Are you looking at the same pictures? There is a voltage source amplifier connected directly to the speaker in both cases. The impedance seen by the amplifier is the ONLY thing affected.
@Dick, Joe is not talking about current drive of speakers, not even a little.
Last edited:
Tourney, you misspoke! You are back! Welcome!!!!
Scott just include the loudspeaker wire resistance. Explains why a constant impedance load has a more uniform frequency response.
Now for the real humor the motor strength in a loudspeaker is related to the number of turns in the voice coil and the current through it. So a sixteen ohm loudspeaker driver might have twice the turns of an eight ohm model thus from the same source voltage, the same output at half the power.
This is often used in high frequency compression drivers to limit the current and to gain a bit of safety in power handling. An issue that is almost forgotten.
Look at auto sound loudspeakers. They go for lower impedances to get more power into the loudspeakers from the limited rail voltage often used in the amplifiers?
A bit more to loudspeaker design than so far considered by many.
Scott just include the loudspeaker wire resistance. Explains why a constant impedance load has a more uniform frequency response.
Now for the real humor the motor strength in a loudspeaker is related to the number of turns in the voice coil and the current through it. So a sixteen ohm loudspeaker driver might have twice the turns of an eight ohm model thus from the same source voltage, the same output at half the power.
This is often used in high frequency compression drivers to limit the current and to gain a bit of safety in power handling. An issue that is almost forgotten.
Look at auto sound loudspeakers. They go for lower impedances to get more power into the loudspeakers from the limited rail voltage often used in the amplifiers?
A bit more to loudspeaker design than so far considered by many.
Not what is shown, anyway out come my 0000 welding cables. 🙂 You know this is not what is being discussed, nonsense about amplifiers and reactive current is. There's something to consider pulling pairs of 0000 welding cable in your auditoriums, pretty expensive and heavy (nice thick sound).
Last edited:
I have stated before current drive works best on loudspeakers that are designed for it. There are none as far as I know.
Many of the comments are based on observations of realized results. These are often explained by understanding the actual usage.
Perhaps you did not pickup that in many full range loudspeakers the high frequency driver is higher impedance than the other drivers. Under current drive that would drastically tilt the frequency response.
Impedance is rarely a single value as frequency sweeps. As the loudspeakers I am familiar with are designed and tested with voltage sources, they would not reproduce those results with a current based signal source.
Now is anyone familiar with single box complete loudspeaker systems that contain amplifiers that use current source drive?
Or in simple form "don't feed the trolls."
Many of the comments are based on observations of realized results. These are often explained by understanding the actual usage.
Perhaps you did not pickup that in many full range loudspeakers the high frequency driver is higher impedance than the other drivers. Under current drive that would drastically tilt the frequency response.
Impedance is rarely a single value as frequency sweeps. As the loudspeakers I am familiar with are designed and tested with voltage sources, they would not reproduce those results with a current based signal source.
Now is anyone familiar with single box complete loudspeaker systems that contain amplifiers that use current source drive?
Or in simple form "don't feed the trolls."
Last edited:
Good summary, Ed. The only system I know of were JBL 4408 modified prototypes I made in the early to mid-1990s, with separate amps for LF/HF. We were JBL dealers, and I was too munificently challenged to risk destroying a more expensive set of speakers, and of course they never went into mass production, which was likely your query.
In order to flatten response around resonances I eq'ed the current sample feedback of each driver using White passive eqs. Due to the use of fixed 1/6th octave equalizers (what I had available) the resultant eq was not perfectly flat as it may have been if dedicated eq exactly matching the impedance curve had been used.
The main performance change from stock was the dynamic performance of the system was subjectively improved. The amplifiers we were using were showing over twice the transient peak power when reproducing the Sheffield Drum and Track CD at approximately equivalent average SPLs. Would this have also been the case with mere biamping? It is possible given the lower parasitic resistances between the amp and driver compared to running through a crossover, but from this vantage I can't answer that.
I know PMA has done a lot of testing with pure current drive and can show the only real change to be a slight drop in distortion under certain conditions (eq change aside). With the approach I took I did like the sound with the current corrected drive better than stock JBL...but then again that opens a can of worms regarding the sound of stock JBL studio monitors.
I built a few versions of this circuit and shopped it around, and it ended up being copied by a famous mobile audio manufacturer who marketed amplifiers integrating the concept. They had signed a confidentiality agreement, but responded to a cease and desist with "bring it on'"...Jerks...I had a lot to learn about IP protection back then, namely unless you can pay an equal legal firm there ain't any.
I would have loved to have continued development on that project, but building an optical disc plant which eventually made over a billion shiny discs ended my free time back then...
Cheers!
Howie
Nelson Pass did some (non-commercial) stuff too. It should be obvious, like VHS vs. BETA, the two distinct systems are not going to co-exist as commercially viable entities.
Yes, in Melbourne Australia, SGR Loudspeakers: SGR High End Loudspeakers | SGR Loudspeakers
There is also a Dutch speaker that is connected with Hypex's Bruno Putzeys, which I have been told that a modified Class D amplifier drives the midrange. Is it *Kii speaker, I am not sure? Maybe anybody here knows more details.
I agree that current drive lends itself to 'active' speakers and especially in the midrange. But the high pass filter, when a waveguide is used, you can get the tweeter working in current mode by using a coupling capacitor value of about 30% of what is normal, then EQ the current just before the series cap, and you have something very close to current drive of the tweeter in the octaves close to the crossover. But around 3kHz crossover is required.
But my own ambition and what I strive for is to get virtually the same benefits from voltage drive that can be had with current drive. There will be some that will say that is not possible, in which case they will never achieve it. But I believe it can be done (I might even have done it, teasing...😉), Earl Geddes has told me emphatically it can't be done (but then again I don't think he likes current drive anyway) and that I am wasting my time. But it is not over until the Fat Lady sings.
*EDIT: Kii Audio Three does.
Last edited:
Yes and they are looking for increased impedance behavior to sound more like a tube amplifier.
Anyone serious about it today would likely use a class D switching amplifier.
On this, I'm with Earl. Joe, seriously, you need to revisit the whole amplifier/cable/speaker system issues with what you are hearing (or think you are hearing).
The Kii3 has current drive, but not as anyone knows it. They use the resistor as feedback sensing driving a wad of DSP. They also have std voltage sensing, and for the woofers motional feedback. So with voltage AND current AND position AND look ahead monitoring of the signal they can make the most of it.
THAT IS WHAT I AM DOING! 😀😀😀
Scott, there maybe something here that has been overlooked. It comes down to what reactive currents are and what triggers them. The Fat Lady has not sung yet (or maybe she has, but behind the curtain). What is the primary advantage of current drive? If you pin it down to a single factor, then that is the key, because it suggests something that is more obvious than you might think, even hiding in plain sight, or close to it.
Neville Thiele did something that pointed towards it five decades ago. If it works, all will come out - and maybe even you will be convinced in the end. It won't be historically the first time things have worked out this way, often things just gets found by chance. BUT... if it does not work, then again somebody still has to brave to attempt it. As it is, the first lot of measurements points to a positive end result - yes, it will be measurement based.
If the end result is not good as hoped for, I will still have figured out this: You can cancel out the output impedance of an amplifier and make the speaker compatible with both voltage and current drive. THAT has already been achieved and a good thing! I have speakers here that can already do that and have amplifiers with output impedances between 0.1 Ohm and 270 Ohm. It locks in the crossover, it locks in the bass alignment. This is not theory, but is a physical fact.
PS: Scott, was there any time when you were told you couldn't do it and then went ahead and did it anyway?
Comic relief
The first IC instrumentation amplifiers were based on open-loop current transfer. I realized that I could get almost all the same functionality out of same simple folded cascode with common mode feedback discrete circuit that I used for my RIAA.
In-amps using open-loop current transfer have some nice features like fairly constant BW vs. closed-loop gain and nearly complete common mode rejection at the input. Typically the IC versions were based on a 100K trans-resistance which made them too noisy for audio (they were never intended for audio anyway). To this day they have been replaced by continually improving versions of the Demrow data amplifier (1968). There are versions of this circuit designed specifically for audio like the SSM2219 and THAT1512. The main issue here is getting low enough noise at low gains since a line stage typically would be at 20dB or so max.
The in-amp consists of a differential input V to I converter with transconductance set by a resistor and an output differential I to V converter. The input stage lends itself to variants which could be discussed later, but here is the output I to V. The little trick (inset) is that by simple jumper the output becomes differential or single ended depending on where you connect the comp cap and the base of Q9. You could also use the base of Q10 as a remote ground sense just as in a typical IA (I think ��). If you look carefully and you don’t load the outputs of the I to V converter you will see that the common mode loop does virtually no work in either connection, so it has little or no effect on the signal path (IMO). If you want, you could substitute a PFET pair with no degeneration. WARNING use of some PFET’s might require attention to breakdown issues or use of additional supply rails. As usual I would make the appropriate current sources trimmable for RTI and RTO offset and forgo device matching except at the grossest level.
Using 10K as transresistance I was surprised how close in noise performance to the IC’s I got. The output is capable of 7V rms and even with a simple degenerated (1k Ohm = ~20dB gain) JFET pair at the input the distortion at this output level is ~-76dB and much less at 2V rms line level (all sims). The output resistance is 10K, so you need to buffer it. Since the input is FET based, you can move the source follower buffers of your choice to this stage. The JFET cascodes are a precaution against possible thermals that you might get especially using discretes in SOIC.
I’ve done some sims using Bob Cordell’s models of common inexpensive bi-polars and have not found big problems. Please remember this is for entertainment, “Muntzing” if you will, and seeing how well simple circuits can be made to perform. More later.
The first IC instrumentation amplifiers were based on open-loop current transfer. I realized that I could get almost all the same functionality out of same simple folded cascode with common mode feedback discrete circuit that I used for my RIAA.
In-amps using open-loop current transfer have some nice features like fairly constant BW vs. closed-loop gain and nearly complete common mode rejection at the input. Typically the IC versions were based on a 100K trans-resistance which made them too noisy for audio (they were never intended for audio anyway). To this day they have been replaced by continually improving versions of the Demrow data amplifier (1968). There are versions of this circuit designed specifically for audio like the SSM2219 and THAT1512. The main issue here is getting low enough noise at low gains since a line stage typically would be at 20dB or so max.
The in-amp consists of a differential input V to I converter with transconductance set by a resistor and an output differential I to V converter. The input stage lends itself to variants which could be discussed later, but here is the output I to V. The little trick (inset) is that by simple jumper the output becomes differential or single ended depending on where you connect the comp cap and the base of Q9. You could also use the base of Q10 as a remote ground sense just as in a typical IA (I think ��). If you look carefully and you don’t load the outputs of the I to V converter you will see that the common mode loop does virtually no work in either connection, so it has little or no effect on the signal path (IMO). If you want, you could substitute a PFET pair with no degeneration. WARNING use of some PFET’s might require attention to breakdown issues or use of additional supply rails. As usual I would make the appropriate current sources trimmable for RTI and RTO offset and forgo device matching except at the grossest level.
Using 10K as transresistance I was surprised how close in noise performance to the IC’s I got. The output is capable of 7V rms and even with a simple degenerated (1k Ohm = ~20dB gain) JFET pair at the input the distortion at this output level is ~-76dB and much less at 2V rms line level (all sims). The output resistance is 10K, so you need to buffer it. Since the input is FET based, you can move the source follower buffers of your choice to this stage. The JFET cascodes are a precaution against possible thermals that you might get especially using discretes in SOIC.
I’ve done some sims using Bob Cordell’s models of common inexpensive bi-polars and have not found big problems. Please remember this is for entertainment, “Muntzing” if you will, and seeing how well simple circuits can be made to perform. More later.
Attachments
You would do well to get an EE to help you write your material. That's not an insult there is just that big of a language barrier.
Scott,
If it is for Vin and Vout, would a Blowtorch like topology not does the same thing but simpler ? (One can of course replace the 4 resistors of the folded cascode with CCS's.)
And for the full differential version, have you try leaving out C1 (in simulations) ? You might find that single phase outputs have lower even harmonics. Afterall you did not use C1 in your phono pre.
Unless of course you want to use it with current input, then ignore what I just said.
Patrick
If it is for Vin and Vout, would a Blowtorch like topology not does the same thing but simpler ? (One can of course replace the 4 resistors of the folded cascode with CCS's.)
And for the full differential version, have you try leaving out C1 (in simulations) ? You might find that single phase outputs have lower even harmonics. Afterall you did not use C1 in your phono pre.
Unless of course you want to use it with current input, then ignore what I just said.
Patrick
That is what sometimes has to happen. It has not so much to do with language, but the results. I will leave the language for others, my job is to convey, not to be 'politically correct' as they say. Besides, it all sounds so stuffy and sterile - and maybe if I don't have YOUR attention, this is not to say that I don't have others with EE qualifications very intrigued. I do! The sterile language has gotten us no progress for at least twenty years. Oh, the measurements I spoke about, they will be very disciplined and tightly documented. Might still make a convert out of you yet, eh? 😀
PS: I would love you to answer the question, what is the most important thing that current drive does - which alone explains all its difficult pluses? It is the key.
Last edited:
Words and their meaning is more a philosophical point, I think a lot of progress has happened in the last two decades.
Again you are not promoting current drive.
- Status
- Not open for further replies.