First: Why are we concerned with H2? H2 is usually deemed to be benign.
Second: -96dB to -90dB or even higher is lost in the noise floor.
Third: To lower H2 even farther will require knowing the mechanism that generates the distortion.
Remember that X-Max is measured at the point that Distortion is =>10%.
D2 is in the current
Resistor-drive reduces D2 (and all others) in the current
So IMDs are reduced
IMDs are undeniably audible
Best regards
Bernd
Resistor-drive reduces D2 (and all others) in the current
So IMDs are reduced
IMDs are undeniably audible
Best regards
Bernd
Last edited:
Did you compare equal FR situations? EQ is most often needed to achieve this.... or they are not comparable as I see it...
My adventure that ended with... not so much...
//
My adventure that ended with... not so much...
//
Look at the fundamentals at the top of the diagram. These are the frequency responses of the exciting currents for the colorful distortion diagram. Please ignore the (distracting) harmonics for a moment.
Overlay = yellow = without 20R (= 0R)
Green = with 20R
These are exactly the fundamentals of my colorful distortion diagram in Posting #1.
Green = with 20R was consistently “louder”.
However, I also measured experimentally with the opposite adjustment error. This adjustment error attempt did not fundamentally change the picture of the distortions.
Best regards
Bernd
PS @All: This original thread remained open:
https://www.diyaudio.com/community/threads/drive-current-distortion-measurement.402566/
Please do not clutter this @mikets42 thread. Please reply there. Thank you!
Overlay = yellow = without 20R (= 0R)
Green = with 20R
These are exactly the fundamentals of my colorful distortion diagram in Posting #1.
Green = with 20R was consistently “louder”.
However, I also measured experimentally with the opposite adjustment error. This adjustment error attempt did not fundamentally change the picture of the distortions.
Best regards
Bernd
PS @All: This original thread remained open:
https://www.diyaudio.com/community/threads/drive-current-distortion-measurement.402566/
Please do not clutter this @mikets42 thread. Please reply there. Thank you!
Last edited:
In my view, such basic hysteresis graph fails to explain these because it assumes wrongly that the coil field simply adds to the permanent field one-dimensionally. In reality, when the permanent field runs horizontally through the air gap, the coil field tends to run vertically in the iron near the gap. Thus, though the iron may be almost (or even fully) saturated horizontally, it may still be relatively free to assume vertical magnetization in both directions, giving rise to the observed odd-order distortions under voltage-drive. This is my explanation, as I haven't come across any better one.
The effectiveness of shorting rings in reducing current-induced distortions is quite proportional to the amount they reduce the driver's inductive impedance by shunting it. A ring at the bottom of the center pole is much less effective in this than a 'sleeve' around the top, close to VC.
Quite true. Treating current-induced distortions is not his forte.
I'm not yet convinced that the level invariance is real and not e.g. some software issue. Can you check this by making some manual point frequency measurements at the various levels?
Thats probably the reason why mikets42 observes few benefits of current drive when applied to HF compression drivers: The better ones have 2 - 2.2 Tesla in the gap, so near full saturation....
But in the end the stablizing action still stands, doesn't it? Opposing the change == keeping things more constant.
I don't agree with that. The various distortion mechanism that are discussed in the various Klippel papers and application notes apply to all VC-based drivers and all frequencies.
Notably Le(x), BL(x) and Le(i), BL(i) strongly affect midrange distortion ("midrange" with respect to the drivers resonant frequency) with complex signals. Single-Sine HD is not very useful way above resonance, but IMD is, with lower stronger tones modulating Le and BL.
Yes, could be. Another point is CDs are usually underhung design (thus Le(x)=const).
I've actually seen distortion decrease at low frequency with increased internal feedback (voltage drive or even negative source resistance) compared to current drive (not feedback). That is because the suspension usually is extremely progressive and a major distortion contributor near (or even below) resonance. Strong internal feedback helps, and for CDs the VC as a sensor is very linear (again, because of underhung design).
But the Y axis span 200 dB so it's a bt hard to judge the FR differences. For FR "we" use 50 dB span typically.
//
D
Deleted member 375592
The vibration amplitude of the cone (and voice coil) changes as (1/f)^2 for the same sound pressure. 1mm at 20Hz becomes 10um at 200Hz and 100nm at 2000Hz, well below the radar.
The cone vibration amplitude changes proportionally to the current. If the distortion-originating curve is smooth, when the main delays by X dB, H2 (relative to the main) must decay the same X dB, H3 must decay by 2*X dB, and H4 by 3*X dB, etc. Is that the case? Let's test Klippel's theory by testing in the real world: i.e. measure Dayton Audio SIG150-4, from 65 dB SPL @1m to 95.
For H2, it seems to work fine. The lowest SPLs of course fall under the noise floor.
However, H3 does not fit into Klippel's theory at all. (@ETM - all measurements are "manual"). Lars von Purify also describes the same effect. It is unlikely that two independent researchers make exactly the same mistake.
For a scientific theory to be proven wrong, you need only 1 (one) experiment contradicting it. It is called "falsification". Here it is. Full stop.
This all can get really complicated really quickly. Good luck writing the MATLAB script.
Just for a peek at a few variables.
Force factor
Air compliance in an enclosure.
Voice Coil excursion
Force factor will change with the position of the Voice Coil in the B-field
as the position of the Voice Coil moves the position of the cone also moves compressing air in the enclosure that then pushes back on the cone, modulating the net force acting on the cone. Mechanical dynamic breaking that has only a limited relationship with magnetic fields.
This is not so much Back EMF as it is "Back Pressure".
The power amplifier may be producing volts, amps, even watts yet the net force pushing or pulling the driver cone may be zero. At that point shorting rings, inductors, resistors... nothing else matters.
You think that you know Force Factor?
?
My aging eyes still can see 6-7 dB increase in H3 from 65 dB to 95 dB.
Also, bear in mind that....
the rise of H2 with SPL, that is.
Take previous two measurements with single-tone stimulus, and compare it with these remarks from Klippel:
"Unfortunately, harmonic distortion measurement gives not a comprehensive picture of the large signal performance of loudspeaker systems. At least a second tone is required to generate intermodulation products... Increasing the number of fundamental components in multi-tone stimulus will generate more and more intermodulation components spreading over the complete audio band. ... THD generates significant intermodulation distortion at higher frequencies ... Thus, harmonic distortion measurements using a single test tone are not sufficient for assessing loudspeakers comprehensively and predicting the large signal performance for complex stimuli like music."
So, "Klippel theory" is not proved to be wrong yet.
Last edited:
D
Deleted member 375592
Please, let's keep focused on midrange and relatively low SPL where we do not understand where the distortions are coming from, and specifically how current driving helps alleviate these problems. Voice coil movements here are measured in microns and currents in milliamperes. I am not saying that there are no problems beyond this narrow focus - there are simply way too many of them.
Let's stay within the scientific definition of what is an acceptable theory and what is not. Of course, you are free to believe whatever you please.
In spite of your focus only on the frequency range of a typical midrange driver, you are overlooking the intermodulation distortion which is much higher than simple single-tone harmonic distortion in your measurements. Do the multitone testing and you will see. Even if we neglect the H3 and higher harmonics, level of H2 distortion is high enough (evidently from your measurements!) to cause much higher intermodulation distortion (IMD). And the current drive is not effective enough in reducing H2... and IMD consequently.
It is well understood where the distortion is coming from, even at relatively low SPL: Nonlinear force factor Bl(x) and inductance Le(x), Le(i) of motor assembly (voice coil, iron path, magnet), nonlinear stiffness Kms(x) of mechanical suspension (surround and Spider), nonlinear losses Rms(v) of mechanical and acoustical system, Partial vibration of the radiator’s surface (surround, cone, diaphragm, dust cap), ...
By that definition, Klippel theory is accepted in the scientific community. Yours is not. You didn't contradict Klippel by your experiments. Of course, you are free to believe whatever you please.
Last edited:
D
Deleted member 375592
I don't have any theory. I am just measuring things and sharing the results. If this is unwelcome - have a good day.
Yes, you have a "theory" - that your experiments contradicts Klippel.
Thanks for sharing your measurements - they surely are welcomed here, to reinforce the known fact - that current drive can be used only as a midrange amplifier for midrange driver in a fully active 3-way loudspeaker. Also, that Faraday rings works equally well in all drivers, if used with voltage drive amplifiers.
Thanks for sharing your measurements - they surely are welcomed here, to reinforce the known fact - that current drive can be used only as a midrange amplifier for midrange driver in a fully active 3-way loudspeaker. Also, that Faraday rings works equally well in all drivers, if used with voltage drive amplifiers.
If you take the pole piece out of the speaker and keep the voicecoil around it, you will see the same 3rd harmonic that follows a similar pattern as crossover distortion in a class AB amplifier, although rising slightly with level rather than falling slightly. This is because the energy stored in hysteresis is roughly proportional to the peak current, and is dumped all at once as a disturbance at zero crossing.
If we were to innocently wrap a coil around a core of the same material and it did NOT exhibit well-known properties of magnetics, we would win a Nobel prize. There is not much else in a speaker besides poor glue junctions that might be able to exhibit similar behavior.
When the voicecoil is carrying a current and it moves near the pole piece, the eddy currents in the pole piece create a nonlinear braking effect which is familiar in demonstrations of powerful magnets dropped on copper plates. But this distortion follows a different pattern, H3 rises quickly with level as you would expect. I think this is more likely to be a problem with equal or underhung designs since the eddy currents appear at the coil edges which are further from steel in overhung designs and have relatively less flux density.
If we were to innocently wrap a coil around a core of the same material and it did NOT exhibit well-known properties of magnetics, we would win a Nobel prize. There is not much else in a speaker besides poor glue junctions that might be able to exhibit similar behavior.
When the voicecoil is carrying a current and it moves near the pole piece, the eddy currents in the pole piece create a nonlinear braking effect which is familiar in demonstrations of powerful magnets dropped on copper plates. But this distortion follows a different pattern, H3 rises quickly with level as you would expect. I think this is more likely to be a problem with equal or underhung designs since the eddy currents appear at the coil edges which are further from steel in overhung designs and have relatively less flux density.
Last edited:
Wolfgang Klippel even overcomplicated current-drive:
From:
https://www.klippel.de/fileadmin/_m...daptive_Controller_with_Current_Sensor_02.pdf
(With this paper he acknowledges that the voltage drive does not work)
The only really noteworthy thing he did was to eradicate the "cup spider":
From:
https://www.klippel.de/fileadmin/_m...daptive_Controller_with_Current_Sensor_02.pdf
(With this paper he acknowledges that the voltage drive does not work)
The only really noteworthy thing he did was to eradicate the "cup spider":
Last edited:
@Sonce
Wolfgang's work, whilst informative, like all science, continues to be developed.
AFAIK, he is yet to discuss Sd(x) as a parameter for distortion. This is something discovered by researchers at DTU some time ago, and re-discovered/uncovered by Carsten Tinggaard and Lars Risbo whilst working on their transducers at Purifi Transducer Technology.
One could argue that the rubber surround causing problems was uncovered and developed into product a long time ago. I'm talking about the Nautilus 801, introduced in 1998,
replacing the rubber surround of the midrange driver with a thin foam edge, later to be named Fixed Suspension Transducer...
Perhaps it was a fluke. Or perhaps unpublished as (academic) research but used as a trade secret.
Whatever marketing you want to call it, that is VERY minimal Sd(x) variation.
And the harmonic and intermodulation distortion is VERY low, I had to acquire new equipment and methodologies to test.
Remember, 1998 is before Farina published his exponential sine sweep method.
I encourage you to maintain a healthy skepticism.
Wolfgang's work, whilst informative, like all science, continues to be developed.
AFAIK, he is yet to discuss Sd(x) as a parameter for distortion. This is something discovered by researchers at DTU some time ago, and re-discovered/uncovered by Carsten Tinggaard and Lars Risbo whilst working on their transducers at Purifi Transducer Technology.
One could argue that the rubber surround causing problems was uncovered and developed into product a long time ago. I'm talking about the Nautilus 801, introduced in 1998,
replacing the rubber surround of the midrange driver with a thin foam edge, later to be named Fixed Suspension Transducer...
Perhaps it was a fluke. Or perhaps unpublished as (academic) research but used as a trade secret.
Whatever marketing you want to call it, that is VERY minimal Sd(x) variation.
And the harmonic and intermodulation distortion is VERY low, I had to acquire new equipment and methodologies to test.
Remember, 1998 is before Farina published his exponential sine sweep method.
I encourage you to maintain a healthy skepticism.
Last edited: