I tried to explain earlier that just series coil makes better performance here, than a notch I have parts for. Impedances illustrated in previous post.
Since distortion is reduced somewhat smoothly through out the breakup region there is still the breakup amplified distortion visible as peak. If I had used a notch instead, the peak would have smoothed out.
Since distortion is reduced somewhat smoothly through out the breakup region there is still the breakup amplified distortion visible as peak. If I had used a notch instead, the peak would have smoothed out.
Jeah 🙂 slight correction, by increasing series impedance with the driver, to reduce current Le(x) makes. Any current in the circuit makes force to cone so reducing current reduces acoustic output of the Le(x).
Instead of series inductor one could use high output impedance amplifier, and poor quality thin speaker cables, series resistor, or passive crossover network. Anything that takes load impedance of driver account, higher. This is distortion reduction mechanism of Current drive, with voltage amp.
I think the series coil is the most elegant solution. It can be removed anytime.
Oh apologies, I think something got lost in translation here, haha
Yeah it would be interesting to see what just a pure notch does as well.
Anyway, this does follow up the idea of the actual impedance that counts, like I mentioned before.
This is also interesting for tweeters in a active speaker system.
A passive attenuator already has the benefits of bringing down the noise of the amplifier (and all other parts), but in this case actually improving the distortion of the tweeter at the same time 😉
So I assume that the reduction in distortion is directly proportional?
So attenuating 6dB will result in 6dB reduction of the distortion.
Would you be able to measure this as well? 😇
In just a finished design, the issue with inductors, is that they have quite a bit of series DC resistance.
Or you have to take the distortion of ferrite inductors into account when they reach certain currents.
This DC resistance does influence the behavior of the speaker, especially the low end.
That doesn't have to be a problem, but it's something to look out for.
So practically speaking, I still think the notch is a little better.
Especially when it's already out of the pass band.
Yes, except, if you also have a protection capacitor in series as well, which would reduce electrical damping at the resonance, which might affect sound. Instead of cap one can use just single resistor and then inductor parallel with driver to make a high pass, relatively low impedance for the electrical damping but high on higher frequencies to reduce any distortion current. This is dicussed on this very thread earlier.
I've got this measured as well, and with audio recording, I'll try to EQ match them to see if there is audible difference.
Yeah it should, if driver impedance was 8ohms, and you increase the series impedance with another 8ohms the distortion should halve compared to situation with just very low amplifier output impedance, right? Also acoustic output halves but you could just boost it back with amplifier, which drops efficiency though as the extra amplifier power is not making sound. But, in some cases like here the output reduction is fine, I need to do it anyway.
I bet this works to some extent, until close to drivers xmax where the distortion perhaps rises more rapidly.
Yeah, and depends on the driver and application. Here with this driver its quite nasty, multiple peaks past that 4.5kHz, also the inductor makes fine response so a nice thing. Air core inductor with perhaps 0.5ohm resistance if I remember. This shows up as boost around Fs, it reduces driver electrical damping a bit allowing it to overshoot compared to datasheet frequency response without the extra resistance 😉 In my application this is fine as well.
Here is compression driver with 5uf series capacitor:
And here the same thing with resistor and inductor making ~similar high pass.
The driver has impedance peak around ~500-600Hz, you can see driver losing electrical damping with the high pass capacitor as peak in frequency response, innocent high pass filter isn't. Using resistor and coil instead of cap, the electrical damping is maintained better, excursion reduced (in this case with no DSP high pass) and 2nd harmonic drops, is my reasoning over this data. I guess, if I matched these with DSP they'd likely look about the same. I bet difference would show up with microphonics if any, the horn picks up sound of rest of the speaker and without electrical damping it's likely gonna make the driver resonate, unnecessary excursion, affecting electrical parameters of the driver and distorting sound on the pass band. I guess multitone test with the whole speaker would show it, if there is any difference.
