I was answering this statement.
I agree a Qms of 3,5 is not that low. I wish there was drivers with a Qms down around 1,0. That would be nice for current-drive without excessive EQ and other methods of linearising the spl response.
Cheers,
Johannes
What is the goal with reducing Bl^2 on a low Qms driver?
I wan't as much Bl^2 as possible to have a high efficiency and good transient response. I can't see any reason to lower Bl^2 in any circumstance when driving it with current-drive. A voltage drive amp will have a tough time driving a very reactive speaker.
Cheers,
Johannes
This thread is about current-drive, and with a 100 ohm out-z from the amp, i don't think your statement is true any more. Please simulate a high Qms driver with a 100 ohm series-resistance in an open baffle and compare to a low Qms driver.
Cheers,
Johannes
It's called flat LF response 🙂
You may of course prefer con belto rather than quality 😀 See previous comments by various people about "voltage amp having tough time driving a very reactive speaker"
Circlomanen, what are you using for a Current Drive amp?
You are right that a true current drive amp will not react to higher Bl^2 / Rdc (don't forget to add the Rdc) But you won't get better transient response either. 😀
At the moment a QSC RMX 1450 with 10 ohm power resistors in series with the drivers.
Previously i used a inverted Triadtron It's About Time&Ultra-Linear Line Stages with 12 - 15 ohm output impedance.
https://www.youtube.com/watch?v=WkZJfTlsvfE
I have been building SUSY circlotrons built with IRF9530 and 2,2 ohm source degenerative resistance as output stage. It sounded very good. Probably my favourite amp so far. I was to build as more powerful version, since the first one had 12 watt output power, but due to health problems I have not had the energy to do so.
Speaker DESIGNED for current drive.
Here's the simplest implementation of my idea which I think luckythedog has sorta latched on already.
If you put a speaker into a sealed box, Qms will drop & fs will rise cos the box adds its own stuff including resistance. But you can treat "speaker in box" like just another unit and measure fs & Qms again.
If you use a Current amp, this Qms will the Q of the complete system and determines the LF response EXACTLY
Q = 2 will give 6dB peak at fs
Q = 0.7 is Butterworth with -3dB at fs. This is good for a long throw 12" unit with fs (in the box) of 50Hz
I recommend Q = 1 for 0dB at fs and a small peak above, for smaller units & boxes.
You adjust the Q by stuffing the box. You must have a LOT of stuffing .. like a VERY firm cushion. Use fibreglass roof insulation to get a bit of adiabatic decrease in fs.
If you can't get the Q low enough by stuffing, glue tightly stretched silk across the unit's chassis openings for more resistance .. but stuff as much as possible first cos stuffing has other good effects.
Bl^2 / Rdc is unimportant but Circlomanen can have it has high as he wants ... that's if he has a true Current Drive amp 🙂
OK. I confess. I would want it high too 😎 but it doesn't affect LF response.
Various caveats about whether you get the full Current drive THD improvement bla bla. You probably do but I would want to check this. Besides the mumbo jumbo, this should be a rather nice speaker too.
________________________
Zo of 10R - 15R is hardly Current Drive. You are lowering Bl^2 / Rdc by 1/2 to 1/3 so it still dominates Mechanical Damping.
Anyone have experience with TRUE CURRENT SOURCE AMPS? Preferably with measurements of Zo over 20Hz - 20kHz.
Nelson Pass is pretty comprehensive about "resistor in series" stuff.
Here's the simplest implementation of my idea which I think luckythedog has sorta latched on already.
If you put a speaker into a sealed box, Qms will drop & fs will rise cos the box adds its own stuff including resistance. But you can treat "speaker in box" like just another unit and measure fs & Qms again.
If you use a Current amp, this Qms will the Q of the complete system and determines the LF response EXACTLY
Q = 2 will give 6dB peak at fs
Q = 0.7 is Butterworth with -3dB at fs. This is good for a long throw 12" unit with fs (in the box) of 50Hz
I recommend Q = 1 for 0dB at fs and a small peak above, for smaller units & boxes.
You adjust the Q by stuffing the box. You must have a LOT of stuffing .. like a VERY firm cushion. Use fibreglass roof insulation to get a bit of adiabatic decrease in fs.
If you can't get the Q low enough by stuffing, glue tightly stretched silk across the unit's chassis openings for more resistance .. but stuff as much as possible first cos stuffing has other good effects.
Bl^2 / Rdc is unimportant but Circlomanen can have it has high as he wants ... that's if he has a true Current Drive amp 🙂
OK. I confess. I would want it high too 😎 but it doesn't affect LF response.
Various caveats about whether you get the full Current drive THD improvement bla bla. You probably do but I would want to check this. Besides the mumbo jumbo, this should be a rather nice speaker too.
________________________
Zo of 10R - 15R is hardly Current Drive. You are lowering Bl^2 / Rdc by 1/2 to 1/3 so it still dominates Mechanical Damping.
Anyone have experience with TRUE CURRENT SOURCE AMPS? Preferably with measurements of Zo over 20Hz - 20kHz.
Nelson Pass is pretty comprehensive about "resistor in series" stuff.
