The amplitude response of the current output amplifier was digitally equalized to prevent huge modulation by the driver impedance curve, yes especially near resonance.
Sorry, I didn't realize those were acoustic measurements. I kept seeing the word "current". :/
I wish we knew what drivers you were using to get an idea of their design.
I wish we knew what drivers you were using to get an idea of their design.
Sorry to everybody, I thought Klipppel, his papers, investigations and his analyzers were sort of a baseline for everybody talking about speaker drivers seriously.
Like every electronic design house has an AP, every speaker design house has a Klippel.
Like every electronic design house has an AP, every speaker design house has a Klippel.
Double p (pp) would be enough 😉
Yes, his work is great and essential and he is also very responsive.
Just shows how amplifier centric most here are. Plus of course real engineering can so eaily get in the way of a good marketing story 🙂
I am still amplifier centric, I just have to learn all this loudspeaker junk now that I'm curious about current drive! 
It's also uncomfortable for some to realise they are spending far too much effort in the wrong place 🙂
I mean what idiot would think a DSP in front of some LM3886 amplifiers and then spending their time to get the speakers right would be a good approach...
Yes.
It is VERY simple to understand. A speaker is a moving motor. Its displacement DO NOT follow exactly the signal. Under a pulse, he will resonate on several periods (see the waterfall) after the signal has returned to 0V. And this movement produce a current across the impedance of the amplifier+cable. This signal will enter in the feedback line of the amp, and the amplifier will try to compensate-it. This added work (added signal in the feedback path) will add distortion, of course.
A way to measure the back EMF should be to glue the moving coil and measure the difference between the glued and not glued coil.Anyway, I do not see any interest in making such a measurement.
I use to flatten the impedance curve of each of my speakers. At the resonance, with a parallel network composed of an inductance+a capacitance+a resistance in serial, in order to get a flat impedance corresponding to the DC resistance. And, for the upper part of the curve, by a network composed of a capacitance in serial with a resistance+an other capacitance in parallel.
La correction d'impédance RC et RLC série
La correction d impédance RLC série
The improvement in the quality of reproduction with most of the amplifiers is so obvious that i don't understand why manufacturers do not apply this method as a requisite.
The reason is easy to understand: The amplifier has no more to compensate a huge difference in current for the same voltage while the frequency of the signal changes. This help a lot to reduce distortion and to dump the speaker by itself.
One thing is interesting is that, once you have made the "motional" (at the resonance) compensation of the boomer in free air, it changes the Thiele/Small parameters of the speakers in such a way that this same compensation applies without any change of the LC values in a bass reflex. You will get a flat impedance curve as well, just tuning the value of the resistance.
In a closed box, just do it once your boomer is in its enclosure.
An other advantage of this is it simplify the calculation of any passive filters: No adjustment to compensate the variations of the impedances of a given speaker. For a 8 Ohm speaker, the DC is around 6 ohms: this will be the impedance all over the audio bandwidth.
To resume, and in my point of view, a good speaker should had a flat acoustic response curve AND a flat impedance curve.
The improvement is easy to listen to. Just plug on and off this parallel network. And you can *see* the differences in the acoustic response curves as well. It works !
Last edited:
I think this is controversial.
Eg: To even out the impedance rise near woofer resonance you add a parallel damper to the woofer to bring the impedance back down towards nominal. This will result in a more consistent load for the amplifier but it will also result in higher energy demand from the amplifier.
So you are arguing that consistent amplifier load with frequency is better even though it places higher power demands on the amplifier?
I was suggesting the use of prediction and control software to calculate where the vc was in it's excursion space, and compensate for the driver non linearities by modifying a transfer function. To the first order, it would linearize all frequencies through the full excursion. With enough horsepower it can also capture velocity modulation issues as well.
In the case of BL drop, the dsp would increase gain to maintain a steady BLI product.
Since most bass boxes with drivers have resonances, the dsp has to span multiple cycles, guessing ten would suffice.
Jn
Ps. You can use the BL product graph as per klipple, and all that is left is vc position prediction software.
Last edited:
We were discussing he Smyth Realiser the other day, I hadn't realised that a new version was on its way. Apparantly a year late even after kickstarter funding but is being reviewed smyth Realiser A16
Tyll likes it! CanJam SoCal 2018 - Smyth Research Realiser A16 Headphone Surround System | InnerFidelity
If the production units come out close to the kickstarter price then I think I could be seriously tempted to start saving for that.
Tyll likes it! CanJam SoCal 2018 - Smyth Research Realiser A16 Headphone Surround System | InnerFidelity
If the production units come out close to the kickstarter price then I think I could be seriously tempted to start saving for that.
I find these statements confusing. Are you implying that one can take any commercial 3 way speaker and simply add traps and whatever at the input with no modifications of the internal crossover components?
I would assume changing the response is audible and if you do isolating the amplifier loading effects is virtually impossible. IMO the only valid test is to compare both techniques on the same set of drivers equalized to have exactly the same acoustic output. There is no evidence that this has ever been done.
To address traderbam's point, at resonance the impedance is real and purely resistive so the class A/B argument is not valid, the trap only makes the amplifier deliver more power again at 0 phase simply a loss of efficiency.
Last edited:
Thank you.
It's interesting the similarities in approach he takes to what I'm doing. Especially the state variable (predictor) approach.
A shame he did this work a decade and a half ago, now I can't claim to be the smart guy in the room. (Well, I have kids so that thought crashed and burned about 20 years ago.)😛
Jn
Yes. And, as I believe your power supply and amplifier is correctly designed to provide more power with low impedance loads, it makes no difference. Of course, if it is a portable bluetooth speaker, battery powered, i will not recommend-it for this application ;-)
- Status
- Not open for further replies.