Current drive for Loudspeakers

Problems arise with full range current drive (or even just non-zero drive impedance) where mechanical resonance causes a rise in acoustic output along with an increase in load impedance. In this case, and at such frequencies where this occurs, the speaker ends up drawing more power and the peak is accentuated. Whether distortion is decreased, response flatness is not inherently improved. Are we considering only woofers or assuming pistonic cone response?

Speaking for myself - when I conceived of #96, I too realized it would be for woofer only.... for the same reasons you give. However, if broken down into tri-amp system, maybe each driver's resonance could be addressed and lower thd of each driver?

Thx-RNMarsh
 
Last edited:
perhaps an interesting thing to do is put a variable Rs (pot) and adjust its Rs for lowest measured LF speaker distortion. Then design the amp to have same output Z without the Rs. [then verify thd is same].

A drop in distortion of 50-60% is quit noticable and seems like a very worthwhile thing to do. For too long we have focused on freq response... we now have many means to flatten the FR. and drivers have improved.... while thd is still too high. I bought the Quad speakers for reference because it had both flat FR and very low distortion. But it still cannot be played loud, its expensive and has limited bass spl at low freqs. Would be nice to finally, make this improvement for dynamic drivers.... speaker designers cant do it all alone.

Since 1985, I modify all my bass amps for this distortion reduction using the simple #96 config. I thought it would have taken off long ago. Wonder why it didnt.

Thx-RNMarsh
 
Last edited:
perhaps an interesting thing to do is put a variable Rs (pot) and adjust its Rs for lowest measured LF speaker distortion. Then design the amp to have same output Z without the Rs. [then verify thd is same].

You mean a variable transconductance amp? We have a couple of those in the neighborhood. Each speaker seems to have a "best" setting. No THD measured (and with Geddes showing that this is a meaningless measure, moot)

dave
 
perhaps an interesting thing to do is put a variable Rs (pot) and adjust its Rs for lowest measured LF speaker distortion. Then design the amp to have same output Z without the Rs. [then verify thd is same].

A drop in distortion of 50-60% is quit noticable and seems like a very worthwhile thing to do. For too long we have focused on freq response... we now have many means to flatten the FR. and drivers have improved.... while thd is still too high. I bought the Quad speakers for reference because it had both flat FR and very low distortion. But it still cannot be played loud, its expensive and has limited bass spl at low freqs. Would be nice to finally, make this improvement for dynamic drivers.... speaker designers cant do it all alone.

Since 1985, I modify all my bass amps for this distortion reduction using the simple #96 config. I thought it would have taken off long ago. Wonder why it didnt.

Thx-RNMarsh

If the Quad speakers are not loud enough maybe you should look for Open baffle Linkwitz LX521 ot Orion which I have.Linkwitz Lab - Loudspeaker Design
By the way you never answered PM about PCB I sent to you.
BR Damir
 
But then the thd is not as low... but plays louder.

I have not received the pcb, yet.

Richard Marsh
2457 Cascade Trail
Cool, CA. 95614
USA
?

Yes I sent it last Friday on that address, probably it will take some time to arrive.

PS. it looks that the girl at the post office wrote "CA. 9561C", could that be a problem?
 
Last edited:
But then the thd is not as low... but plays louder.
The Quad ESL63 has the lowest measured (& audible) THD at LF within it's output capacity of ANY speaker I've measured. When I first ran one up on an oscillator at 'high' level, I thought I'd done something wrong cos I couldn't hear any LF distortion. Similarly when I measured one.

This is a little known fact about push pull constant charge electrostatics .. of which the ESL63 is probably the best.

In the early 70's, advanced implementations of Rice & Kellog's invention overtook electrostatics in Mid & HF performance. Today, for the BEST mid & HF sound, moving coil units rule.

Hence one candidate for my ULTIMATE speaker is moving coil mid & HF and Electrostatic bass. It will be active and might incorporate Current Drive for the bass. It will also be very large to get around the output limitation for electrostatics at LF.

There are clues to the reasons in Peter Walker's original paper, New Developments in Electrostatic Loudspeakers
_______________

Mr. Marsh, how do/did you measure acoustic distortion?
 
The Quad ESL63 has the lowest measured (& audible) THD at LF within it's output capacity of ANY speaker I've measured.

Hence one candidate for my ULTIMATE speaker is moving coil mid & HF and Electrostatic bass. It will be active and might incorporate Current Drive for the bass. It will also be very large to get around the output limitation for electrostatics at LF.

There are clues to the reasons in Peter Walker's original paper, New Developments in Electrostatic Loudspeakers
_______________

Mr. Marsh, how do/did you measure acoustic distortion?