Last edited:
As long as the Qms is above 0,8 I don't worry about flat LF response.
With current drive in open baffles, BIBs or large closed boxes it does not matter if you have a Qes of 0,001 and a Bl^2 / Rdc of 1000. Qms will totally dominate the LF-response anyway.
I can't see the problem with a very low Qes and high Bl (and yes i know there is a square and RDC in there) as long as the amp has a high output impedance.
Cheers,
Johannes
Yes. I obtained my prototype by taking a robust stable voltage amp and altering its feedback such that a current spy resistor provides the majority of feedback sensing. After all, low output impedance of most voltage amps derives from feedback.......simulation says output impedance is c 4k over the audioband, but I haven't confirmed.
This is pretty much what i have been saying all along.
And the Qms of 2,9 is not low enough for a true current drive. Since the drivers are unsuitable for current drive, i have to use as much as possible without a huge peak at Fs.
I consider a output impedance larger then the driver impedance to be more or less current drive. I don't aim for perfect infinite output impedance. I would happily use 1000 ohms, but i need a low Qms driver first.
Cheers,
Johannes
If EQ can be used to tame and tailor the LF response then why insist on low Qms drivers?
Addressing a simple peak is an easy task with parametric EQ, and as long as it is not to steep (ie not crazy high Qms together with crazy high Zout) this should not be difficult to do consistently.
With digital EQ a simple Linkwitz transform should be able to turn any 2nd order alignment into any other 2nd order alignment.
Addressing a simple peak is an easy task with parametric EQ, and as long as it is not to steep (ie not crazy high Qms together with crazy high Zout) this should not be difficult to do consistently.
With digital EQ a simple Linkwitz transform should be able to turn any 2nd order alignment into any other 2nd order alignment.
In theory this is correct, but any high Q resonant peak will absorb and store energy from harmonic content to the fundamental. Shake (or tap with your finger) a guitar or piano well below any fundamental of the strings and you will be able to hear a lot of twang emanate from it.
Even if you EQ out all the energy of the fundamental resonant frequency it would still resonate disharmoniously from the energy in the music.
Here is a Hornresp simulation comparison of the impulse-response from a B&C 12NW100 (QMS 3,6) Black line, with a Beyma 12LX60 v.2 (Qms 15,3) red line, both in a 150 liter closed box and 100 ohm Z-out from the amp.
I don´t believe a lower voltage around Fs (EQ) will "cure" the ringing impulse response from the Beyma 12LX60 v.2.
Last edited:
Well, I've just told you how to make one with existing parts and high quality.
Making the unit itself low Qms is much more difficult without introducing THD. The silk cloth is about only thing which works.
pos, doing this with loadsa fibreglass stuffing has other advantages like much better midrange. The matching is 'perfect'. Some of the resonances exacerbated by Current Drive are also treated.
luckythedog, can you post the SPICE file for your Current Amp? Load Stability is difficult with current amps .. kinda like the Holy Grail
Last edited:
Thanks for that - that is similar to typical models of phono cartridge response FWIW. I believe it to be audibly fine, preferable even, though the 3.9 Q probably preferred.........between 2 and 3 I think is the mark.
Well for these 2 examples with Zo = 100R ...
Qms=2 gives a 6dB peak at fs
Qms=3 a 9.5dB peak 🙂
Qms=15.3 23.7dB peak 😱
More boom = Good 😀
Well I'm not noting any such audible boom, with fs near 40Hz and Qts near 2.......I really find it OK without EQ, and perceive a nicely timed well controlled bass end.
Minimum-phase EQ will exactly compensate this, including any ringing (what you call stored energy). Any minimum-phase phenomenon can be compensated for using EQ. As long as we are not talking about breakups we are talking about minimum-phase phenomenons here.
External stimulus like taping the cone will of course not be addressed, but signal coming from the recording through the EQ will be.
POS,
while you can indeed use EQ to electrically attenuate a resonance in a device that only partially works in a physical device. The problem is the actual physical harmonics, 2nd, 3rd, etc. produced by the source material can match the peak you are tying to electrically cut. These harmonics will again physically excite the resonance in the device and though it should be lower in level the resonance peak will again appear if you look for it acoustically. Any high Q resonance in the device should be address in the device with mechanical correction.
I'll let kgrlee chime in on this one but that is my understanding of the problems you face with mechanical resonances. This become obvious in waterfall plots with long decay rates at resonance, the lower in level and the faster you can dissipate that energy the better.
while you can indeed use EQ to electrically attenuate a resonance in a device that only partially works in a physical device. The problem is the actual physical harmonics, 2nd, 3rd, etc. produced by the source material can match the peak you are tying to electrically cut. These harmonics will again physically excite the resonance in the device and though it should be lower in level the resonance peak will again appear if you look for it acoustically. Any high Q resonance in the device should be address in the device with mechanical correction.
I'll let kgrlee chime in on this one but that is my understanding of the problems you face with mechanical resonances. This become obvious in waterfall plots with long decay rates at resonance, the lower in level and the faster you can dissipate that energy the better.
Last edited: