And, because of that, we are starting to see some light shining thru the hegemony of the "crazy impedance/voltage amp". Much of that hegemony a side effect of competing in a mass-market world. With smaller, more nimble vendors, and more direct routes to the end user (ie diy, interent), it is no longer necessary to toe the CIVA line -- "everyone else does it this way, i guess i have too". One is free to explore turns not taken.
I got the 1st taste of input re the myth of high damping factors from an old WE engineer that was horrified at the thot that an amplifier's output impedance wasn't equal to the loudspeakers.
dave
It is certainly self-evident that the force on a voice-coil is proportional to the current going through it. So a current amplifier makes way more sense as a first approximation! 
Nevertheless, you can't do pure high impedance current drive. That will blow the speaker at high impedance and wreck the frequency response. You need to reduce the output impedance to something sensible to match the DC resistance of the voice-coil. Which is what I see PassLabs do.
I'm really not sure why valves often sound better, and I have also heard some very good sounds from bridged transistor voltage power amps. So I'm sitting on the fence here. But enjoy this topic and trying things out!
Nevertheless, you can't do pure high impedance current drive. That will blow the speaker at high impedance and wreck the frequency response. You need to reduce the output impedance to something sensible to match the DC resistance of the voice-coil. Which is what I see PassLabs do.
I'm really not sure why valves often sound better, and I have also heard some very good sounds from bridged transistor voltage power amps. So I'm sitting on the fence here. But enjoy this topic and trying things out!
Some explanation as to why current drive sounds better can be found here:
http://www.diyaudio.com/forums/multi-way/132777-flux-modulation.html#post2757872
Check out the link provided in the post by SpeakerDave. He has actual measurements showing the drop in flux mod distortion in the midband (not near resonance).
As Dave here says, the speaker and amp together contribute towards the total damping provided to a driver/speaker. As long as it is designed in, either approach can work. Except the benefit of lower distortion from current drive.
http://www.diyaudio.com/forums/multi-way/132777-flux-modulation.html#post2757872
Check out the link provided in the post by SpeakerDave. He has actual measurements showing the drop in flux mod distortion in the midband (not near resonance).
As Dave here says, the speaker and amp together contribute towards the total damping provided to a driver/speaker. As long as it is designed in, either approach can work. Except the benefit of lower distortion from current drive.
Yes, of course speakers can be designed for drive impedances above zero. You just have to ensure that more of the necessary LF damping is provided mechanically/acoustically instead of electrically. However, having a simple standard (voltage drive) does make things a bit easier as speaker and amp design can be done by different people.
The force on a voice coil, other things being equal, is proportional to the current. Other things are not always equal. Cone displacement is not necessarily proportional to force - for a start force causes acceleration. So speaker design is a huge balancing act, with non-linear and non-ideal components all interacting together. Electronics is much easier!
The force on a voice coil, other things being equal, is proportional to the current. Other things are not always equal. Cone displacement is not necessarily proportional to force - for a start force causes acceleration. So speaker design is a huge balancing act, with non-linear and non-ideal components all interacting together. Electronics is much easier!
So what did yourself answer? Would yourself agree that lowering the output impedance with multiple devices in parallel or lowering the impedance of the power supply could have an effect on damping speaker drive units from flapping in the wind.
Now think of the following, a huge woofer flapping around in the wind, coupled to a midrange on the exact same pair of wires. Do you think that there may be current induced in the midrange causing it to make sounds that should not be there?
Now consider an almost short circuit between the two pairs of wire connecting the two speakers together, would you consider that the flapping speaker would react less with the other on the same circuit if it is dynamically shorted? In fact would the flapping speaker even flap?
I don't agree that paralleling output devices lowers output impedance. For BJTs two devices in parallel, each carrying half the current, would each have twice the impedance so the total remains unchanged. Simple device physics. Paralleling devices increases peak current capability, it doesn't reduce output impedance.Nico Ras said:
Reducing the PSU impedance would have little effect, unless it was severely inadequate, as the output impedance is determined primarily by output stage topology and the feedback loop.
So I don't accept the premise of your question.
It's possible yes but not a good idea.... Firstly there would be less electrical damping and more expensive motor assemblies would need to be used in drive units plus the type of bass loading would need to be one suited to this such as horn loading. Next there is the problem that the speaker would only sound correct with a specific amplifier driving it!
I have articles by Stanley Kelly and Arthur Bailey showing the problems of non perfect voltage sources interacting with speakers, and more specifically their crossovers, to produce non ideal responses.
Paralleling the output devices should indeed half the output impedance because you have doubled the transconductance. This assumes that the total bias current has doubled and not been reduced to equal that of a single pair.
However, this mainly applies to zero NFB or very low NFB designs as in any normal amp the output impedance is entirely dependent on the NFB factor as was shown experimentally by Professor Edward Cherry. Obviously output wiring, any relay contacts etc add to this in practice...
Last edited:
1) It's well known that a good big transistor amplifier tends to work better than a good small one. Especially in the bass, so I'd reckon doubling power supply reservoir capacitors and output transistors is going to extend bass control and ability to drive awkward low impedance loads. Don't think anyone will argue with that!
2) The question I've often asked myself is whether valve amps have serious advantages over solid state. After all, you can give either topology an 8 ohm output impedance, woolly bass an' all:
Even when you replace the 8 ohm speaker load with an inductance L, the ratio of voltage at the load to voltage at the source remains jwL / (R + jwL) for both topologies where j is the familiar engineer's complex number.
Unless I've missed something, you can therefore get a valve sound using transistor voltage drive and a 8 ohm wirewound ballast resistor!
(Waits to be shot down in flames...)
2) The question I've often asked myself is whether valve amps have serious advantages over solid state. After all, you can give either topology an 8 ohm output impedance, woolly bass an' all:
An externally hosted image should be here but it was not working when we last tested it.
Even when you replace the 8 ohm speaker load with an inductance L, the ratio of voltage at the load to voltage at the source remains jwL / (R + jwL) for both topologies where j is the familiar engineer's complex number.
Unless I've missed something, you can therefore get a valve sound using transistor voltage drive and a 8 ohm wirewound ballast resistor!
(Waits to be shot down in flames...
)
somebody was debating why damping factor may (or may not) be better with paralleled output devices and larger caps in the power supply. the fact is that output impedance is affected by output resistance (i.e., the "raw" output resistance without the effects of feedback in the amplifier). the simplest way to calculate the output impedance is Rout/(Aol/Acl) (simple, maybe not as accurate as other methods). basically, the output resistance divided by the feedback ratio. so the output resistance has some effect on the output impedance, but not that much. the real effects of low output impedance are from the feedback ratio, primarily the open loop gain of the amp. this also in practice shows that the damping factor is much higher at low frequencies than it is at higher frequencies, because the open loop gain starts dropping as the frequency increases. an amp that has a DF of 5000 at 200hz, might only have a DF of 500 at 2khz. but an amp that has a DF of 50at 200hz will only have a DF of 5 at 2khz.
"So I don't accept the premise of your question."
You talk good, but the reality is that adding bypass caps to power supplies changes how they sound in the bass, without making measurable changes in frequency response or output impedance (DF).
We must not have figured out what to measure?
"It's well known that a good big transistor amplifier tends to work better than a good small one. "
That one leaves me laughing.
The first time I heard a 25W Electrocompaniet Ampliwire I was stunned. It sounded like it was infinitely powerful, right up to the point it clipped.
"Unless I've missed something, you can therefore get a valve sound using transistor voltage drive and a 8 ohm wirewound ballast resistor!"
You can mimic the DF of a typical tube amp with a 1Ω resistor, or you can monkey with the feedback loop (as did some Mark Levinson and Carver models).
You talk good, but the reality is that adding bypass caps to power supplies changes how they sound in the bass, without making measurable changes in frequency response or output impedance (DF).
We must not have figured out what to measure?
"It's well known that a good big transistor amplifier tends to work better than a good small one. "
That one leaves me laughing.
The first time I heard a 25W Electrocompaniet Ampliwire I was stunned. It sounded like it was infinitely powerful, right up to the point it clipped.
"Unless I've missed something, you can therefore get a valve sound using transistor voltage drive and a 8 ohm wirewound ballast resistor!"
You can mimic the DF of a typical tube amp with a 1Ω resistor, or you can monkey with the feedback loop (as did some Mark Levinson and Carver models).
"You talk good, but the reality is that adding bypass caps to power supplies changes how they sound in the bass, without making measurable changes in frequency response or output impedance (DF)."
djk. I reckon you may be alone in that finding... I have never noticed any such effect in many years of designing and building amps.
unclejed613. Yep we are reading from the same hymn book!
In Cherry's experiment he made two amps of identical design except one was common emitter output stage and the other emitter follower. Now obviously the follower version will have a low output impedance of an ohm or two even without NFB whilst the common emitter will approximate to a current output of very high impedance without NFB but will have gain. When the loop is closed the extra open loop gain of the common emitter version results in both amps having the same output impedance.
djk. I reckon you may be alone in that finding... I have never noticed any such effect in many years of designing and building amps.
unclejed613. Yep we are reading from the same hymn book!
In Cherry's experiment he made two amps of identical design except one was common emitter output stage and the other emitter follower. Now obviously the follower version will have a low output impedance of an ohm or two even without NFB whilst the common emitter will approximate to a current output of very high impedance without NFB but will have gain. When the loop is closed the extra open loop gain of the common emitter version results in both amps having the same output impedance.
Do tell me what I've missed, DF96! Enquiring minds want to know!
On the subject of damping factor, our hectoring djk suggested these findings as being useful:
DAMPING FACTOR DEBATE
They certainly make the point that an amplifier output impedance below 1 ohm is almost indistinguishable from an ideal voltage source. I would expect almost no tonal difference.
djk and I are not going to agree about valve amplifier output impedance, which in my view is variable by design, depending on topology and feedback on the various configurations. Single Ended Triode designs use no feedback in the classic setup and have high measurable distortion, but are well liked. Most strange!
On the subject of damping factor, our hectoring djk suggested these findings as being useful:
An externally hosted image should be here but it was not working when we last tested it.
DAMPING FACTOR DEBATE
They certainly make the point that an amplifier output impedance below 1 ohm is almost indistinguishable from an ideal voltage source. I would expect almost no tonal difference.
djk and I are not going to agree about valve amplifier output impedance, which in my view is variable by design, depending on topology and feedback on the various configurations. Single Ended Triode designs use no feedback in the classic setup and have high measurable distortion, but are well liked. Most strange!
Your argument is circular. It is only valid in a world where all amps are voltage sources & all speakers are designed to be driven by a voltage source.
High damping is needed to control a loudspeaker system in the LF with a high mechanical Q. And because of that goal, we pay by putting up with inherently gtrater distortion every where else.
All speakersonly sound best with a specific amplifier driving it. Many a speaker sounds too lean because of too much damping. It is very enlightening to sit down with a pile of speakers & a veriable transconductance amplifier (has a knob that varies damping from very high to very low). There is an optimal setting for every speaker.
It is certainly possible to design an XO for a current source amp, but XOs are inherently evil and should be avoided, if they can't be avoided, active is generally preferred, and if not active should not be in the ~400-4k range (that last point has nothing to do with current vrs voltage amps)
dave
It is actually not well known. There is a general tendency for good small amps to sound better than good big amps.
Some of the best bass i've heard was from a 3.2 w Class A PP triode amp
dave
You have missed something. A transistor does not have the same transfer curves as a triode (one of the reasons Nelson dropped 6 figures to get a batch of SITs which do have triode-like curves)
dave
"Your argument is circular. It is only valid in a world where all amps are voltage sources & all speakers are designed to be driven by a voltage source."
So it is a perfectly valid argument then as all speakers I know of (commercially) ARE designed to be driven by voltage sources and all amps, again commercially, ARE voltage sources!
As an aside, current amps are most commonly used to drive "AFILS" induction loop systems for the hard of hearing (they set their hearing aids to the "T" position).... I was once employed to design such amps....
I have heard (possibly apocryphal) stories of speakers being designed mainly by ear using the designers favourite valve amp that had a DF of only 3. The resultant product then sounded awful to all and sundry who had an amp with a more normal DF of 30+ !
As far as SET amps go I have yet to hear one that I would consider to meet even the basic requirements to be called hi-fi!
Active crossovers are indeed a big step forward.... it's just a pity they still have not received universal acceptance after all the years that the obvious benefits have been known.... Other than in pro audio.
So it is a perfectly valid argument then as all speakers I know of (commercially) ARE designed to be driven by voltage sources and all amps, again commercially, ARE voltage sources!
As an aside, current amps are most commonly used to drive "AFILS" induction loop systems for the hard of hearing (they set their hearing aids to the "T" position).... I was once employed to design such amps....
I have heard (possibly apocryphal) stories of speakers being designed mainly by ear using the designers favourite valve amp that had a DF of only 3. The resultant product then sounded awful to all and sundry who had an amp with a more normal DF of 30+ !
As far as SET amps go I have yet to hear one that I would consider to meet even the basic requirements to be called hi-fi!
Active crossovers are indeed a big step forward.... it's just a pity they still have not received universal acceptance after all the years that the obvious benefits have been known.... Other than in pro audio.
So it is a perfectly valid argument then as all speakers I know of (commercially) ARE designed to be driven by voltage sources and all amps, again commercially, ARE voltage sources!
You are obviously only familiar with a subset of commercially available loudspeakers and are ignoring that we are diyers here and build our own to suit. Often because there is no commercially available stuff that fits our needs.
It could be dsigned with measures and still have the same issues. I repeat, every loudspeaker has an optimum amplifier (output impedance). The speakers you mentioned were designed to be used with highish output impedance. Conversly, a loudspeaker designed for a strong voltage source will sounf like crap driven by that SE amp (not the amps fault, not the speakers fault, the fault of the joker who ignored that speakers and amps are an inseparable system)
Maybe you have only heard SETs set up by the joker mentioned above? There are certainly some very good systems with SETs in them. Put a typical SS amp into those systems and that sounds like crap (typically the amp gets blamed, but it is really the joker again, althou i haven't heard very many truly awesome SS amps)
And in the world of subwoofers which almost universally employ line level XOs. Not a big step from there to a woofer augmented single driver system (FAST).
dave
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.