Hi,
I'd like to learn more about current drive amplifiers and where to buy them. A big reason for me to use them is to avoid thermal compression, which is a big problem for the active speaker I'm designing due to huge bass boosts. There is a limiter that'll prevent the woofer from dying, but the effects of thermal compression is significant and current drive can eliminate this problem.
Thanks
I'd like to learn more about current drive amplifiers and where to buy them. A big reason for me to use them is to avoid thermal compression, which is a big problem for the active speaker I'm designing due to huge bass boosts. There is a limiter that'll prevent the woofer from dying, but the effects of thermal compression is significant and current drive can eliminate this problem.
Thanks
Last edited:
current driving amplifier
This book is the best what I have found of the theory of current driving. And it is easy to build an amp with a LM3886 f.e.
I would advise you to make an effort to understand the serious new problems that current drive will create before jumping in...
Is there a page that I can read about the problems of current drive? The book by Esa looks very complicated and would take weeks, if not months to digest.
If you read Nelson's article you will see some impedance plots. Absent any compensation, the ratio of impedances of any two frequencies are also the ratio of voltage that will be fed to the driver at those frequencies (following Ohms Law V=IR) because CURRENT is constant. This is just like a new EQ network applied to your system. Nelson provides some compensation networks that can flatten some of this, however, the peak around resonance is caused by only the mechanical parts of the driver, and some of them (suspension compliance) are pretty non-linear so they vary as a function of cone position. This means that the resonance peak can move around as a function of cone travel, and will do so much more than when there is electrical damping acting in concert - remember, all electrical damping is lost under 100% current drive. How can you EQ out a high Q peak (e.g. with Nelson's network) that is constantly changing in frequency and Q? Other parameters like the voice coil inductance are also changing 10-20% at large excursions. You do not want a system with a cone-travel-modulated frequency response but it seems that is exactly what you will get. To me there are too many possible side effects that are not easily fixed.
If your main concern is thermal compression, I would first try to figure out if that will truly be a problem in your system and for which drivers. If you can get info on driver Re you can adjust the gain for that driver to compensate for the sensitivity loss without having to resort to current drive and all of the new problems it brings. That is all that current drive amplification is doing anyway - when the VC heats up and Re increases, by Ohm's Law the constant current source increases the voltage into the load so that the current remains the same. Turning up the gain to increase the output voltage of a voltage source is doing the same thing in that regard.
If your main concern is thermal compression, I would first try to figure out if that will truly be a problem in your system and for which drivers. If you can get info on driver Re you can adjust the gain for that driver to compensate for the sensitivity loss without having to resort to current drive and all of the new problems it brings. That is all that current drive amplification is doing anyway - when the VC heats up and Re increases, by Ohm's Law the constant current source increases the voltage into the load so that the current remains the same. Turning up the gain to increase the output voltage of a voltage source is doing the same thing in that regard.
Last edited:
Agree and add:
1) constant current drive will *dramatically* change speaker sound and behaviour, way more than the change in impedance/loss of efficiency will do, and ... at ALL power levels, so it´s at best a mixed blessing
2) power compresson must be factored in in systems which work full power (or nearby) for a long time , read it as "PA system" and you won´t be far from truth.
Include some DJ equipment in the deal.
No reasonable home Hi Fi will be even remotely used in such way.
3) there is no free lunch ... if power compression is 3 dB, you´ll need double the power amp size just to cope with this, power must come from *somewhere*.
If it´s higher, then you need even more extra power.
To boot, t´s *easy* to thermally destroy speakers this way: under normal circumstances (voltage drive), speakers self protect (like any PTC resistor does 😉 ), you overdrive it and it takes less power ... but if you feed it constant current:
P=I*I*R
you keep I constant, the more a speaker compresses the more it keeps heating up ... a potentially disastrous situation.
So although in theory it looks good ... it creates a set of new problem.
1) constant current drive will *dramatically* change speaker sound and behaviour, way more than the change in impedance/loss of efficiency will do, and ... at ALL power levels, so it´s at best a mixed blessing
2) power compresson must be factored in in systems which work full power (or nearby) for a long time , read it as "PA system" and you won´t be far from truth.
Include some DJ equipment in the deal.
No reasonable home Hi Fi will be even remotely used in such way.
3) there is no free lunch ... if power compression is 3 dB, you´ll need double the power amp size just to cope with this, power must come from *somewhere*.
If it´s higher, then you need even more extra power.
To boot, t´s *easy* to thermally destroy speakers this way: under normal circumstances (voltage drive), speakers self protect (like any PTC resistor does 😉 ), you overdrive it and it takes less power ... but if you feed it constant current:
P=I*I*R
you keep I constant, the more a speaker compresses the more it keeps heating up ... a potentially disastrous situation.
So although in theory it looks good ... it creates a set of new problem.
OK I see current drive is not the solution to thermal compression. Thank you guys for the explanation of the problems.
Charlie, thermal compression is a real and big issue for my next speaker. I'm applying up to 20dB of bass boost and up to 200W in a speaker with only 5L internal volume, and that would barely exceed woofer Xmax only at certain frequencies. The 2 woofers will be cooking in there even at moderate volumes that will result in at least 1-2dB of sensitivity loss in the lower midrange within just a few minutes of bass heavy music, which then will cause the speaker to sound audibly thin. I won't be driving them loud for an extended period of time, so I'm not worried about thermal damage.
The problem is that I need dynamic compensation for thermal compression because it varies depending on volume, and this can't be done by DIY, which is why I originally wanted to explore current drive amps for the woofers like the Kii Audio Three.
Charlie, thermal compression is a real and big issue for my next speaker. I'm applying up to 20dB of bass boost and up to 200W in a speaker with only 5L internal volume, and that would barely exceed woofer Xmax only at certain frequencies. The 2 woofers will be cooking in there even at moderate volumes that will result in at least 1-2dB of sensitivity loss in the lower midrange within just a few minutes of bass heavy music, which then will cause the speaker to sound audibly thin. I won't be driving them loud for an extended period of time, so I'm not worried about thermal damage.
The problem is that I need dynamic compensation for thermal compression because it varies depending on volume, and this can't be done by DIY, which is why I originally wanted to explore current drive amps for the woofers like the Kii Audio Three.
Last edited:
OK, that is not a good idea! Remember, for a loudspeaker driver to be 1% efficient (meaning 99% of the input power will be converted to HEAT in the voice coil) the sensitivity is something like 92dB/W. If your driver has an efficiency of only 85dB/W, it's only about 0.2% efficient.
The driver will have a power input rating, Pe. This is obtained by powering the driver with this many Watts of power for a long time. The MFG is asserting that the driver will not sustain damage under that condition, however, this may be in free air. If you connect this same driver to more power and do not use some kind of duty cycle you run the risk of delivering more power to the driver that it can handle, meaning the VC temp will get too high and you risk burning the VC former or even melting the VC together it if is thin. This doesn't mean you can't use a 200W amplifier with a 20W driver - you CAN. But you must only apply full power for a short time before shutting off or dramatically reducing the power input to allow the system to cool. Some drivers like woofers designed for PA use have sophisticated designs that try to promote cooling of the VC, however, I would guess that is not the case in your system.
Trying to compensate for thermal compression by boosting power will just make the situation worse. Instead you could consider reducing the gain of the other drivers in the system to match the lower sensitivity of the "hot' woofer if that is of concern to you. This is something that I am considering in my own system - monitor the woofer and given some knowledge about how it changes when hot, adjust the crossover functions for the OTHER drivers to match.
The issue with current drive being responsible for overheating voice coils and destroying drive units is not observed practically.
Firstly music programme material has a relatively high crest factor compared to sine waves. Peak power requirements are therefore typically well in excess of rms power requirements.
Sadly in order to make their drive units look good many manufacturers - and even more sadly now official specifications - use maximum power measures that represent peak not rms figures. If you attempt to run most drivers at an rms power equal to the quoted maximum power rating they will not last long (in the order of a few minutes at best for most).
The reasons are manifold but an extra 3dB power output due to voice coil heating is seldom the reason for destruction. Insufficient thermal mass, eddy current heating and cheap-to-manufacture 'dry' wound coils is where you will normally find the culprit.
In summary current drive has some outstanding merits and the worry over drive unit destruction is not normally one of them.
I would also like to add a note on the audibility of thermal compression that is seldom stated...
Whilst thermal compression is deemed inaudible in single drive unit cases due to the relatively long time constant involved, the same is not true for multi-way systems. Here heating in one channel can lead to audible changes in the frequency response that exceed the minimally audible and often cited +/-0.1dB criterion.