Hi,

For the Quads (newer model) I relied on the mfr data for distortion and saw it was very low and even lower than some amps. Just my kind of idea for a speaker. Using MLSSA (DRA Labs) the FR is also ruler flat. Great speaker.
'
But I have them put in the closet because of the bass issue. I dont want speakers far out into the room. And, I want to get very low at realistic spl levels. Polite' levels are not for me. Something the QUADS cant do to my satisfatction. I may do a Bi or TRI amp (Again) but not eager for the effort and hope someone else has done it all for me. Right now I have 4 each 15 inch bass drivers which are highish effeciency. But, I plan an upgrade to build into the building a pair of 18 inchers. The 18 inchers have thd and FR that are close to best in class. But it is a LOT of work. I could then use the Quads with these. Current drive or motional feedback is in the cards for the bass.

Bass distortion is somewhat relative to others but I measure on my back deck (which goes out to infinity space). Use various instruments to mesure Fr and THD etc.

The first and last tri-amped system consisted of a pair of 18 inch Hartley bass drivers in a folded transmission line ( as per TAA) which were the size and weight of a refrigerator, each. Both were laid on their side. I put Quad (older) ESL on them - as mid-range drivers at ear level and mounted on the top edge of the QUAD's were a few RTR electrostatics driven off the plates (direct-drive) of a DYNACO tube amp. The X-over was my own design.... bass-mid electronic and the mid to treble was passive. Sounded great but what a chore it was to build and I dont want to go thru that ever again.... I would rather buy something made and modifying it.

Still, cutting the bass driver(s) distortion by at least 1/2 is audible on most any system IMNSHO and experience. Should have similar results on mid and tweeter distortion.

THx-RNmarsh
 
Last edited:
But I have them put in the closet because of the bass issue. I dont want speakers far out into the room. And, I want to get very low at realistic spl levels. Polite' levels are not for me. Something the QUADS cant do to my satisfaction.
My 'Ultimate' speaker would have the equivalent of 4 x ELS63 on each side to do the bass. About 12dB more capability at LF

Bass distortion is somewhat relative to others but I measure on my back deck (which goes out to infinity space). Use various instruments to mesure Fr and THD etc.
I know you have SOTA methods to measure amp. THD but what do you use to measure Speaker THD?

The latest AP uses Prof. Angelo Farina's method which is what I would use.
 
I think it will be very hard to make a well-defined, linear 'intrinsic high output Z'.
With feedback, it's a no-brainer to make it accurate and linear whatever value you want.

Depends on what you mean by "high". Without recourse to loop feedback,
it's pretty easy to get several hundred ohms, which the F1 would have but
for the resistance I provided parallel to the output at about 80 ohms.

😎
 
My 'Ultimate' speaker would have the equivalent of 4 x ELS63 on each side to do the bass. About 12dB more capability at LF

I know you have SOTA methods to measure amp. THD but what do you use to measure Speaker THD?

The latest AP uses Prof. Angelo Farina's method which is what I would use.


I have not done any serious speaker testing in years. When I did, it wasnt with the SOTA stuff I have now.... just HP THD+N thd analyzers and 'greened' oscillator and a flat mic/preamp of low distortion. I now have plenty of equipment but little time to use them as I seem to be all over the map with this and that thing. I need a company of people to work with or not much gets done on the audio front. I have drivers in boxes, several electronic crossovers with DSP never opened, Audyssey PRO and the like. Lots of good measurement microphones etc. I could outfit a factory and keep people busy for months if not years. But, its just me and the old Apache lady... she looks after me. So, I work in spurts of R&D. Tonight I ripped open the chassis bottom for access to the IC in the MM7025 Marantz. I am changing out the IC and compensation cap just to see what happens... measure and listen.
Got to drive to L.Angeles next week to work on a visa to go exploring in Asia. Life is too short to do everything, I guess. But, I'm still trying 🙂
Can you give me a link up to the Prof. A.Farino method so i can try to catch up with you on speaker measurments.


THx-RNMarsh
 
Last edited:
Depends on what you mean by "high". Without recourse to loop feedback,
it's pretty easy to get several hundred ohms, which the F1 would have but
for the resistance I provided parallel to the output at about 80 ohms.

😎

Hi Nelson,
Yes, true of course, and the output transformer in that configuration will down-transform that to a lower value.
When I wrote the post I was thinking about an more standard amp with a class AB output stage without feedback and they often have an intrinsic Zout between several up to 10 ohms, but these values vary a lot between copies of the same amp and are difficult to predict.
I wonder how much variation you are seeing with your configuration?

jan