Loudspeaker for current drive

At high frequencies it's a different matter because of inductance that changes depending on cone position*, and that's where you want the amplifier's output impedance to be as high as possible to keep the current steady.
In this 'scenario' to achieve your result, you pick a high efficiency tweeter and introduce resistance
in the crossover and NOT use the amplifier to introduce the higher impedance/resistance,
because high amplifier impedance has a distinctly negative effect on bass 'tightness' > Attack & Decay.
For the 'high frequencies' of a woofer, you use an inductor to increase impedance/resistance.
(active Bi or Tri amping is a different story)
 
Last edited:
  • Like
Reactions: 1 user
About the only thing I can vaguely conclude is that it is very inefficient cos hardly any wriggles can be seen including the 'bass' resonance. ie Electrodynamic Damping is very small.
Just to let you know >
Variations of impedance (curve) is not a true indicator of a speakers efficiency,
nor is a drivers 'electrodynamic damping' - it can even be quite the reverse.
 
In this 'scenario' to achieve your result, you pick a high efficiency tweeter and introduce resistance
in the crossover and NOT use the amplifier to introduce the higher impedance/resistance,
because high amplifier impedance has a distinctly negative effect on bass 'tightness' > Attack & Decay.
For the 'high frequencies' of a woofer, you use an inductor to increase impedance/resistance.
(active Bi or Tri amping is a different story)
Maybe that's what you do, but your "attack and decay" sounds a lot like the nebulous 'PRAT' (pace, rhythm and timing?). Unless you've actually checked one set of room correction filters against another, how can you be sure?

Also, your recommendation to just change the impedance passively in a multi-way speaker seems to be missing the point. If the useful range can be extended, then that changes the calculation on the 'necessary' number of ways.
 
Just to let you know >
Variations of impedance (curve) is not a true indicator of a speakers efficiency,
nor is a drivers 'electrodynamic damping' - it can even be quite the reverse.
Bl is what moves stuff between the Mechanical (including Acoustic) world and the Electrical world.

If Bl = 0, you only see Rdc in the flat Impedance curve.
As you increase Bl, Mechanical resonances start to be reflected in the Impedance curve,. The most obvious is the 'bass' resonance but other stuff also appears. I gave a short explanation earlier how the 'height' of the bass resonance is an EXACT measure of the RELATIVE values of Rem & Rm; (Electromagnetic & Mechanical damping)

This holds for ALL Electrodynamic transducers including MC speakers & pickups, ribbons, moving iron stuff, the AMT, original Magneplanars bla bla. Are you saying the LRS is different?

What speaker are you thinking of .. that has no 'bass' resonance or other impedance wriggles? AND has Rem > Rm and is more efficient than other comparable size speakers of the same 'type' (eg Open Baffle or sealed box)?
 
Last edited:
Maybe that's what you do, but your "attack and decay" sounds a lot like the nebulous 'PRAT' (pace, rhythm and timing?). Unless you've actually checked one set of room correction filters against another, how can you be sure?

Also, your recommendation to just change the impedance passively in a multi-way speaker seems to be missing the point. If the useful range can be extended, then that changes the calculation on the 'necessary' number of ways.
Filtering is Filtering AND Amplifying is Amplifying.
 
  • Like
Reactions: 1 user
There is no "bump".
Now why could that be? I've also encountered a phenomenon where, after adding 10 ohm series resistance to an 8 ohm woofer, a 40Hz notch filter for a prominent 'room' (?) mode had to be dialed down to use less attenuation and lower Q. (The exact opposite of what conventional 'wisdom' would suggest ought to happen, especially as it often fails to test to see if reality matches the theory.)

No explanation that I can come up with makes complete sense without considering a possibility the usual T/S based models are borderline broken if the room isn't taken into account.

It seems like the 'overshoot' that (nearly) everyone is always so worried about when Q goes above 0.707 is easily swamped by ringing echoes, but if the speaker Q is raised it could act as an active bass trap. But for that to make sense, there should be some significant loading effects that pressurise the speaker from the outside.
 
  • Like
Reactions: 1 user
I remember I had in the 80ies a small reflex two way Celestion DL4 (70hz-20khz) and let it listen to a friend in his home.

It was a small square room which must have had a prominent resonance at 50hz / 60hz.

It ideally made the small speaker sound like a big box with a much lower and louder bass than in my home.

Really no need for a subwoofer and some rare good matching between box and room.
2142.jpg
 
Key is either flat-ish impedance or a frequency response that ends up such that the affects of the impedance curve end up giving you a flat FR. The latter would be more limited, depending on the specific magnitutde of their Rout]/sub] | impedance interaction.

dave
in post #88 under
https://www.diyaudio.com/community/threads/chip-amp-modification-to-current-drive.389985/page-5
I ask this:
Are there any benefits by use of current drive, when the impedance plot of the loudspeaker is flat (i. e. like a normal resistor of 6-8 ohms - that is done by use of RC/RCL network to equalize the increasing impedance through voice coil inductance and the impedance maximum around fo) ??
I estimate that in most cases a perceived sonic improvement in a listening test by comparsion "current drive" vs. "voltage drive" arises from the more flat frequency response in the case of current drive (e. g. reducing mid range peak due impedance increasing in that area).

check out also this threads:
https://www.diyaudio.com/community/...e-current-drive-power-amplifier.330157/page-4
https://www.diyaudio.com/community/threads/current-drive-power-amp.22959/
https://www.diyaudio.com/community/threads/help-to-understand-current-drive.390405/
https://www.diyaudio.com/community/threads/direct-drive-amplifier-for-diy-ribbon.154690/
 
Last edited:
^you can think that through and successfully analyze any system before you, if you first realize what is the benefit of current drive and how it plays out.

Namely, to reduce effect of voice coil impedance to acoustic output. Acoustic output roughly equals current, which is responsible making force in the motor, so increasing series impedance with the voice coil does it by reducing driver "ability to effect" current. High output impedance of current amplifier helps to make but could be just a series resistor.

The shunt components you describe, that would flatten amplifier load impedance, would help equalize acoustic response with current amplifier as the driver voice coil doesn't regulate current anymore due to high series impedance (compared to datasheet, voltage amplifier). But, if you then analyze impedance in series with voice coil you see the same shunt components are in parallel with amplifier output impedance reducing it, which allows the voice coil impedance have effect on current again and some effect on acoustic output.

So, yeah there is some benefit, but not as much as without the shunt components, case when driver is connected directly to a high output impedance amplifier.
 
  • Like
Reactions: 1 user
There is no "bump".
[...] Now why could that be? [...]
For me, "bump" doesn't mean anything good. So I said: "no "bump"".

Look at the impedance measurement. The impedance is linear at practically 29R, with very very small deviations. The current-frequency response follows this impedance.
The sound transducer/housing fundamental resonance of the ~3.6 cft closed box is dampened with ~10 lbs of rock wool. This is a well-known approach?

By the way, the loudspeaker works pretty well, even when I connect it directly to the amplifier without the 20R series resistor. Then it has 10dB louder harmonics from non-linear function, which I don't like.
In the entire measured frequency range from 30Hz to 5kHz


Please don't take it personally, I could find dozens of posts to write this answer.
[...] encountered a phenomenon [...]
Have you measured the resulting SPL frequency response(s)? For example at your listening position?
 
I agree with tmuikku above. All the current must flow into the voice coil for it to have any effect on the SPL and shunting would steal some of it. So, if a speaker has flat(ish) impedance, then congrats (a smaller amplifier and simpler EQ), but flattening of the impedance using a Zobel network somewhat defeats the purpose of current drive.
 
^you can think that through and successfully analyze any system before you, if you first realize what is the benefit of current drive and how it plays out.

Namely, to reduce effect of voice coil impedance to acoustic output. Acoustic output roughly equals current, which is responsible making force in the motor, so increasing series impedance with the voice coil does it by reducing driver "ability to effect" current. High output impedance of current amplifier helps to make but could be just a series resistor.

The shunt components you describe, that would flatten amplifier load impedance, would help equalize acoustic response with current amplifier as the driver voice coil doesn't regulate current anymore due to high series impedance (compared to datasheet, voltage amplifier). But, if you then analyze impedance in series with voice coil you see the same shunt components are in parallel with amplifier output impedance reducing it, which allows the voice coil impedance have effect on current again and some effect on acoustic output.

So, yeah there is some benefit, but not as much as without the shunt components, case when driver is connected directly to a high output impedance amplifier.
interesting facts. But here arise the question, how make one an equalizing in a passive crossover network without the shunt components - e. g. on a full range driver, which have peaks in the midrange frequency area.
 
Hi, if you have a driver that is designed for voltage drive, you could put the passive components before power amplifier to maintain high impedance in series with the driver ;)

If you must have passive filter between amplifier and driver then you must use the shunts to divert some current past the driver, and that's it. It is more important to make nice frequency response than have maximal distortion reduction. Typical compromise situation, can't have it all, and there is no need to, have what is important and leave what is not that important.

Or, perhaps you could make driver (and enclosure) for current drive so that frequency response results pretty nice without any filter.
 
For me, "bump" doesn't mean anything good. So I said: "no "bump"".

Look at the impedance measurement. The impedance is linear at practically 29R, with very very small deviations. The current-frequency response follows this impedance.
The sound transducer/housing fundamental resonance of the ~3.6 cft closed box is dampened with ~10 lbs of rock wool. This is a well-known approach?

By the way, the loudspeaker works pretty well, even when I connect it directly to the amplifier without the 20R series resistor. Then it has 10dB louder harmonics from non-linear function, which I don't like.
In the entire measured frequency range from 30Hz to 5kHz


Please don't take it personally, I could find dozens of posts to write this answer.

Have you measured the resulting SPL frequency response(s)? For example at your listening position?
I wasn't disagreeing at all. The speakers and room have changed since I made the original observations, and the haphazard measurements are lost on old hard-drives somewhere. But adding series resistance seems to give pretty consistent improvements at high frequencies. At low frequencies I'd have to do more experiments.

~~~
I keep working myself up to build a current amplifier when I have a bigger block of spare time (so, probably never... 😉 ).

As usual, the various practical considerations like bias setting, DC servo, or using dual power supplies vs single-ended or some kind of battery solution with intermittent charging (overnight?) seem like bigger hurdles than whatever topology it's built around.

I won't know the raw frequency response of speakers until it's built, but at a minimum it should have 2 high quality buffer stages and modular space for active filters.
 
But here arise the question, how make one an equalizing in a passive crossover network without the shunt components - e. g. on a full range driver, which have peaks in the midrange frequency area.
The solution depends on where you install the distortion-reducing series resistance. Consider what it is supposed to do. It is the dynamic sound transducer that produces distorted voltage. This distorted voltage should not be converted into distorted current that then flows through the voice coil... so an insulator is needed on the terminal of the dynamic transducer.

Even a poor insulator with 20R reduces all "harmonics" by 10dB. As long as you don't interfere with this insulator while insulating, you can do anything.

In your case you may add an RLC in series.
But don't disturb the insulator ;)
 
Last edited:
Today listening test with fullrange drivers which have a Faraday ring and paper cones sandwiched with aluminum foil in a marble box, reflex loaded to 35hz.

Two 8 ohm drivers in parallel (=3 to 4 ohms) with in line 2.5 ohm resistor and 1.5 microfarad in parallel.

Small BTL TDA2003 Mono amplifier.

Clearly this resistance lowers distortion audibly.

Without resistance it sounds "harder" and more distorted.

IMG_20231012_185906.jpg
IMG_20231012_185941.jpg
 

Attachments

  • IMG_20231012_190910.jpg
    IMG_20231012_190910.jpg
    195.1 KB · Views: 38
  • IMG_20231012_190920.jpg
    IMG_20231012_190920.jpg
    272.9 KB · Views: 40