[JoshK]

So there seemed to be a spark of interest in the topic of current amplifiers rather than traditional voltage amplifiers and their potential benefits to driving speakers that arose from this thread (starting around post 86 by Roger).  Rather than derail that topic further, it would be interesting to discuss the topic on its own.   This topic is something I've heard and read a tiny bit about but is still relatively new to me.

Some observations (theoretical, empirical, practical):

1)  To quote Roger,  "Those who understand the physics of the moving coil driver know that it is a current controlled device. To state it simply the current in the voice coil produces a force that acts on the mass of the cone producing constant SPL over the usable range of that driver."

2) Traditional crossovers (more than first order) mess up the ideal, thus we had discussed active multi-way systems with current amps for each driver.

3) benefits of current drive better at higher frequency?  [source needed]

The above is just a start.  I thought it might be helpful to collect some general characteristics and even put links to where to read up. 

Some resources on the topic:
http://www.current-drive.info/   
http://www.firstwatt.com/pdf/art_cs_amps.pdf     (Pass's current source amps)
http://www.audioasylum.com/forums/magnequest/messages/5167.html   (John Swenson's 9w Transconductance amp)
http://www.tubecad.com/2012/03/blog0226.htm  (Broskie talk a lot about current driven loudspeakers and mentions the book of the first link's author)
http://www.edn.com/design/consumer/4423155/Loudspeaker-operation--The-superiority-of-current-drive-over-voltage-drive   (article describing the first link's concepts)
http://www.epanorama.net/newepa/2014/10/10/loudspeaker-operation-current-drive-better-than-voltage-drive/

I had discussed going active, quite possibly with PLLXO, so topics about how to implement XOs in a multi-way are germane to this topic.

[G Georgopoulos]

current drive...no thanks...

[sts9fan]

I love the F1

[JoshK]

current drive...no thanks...

Is that all you have to offer?  If so, we could do without. 

[JoshK]

I have a few things I have to take care of first and my hip means I move quite a bit slower, but I'd like to knock up a breadboard of the 9w gm amp.  I have most all the parts on hand, I think and I have a couple of Gary Pimms' SBCCS boards that have external jumpers to the tube socket as they were provisions to be hybrid CCS for active loads and such as well.

[Roger A. Modjeski]

I hope this thread gets some attention. It is always nice to have someone out there to discuss these things with. I just the other day did a bredboard of a current driven headphone amplifier.

Josh, I looked at the amp your are referring to and it seems overly complex.  All those mosfets cant help the sound. Why not just make straight-forward pendtode output stage with no feedback.. of course.  It will have pretty darn high output impedance.

[BobRex]

Weren't the old Quad amps current drive?

I'm also thinking the at Atma-Sphere amps are designed around current drive.

[GEC]

I have no experience with this, but this link may be of interest.

http://www.diyaudio.com/forums/chip-amps/239321-joe-rasmussen-trans-amp-40-watt-transconductance-current-amplifier.html

[richidoo]

Atmasphere and every other commercial amp in the world is voltage source, except the First Watt amp. A current amp requires a passive crossover that is designed for use with current source amp. All commercial speakers with passive crossovers are designed for a low impedance voltage source.

A current source amp has infinite output impedance, which makes it impervious to back EMF from voice coil, caused by internal box echos, cone overshoot, etc.  So there is no noise coming back fromthe speaker that the amplifier has to fight against, making a very pure musical signal. But this is also the weakness of a current source amp, because, having infinite output impedance, there is no electrical damping at all, damping factor is zero, and the back emf from voice coil can't draw current from what is essentially an open circuit. So the speaker must damp itself mechanically and the ideal driver Qms= 1. The Tang Band full rangers, especially W8-1772   Qms=1.02! It should be an ideal driver for CS amp.

Current source amps are particularly good for full range drivers. It cures the midrange megaphone shouty thing that is common with those drivers. And no passive crossover required.

I've read Esa's Current Drive book, highly recommend it.

I have a diy CS amp based on LM3875. I have only listened to it using a FR driver with Qms= 7.   I have not used it with a low Q speaker yet. But it did help the drivers' mild shoutiness.  It sounded like a gentle mannered tube amp, but it is very musical.

[*Scotty*]

I would have thought that a sealed system Qts of 0.5 would be ideal as the speaker system is critically damped and would exhibit zero ringing.
Scotty

[richidoo]

Yup sealed is better than ported. Add aperiodic port to dampen the sealed resonance too.  Open baffle would work well with low Qms driver. That might be ideal.

With CS amp, the load impedance determines the FR. Response rises with load impedance. So any impedance peaks like system/driver resonance or rising impedance of voice coil will increase the SPL. So zobel filter on the treble and a series notch filter for the bass, if necessary.

Since it's hard to flatten the bass impedance peak without a soul robbing coil, it's usually better to give the bass to a voltage powered woofer. FR drivers usually aren't big enough to do justice to low bass anyway.

But with the right driver, box and mechanical and electrical damping strategies you can use the impedance hump to boost the bass for lower extension. The TB 1772 has a low Q resonance peak so a big box plus damping with aperiodic port plus a very mild notch could be the ticket. Or active EQ with no coil.

Edit: W8-1772 Fr and impedance plot. See the wide fat resonant hump? That's the low Q of self damping suspension. But the impedance amplitude is still too high in free air, so OB would need a notch filter, or active EQ. But aperiodic port in a large box will reduce the peak amplitude mechanically, though hard to make it flat. George Short of North Creek Audio is the man regarding aperiodic ports. Some of his cone speaker designs have some cool techniques to flatten Z.

[*Scotty*]

Here's Stereophile's impedance and phase measurements of the Dunlavy SC-IV/A. A sealed critically damped system

Fig.1 Dunlavy SC-IV/A, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).
As you can see from the graph there appears to be a maximum worst case impedance rise of 3 ohms at the bass resonance frequency.
A higher system Qts with the associated impedance rise resulting from a higher driver Qts could be easily be reduced to the same magnitude or lower by using an aperiodic vent. The system Qts could be about .5 with reasonable bass extension if the driver Qts allows it. If good bass extension can not be achieved with an aperiodic alignment, then an aperiodic alignment can still be used and the bass supplemented with an added sub-woofer.
Scotty

[OzarkTom]

Here's Stereophile's impedance and phase measurements of the Dunlavy SC-IV/A. A sealed critically damped system

Fig.1 Dunlavy SC-IV/A, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).
As you can see from the graph there appears to be a maximum worst case impedance rise of 3 ohms at the bass resonance frequency.
A higher system Qts with the associated impedance rise resulting from a higher driver Qts could be easily be reduced to the same magnitude or lower by using an aperiodic vent. The system Qts could be about .5 with reasonable bass extension if the driver Qts allows it. If good bass extension can not be achieved with an aperiodic alignment, then an aperiodic alignment can still be used and the bass supplemented with an added sub-woofer.
Scotty

I sold a pair of the Duntech Sovereigns back in the 80's. Those were incredible if set up in the right room. Dunlevy always used a first order, 6db per octave crossover.

[richidoo]

That's a sweet looking plot, a tube amp's dream speaker with phase >0.

Midrange and up, drivers have good acoustic impedance, they are big enough to grip the air well and minimize harmonic distortion from air slipping out of the cone. So in these upper freqs a current source amp's poor damping is not a big concern. But in LF, the acoustic impedance drops because driver size doesn't increase in proportion to wavelengths played, and we rely on electrical damping of VS to tighten up the bass as much as possible. With CS' poor elec damping slam and detail suffer in LF.

By using CS in the upper ranges, crossing to VS in the bottom driver, CS can avoid both the system resonance and also the low acoustic impedance drivers. Then the bottomest driver in the speaker can benefit on both issues by using voltage source amp.

[*Scotty*]

richidoo, if its not too far off topic, what changes did you make to the build on the LM3875 to turn it into a current drive amp. I am running a LM4780 as voltage amplifier which is the usual implementation of the Overture Series chips. I am running + and - 35V rails for 60 watts into 8 ohms and 120 watts into 4 ohms.
Scotty

[Davey]

A big-ass resistor in series with the LM3875 output?

If you use Magnepan speakers with a textbook passive crossover does your voltage source amplifier become a current source amplifier? 

Dave.

[richidoo]

A big enough series resistor does change any voltage source amp into a current source amp, but rather inefficiently.
Better way: http://www.current-drive.info/

Scotty, I didn't design my amp, but I'll try to find the schematic. The link above spells out pretty clearly how it works. Current feedback with the load in the feedback loop, which is connected to ground by .5R current sensing resistor - instead of voltage feedback with the load outside loop. His design example uses TDA2040, a similar power opamp as 3875.
and...
http://e2e.ti.com/support/amplifiers/audio_amplifiers/f/6/t/284951
http://sound.westhost.com/project56.htm

[Davey]

Yeah, that's the standard configuration.  Current source amps are interesting, but the problem is, the vast majority of driver/transducers on the market are not suitable for this type of source.

To me, it's stretching the definition to even call the Pass F1 a "current source" amp.  There's a rather large gray area between "current source" and "voltage source with high output Z" I guess. 

I always got a kick out of the Sunfire amps with their "current source" output........which was simply a 1 ohm resistor in series with the "voltage source" output. 

Dave.

[JoshK]

Current source amps are interesting, but the problem is, the vast majority of driver/transducers on the market are not suitable for this type of source.

Why is that? 

[Davey]

The average speaker system (driver) has large impedance swings and is too low efficiency.  The SPL curve looks like the impedance curve if driven with a transconductance amplifier source.  We can electronically equalize things pretty easily nowadays, but the efficiency issue still remains.

Current source amps are aimed at high-efficiency driver systems.  Horns, Fostex, Lowther, etc.

Dave.