I would like to protect a sub from overheating by monitoring the power going into it.
The amplifier used on the sub would be capable of substantially larger power than the subs continous rating to allow the peaks. This circuit would need to limit the average power to a lower level than the peak power.
I am thinking of putting the measured power through a filter, say a single time constant, then a comparator to decide if its too much.
The question is what to make the time constant?
The amplifier used on the sub would be capable of substantially larger power than the subs continous rating to allow the peaks. This circuit would need to limit the average power to a lower level than the peak power.
I am thinking of putting the measured power through a filter, say a single time constant, then a comparator to decide if its too much.
The question is what to make the time constant?
It could be much easier to measure coil temp directly, or perhaps you could infer by comparing Voltage and current?
I'm trying to avoid cutting into the subwoofer itself. What I'm thinking of is essentially taking the voltage, squaring it (to get VxV= VxIxR) and then working on the worstcase assumption that R=8Ohms always I have an approximation of the power in. Putting this through a low pass filter before deciding if its too much would then allow peaks through but hopefully prevent it from overheating (as the voice coil has thermal inertia). I just dont have a feel for even an approximate idea of the thermal inertia.
I can use approximations as I'm not trying to squeeze every ounce of power out of the subwoofer, just don't want to fry it or buy a $$$ super power capable model.
It takes about a minute or so for the over-temp-warning lights to come on when I way-overdrive my Vandersteen 2Ce speakers. Then they don't go all the way off unless I turn off the input completely, or turn it WAY down, immediately. (Just turning it down to the max level that normally would never make the lights come on makes them stay on and I haven't wanted to leave them like that to see how long it would take for them to go off.) I don't know if that will help but it's the only data point I have.
Wouldn't a worst-case assumption be LOWER than 8 Ohms?
Wouldn't a worst-case assumption be LOWER than 8 Ohms?
I have some papers by Penkov dealing with such circuits.
They are in french but the figures may suffice for your application.
They look more or less like the attached picture and are available from Selectronic in France :
Kit de protection pour haut parleur pas cher | Acheter Kit de protection pour haut parleur discount
They are in french but the figures may suffice for your application.
They look more or less like the attached picture and are available from Selectronic in France :
Kit de protection pour haut parleur pas cher | Acheter Kit de protection pour haut parleur discount
Attachments
There are various protective gizmos sold by kit houses.
Please correct me, but doesn't Class D kind of amps handle big peaks but poop out when longer notes are played?
I believe there is also a spec about headroom in amps which is usually consulted in order to buy an amp which has muscular power rails so that it doesn't droop on long notes. The OP is worried about those long notes being too strong.
Without endorsing Class D (which, as it happens, power all 6 of my amps), the logic of buying an amp which can play loud brief notes but which DOES droop on longer notes, makes sense for speaker protection and sensible enough given the millisecond peaks in music waveforms (such as soprano choristers).
Ben
Please correct me, but doesn't Class D kind of amps handle big peaks but poop out when longer notes are played?
I believe there is also a spec about headroom in amps which is usually consulted in order to buy an amp which has muscular power rails so that it doesn't droop on long notes. The OP is worried about those long notes being too strong.
Without endorsing Class D (which, as it happens, power all 6 of my amps), the logic of buying an amp which can play loud brief notes but which DOES droop on longer notes, makes sense for speaker protection and sensible enough given the millisecond peaks in music waveforms (such as soprano choristers).
Ben
Last edited:
I looked at one of the videos, looks like it works well at protecting speakers, but would require proper programming which requires destructive testing if you want to get the most out of a speaker.
Spot's original question:
"The question is what to make the time constant? "
still needs to be answered, the time constant is dependent on voice coil heat dissipation capability, mass, motor mass, and frequency (for starters).
Some drivers are far better at getting rid of heat than others.
An RMS time constant of a second or two may be appropriate for a smallish woofer, but would be too long for a light tweeter.
Also, the question of what is "substantially larger power" than a driver can take on peaks comes into play, what works at 3 dB above the RMS level won't work for 6 dB.
Eminence designers have probably figured out the protection parameters for their drivers, using other drivers similar to theirs could use similar parameters with their device.
Hi,
Ball park numbers for a 2" voicecoil driver are a time constant of ~ 20 seconds,
and thermal resistivity of 3 degrees C per watt, with something like 300 degrees
C peak allowed and something like 250 degrees C continuous allowed.
These numbers of course will vary between drivers. However it all depends on
how high in frequency the subwoofer is used. Excursion not thermal issues are
the deciding factor in the low bass. Typical sub amps will take the driver to
its excursion limits and there is no point in having any extra peak power.
Might be for the bass unit in a 3-way, as that enters the thermal region.
But IMHO for a sub you should forget about thermal issues for excursion.
A sub drivers continuous thermal rating practically will never be approached.
The more "hifi" the bass, with higher peaks, the lower the average power.
The low bass peaks should not exceed the excursion capabilities of the
driver, this can have very little to do with the drivers thermal rating.
rgds, sreten.
For AV some subs predict driver excursion and apply progressive compression
to prevent overexcursion, not really hifi, but the sub goes a lot "louder" FWIW.
Ball park numbers for a 2" voicecoil driver are a time constant of ~ 20 seconds,
and thermal resistivity of 3 degrees C per watt, with something like 300 degrees
C peak allowed and something like 250 degrees C continuous allowed.
These numbers of course will vary between drivers. However it all depends on
how high in frequency the subwoofer is used. Excursion not thermal issues are
the deciding factor in the low bass. Typical sub amps will take the driver to
its excursion limits and there is no point in having any extra peak power.
Might be for the bass unit in a 3-way, as that enters the thermal region.
But IMHO for a sub you should forget about thermal issues for excursion.
A sub drivers continuous thermal rating practically will never be approached.
The more "hifi" the bass, with higher peaks, the lower the average power.
The low bass peaks should not exceed the excursion capabilities of the
driver, this can have very little to do with the drivers thermal rating.
rgds, sreten.
For AV some subs predict driver excursion and apply progressive compression
to prevent overexcursion, not really hifi, but the sub goes a lot "louder" FWIW.
Last edited:
The voice coil impedance in normal high power use will double just before permanent damage occurs typically.
The voice coil blows out according to I2 x T. That depends on how thick the wire is.
Over excursion occurs from too much level at too low a frequency but drops by a factor of 4 for each doubling of frequency.
So you need to monitor the I & V & phase angle to measure the impedance and do an rms limit based on that. Monitor the current square that and us it to limit peak voltage. Then you need a low passed filter voltage monitor to track the excursion limit.
You can get impedance from monitoring the actual driver, excursion limit from the data sheet or monitoring distortion. You get the I2 x T limit from experiments.
The voice coil blows out according to I2 x T. That depends on how thick the wire is.
Over excursion occurs from too much level at too low a frequency but drops by a factor of 4 for each doubling of frequency.
So you need to monitor the I & V & phase angle to measure the impedance and do an rms limit based on that. Monitor the current square that and us it to limit peak voltage. Then you need a low passed filter voltage monitor to track the excursion limit.
You can get impedance from monitoring the actual driver, excursion limit from the data sheet or monitoring distortion. You get the I2 x T limit from experiments.
Anyone try designing a realtime impedanc tracking circuit?
Something that compares current and voltage to detect the impedance rise at high power.
Has anyone the schematics for the EV Deltamax circuits? It predicted thermal and excursion related stress, and had a sense line on the amplifier output.
Something that compares current and voltage to detect the impedance rise at high power.
Has anyone the schematics for the EV Deltamax circuits? It predicted thermal and excursion related stress, and had a sense line on the amplifier output.
I2xT is only really valid where T is very small such that negligible heat will escape. I would have thought that the time constants in speakers were much larger as they are designed to dissipate a lot of power unlike fuses.