There are many other advantages too that are detailed elsewhere and that are very often more audibly beneficial than the removal of thermal modulation. I would cite the removal of third order distortion due to modulation of the impedance by eddy currents in magnetically non-linear pole pieces as the biggest advantage.
Whilst thermal compression is deemed inaudible in single drive unit cases due to the relatively long time constant involved, the same is not true for multi-way systems. Here heating in one channel can lead to audible changes in the frequency response that exceed the minimally audible and often cited +/-0.1dB criterion.
There are many other advantages too that are detailed elsewhere and that are very often more audibly beneficial than the removal of thermal modulation. I would cite the removal of third order distortion due to modulation of the impedance by eddy currents in magnetically non-linear pole pieces as the biggest advantage.
Last edited:
As we have discussed previously in the thread "sensorless DSP cone excursion limiter" both Klippel and Birt show how to do this to great effect by pre-processing. Birt also shows an even better method via his self-balancing bridge that is eminently suitable for anyone implementing an active system.
Within the aluminium voice coil former? That aluminium safes more heat by allowing both sides of the coil to become cooled than it wastes by eddy currents.
Else thou are reit, some source R has rarely hurt a loudspeaker if designed for it. Full curent drive (source R >> load R) is impractical, but source R = load R is feasible.
Got a link to this?
IIRC, his AES paper is short on detail. But has an excellent discussion of the thermal problems.
I'm now a beach bum so buying AES papers is a big expense. 😡
Last edited:
No, within the pole pieces - specifically steel re the distortion mechanism. Also any eddys in the pole pieces limit what is a significant thermal sinking capability via the motor. This is particularly evident in short coil geometries.
Not so. It is practically very easy by using a transconductance amplifier.
Birt's paper is indeed an exceptionally good read - and if I dare say so - real DIY audio engineering.
The self-balancing bridge is discussed thoroughly in two UK patents and may well be available with different number in other territories:
https://worldwide.espacenet.com/pub...ES&FT=D&date=19881019&CC=GB&NR=2203609A&KC=A#
https://worldwide.espacenet.com/pub...ES&FT=D&date=19910213&CC=GB&NR=2234880A&KC=A#
I have used a modified version of the bridge for many years both for full online motional feedback (transitioning to current feedback at higher frequencies) and offline as a means for equaliser calibration (with online monitoring only for protection).
As expressed in the other thread I have referenced here (plus a couple of others I seem to remember), it is not a cure-for-all-ills and, in particular, for poorly engineered drive units - but then nothing is...
So eddy currents are, what makes the driver clang? If so, then eddy power should still be much smaller than complete applied electrical power and should not limit thermal handling capability.
But building such an amplifier is not easy, is it? What DIYs fine is moderately hi source R archieved by negatively feeding back voltage occuring over a shunt, which measures load current, see [url="http://www.diyaudio.com/forums/solid-state/295996-black-devil-improved.html#post4832117]Black Devil Improved post 7[/url].
I do not understand this term, but the removal of third order distortion due to eddys in non-linear magnetic materials gives (IMO) a noticeable reduction of what I would describe as an 'electronic glare' and is particularly audible in the mid-band.
What I was trying to say was that eddy heated pole pieces make poorer heat-sinks than unheated pole pieces because of the reduced thermal gradient. And for the same (acoustic) output power, more power must be applied to the input to make up the loss of eddy current heating - hence we see an inductive versus semi-inductive impedances that rise proportionally to different powers of frequency (unity and a half respectively).
It does mean building your own but there are also good reasons why it is easier than building your own conventional amplifier.
Last edited:
Clanging plates and klirrendes Glas involve non-linear distortion. As mechanic force built up by an electromagnet depends on current only, not on voltage, some current drive is good for common loudspeakers indeed.
Please explain to me, why a current source were easier to build than a voltage source.
Please explain to me, why a current source were easier to build than a voltage source.
The method I use employs a floating power supply driven from MOSFET drains with their sources (or source resistors) and the feed back current sensing resistor commoned at ground. This means their is a simple resistive load that avoids the complications of a speaker in the feedback loop. That in turn make stabilisation an easier task and the necessity of a separate supply for the output stage is something of an unexpected bonus...
- Status
- Not open for further replies.