Depends. Speakers and their xover filters are designed so that they get the most flat and best response when fed from a voltage source. Now, if you deliberately introduce a varying series impedance in the link between amp and speaker (because that's what you do effectively), all bets on the response flatness and damping are off.
It may be different if you design a speaker from the ground up with combined voltage/current drive, but I doubt that this would be an ideal DIY project.
Jan Didden
It may be different if you design a speaker from the ground up with combined voltage/current drive, but I doubt that this would be an ideal DIY project.
Jan Didden
plus ca change...
This was all the rage back in the '50s. There are quite a few articles in Audio Engineering about it, though using tubes and not quantified in the way Mills and Hawksford did. It can be a useful, if inefficient, technique for warming up overly tight-sounding woofer systems.
This was all the rage back in the '50s. There are quite a few articles in Audio Engineering about it, though using tubes and not quantified in the way Mills and Hawksford did. It can be a useful, if inefficient, technique for warming up overly tight-sounding woofer systems.
There is nothing to argue. Electrodynamic speaker must be supplied from voltage output to maintain flat amplitude characteristic and appropriate transient response. Driving electrodynamic speaker from current output is a total nonsense. There are enough of simplified speaker circuits based on electro-mechanical analogy, I hope.
I do partially disagree with PMA.
You do of course need a voltage source to achieve proper damping around a driver's fundamental resonance, unless you are using MFB (which can be achieved more easily whith current drive).
But above resonance things are different.
Current drive will:
-give less nonlinear distortion.
-eliminate the effect of Lvc.
-completely eliminate VC - heatup releted SPL loss.
I do not claim that it is easily applied, but for a cleverly designed active speaker it would have it's merits.
Regards
Charles
You do of course need a voltage source to achieve proper damping around a driver's fundamental resonance, unless you are using MFB (which can be achieved more easily whith current drive).
But above resonance things are different.
Current drive will:
-give less nonlinear distortion.
-eliminate the effect of Lvc.
-completely eliminate VC - heatup releted SPL loss.
I do not claim that it is easily applied, but for a cleverly designed active speaker it would have it's merits.
Regards
Charles
Charles,
you will probably agree with attached speaker's simplified schematics. From this frequency characteristics under resonance, between resonance and critical frequency and above critical frequency are derived. Now what happens, if we increase R' to infinity (or very big value)? This will mean in fact current drive of the speaker.
Pavel
you will probably agree with attached speaker's simplified schematics. From this frequency characteristics under resonance, between resonance and critical frequency and above critical frequency are derived. Now what happens, if we increase R' to infinity (or very big value)? This will mean in fact current drive of the speaker.
Pavel
Attachments
phase_accurate said:
-give less nonlinear distortion.
Why?
-eliminate the effect of Lvc.
What about the xover filter? The Lvc is part of that. Again, all bets on the result are off.
-completely eliminate VC - heatup releted SPL loss.
Correct. Is this significant enough in hi-fi apps to warrant a complete strategic redesign of the whole system?
Jan Didden
Hi Jan
To me (which is of course a matter of taste), decent speakers are always active speakers (with line level-crossovers).
It wouldn't make much sense to use current drive on passive ones.
This way, my statement about Lvc might make more sense to you.
The distortion reduction comes from the elimination of (B*L)^2 in the denominator that is there for the voltage to velocity transfer function but is absent for the current to velocity transfer function of a driver.
I have never attempted to figure out the actual differences though.
Pavel
I see what you mean but you have to take care not to make a mistake.
A current source presents indeed an infinite output resistance, but compared to just increasing the Rdc of a driver (or any resistor connected in series) that is connected to an ordinary amplifier, a real current source wouldn't reduce efficiency in such a dramatic way.
Moreover, current drive would even make Rdc "invisible".
Regards
Charles
To me (which is of course a matter of taste), decent speakers are always active speakers (with line level-crossovers).
It wouldn't make much sense to use current drive on passive ones.
This way, my statement about Lvc might make more sense to you.
The distortion reduction comes from the elimination of (B*L)^2 in the denominator that is there for the voltage to velocity transfer function but is absent for the current to velocity transfer function of a driver.
I have never attempted to figure out the actual differences though.
Pavel
I see what you mean but you have to take care not to make a mistake.
A current source presents indeed an infinite output resistance, but compared to just increasing the Rdc of a driver (or any resistor connected in series) that is connected to an ordinary amplifier, a real current source wouldn't reduce efficiency in such a dramatic way.
Moreover, current drive would even make Rdc "invisible".
Regards
Charles
, it is difficult to find better reminiscence As already mentioned, you will loose control over the driver's motion around fs. So you have to take measures to avoid this. One is the implementation of MFB, the other one is a frequency dependant mixed situation (i.e. voltage and current feedback).
If you apply the principle to a passive multiway speaker the resulting FR deviations will be even worse.
Regards
Charles
If you apply the principle to a passive multiway speaker the resulting FR deviations will be even worse.
Regards
Charles
phase_accurate said:
I was thinking of the voltage gain as a function of frequency. Due to the inductance, the speaker impedance becomes infinetely high as frequency increases. This means the amp will amplify RF noise as much as it can. To avoid this, you need to bring the HF load down. The easiest way is to connect the series combination of a resistor and a capacitor in parallel to the speaker.
I believe this is called a Zobel network in English. The German term should be "Boucherot-Glied".
Regards,
Eric
Some years ago, I had an idea that I would like to share with you. I never tried it, but maybe someone who is good at digital signal processing might want to have a go at it.
Normal loudspeaker boxes are designed to have a more or less flat response under voltage drive, but the distortion decreases when they are current driven, according to Mills and Hawksford. (I am actually not sure if their arguments are valid near the resonance frequency, but that's a different matter.)
The frequency response with current drive is simply the product of the impedance curve and the response with voltage drive, as the voltage which will develop across the loudspeaker terminals is the product of the current and the loudspeaker impedance.
Suppose you make an amplifier with a current output which has a self-adjusting equaliser built in. The equaliser measures the impedance curve of the loudspeaker after switch-on and adjusts its response to be the reciprocal of the loudspeaker impedance curve. This way, it is possible to drive a normal loudspeaker box designed for voltage drive with current, while still getting the normal frequency response and reduced distortion.
For loudspeaker boxes which behave as lumped systems at low frequencies, such as closed boxes and bass reflex boxes, it is even possible to compensate for a box that is boomy under voltage drive, or to extend the bass response. In this case, put the measured impedance curve in a pole-zero extraction algorithm. As a linear time-invariant lumped system has only one characteristic polynomial, the poles of the impedance (current to voltage transfer) are also the poles of the current to sound pressure transfer. Adjust the equaliser such that they are covered by equaliser zeroes and let the equaliser add an equal number of poles in desired locations (Butterworth, for example). This way, you can let the amplifier taylor the bass response, automatically compensating for boomy boxes or automatically extending the bass response if you press the right button on the front panel.
Normal loudspeaker boxes are designed to have a more or less flat response under voltage drive, but the distortion decreases when they are current driven, according to Mills and Hawksford. (I am actually not sure if their arguments are valid near the resonance frequency, but that's a different matter.)
The frequency response with current drive is simply the product of the impedance curve and the response with voltage drive, as the voltage which will develop across the loudspeaker terminals is the product of the current and the loudspeaker impedance.
Suppose you make an amplifier with a current output which has a self-adjusting equaliser built in. The equaliser measures the impedance curve of the loudspeaker after switch-on and adjusts its response to be the reciprocal of the loudspeaker impedance curve. This way, it is possible to drive a normal loudspeaker box designed for voltage drive with current, while still getting the normal frequency response and reduced distortion.
For loudspeaker boxes which behave as lumped systems at low frequencies, such as closed boxes and bass reflex boxes, it is even possible to compensate for a box that is boomy under voltage drive, or to extend the bass response. In this case, put the measured impedance curve in a pole-zero extraction algorithm. As a linear time-invariant lumped system has only one characteristic polynomial, the poles of the impedance (current to voltage transfer) are also the poles of the current to sound pressure transfer. Adjust the equaliser such that they are covered by equaliser zeroes and let the equaliser add an equal number of poles in desired locations (Butterworth, for example). This way, you can let the amplifier taylor the bass response, automatically compensating for boomy boxes or automatically extending the bass response if you press the right button on the front panel.
If you employ DSP then you can also adjust the final acoustical response instead of just the response as a function of diver impedance.
Then you will end up with something like that:
http://www.tactaudio.com/RCS22X/index.html
I can confirm that it is amazing.
Regards
Charles
Then you will end up with something like that:
http://www.tactaudio.com/RCS22X/index.html
I can confirm that it is amazing.
Regards
Charles
Phase_accurate, do these things also have an amazingly small sweet spot? After all, the transfer from loudspeaker to listener will be different depending on the listener location.
Anyway, an advantage of using the loudspeaker impedance is simply that you don't have to put a measurement microphone anywhere. To the user, the amplifier simply behaves as a normal amplifier, except for the funny noises produced during calibration after switch-on (tone sweep, maximum length sequence noise, whatever), and the reduced distortion and improved bass. Perhaps you could make a combined version, which uses a measurement microphone if connected and otherwise uses the impedance curve.
Anyway, an advantage of using the loudspeaker impedance is simply that you don't have to put a measurement microphone anywhere. To the user, the amplifier simply behaves as a normal amplifier, except for the funny noises produced during calibration after switch-on (tone sweep, maximum length sequence noise, whatever), and the reduced distortion and improved bass. Perhaps you could make a combined version, which uses a measurement microphone if connected and otherwise uses the impedance curve.
MarcelvdG said:
it is an interesting concept but its implementation could be probamatic. For example, speaker's impedence changes with signal (frequency is an obivious one).
so you will have to test over a full spectrum of sound. Now, when you get a non-sine signal, you need to do a quite FFT (which has its own shortfalls), figure out relative weight of each harmonics, and come up with a weighted gain for that particular composition of harmonics.
and so on and so forth.
a little bit too complicated, in my view.
Konnichiwa,
Isn't it obvious?
You have a coil filled with solid Iron. Eddy losses give rise to a cubic distortion with applied voltage, for voltage feed, as the current is distorted. Current Feed, no distortion (from this source).
Well, I believe current feed is applicable ONLY for active speakers (or fullrange), but again, this should be obvious.
ABSOLUTELY.
One UK studio mag tested a number of "HiFi" and "Studio" monitor designs of the classic "small 2-Way" (aka LS3/5 & grown up company). They found at rated input power 5 - 6db compression and 10 - 20% distortion (3rd harmonics - eddy current created) in the midband for the socalled "HiFi" Speakers.
In fact, if we sit down and instead of slavishly copying again and again the same fundamental mistakes made by virtually ALL Designers of HiFi Gear (such as passive crossover speakers in reflex enclosures, with **** poor radiation pattern control and riddeled with high distortion and severe compression due to a voltage interface), we can actually sit down and design a system that ACTUALLY works.
May I propose a few ideas?
A lot of what ME Gaithein does makes eminent sense. They use a "pseudo coaxial" driver with a modest "open baffle" fitted with a 5.25" Midrange and 1" Tweeter, mounted in front of a 15" Woofer.
How about using as driver the Seas 6.5" Coaxial with the XP Cone, on a modestly wide open baffle and driven via current drive (both tweeter dome and lf cone)? With a Qm of around 1.3 and an Fs around 40Hz this will do nicely as extended midrange driver with co-incident tweeter on an open baffle.
Also, the combination of open baffle and hornloading the tweeter with the cone produced an excellent dispersion control (the exact opposite of the stupid "wide dispersion" designs so commonly turned out by Idiots as "HiFi" speakers).
Depending upon the exact baffle size the crossover between cone and large "woofer" would likely fall between 250-500Hz if we used the semi-coaxial but can probably be pushed down to 125Hz with a suitable width baffle (~ 20" wide). At such a low frequency we can place the 15" Woofer below the main driver.
The woofer now could be a dipole or a "cardiod" system with suitable flow resistances (or by adding a 15" dipole and a modest size monopole woofer) and again use a low Qm driver with current feed. If a low Qm driver is not available off shelf we make one by packing the area directly behind teh driver with a "flow resistance", lowering Qm untill we can use current drive.
Now the resultant system should play loud without strain, distortion or compression, work well in normal small rooms and do all of this for a quite piddly budget, compared to many "Super High End" projects.
Sayonara
janneman said:
Isn't it obvious?
You have a coil filled with solid Iron. Eddy losses give rise to a cubic distortion with applied voltage, for voltage feed, as the current is distorted. Current Feed, no distortion (from this source).
janneman said:
Well, I believe current feed is applicable ONLY for active speakers (or fullrange), but again, this should be obvious.
janneman said:
ABSOLUTELY.
One UK studio mag tested a number of "HiFi" and "Studio" monitor designs of the classic "small 2-Way" (aka LS3/5 & grown up company). They found at rated input power 5 - 6db compression and 10 - 20% distortion (3rd harmonics - eddy current created) in the midband for the socalled "HiFi" Speakers.
In fact, if we sit down and instead of slavishly copying again and again the same fundamental mistakes made by virtually ALL Designers of HiFi Gear (such as passive crossover speakers in reflex enclosures, with **** poor radiation pattern control and riddeled with high distortion and severe compression due to a voltage interface), we can actually sit down and design a system that ACTUALLY works.
May I propose a few ideas?
A lot of what ME Gaithein does makes eminent sense. They use a "pseudo coaxial" driver with a modest "open baffle" fitted with a 5.25" Midrange and 1" Tweeter, mounted in front of a 15" Woofer.
How about using as driver the Seas 6.5" Coaxial with the XP Cone, on a modestly wide open baffle and driven via current drive (both tweeter dome and lf cone)? With a Qm of around 1.3 and an Fs around 40Hz this will do nicely as extended midrange driver with co-incident tweeter on an open baffle.
Also, the combination of open baffle and hornloading the tweeter with the cone produced an excellent dispersion control (the exact opposite of the stupid "wide dispersion" designs so commonly turned out by Idiots as "HiFi" speakers).
Depending upon the exact baffle size the crossover between cone and large "woofer" would likely fall between 250-500Hz if we used the semi-coaxial but can probably be pushed down to 125Hz with a suitable width baffle (~ 20" wide). At such a low frequency we can place the 15" Woofer below the main driver.
The woofer now could be a dipole or a "cardiod" system with suitable flow resistances (or by adding a 15" dipole and a modest size monopole woofer) and again use a low Qm driver with current feed. If a low Qm driver is not available off shelf we make one by packing the area directly behind teh driver with a "flow resistance", lowering Qm untill we can use current drive.
Now the resultant system should play loud without strain, distortion or compression, work well in normal small rooms and do all of this for a quite piddly budget, compared to many "Super High End" projects.
Sayonara
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.