These were not matched with DSP so it's not again straight forward comparison, but shows that on an active setup there is likely benefit manipulating circuit impedance, and there is difference how you do it. Active setups usually just short the driver (with the amplifier output impedance), but could be done more elaborately as well 🙂
edit. here is impedance plot for the driver
And here the same thing with resistor and inductor making ~similar high pass.
The driver has impedance peak around ~500-600Hz, you can see driver losing electrical damping with the high pass capacitor as peak in frequency response, innocent high pass filter isn't. Using resistor and coil instead of cap, the electrical damping is maintained better, excursion reduced (in this case with no DSP high pass) and 2nd harmonic drops, is my reasoning over this data. I guess, if I matched these with DSP they'd likely look about the same. I bet difference would show up with microphonics if any, the horn picks up sound of rest of the speaker and without electrical damping it's likely gonna make the driver resonate, unnecessary excursion, affecting electrical parameters of the driver and distorting sound on the pass band. I guess multitone test with the whole speaker would show it, if there is any difference.
These were not matched with DSP so it's not again straight forward comparison, but shows that on an active setup there is likely benefit manipulating circuit impedance, and there is difference how you do it. Active setups usually just short the driver (with the amplifier output impedance), but could be done more elaborately as well 🙂
edit. here is impedance plot for the driver
Last edited:
I haven't used a series cap in most designs and it is totally fine.
Even in professional designs, not one single tweeter that has blown up because of a DC issue.
Using a inductor doesn't make much sense to use.
I am gonna be honest, I am still extremely skeptical about this, 😀 😀 because it feels way to good to be true.
But at least we have a lot more data to go off. I will do some testing myself as well at some point.
I just also really wanna ask people to do the same as well!
I have to say it kinda clicks in my head when we look at this from a transmission line point of view.
(I am talking about electronics, not the speaker cabinet).
Where you don't have a proper impedance match at the termination of the end of the line.
Since a voltage source (= amplifier) is just fully wide open, it reflects back.
This is now being absorbed by the passive network.
Obviously at a loss of energy (and efficiency), like even a real transmission line.
edit: But this would still mean that a constant current amplifier would do equally well or actually better!!
Since there won't be any signal loss.
But going back to the tweeter, the total power and energy at these higher frequencies is pretty low.
Besides, compared to a fully passive crossover, you don't loose anything.
Last edited:
Yeah you must test it yourself, but it totally makes sense 🙂 This is same mechanism as with current drive, which is just high circuit impedance.
Explained in various threads multiple times, one more 😀 the acoustic distortion reduction is due to following train of though:
any acoustic sound is from cone movement, which is moved by current in the voice coil making force with the magnetic field in the motor.
As our audio power amplifier makes the cone move, the electro-magnetic motor makes back EMF voltage, it's a generator and motor at the same time. Back EMF voltage makes current over the circuit impedance, simplified, over the impedance between driver terminals. Think, as our input signal makes cone move, the driver itself with it's parameters changing with voice coil position makes another (error) voltage source in the circuit, simultaneously in real time making current in the circuit which immediately also makes force in the motor moving the cone. This is the distortion mechanism we see reducing with increased impedance.
If you raise circuit impedance, the current generated by the driver itself is reduced per Z = U / I. Thus, even the distortion mechanism exists as before as voltage amplifier we got less current from it, less force in the motor, less acoustic sound.
It is simple to think there is voltage amplifier inside the driver in series with it, and if allowed to work on low impedance between driver terminals it makes maximal current in the circuit, so maximal effect into acoustic domain, maximal distortion. So, increasing impedance prevents the driver emit some of it's own motor non-linearities into acoustic domain 😉
Explained in various threads multiple times, one more 😀 the acoustic distortion reduction is due to following train of though:
any acoustic sound is from cone movement, which is moved by current in the voice coil making force with the magnetic field in the motor.
As our audio power amplifier makes the cone move, the electro-magnetic motor makes back EMF voltage, it's a generator and motor at the same time. Back EMF voltage makes current over the circuit impedance, simplified, over the impedance between driver terminals. Think, as our input signal makes cone move, the driver itself with it's parameters changing with voice coil position makes another (error) voltage source in the circuit, simultaneously in real time making current in the circuit which immediately also makes force in the motor moving the cone. This is the distortion mechanism we see reducing with increased impedance.
If you raise circuit impedance, the current generated by the driver itself is reduced per Z = U / I. Thus, even the distortion mechanism exists as before as voltage amplifier we got less current from it, less force in the motor, less acoustic sound.
It is simple to think there is voltage amplifier inside the driver in series with it, and if allowed to work on low impedance between driver terminals it makes maximal current in the circuit, so maximal effect into acoustic domain, maximal distortion. So, increasing impedance prevents the driver emit some of it's own motor non-linearities into acoustic domain 😉
Well, for other back EMF issues, this is exactly the model and equivalent circuit you make 😉
Including transmission line termination.
But that's why I always said that a constant current amplifier must do the same or better.
It just doesn't make any sense why it wouldn't.
However, I am still confused about this:
I had memory there was some word play around this subject on other thread(s) earlier, about how the breakup is amplifier in acoustic domain and how the peaks in distortion are because the breakup amplifies the distortion and why a notch filter would remove the peak in distortion plot.
The peak(s) appear only on-axis distortion measurement, because the breakup makes peak only on-axis, but there could be dip somewhere off-axis instead. The breakup makes erratic response to all directions and the distortion follows.
If one is concerned about the distortion peaking due to breakup, just listen off-axis and bum it's gone 😉
Uploaded files to listen the woofer with series inductor, and without: https://www.diyaudio.com/community/threads/listening-about-things.402922/post-7444574
There are links to files to listen the compression driver setup in post #88 as well.
There are links to files to listen the compression driver setup in post #88 as well.
Hi, forgot to comment on this. Yes of course.
Difference is that backEMF at driver resonance is useful, while above it isn't. Now, in the world that revolves around voltage amplification, including drivers we can buy, it is pratical to use voltage amp and driver driver designed for it, and then reduce backEMF generated current above the resonance.
You could design driver and amplifier for proper current drive if you wish. or just use off the shelf parts to build something that doesn't rely on electrical damping driver resonance, or allow the damping with series notch paralöeö with driver, or something.
Analyzing the circuit impedance both directions allows you to get best performance what ever it is that is on your table 😉 How about ironless high impedance driver? 🙂
Nelson Pass thought it was good at higher frequencies when using it to give equalisation to fullrange drivers. Do you have a problem with it?
Hi,
I'm not having problem with other people systems, of course not, use how ever suits the application and what you are trying to achieve.
Are you refering Nelson Pass using current drive amplifier with fullrange driver? High circuit impedance gives rise to high frequencies because circuit impedance rises relatively less on highs than on midrange! The increased response of high frequencies is not due to what happens to back EMF but how much current goes through the driver relative to mid frequencies. On current drive it would be flat current regardless of driver impedance, or how ever you want to equalize, so this is natural response of the driver! With voltage amplifier there is more current through midrange where impedance is very low, and less current on high frequencies where Le makes impedance high. What makes frequency response rise with current drive amplifier, or high circuit impedance, is because the driver was constructed with assumption of very low circuit impedance (voltage amplifier) and frequency response is tailored flat for voltage amplifier use! This means that with current drive such driver would have rising response. Peak at woofer resonance is different because also the electronic damping, or useful back EMF, is involved.
What I'm referring to is explained here:
https://www.edn.com/loudspeaker-operation-the-superiority-of-current-drive-over-voltage-drive/
Simply, phase of current from backEMF is perpendicular to cone motion/current so the back EMF has no effect on "controlling movement" above driver main resonance, so back EMF is not useful to anything on these high frequencies and makes acoustic distortion so could be reduced. At drivers resonance the current would be opposite phase and useful as electronic damping, useful back EMF, but even that isn't necessary. You could choose to use it or not, what ever it is you try to achieve.
Read the article as many times as necessary. Have fun!🙂
edit. also part 2 of the article of course https://www.edn.com/loudspeaker-ope...y-of-current-drive-over-voltage-drive-part-2/
I'm not having problem with other people systems, of course not, use how ever suits the application and what you are trying to achieve.
Are you refering Nelson Pass using current drive amplifier with fullrange driver? High circuit impedance gives rise to high frequencies because circuit impedance rises relatively less on highs than on midrange! The increased response of high frequencies is not due to what happens to back EMF but how much current goes through the driver relative to mid frequencies. On current drive it would be flat current regardless of driver impedance, or how ever you want to equalize, so this is natural response of the driver! With voltage amplifier there is more current through midrange where impedance is very low, and less current on high frequencies where Le makes impedance high. What makes frequency response rise with current drive amplifier, or high circuit impedance, is because the driver was constructed with assumption of very low circuit impedance (voltage amplifier) and frequency response is tailored flat for voltage amplifier use! This means that with current drive such driver would have rising response. Peak at woofer resonance is different because also the electronic damping, or useful back EMF, is involved.
What I'm referring to is explained here:
https://www.edn.com/loudspeaker-operation-the-superiority-of-current-drive-over-voltage-drive/
Simply, phase of current from backEMF is perpendicular to cone motion/current so the back EMF has no effect on "controlling movement" above driver main resonance, so back EMF is not useful to anything on these high frequencies and makes acoustic distortion so could be reduced. At drivers resonance the current would be opposite phase and useful as electronic damping, useful back EMF, but even that isn't necessary. You could choose to use it or not, what ever it is you try to achieve.
Read the article as many times as necessary. Have fun!🙂
edit. also part 2 of the article of course https://www.edn.com/loudspeaker-ope...y-of-current-drive-over-voltage-drive-part-2/
Last edited:
It is key to understand that driver datasheets show frequency response swept with constant voltage, which means the graph shows frequency response with low circuit impedance. Think about that for a second, the driver is designed and manufactured so that it's impedance makes equalization with voltage amplifier so that acoustic output is flat (or what ever the target was), the driver defines current in the circuit which means it regulates the acoustic output. Rising voice coil impedance reduces current toward high frequencies, so the equalization actually happens with voltage amplifier, and current amplifier woud reduce the equalization!🙂
Driver dominating circuit impedance with voltage amp is also the reason why there is lot of distortion because it all comes through to acoustic domain, and why increasing circuit impedance reduces the distortion, it simply reduces driver domination over the circuit impedance.
Now it's easy to see why frequency response differs with high circuit impedance, why high (adjustable?) output impedance amplifier would be nice with fullrange drivers, smiley face EQ and distortion reduction, what's not to like. You can play with this in simulator.
I suppose it would be possible to make a driver whose frequency response was nice with high circuit impedance. But how high?🙂 It's just easier, and perhaps how the drivers work out naturally, to assume roughly 0ohm circuit impedance and relying driver impedance dominating and making flat frequency response. I don't know how drivers are made and why, this is just reasoning based on observation. I bet one could make driver and amplifier work as unit for what ever performance goal. This wouldn't be sexy in the market though, as you couldn't upgrade parts of it, that's bad for money flow.
Driver dominating circuit impedance with voltage amp is also the reason why there is lot of distortion because it all comes through to acoustic domain, and why increasing circuit impedance reduces the distortion, it simply reduces driver domination over the circuit impedance.
Now it's easy to see why frequency response differs with high circuit impedance, why high (adjustable?) output impedance amplifier would be nice with fullrange drivers, smiley face EQ and distortion reduction, what's not to like. You can play with this in simulator.
I suppose it would be possible to make a driver whose frequency response was nice with high circuit impedance. But how high?🙂 It's just easier, and perhaps how the drivers work out naturally, to assume roughly 0ohm circuit impedance and relying driver impedance dominating and making flat frequency response. I don't know how drivers are made and why, this is just reasoning based on observation. I bet one could make driver and amplifier work as unit for what ever performance goal. This wouldn't be sexy in the market though, as you couldn't upgrade parts of it, that's bad for money flow.
Last edited:
That's quite obvious I think.
Even Voice Coil magazine came to the same conclusion over a decade ago.
(I can look up the year and month they tested it, if you want)
I personally don't understand the thought why one driver has to do the entire frequency response.
I am always talking about a driver that's being used above its Fs and ideally with a (relatively) flat impedance curve.
Steep rising impedance is obviously very problematic for current drive.
Anyway, my point was about tackling distortion, not another discussion about constant current drive, lol
Previously it has been said that it's not the same as putting some passive filter componentes in series.
But that argument still doesn't make any sense to me.
Yeah, it doesn't make sense, which text you refer to? It is about load impedance for the driver, be it amplifier output impedance, kilometer of cable, or any combination of impedance network between driver terminals. Only difference I see between "current drive" amplifier or "voltage amplifier" is just their output impedance within this context. All the work is done within the driver so to speak, it generates it's own distortion current and emits it acoustically assuming low impedance between terminals. Actual power amplifier has nothing to do with the distortion we see here, other than what it's output impedance is and how much it affects load for the driver, how much amplifier allows driver to affect current in the circuit. Amplifier can be simplified away from the issue, replaced by a resistor equivalent to it's output impedance.
ps. In this regard Esa Meriläinen articles I linked earlier have kind of weird tone, blaming amplifiers on things, when the actual culprit is just the circuit impedance. Perhaps this stuff would be better adopted if the message was expressed differently and not through accusation of current industry preventing people thinking / realizing the core idea behind it. Analyzing circuit impedance one can utilize "current drive" on what ever speaker system no matter what the amplifier or drivers are.
I can understand the thought, and why it is appealing, but it just doesn't deliver. Single fullrange driver system is both bandwidth and output limited, tight sweetspot, erratic polar pattern to name few.
Make a thought experiment, think any single source for sound, ideal flat disk piston or pulsating point in space or anything you can come up with, and it just doesn't work 🙂 Except if the pulsating point in space is in another dimension somehow so that it's size in this realm stays relatively small while still exciting molecules a lot, but even that wouldn't work in room due to sound wavelength being large in relation to the room, making modes. You'd need multiple sources, and to preserve stereo you need two for the highs and multiple for the lows, bam, multi-way system no matter how idealized if you have to deal with acoustic domain. "Fullrange" is misleading as name, because fullrange driver cannot make fullrange system. Or can, but it's performance is limited. Perhaps performance can be good enough, I enjoyed a fullrange driver system for long time, but multiway system takes performance much further.
Problem, or issue, is that sound wavelength varies dramatically while physical objects size doesn't, or if it would you'd need to have a safety cage not to get injured. And it's a problem even without any electric and mechanical issues that come with real implementation of a transducer. Even imaginary concepts don't quite work ideally. Neither for multiway speaker, but one can go little bit further with multiway speaker, pushing issues outside pass band, separation of concern, and tailor fit for the environment you are at.
Last edited:
Thinking about it, everything doesn't need to be ideal or perfect, just good enough and realistic in a way that the sound is somehow what is expected, what the past experiences and real world interaction has defined to you what is "good sound". If real world single fullrange driver system is enough performance, then why not, couldn't be simpler, enjoy music 🙂 Want more? Go multiway, big, optimize for the room or what ever situation, optimize as far as possible to get every bit of perfomance in the particular scenario. If not willing to go that far, then make some suitable system that works averagely quite nicely on most scenarios. Or is practical. Or what you've been sold by marketing teams trying to get into your pockets 😀 What ever, everything is fine for me, goal is to have fun time and we are fortunate enough to have the possibility.
Last edited: