Hello,
Here is what I feel: The loudspeaker (LS) impedance being the culprit, We know that the power delivered into the LS increases with frequency if we have a feedback loop that uses current flowing through the LS and power delivered into the LS decreases with frequency if we have a feedback loop that uses voltage across the LS.
If we are interested in output POWER that is dependent on input signal only, then we need to feed the product of voltage of across the LS and current through the LS back to complete the loop.
An analog multiplier of some sort needs to come in the picture.
By the modern digital way of looking at it needs us to feed the impedance plot of a particular LS into a digital signal processor (DSP) and the DSP will then output the amplifier drive signal by 'looking up' the latest incoming audio sample and 'computing' it with the impedance plot. Needless to say the processing goes into frequency domain, the amplifier then only needs to be of the driven type.
i.e. the amplifiers then, will neither need to have any feedback or feed-forward correction. It then merely needs to power amplify the drive signal.
Since a DSP is anyway coming in the play, it can also compensate for the amplifier part of the signal chain on a per channel basis. Just the way we do it in industrial controls.
May be this could result in a couple of patents. Maybe not.
Maybe I misinterpreted the requirements of this thread altogether.
With a DSP based approach, you can only modify frequency response.
What current drive does is change the wave shape of the voltage produced by the amplifier. Thus, non-linearities of the driver can be partially corrected, dixit Hawksford.
In other words, with a DSP you can straighten out the FR of a loudspeaker, with current drive, you ideally kill part of the driver distortion.
What current drive does is change the wave shape of the voltage produced by the amplifier. Thus, non-linearities of the driver can be partially corrected, dixit Hawksford.
In other words, with a DSP you can straighten out the FR of a loudspeaker, with current drive, you ideally kill part of the driver distortion.
Exactly. Specifically the distortion caused by the back emf playing into a short. I have captured this distorted waveform and posted screen shots of the distorted current earlier in this thread.
DSP could solve the current distortion IF there were a shunt being sampled by the DSP. The caveat is that the DSP would be "behind" by the amount of delay in the A/D and D/A chain as well as the delay in the DSP itself. If it is a USB connected DSP, then you add the Windows USB buffering delay etc etc etc.
Current drive for LS
Well, yes. it is changing the frequency response in a way.
In a current drive, however, we would need to reduce Vin with increasing frequency, in order to maintain output power proportional to input signal only.
This reduction in Vin will lower distortion further that what the current drive is inherently doing.
Well, yes. it is changing the frequency response in a way.
In a current drive, however, we would need to reduce Vin with increasing frequency, in order to maintain output power proportional to input signal only.
This reduction in Vin will lower distortion further that what the current drive is inherently doing.
The output rise that occurs due to impedance rise at high freq can be reduced by a zobel. No DSP required.
I have not tried a capacitor at the output. What would be the reason to try this? Much of my testing is a speaker at Fs where phase equals zero degrees.
just a personal experience i had with the kit from ebay...as i mentioned previously i use it for mid/tweeter in biamped system and it is cap coupled (and sounds sublime)
when playing with various fullrange drives and this kit, they always sounded better with cap, that's all
when playing with various fullrange drives and this kit, they always sounded better with cap, that's all
this is easy current amp: http://www.diyaudio.com/forums/solid-state/189766-square-law-class-amps-2.html#post2956849
KGRlee (#42) said: "The type of 'normal' speaker most likely to show improved audible benefits are small high efficiency full-range drive units in big sealed boxes. These will have little bass and often poor HF too."
I am new (very new) to idea of negative impedance. Also I am near-zero in h/w skills 🙂
I cite the above text because it'd be fascinating to know how a current source would work with my preferred speakers (are you sitting down?) -- Bose 901 series I or II. While they surely don't meet the aurum eardrum criteria 🙂 they would be the perfect sample as (sans EQ) they need help on the low and high ends.
For now, my interest is in current drive for my (are you still sitting down?) Behringer iNuke NU3000 driving a DIY 2x15" sub.
You may now take a deep breath, and change your underwear 🙂
I am new (very new) to idea of negative impedance. Also I am near-zero in h/w skills 🙂
I cite the above text because it'd be fascinating to know how a current source would work with my preferred speakers (are you sitting down?) -- Bose 901 series I or II. While they surely don't meet the aurum eardrum criteria 🙂 they would be the perfect sample as (sans EQ) they need help on the low and high ends.
For now, my interest is in current drive for my (are you still sitting down?) Behringer iNuke NU3000 driving a DIY 2x15" sub.
You may now take a deep breath, and change your underwear 🙂
Practical current drive is where amplifier output impedance is much greater than what you are driving. A perfect current source has infinite output impedance.
A negative output impedance amp is the opposite. Some people use it to remove the speaker cable impedance. This would lower the apparent Qt of the cable/speaker combination.
Current driving you Bose 901 would give a bump at Fs of the speaker. It would also give you an increase at high freq.Do you have an impedance plot of your 901?
My experience with a current drive amp driving a naked speaker is that the amp clips at high freq.
A negative output impedance amp is the opposite. Some people use it to remove the speaker cable impedance. This would lower the apparent Qt of the cable/speaker combination.
Current driving you Bose 901 would give a bump at Fs of the speaker. It would also give you an increase at high freq.Do you have an impedance plot of your 901?
My experience with a current drive amp driving a naked speaker is that the amp clips at high freq.
Here is a plot of the PE replacement driver. Fs about 89 Hz. You will get LOTS of 89 Hz with current drive. I would not call this bass. Actually, in a box, Fs will increase, so you will get lots of something higher than 89 Hz with current drive.
Last edited:
Just EQ it to target and everthing's fine, no matter what ouput impedance your amp is presenting to the speaker. Proper EQing is the key to any succesful woofer/speaker design. People who refuse to EQ and rather restrict driver and cabinet choices to things that "work without EQ" are missing a lot of chances.
EQing a BR (high Q) impedance peak can result in an "instable" system as power compression will change it.
I completly agree with the principle of EQing the driver to flat, whatever the box/amp/horn, but I think current drive is best to be used with closed box.
I completly agree with the principle of EQing the driver to flat, whatever the box/amp/horn, but I think current drive is best to be used with closed box.
With current drive there is no power compression anymore. But generally agreed, don't EQ sharp resonances unless they were absolutely linear and time-invariant.... which they are not.
Yep, closed box is well suited for current drive... but pure current drive might not be the best strategy anyway, down at low frequencies around any system resonance. It all depends how the specific driver reacts when it is pushed hard or even overdriven and you can control / fine-tune that behavior somewhat by drive impedance.
Last edited:
Bose 901 Mk1 is an example of small high efficiency units in small boxes.
Current drive will give it better HF but the extra bass will be at frequencies higher than 100Hz so will sound VERY boomy. EQ will be necessary.
901 Mk2 & later are ported (reflex) speakers. Current Drive is inappropriate even with fancy EQ. To obtain the same advantages as current drive with BR, you need to do even fancier stuff.
AES E-Library Synthesis of Loudspeaker Mechanical Parameters by Electrical Means: A new Method for Controlling Low-Frequency Loudspeaker Behaviour
Compression is a problem with this but there are ways to get around this.
Last edited:
I would imagine that current-drive could also work with a dipole sub.
Maybe I'll test this with my dipole sub.
If I found time...
...and drive 😀
Maybe I'll test this with my dipole sub.
If I found time...
...and drive 😀
Sound tests
Has anybody tried pure current drive with a fullrange speaker in a sealed box?
Can they share some of their experiences please?
What happens to the Bass? Midrange? Treble? Dialogue, Music, Effects? as compared to a voltage drive in an A/B test?
Is it best to be used as a subwoofer and a satellite, separately or as a fullrange driver?
Thanks in advance!
Has anybody tried pure current drive with a fullrange speaker in a sealed box?
Can they share some of their experiences please?
What happens to the Bass? Midrange? Treble? Dialogue, Music, Effects? as compared to a voltage drive in an A/B test?
Is it best to be used as a subwoofer and a satellite, separately or as a fullrange driver?
Thanks in advance!
As many working with speakers may have noticed, padding down a tweeter can make it sound sweeter. Since I am presently developing a series of active speakers, normally one would not pad down any of the drivers.
Curious to find out whether that would nonetheless be a good idea, I decided to do some measurements today. After all, putting a resistor in series with a tweeter makes it partially current driven, so according to theory and observation, this should drive distortion down..
Time for some bloody evidence.
The brown curve is a 4 Ohm tweeter with a 10 Ohm resistor in series, the blue curve without. In both measurements, the input to the driver was adjusted to provide approximately the same acoustic output.
In the critical band between 2 and 5K, the distortion is appreciably lower with partial current drive, up to 15 dB improvement. However, there is a cross over point, where partial current drive provides more distortion. But frankly, this doesn't bother me to much, because at this point, the distortion products are at the verge of ultrasonics anyways.
Another effect that is obvious, is the 6 dB rise between 10 and 20 K. The reason is that this tweeter has no shortening ring and thus has a fairly high inductance. It was engineered to be straight with voltage drive. Easily remedied by putting a small cap in parallel with the tweeter.
The difference in distortion is mainly in the third harmonic.
Note: the distortion figures are normalized to a 0dB signal, as those familiar with Holm impulse will know.
Curious to find out whether that would nonetheless be a good idea, I decided to do some measurements today. After all, putting a resistor in series with a tweeter makes it partially current driven, so according to theory and observation, this should drive distortion down..
Time for some bloody evidence.
The brown curve is a 4 Ohm tweeter with a 10 Ohm resistor in series, the blue curve without. In both measurements, the input to the driver was adjusted to provide approximately the same acoustic output.
In the critical band between 2 and 5K, the distortion is appreciably lower with partial current drive, up to 15 dB improvement. However, there is a cross over point, where partial current drive provides more distortion. But frankly, this doesn't bother me to much, because at this point, the distortion products are at the verge of ultrasonics anyways.
Another effect that is obvious, is the 6 dB rise between 10 and 20 K. The reason is that this tweeter has no shortening ring and thus has a fairly high inductance. It was engineered to be straight with voltage drive. Easily remedied by putting a small cap in parallel with the tweeter.
The difference in distortion is mainly in the third harmonic.
Note: the distortion figures are normalized to a 0dB signal, as those familiar with Holm impulse will know.
Last edited: