Yes, bummer when the lunatics take over the asylum, but pay attention and you might learn something.
David
David
As a physicist I'll be paying careful attention to how little some people know about quantum electrodynamics and superconductivity.
I've just got the preliminary tests back,
and I'm afraid everyone in the class won't be moving ahead this year.
Hi,
This thread had no grounded starting point, as such it can't lead to
any grounded end point, and will end up as pointless ego-stroking.
rgds, sreten.
This thread had no grounded starting point, as such it can't lead to
any grounded end point, and will end up as pointless ego-stroking.
rgds, sreten.
I've done the general analysis.
Next would analysis of specific cases,
with a thorough investigation and quantification of construction, frequency crossover points, drivers, damping materials and some measurements.
This cannot take place without people putting up examples for testing,
that can be deconstructed and documented,
and some equipment for measurements of phase effects etc.
Ideally a soundproof anechoic chamber would do nicely.
Are you willing to put in some speakers, equipment, or premises?
Thats where data comes from.
So you want us to volunteer to do all the work for you to confirm your pet theory ? Very clever, but I don't see that happening somehow 😉
The onus of proof is on you. If you have a theory which is different to the generally accepted understanding of how something works, it's up to you to prove it, not for us to do the work to disprove your theory.
The onus of proof is on you. If you have a theory which is different to the generally accepted understanding of how something works, it's up to you to prove it, not for us to do the work to disprove your theory.
Funny, I came to a thread entitled "woofer distortion 101" thinking I might learn something about the named topic. After hearing a flawed argument about divided cabinets (my comments never answered) it has turned into a series of insults about our lack of physics knowledge.
Well, nazeroo, I hope you've got something out of a slightly rough introduction to the utter horror that is the diyaudio forum when you speculate or try new stuff. The old guard don't like it. 😉
Truth is no-one here has ever built the perfect speaker. 😀
We've scratched our heads over Tannoy's, Ditton 66's, Gale 401's, Quad Electrostatics, BBC Ls3/5A and the like. Why are they good? Only the merest inkling. 😕
I do sometimes wonder if Quantum Mechanics and the Uncertainty Principle is telling us that when you get one thing exactly right, something else gets blurry...😡
Of course, like any interested person, I have moved on to 10 dimensional string theory in search of answers. But, frankly, it is even more baffling. Perhaps you can delve into it for us, being smarter.
Lecture 1 | String Theory and M-Theory - YouTube
I expect there are some valuable Nobel prizes to be won, which will buy many tins of baked beans. 😎
Truth is no-one here has ever built the perfect speaker. 😀
We've scratched our heads over Tannoy's, Ditton 66's, Gale 401's, Quad Electrostatics, BBC Ls3/5A and the like. Why are they good? Only the merest inkling. 😕
I do sometimes wonder if Quantum Mechanics and the Uncertainty Principle is telling us that when you get one thing exactly right, something else gets blurry...😡
Of course, like any interested person, I have moved on to 10 dimensional string theory in search of answers. But, frankly, it is even more baffling. Perhaps you can delve into it for us, being smarter.
An externally hosted image should be here but it was not working when we last tested it.
Lecture 1 | String Theory and M-Theory - YouTube
I expect there are some valuable Nobel prizes to be won, which will buy many tins of baked beans. 😎
Coming back to this, it is not all wrong, of course. It's more a question of semantics and of the choosen application, too.
Without doubt the transducer's motion is purely current induced (not: controlled), as only current is found in the motional equation. Current is the root cause of motion by establishing force.
Now it depends on the application, on how we control that current.
Pure current drive is easy to understand, we just inject the current (force it, that is) and that's it. So we have a pure current controlled system, which is basically equivalent to a purely force controlled system.
When the driving impedance is not infinity, voltage comes into play, the total system voltage accross the driver AND any external impedance. Lets assume a real impedance, a pure ohmic resistance for the moment and name it Reff (effective system resistance comprised of Rdc of the VC and anything additional to that, be it positive, zero or negative).
System voltage is given by :
e = Reff*i + BL*v
This is simply I*R drop plus voice coil microphonic voltage (proportional to VC velocity).
Assume that Reff is really small, say a fraction of an Ohm. Then this establishes a strong local feedback for the cone motion : The driver (as a system) "compares" its own generated voltage with the one externally applied. Any difference will cause a huge current flow due to the small value of Reff, correcting the driver's motion in the proper direction. The driver itself is still operated by current but the way the current is established is completely different (and that has huge impact on the type and amount of distortion since now several transfer stages with feedback are at work instead of a single one without feedback, voltage into current into force, then into velocity into voltage again. Compare this with current into force for the current controlled case).
So now we could say we have an (almost) purely voltage controlled driver, and this voltage must be proportional to the cone velocity we need for a given output, so we might speak of a purely velocity controlled system.
It is important to note that for the same overall acoustic output the driver terminal voltage is the same too, regardless how we choose to drive the transducer. Of course only within linearity limits (which do change by the way of driving).
Real-world systems have non-zero Reff, but it can be tailored to the needs in some extend. One of that needs is that we want to have a flat acoustic output for a flat electric voltage input. This is done by dialing in the right amount of damping so that the increased output around system resonance is tailored to the exact level needed. Also, overall distortion can be mimized by some (non-intuitive) cancelling of force distortion with suspension distortion. That's (a tiny portion of all the aspects, that is) were the real brains went into driver design by all the clever guys.
Flat and voltage input is the classic convention but it is becoming less and less important as voltage headroom and system EQ for active design is a non-issue today and quite the norm in studio/pro/PA applications. Now, I do design electronics for active speakers and neither the input is always flat nor is it always classic voltage-only drive (zero amplifier output impedance).
Last edited:
Your equations are true but don't let it lead you astray! You're look at the driver is similar to my discussion of motional impedance and its size relative to voice coil impedance: when motional impedance is high relative to coil resistance then we have very high efficiency and more of the power is converted to sound rather than heat. Per your equation the BL*v term dominates IR drop.
When you consider the case where Ref is low or BL is high these lead to desirably high efficiency but, in the limit, force velocity to be in proportion to input voltage. This is not desirable as it will not result in flat response. It is approximated with a heavily overdamped woofer giving 6dB per Octave response in the vicinity of resonance.
Key to all of this is considering the driver's air load, where we find that its acoustic impedance rises 12 dB per Octave for a wide range (until about ka = 1 where it flattens out). From this we see we really want our driver to have constant acceleration rather than constant velocity. That was the discovery of Rice and Kellogg (apperantly prior to their entrance into the cereal industry), that a direct radiator can have a wide range of flat response if used in the region above resonance and having a moderately damped resonance. The F = BLi and F = ma relationship allows a region of constant acceleration and thereby flat response.
In the end we can't have flat response and extreme efficiency. Input voltage becomes pure back EMF. Per your equation our superconductor coil will overdamp the woofer and force constant velocity operation.
Although, I suppose if we can afford the refrigeration system we can probably afford some bass EQ.
David
While it is true that a heavily overdamped driver will need a lot of LF EQ to compensate for the velocity slope to get a flat output it is not true that this eats a lot of power because most of the passband happens in a region of high impedance modulus. It requires a lot of voltage headroom and therefore the only resonable choice for amping is Class-D which does not suffer from low efficiency when voltage is low and current is high, which is i) true for higher frequencies in term of magnitude and ii) true for momentary voltages and currents in case of load points of high reacitivity. Also the amp must have a good margin of peak current because the maximum instantaneous current can be as high as three times the peak current U/R given by peak output voltage and DC resistance. A margin of two is sufficient in most practical conditions.
I'm using Class-D amps with synthesized negative output impedance in the resonance region to lower the Qes of the drivers even more in order to get more motional feedback (note that I prefer to start off with lowest Qts woofers but with moderate stiff suspension), to effective values way below 0.1. Works like a charm and is as efficient as it gets. I'm talking sealed and open baffle designs, ported has some things to consider when increasing the electrical damping (velocity control, that is).
The following two papers were the real eye-openers to that approach :
www.extra.research.philips.com/hera/people/aarts/RMA_papers/aar03c4.pdf
www.xlrtechs.com/dbkeele.com/PDF/Keele (2003-10 AES Preprint) - Nom vs True Eff High BL.pdf
The negative output impedance approach is very interesting for effectively lowering Q. Remember that it is fundamentaly different than lowering Q by increasing BL. The response curves look similar in the end but the negative output Z doesn't actually increase efficiency. The voltage or current needed to achieve a certain SPL at a given frequency doesn't change (at the driver terminals) although the amplifier/speaker combo frequency response does change in the lower Qt direction.
In the end I'm not sure it is any different than achieving the low Q curve with careful EQ.
Thanks for the references, those are both very good.
David
In the end I'm not sure it is any different than achieving the low Q curve with careful EQ.
Thanks for the references, those are both very good.
David
It's EQ "by the driver itself" vs. nomal EQ. Overall response can be dialed in to be the same, then current, and voltage resp., is the same, too (as it must be). The difference is in the distortion and other ill effect. Negative drive impedance can make DC offset phenomena and recovery very nasty, and it worsens power compression and response shift from VC heating, so it is not a simple fix for a bad driver. It can be used to make a good (and thermally oversized) driver even better, sometimes.
The critical surface for any super also includes temperature, it is a 3d surface which bounds operation. The vast bulk of superconductor applications are not what we call cryo stable, meaning the system is not typically going to return to supreconducting instantly, but instead will runaway.
The snap is what we call a "quench". When it occurs, the conductor will immediately dissipate power in response to an IR drop. This power dissipation will continue the quench process on the conductor. The "quench propagation velocity" is a measure of how fast this normal zone travels down the conductor. (edit: remember, supers are generally used at current densities 3 orders of magnitude higher than metals at room temp). One very bad aspect for temperatures below about 50 kelvin is that most metals lose almost all their heat capacity, so the rise in temp can be very quick. For HTS wires, the onset of quenching is significantly slower and more gentle than LTS conductors though.
Awwww..I was so much hoping to learn a little bit about superconductivity..🙁
Yah, but liquid nitrogen is cheaper than beer. Just schedule the welding supply truck for daily deliveries...
jn
Last edited:
KSTR and speaker dave:
When drivers are measured for L/R, is there any consideration of the frequency dependence on both parameters?
When I measure coils wrapped around metal using the LsRs model, I find that both change in a non linear manner... In particular, the equivalent resistance of the coil changes due to proximity effect, the resistance rises as a result of eddy losses in the magnetic circuit, and the inductance drops as the coil field starts to be excluded from the magnetic circuit (Lenz effect).
I've only seen linear models of drivers, but am wondering..
jn
When drivers are measured for L/R, is there any consideration of the frequency dependence on both parameters?
When I measure coils wrapped around metal using the LsRs model, I find that both change in a non linear manner... In particular, the equivalent resistance of the coil changes due to proximity effect, the resistance rises as a result of eddy losses in the magnetic circuit, and the inductance drops as the coil field starts to be excluded from the magnetic circuit (Lenz effect).
I've only seen linear models of drivers, but am wondering..
jn
Yes, it drives me a little bonky when someone puts a number on woofer Le and thinks it is a precise value.
Most in depth studies think coil inductance as a loosely coupled transformer. Dick Small did some research for us at KEF and found a frequency dependent resistor inductor combo best tracked a measured curve for the widest range. Otherwise you can do an impedance fit to a cascade of 2 or 3 L parallel with R sections and get a decent fit for a decade or so.
As you say it is a matter of level and greatly varies with woofer position as the iron pole piece moves in or out of the coil. Under cut core pole can reduce the Le variation with excursion. I have seen cases so bad that, with high level pink noise, you can hear the treble come and go with excursion.
I think the Klippel web site covers the nonlinearity aspect fairly well and Linkwitz goes into flux modulation effects and Le effects in some depth. All T/S parameters are meant to be thought of as small signal effects and will vary with level. Fs varies with level very strongly, which throws off Qt and other dependent parameters.
David
Apart from lowering the Qt, negative resistance has been said to have good effect on linearity in the resonance region (Jeff Macaulay, "MFB Reply" in Letters, Electronics World, January 2003, p43).
Last edited:
- Status
- Not open for further replies.