If you are concerned about audibility of distortion in speakers in general, think about it: there is usually quite high Q peak with the breakup. If you have low pass filter below the breakup, that means there is not much "main signal" at the breakup left to mask any harmonics boosted by the breakup. Harmonics are generated within the passband but extend beyond, inevitably some of it lands on the breakup. It is like driver distortion waving hand "here I'm, boosted by 10db and unmasked, unrelated high Q peak of noise, nice to meet ya all"
Your passive low pass filter might increase the impedance already, so its already equalized and fine, but for example on a active system there is no such thing.So, the worse the motor the more there is distortion as current for given excursion. The lower the impedance of circuit between driver terminals at the breakup frequency the more distortion current gets into the circuit. So, active speakers are more affected than passive, high damping factor amplifiers perform worse than old tube amps as they provide lower circuit impedance, thick and short audiophile grade cables work worse than lamp cord. Also the more severe the breakup the more it amplifies and the worse the motor the more there is distortion current ready to be amplified by the breakup. So this would be most audible with active system that sends high breakup bad motor driver to high excursion, spice it up with audiophile amp and cabling of high damping factor, iow low impedance
You can reduce audibility of it by designing a system that reduces excursion, having a driver with good motor and low breakup peak, also by increasing impedance of the circuit in series with the driver a the breakup. Listening off-axis and so on.
Having said it, I'm not sure if I can hear any difference with series impedance or not. Perhaps I do side by side comparison one day. It is quite low distortion setup anyway so perhaps it is not audible at levels I listen to. I intend to get very good sound quality so I see no reason to overlook this, as it is potentially very audible, thus having passive parts mixed in to active system just in case
Last edited:
Basically, if your project has a driver with "severe" breakup peak and you are not on tight budget, it might be good idea to invest for parallel notch in series with the driver, unless the xo already provides elevated impedance.
For tweeters with breakup around 30kHz, why not, I'm not sure if that is perceivable though. My hearing drops out octave below, except for some reason very high frequency sounds behind are somehow perceived through cranium or something
For tweeters with breakup around 30kHz, why not, I'm not sure if that is perceivable though. My hearing drops out octave below, except for some reason very high frequency sounds behind are somehow perceived through cranium or something
It's exactly what I ment, because series notch in parallel with driver filter doesn't work The impedance must increase in series with voice coil, so the notch filter components are hooked up in parallel with each other, but in series with the driver If only series or parallel is mentioned, it is easy to confuse which one it is
To illustrate what they look like:
Parallel notch is name of this filter, it's in series with the driver.
Here filter whose name is series notch, but it is in parallel with the driver. This does not work, because amplifier short circuits it if you analyze the circuit from driver perspective.
The impedance must increase in series with voice coil, so the notch filter components are hooked up in parallel with each other, but in series with the driver If only series or parallel is mentioned, it is easy to confuse which one it is To illustrate what they look like:
Parallel notch is name of this filter, it's in series with the driver.
Here filter whose name is series notch, but it is in parallel with the driver. This does not work, because amplifier short circuits it if you analyze the circuit from driver perspective.
Last edited:
There are plenty of very expensive drivers that might be good for many other reasons, but actually have severe break-up
In fact, the majority of alu or ceramic cones have this, where (3rd order) distortion is often around de 800-2000Hz region.
So crossing lower isn't an option either.
But anyway, subjective pricing or audibility is not very relevant here.
It's about a general solution that might work for plenty of other situations.
IF it works in plenty of other cases, this is even an interesting solution to maybe spend resources or budget otherwise.
Since not every design lends itself for infinite budget and money.
(some people seem to have a never ending amount of that growing on a tree or so?)
Btw, your whole explanation is exactly the reason why I briefly mentioned CC amplifiers.
It's being said that (maybe) not everything has an effect on this.
But since a series notch works on a very similar way (seen from a current point of view), this leaves quite some gaps still in the explanation as well as more questions in sense of actual practical results.
At this current stage, results are just far to vague to draw any brought general conclusions that are useful as building blocks.
Extremely promising, but not quite all set in stone just yet.
Series notch in parallel?
Draw simple circuit consisting of power amplifier with low output impedance, and then a woofer connected to it. Now, add another voltage source next to the driver, backEMF through which some of the driver motor distortion mechanisms affect! Now, using thevenin equivalent and replace the power amplifier with a short (or very low impedance). Now you have circuit with distortion voltage source in series with driver. You can add series impedance any point in the circuit to reduce current from the distortion voltage source, but if you have the series notch in parallel it does nothing as it is shorted (through power amplifier very low output impedance you replaced with a short).
It is the distortion mechanisms that make back EMF voltage that are affected by all this. The voltage gets varied as the driver moves, which immediately appear as current over what ever impedamce there is which immediately affects force what makes cone move and you measure acoustic distortion.
Breakup does its thing like always, but we've just reduced all current in the circuit by addin parallel notch in series
you could use current control amp instead of voltage amp, doing the same analysis again you'd see the circuit impedance for back EMF is already high and distortion is reduced. But also electrical damping is gone for woofers resonance. You could lower impedance by adding series notch to allow electrical damping at the driver resonance, if you wish But yeah its just explanations without proof, but this is how it appears to work. It is quite easy concept in a way, easy to utilize with what ever project, just analyze circuit impedance from driver perspective, driver as voltage source.
