LM3875, this work by Joe came after the current amps i listened too, and were vetted by Daniel, the designer of the variable transamps i got to play with (and that built my T/S current amp for measurement.
Joe just had the guts to publish a similar diy project. And a simple one so people could try it. Does need to be revised to use LM3886 too, althou conversion from voltage to current amp is all in the feedback loop AFAIK.
dave
The key to using current amps (or SETs) is getting a speaker that has an appropriate impedance. There are vanishingly few of those on the current commercial market, most of the applicable ones are diy.
dave
This is very off-topic, but so is most of this thread, as one can make a decent voltage output with valves and a decent current output with transistors:
Using DSP techniques (Ian Hegglun has some ideas to even do it without those), you could in principle make an amplifier that measures the loudspeaker impedance curve (at power-on, or when you press the calibrate button) and automatically sets an equalizer with the reciprocal response. You could even automatically have it extract the poles from the impedance, cover those with zeros and place new poles at desired locations, that is, have it automatically apply a kind of Linkwitz transform.
With such a set-up, one could make a current drive amplifier that works with normal loudspeakers without the sound getting boomy.
Using DSP techniques (Ian Hegglun has some ideas to even do it without those), you could in principle make an amplifier that measures the loudspeaker impedance curve (at power-on, or when you press the calibrate button) and automatically sets an equalizer with the reciprocal response. You could even automatically have it extract the poles from the impedance, cover those with zeros and place new poles at desired locations, that is, have it automatically apply a kind of Linkwitz transform.
With such a set-up, one could make a current drive amplifier that works with normal loudspeakers without the sound getting boomy.
I guess you mean an LRC series network in parallel with an LR series network to flatten the impedance around resonance and for high frequencies.
With a circuit like that and a high amplifier output impedance, the loudspeakers are still driven from a higher impedance than with pure voltage drive, particularly in the flat part of their impedance characteristic, so distortion is likely to be somewhere in between the pure voltage and pure current drive cases.
Around the fundamental resonance, the impedance of the impedance correction network will be much lower than the loudspeaker's impedance, so around resonance you essentially have voltage drive. I'm not sure if that is bad, though, because around resonance the loudspeaker's impedance mainly comes from the back-EMF. That is, around resonance, the voltage across the loudspeaker is a measure for the speed of the voice coil. At other frequencies, it is just the product of the current and the voice coil impedance, which has nonlinearities due to magnetic effects and resistance changes with self-heating.
The argument usually given for current drive is that the force on the cone is proportional to the current, with Bl as factor of proportionality. Driven with voltage at frequencies far from resonance, the voltage is first converted to current by the voice coil, which is not perfectly linear, and then the current drives the loudspeaker. So far so good.
Things are more complicated around resonance, though. Around resonance, the voltage across the loudspeaker is Bl times the velocity and the force is Bl times the current. Do you want to control the force or the velocity? The velocity has a more direct relation with the acceleration that determines the sound pressure. Or is it best to have a driving impedance that is close to the loudspeaker's impedance, so the Bl dependencies cancel? With current drive, force goes up with Bl, while with voltage drive, velocity goes down with Bl, so there must be some cancellation when the driving impedance is somewhere in between 0 and infinity.
Things are more complicated around resonance, though. Around resonance, the voltage across the loudspeaker is Bl times the velocity and the force is Bl times the current. Do you want to control the force or the velocity? The velocity has a more direct relation with the acceleration that determines the sound pressure. Or is it best to have a driving impedance that is close to the loudspeaker's impedance, so the Bl dependencies cancel? With current drive, force goes up with Bl, while with voltage drive, velocity goes down with Bl, so there must be some cancellation when the driving impedance is somewhere in between 0 and infinity.
There were FRs made in Germany (Eat at the time) that had close to ruler flat impedance ie Qm near or less than 1.
Evertime i see i see a new FR or driver i look at that Qm number. You can get some with Qm < 2 but i’ve not seen any < 1. Any of those, with appropriate enclosure, would be candistaes for use with current amplifiers.
dave
Evertime i see i see a new FR or driver i look at that Qm number. You can get some with Qm < 2 but i’ve not seen any < 1. Any of those, with appropriate enclosure, would be candistaes for use with current amplifiers.
dave
Those are sought after by SET afficionados for a reason 😉 How about Dynaco A20, it's a-periodic enclosure as well (lots around for grabs). Are you familiar with the 800 ohms (EL86) OTLs Dave? That's current driven too then.
Current drive is best into low impedance loads such as ribbons. Total power dissipation and device distortion becomes exponentially improved when voltages approach nothing and gain approaches zero.
If designing a dynamic driver from the ground up for a current source amp, I think it would be ideal to make it as close to zero ohms as possible.
The ideal load for a voltage source is infinite, the ideal load for a current source is zero. It's a lot easier to get closer to zero than infinite when working with current driven devices. In fact, infinite would never work for electromagnetic drivers because of the fact that the devices are current driven, so zero is the only ideal you can strive toward. Currently the industry mucks around somewhere in between, in nomans land. It never made sense to me.
If designing a dynamic driver from the ground up for a current source amp, I think it would be ideal to make it as close to zero ohms as possible.
The ideal load for a voltage source is infinite, the ideal load for a current source is zero. It's a lot easier to get closer to zero than infinite when working with current driven devices. In fact, infinite would never work for electromagnetic drivers because of the fact that the devices are current driven, so zero is the only ideal you can strive toward. Currently the industry mucks around somewhere in between, in nomans land. It never made sense to me.
Last edited:
Yes, i have had a few of the 800Ω jobbies thru here. Not quite an OTL, the ones i had had special OPTs.
I have also seen lots of the direct drive single ended ESL tweeter sthe Germans liked to make. Tapped right off the anode of the EL84.
dave
I have also seen lots of the direct drive single ended ESL tweeter sthe Germans liked to make. Tapped right off the anode of the EL84.
dave
I don’t think that has to do with low impedance, but that ribbons typically have very flat impedance.
dave
It will probably depend on the output impedance of the amplifier (optimal infinite Z out). I reasoned the advantage (among many disadvantages) might be in the relative stronger cohesion of impedances with high impedance loudspeakers. But that might be a pure theoretical idea as the bigger units a re hard to obtain.
what is the threshold?
thevenin and norton's theorems are such that the two have an equivalency, where do we cross the line?
The theory says that the line for the purposes of classification when the speaker impedance = the amplifiers output impedance (damping factor >1. But as in anything that is really a continuumeffect start showing up when you get close to the line (ie typical SETs).
An analogy would be the transition of a bass reflex box into a mass-loaded TL (by stretching one of the dimensions (and keeping volume constant)) where it is hard to draw a line. A hybrid as it were with character of both if you are near the line.
In one amp we are trying to keep the voltage flat (modulating current), the other the current (modulating voltage).
dave
An analogy would be the transition of a bass reflex box into a mass-loaded TL (by stretching one of the dimensions (and keeping volume constant)) where it is hard to draw a line. A hybrid as it were with character of both if you are near the line.
In one amp we are trying to keep the voltage flat (modulating current), the other the current (modulating voltage).
dave
A number of old tube amp designs provided a control to set the output Z (using a current sensing feedback with a low Ohm pot in the speaker return path) so that critical speaker damping could be set. Not hard to include in a design. Could cover the whole range from Vout to Iout control with an extra pot for V feedback adjustment too.
Last edited:
Much relieved.
Seems the discussion became much more substantive overnight (it is AM here now). I am fine with disagreements, but I also see areas of agreement.
The topic is basically that amplifiers sound different? I believe that they do, but some say they all virtually sound the same. That is not what I hear.
What I am suggesting is that there may be a way to find an explanation(s) for that.
Thank you Scott. I actually agree with that post, but that 'evidence' will be published alright, but it will be done in cooperation and vetting of those with whom I am collaborating. They will see the 'evidence' first. This is not the forum for that. All t's will be crossed and all i's will be dotted. I hope you can appreciate that. In time, everybody will be able to have their say on it. But there is a process to be followed first.
Thank you for clarifying my position on current drive. I get shoved into that camp whenever I bring up the topic of current and dynamic drivers. My position is different from Esa's (that Finnish guy). But I believe that current drive does reveal some valuable things. In fact, I have gradually formed the view that there is no such thing as voltage drive. Maybe we should call it an attempt at voltage steering versus current steering. Lynn Olson said something that made me think along those ways. But please note, the position is very different when dealing with electrostatic speakers.
As for mixed-mode, this can be done (sort of) with a simple resistor and a voltage source, because in a crude way (wasted energy) it achieves the same thing: A lower current phase angle. If the theory is correct, then this will reduce the induced current distortions of the speaker. Go back to the 'two sine waves' in Post #152, and those two sine waves will start to get closer in time. This is because you have limited the current phase angle and you have limited the amplifier's current response.
The parallel resistor has been heard by so many, Richard Marsh on blowtorch said the same thing. Even a series resistor does it to some extent, except we now need more voltage swing, where before we needed more current. The series resistor trick was performed by Menno Vanderveen in a room with a number of people at ETF17 and I was there. The series resistance was 18 Ohm (actually the source impedance of the Hypex amplifier was modified) and all in the room heard a favourable improvement, like a Class D amplifier sounding more tube like? Maybe not quite, but something was going on.
But I would use an 8-15 Ohm parallel resistor? Even one of those cheap 10W ceramic resistors is worth a try.
Here is something almost everyone here could try - this is after all DIY, right?
Please consider these things without instant dismissal, that is all I ask for. Make any reply substantive and nobody will get upset. Not me at least. 🙂
Thank you.
Cheers, Joe
Seems the discussion became much more substantive overnight (it is AM here now). I am fine with disagreements, but I also see areas of agreement.
The topic is basically that amplifiers sound different? I believe that they do, but some say they all virtually sound the same. That is not what I hear.
What I am suggesting is that there may be a way to find an explanation(s) for that.
Thank you Scott. I actually agree with that post, but that 'evidence' will be published alright, but it will be done in cooperation and vetting of those with whom I am collaborating. They will see the 'evidence' first. This is not the forum for that. All t's will be crossed and all i's will be dotted. I hope you can appreciate that. In time, everybody will be able to have their say on it. But there is a process to be followed first.
Thank you for clarifying my position on current drive. I get shoved into that camp whenever I bring up the topic of current and dynamic drivers. My position is different from Esa's (that Finnish guy). But I believe that current drive does reveal some valuable things. In fact, I have gradually formed the view that there is no such thing as voltage drive. Maybe we should call it an attempt at voltage steering versus current steering. Lynn Olson said something that made me think along those ways. But please note, the position is very different when dealing with electrostatic speakers.
As for mixed-mode, this can be done (sort of) with a simple resistor and a voltage source, because in a crude way (wasted energy) it achieves the same thing: A lower current phase angle. If the theory is correct, then this will reduce the induced current distortions of the speaker. Go back to the 'two sine waves' in Post #152, and those two sine waves will start to get closer in time. This is because you have limited the current phase angle and you have limited the amplifier's current response.
The parallel resistor has been heard by so many, Richard Marsh on blowtorch said the same thing. Even a series resistor does it to some extent, except we now need more voltage swing, where before we needed more current. The series resistor trick was performed by Menno Vanderveen in a room with a number of people at ETF17 and I was there. The series resistance was 18 Ohm (actually the source impedance of the Hypex amplifier was modified) and all in the room heard a favourable improvement, like a Class D amplifier sounding more tube like? Maybe not quite, but something was going on.
But I would use an 8-15 Ohm parallel resistor? Even one of those cheap 10W ceramic resistors is worth a try.
Here is something almost everyone here could try - this is after all DIY, right?
Please consider these things without instant dismissal, that is all I ask for. Make any reply substantive and nobody will get upset. Not me at least. 🙂
Thank you.
Cheers, Joe
Last edited:
I got thinking about this subject and the ACA.
Nelson says ACA has damping factor 3 which means, i assume 8/3 ~ 2.5Ω output impedance, twice that in bridge mode, half in parallel, putting this amp into the same class as a typical SET, from the lowish output impedance to the highest, so a cheap way to play.
And given the number of kits sold (multiply that by many), there are a lot of people out there with these SS “SETs”
And it msy explain why driving Tony’s woofers in the OB sounded so good. The interaction with the amps output impedance at the impedance rise will be compensating for at least some of the natural OB roll-off.
I 1st started thinking about the ACAs when in bridge mode, they did not drive the TLs well in the bass… it wa sdirectly compared to Tom’s (really good) 200w multi-LM3886 amp. It’s output impedance was, IMO, a bit low and made the speakers sound just a bit too damped (still jaw dropping). 1 ACA works fine with them, have yet to try in parallel.
dave
Nelson says ACA has damping factor 3 which means, i assume 8/3 ~ 2.5Ω output impedance, twice that in bridge mode, half in parallel, putting this amp into the same class as a typical SET, from the lowish output impedance to the highest, so a cheap way to play.
And given the number of kits sold (multiply that by many), there are a lot of people out there with these SS “SETs”
And it msy explain why driving Tony’s woofers in the OB sounded so good. The interaction with the amps output impedance at the impedance rise will be compensating for at least some of the natural OB roll-off.
I 1st started thinking about the ACAs when in bridge mode, they did not drive the TLs well in the bass… it wa sdirectly compared to Tom’s (really good) 200w multi-LM3886 amp. It’s output impedance was, IMO, a bit low and made the speakers sound just a bit too damped (still jaw dropping). 1 ACA works fine with them, have yet to try in parallel.
dave
- Status
- Not open for further replies.