So if the typical voice coil in an 8 Ohm driver has an impedance of 6.6 Ohms, and ignoring impedance changes with frequency, which could be significant, then over-temp damage would be likely when the impedance seen from outside the box went from 8 Ohms to 14.4 Ohms (i.e. 8 + 6.6). Allowing a safety margin by limiting the voice coil impedance change to only an additional 50%, for example, or 3.3 Ohms, would mean that we would want the safety mechanism to be triggered when the impedance seen from outside the box reached 11.3 Ohms.
It should be relatively easy to use a small current-sense resistor in series with the speaker input, to convert the current to a voltage by connecting a differential amplifier's or instrumentation amplifier's inputs across the current-sense resistor. A similar amplifier would be used to grab the voltage across the input port.
Those could be scaled appropriately (and the amps should be made to be phase-matched, or just identical) and sent into a divider circuit made from something like an AD633 multiplier chip and an opamp (see AD633 datasheet), to give a voltage corresponding to the impedance = V/I, in real time.
A low-pass filter and a comparator with appropriate trigger level would probably give a finished circuit. You could then add a circuit to either trigger a relay driver to disconnect the speaker, or send feedback to your remote-controlled attenuator or whatever else you had in mind, perhaps an explosive-powered guillotine for the speaker cable. 🙂
You'd still also need to worry about the excursion limiting.
Cheers,
Tom
It should be relatively easy to use a small current-sense resistor in series with the speaker input, to convert the current to a voltage by connecting a differential amplifier's or instrumentation amplifier's inputs across the current-sense resistor. A similar amplifier would be used to grab the voltage across the input port.
Those could be scaled appropriately (and the amps should be made to be phase-matched, or just identical) and sent into a divider circuit made from something like an AD633 multiplier chip and an opamp (see AD633 datasheet), to give a voltage corresponding to the impedance = V/I, in real time.
A low-pass filter and a comparator with appropriate trigger level would probably give a finished circuit. You could then add a circuit to either trigger a relay driver to disconnect the speaker, or send feedback to your remote-controlled attenuator or whatever else you had in mind, perhaps an explosive-powered guillotine for the speaker cable. 🙂
You'd still also need to worry about the excursion limiting.
Cheers,
Tom
I don't think that means that the time-constant is small. I think that it only means that I2xT is not the whole story. I2 is proportional to the power dissipated in the voice coil, I2R. But the power only gives you the RATE of heating. To get from there to something like the voice coil temperature, you'd have to integrate I2R with respect to time, which, for a constant I, would be I2RxT. But the current will be constantly changing (not to mention the R vs temperature) so we'd need to actually integrate (to get the area under the curve) over time, instead of just multiplying to get the area of a rectangle like we could if I were constant.
But even if we did that, with a current-sense circuit and an opamp integrator, we'd still be ignoring the rate at which the voice coil was LOSING heat, at the same time. That rate would probably change with the voice coil's temperature. So we would need to integrate that over time, too, and then subtract that from the integral of I2R, to get something proportional to the temperature of the voice coil.
But in this case, we don't know the thermal characteristics of the voice coil, anyway. And it would probably be much easier to just monitor the actual temperature than to try to predict it. And if monitoring for a rise in impedance would work, that would probably be much simpler and more reliable than trying to use integrators and depending on assumptions about the cooling rate and worrying about error accumulation. And we'd probably have to monitor the impedance, anyway, to include the current R value in the I2R we were integrating. So I'd be inclined to try the impedance-monitoring method, first.
This is a DC output passive protector which could be mount in a speaker enclosure or amp.
Number of Z diodes and their values should be calculated. The text is in polish rather than french.
Number of Z diodes and their values should be calculated. The text is in polish rather than french.
I2 x T is important so that a spike doesn't blow things up before the voice coil has time to heat up. However if you are playing CD's there are no such spikes, unless you plug or unplug input connections with power on.
For live sound you will need a spike limiter, but you can probably just square the voltage and set a limit at say 10 db (power) above the rated power level for the loudspeaker.
If you are using a full range multiple driver system you will need to use a crossover to multiple limiter circuitry.
P.S. Actual voice coil impedance is used for protection, rated just means it will not dip to 1/2 of that value across the bandwidth of interest.
- Status
- Not open for further replies.