops moving and that causes convection to air collapse (not getting the fan effect). It's hard to see that the surroundings outside of the gap would be so hot that radiation would collapse. If the enclosure is vented, like for bass drivers, the air exchange can contribute a great deal to dissipating heat.

 

[mike7877]

What about a bass driver that has VC popping out from the gap? Well, I'm not entirely sure, but when looking at ways that heat can dissipate, my only idea is that the VC stops moving and that causes convection to air collapse (not getting the fan effect). It's hard to see that the surroundings outside of the gap would be so hot that radiation would collapse. If the enclosure is vented, like for bass drivers, the air exchange can contribute a great deal to dissipating heat.

One thing to know about the subwoofer that I unfortunately didn't bring up, is that it was rated for 400W RMS and the amplifier driving it, 500WRMS.

The average level of the music was loud, but it wasn't any louder than it had been on numerous occasions previously. The song was unfamiliar though, which is why the volume wasn't adjusted for the transient, which is what caused the accident.

I know that the woofer moves a lot of air when its coil is properly in the gap, but it's not a lot of new air flowing over the coil - the diaphragm only moves about an inch, so it's largely the same couple litres of air.

Also, a non-vented woofer would heat the air in the box to insane levels very quickly if air is where most of the heat went (think toaster-oven hot. Most of the heat has to be going into the magnet structure. It takes quite a bit of energy to heat 20-30lbs of metal, it takes almost none to heat 20-30 litres of air

 

[Jukka]

One thing to know about the subwoofer that I unfortunately didn't bring up, is that it was rated for 400W RMS and the amplifier driving it, 500WRMS.

The average level of the music was loud, but it wasn't any louder than it had been on numerous occasions previously. The song was unfamiliar though, which is why the volume wasn't adjusted for the transient, which is what caused the accident.

I know that the woofer moves a lot of air when its coil is properly in the gap, but it's not a lot of new air flowing over the coil - the diaphragm only moves about an inch, so it's largely the same couple litres of air.

Also, a non-vented woofer would heat the air in the box to insane levels very quickly if air is where most of the heat went (think toaster-oven hot. Most of the heat has to be going into the magnet structure. It takes quite a bit of energy to heat 20-30lbs of metal, it takes almost none to heat 20-30 litres of air

Yes true. It puzzles me. I personally own DIY speakers that have 15" woofers that are in their own separate, sealed cabinets. The nominal AES power handling (2 hours test made with continuous pink noise signal within the range Fs-10Fs. Power calculated on rated minimum impedance. Loudspeaker in free air.) for the driver is 500 W, but I got no idea how well it would last if driven hard in my current boxes, but I used them for a private disco party in a gymnastics hall for one night. The woofers did not give up.

I belittle the effect of the magnet for cooling because of the fact, that by convection the heat has to first go into the air and then to the magnet, and that happens slowly. The VC does not touch the magnet, so it has to go through air. By radiation it goes the speed of light, but radiation power depends on a few factors, like surface area, emission/absorption coefficient and heat difference. It's negligible when both are in room temperature, but if the voice coil has heated up by 100 degrees, radiation power has become strong. Either way, the magnet is inside of the box and as its mass gets heated, its cooling effect will reduce. FWIW, my woofers use neodymium magnets, which are light, net weight for the whole driver is only 4 kg (8.82 lb). That can't rely on mass for cooling capability.

There is also the chance that when the VC pops out of the gap, its magnetic environment changes and that will affect the rate at which the VC generates heat. I don't know enough details to tell how much so. Or maybe it just reached it limit right there. The fact that it popped out is a clear sign it that was driven too hard

 

[fpitas]

There is also the chance that when the VC pops out of the gap, its magnetic environment changes and that will affect the rate at which the VC generates heat.

Well, yes. At that point, all the power delivered goes into heating the voice coil.

 

[fpitas]

Well, yes. At that point, all the power delivered goes into heating the voice coil.

Also, I have to think the impedance drops significantly as that happens, adding to the danger.