Last edited:
You keep repeating the same thing.Series notch in parallel?
Draw simple circuit consisting of power amplifier with low output impedance, and then a woofer connected to it. Now, add another voltage source next to the driver, backEMF through which some of the driver motor distortion mechanisms affect! Now, using thevenin equivalent and replace the power amplifier with a short (or very low impedance). Now you have circuit with distortion voltage source in series with driver. You can add series impedance any point in the circuit to reduce current from the distortion voltage source, but if you have the series notch in parallel it does nothing as it is shorted (through power amplifier very low output impedance you replaced with a short).
It is the distortion mechanisms that make back EMF voltage that are affected by all this. The voltage gets varied as the driver moves, which immediately appear as current over what ever impedamce there is which immediately affects force what makes cone move and you measure acoustic distortion.
Breakup does its thing like always, but we've just reduced all current in the circuit by addin parallel notch in series
But yeah its just explanations without proof, but this is how it appears to work. It is quite easy concept in a way, easy to utilize with what ever project, just analyze circuit impedance from driver perspective, driver as voltage source.
I am familiair what lumped circuits are, thanks (I even showed one couple of posts back).
It's obvious that we are talking about a parallel notch in series.
No idea why there is confusion about this even, this was already clear from de beginning?
As far as I am aware there is not even another way of making a passive NOTCH filter
It's all about impedance, so as @markbakk rightfully wondered, this would also work with an high(er) order passive filter.
Since you're blocking the same things, as well as a CC amplifier, since it does the same thing as well.
(thevenin equivalent would be open instead of a short)
You literately describe this in your explanation.
The mechanism behind it is and was already very clear.
The ins- and outs of it, investigated with practical measurements, are not.
Didn't notice this was my quote.
Purifi doesn't show a different result, so I don't get the question?
With series notch I mean parallel notch filter in series with the woofer.
Which is in short a series notch filter.
In a passive filters you can either have a series notch or parallel notch (which requires an addition series impedance or resistance).
Otherwise it's not a notch but a (very small) bandpass.
I think that the general conclusion could be that the distortion-induced currents from the motor of an electrodynamic transducer should be dealt with if those currents induce secondary distortion products that are worrisome. In the Purifi example we see -60dB vs -70dB, 0,1% vs 0,03%, very narrowband, and we have to remember that was due to a nasty response peak only 1,2 octave or so outside the passband.
You could ask yourself the question: why use this driver up to 2kHz at all? In the (not abundant) designs I did with metal cones I kept farther away from such problems and likely this 'problem' never came up. So is this relevant in everyday practice?
Current steering on the other hand seems to become the perfect panace, working not only for better low frequency behaviour but also to counter these high frequency problems. We should get into that more, I guess. KEF and the like researched this more than 30 years ago... we probably are afraid of the feedback- or feedforward challenges?
You could ask yourself the question: why use this driver up to 2kHz at all? In the (not abundant) designs I did with metal cones I kept farther away from such problems and likely this 'problem' never came up. So is this relevant in everyday practice?
Current steering on the other hand seems to become the perfect panace, working not only for better low frequency behaviour but also to counter these high frequency problems. We should get into that more, I guess. KEF and the like researched this more than 30 years ago... we probably are afraid of the feedback- or feedforward challenges?
That is only in this particular case.
But there is a strong case to make to choose another driver that meets other criteria but can't be picked because of just a nasty 3rd order cone breakup mode in the distortion graph.
There are quite a few woofers that fall within that group; performing and looking great but have a nasty leftover from cone breakup.
I don't wanna discuss this in this topic again, but see the paper posted a couple of posts/pages back.
Current steering actually does NOT work well for lower frequencies, mostly also from a practical point of view.
Voice coil magazine even did a pretty interesting article about this many years ago (I will find it if you're interested) and came to the same conclusion (also shows this).
But it works super well above say 2xFs.
That being said, finding an amplifier that works well and is stable with current steering is quite the challenge.
Yes I am aware that certain brand here are capable, but not every project lends itself for that kind of budget or space/size.
Btw, what is kinda confusing is that other measurements also show slightly different behavior when it comes down to distortion;
https://www.erinsaudiocorner.com/driveunits/purifi_ptt6.5x04-naa-08/
https://hificompass.com/en/speakers/measurements/purifi/purifi-ptt65x04-naa-08a
Haha, yeah just having fun with it
edit. I've got woofer in use with quite serious breakup peak. Currently having series inductor with it and impedance is quite high already and not sure if I hear any of it with or without. Ordered some more parts to try a notch for fun. Parts should arrive soon but I'm not sure when I have chance to measure. I need to record how they sound, in order to AB, gonna use dynamic directional mic.
I did measure distortion with or without the series inductor and later on noticed the distortion was mostly below noise floor haha, not a successful experiment. Well, it's rare to have everyone out the house to take such measurements.
Last edited:
If you don't mind, I'd like you to explain to me what the D2 does at 1kHz, please
Background:
https://www.diyaudio.com/community/threads/drive-current-distortion-measurement.402566/#post-7434990
Last edited:
probably some cone breakup distortion reflected back electrically. Was the amplifier drive voltage the same in the first plot and the overlay with 20 ohms in series?
That is my best guess too.
Vice versa ;-)
No, the measured current was about the same.
When I finally had the plots on the screen, I just thought: That's very interesting.
Last edited:
