ProSoundWeb Community
Sound Reinforcement - Forums for Live Sound Professionals - Your Displayed Name Must Be Your Real Full Name To Post In The Live Sound Forums => Audio Measurement and Testing => Topic started by: Frank Koenig on December 09, 2019, 05:42:52 PM
-
Splitting this off from "Lab Lounge -- Distortion in Front Loaded Subs". I tried a simple experiment which quickly revealed a gap in my reasoning, although I still have some hope for the general approach.
The trouble is that when I (asynchronously) open the circuit from the amplifier output to the speaker, the resulting step, being broad-band, excites the natural frequencies of the system. This results in a kind of step response superimposed on the desired exponentially decaying sine wave envelope, which seriously confounds the measurement. In the example below the speaker was excited at about 10 V rms at 100 and 60Hz. The scope was triggered off the switch drive which was not synchronized with the sine wave. I put the scope into infinite persistence mode so as to generate an envelope as the speaker's voltage waveform "crawled" across the screen. Clearly, these are anything but exponential decays.
Simply windowing the tone-burst prior to the switch opening, while narrowing the bandwidth, will damp the motion of the speaker as long as it's still driven from a voltage source. Perhaps a smooth increase in the source resistance, a sort of "source impedance window" would have the desired effect. The question is, does there exist a window that will not excessively damp the system while not excessively exciting it with out-of-band energy?
I seriously doubt I'm the first to try to go down this rabbit hole, and not sure I'll pursue it further at this point. Thanks for indulging me and ideas are welcome.
--Frank
(http://40Hz.jpg)
I used other frequencies, too, but can't get the forum to swallow the images. I'll post separately if anyone wants. -F
-
sounds like you have too much time on your hands...
JR
-
sounds like you have too much time on your hands...
Touché! But I might not be alone :)
-
that was just easier than answering your question.. (as if I know the answer).
Lots of moving parts in there.
JR
-
Frank, I really wish I could chime in but honestly I have a very hard time even understating what you're trying to do.
What I'd give for some formal engineering/academic training sometimes. Sorry, it's not fair....you're always helping...me is always digging to understand.
Anyway, are you trying to measure the decay of the driver by reading its voltage after both a drive signal and connection to the amplifier is removed?
Kinda sounds like it, but that doesn't make a lot of sense not to have amp damping in place...so I guess I'm missing things.
Missing ….like whatever your after, what is the purpose as opposed to say a set of harmonic measurement sweeps with REW?
-
Splitting this off from "Lab Lounge -- Distortion in Front Loaded Subs". I tried a simple experiment which quickly revealed a gap in my reasoning, although I still have some hope for the general approach.
The trouble is that when I (asynchronously) open the circuit from the amplifier output to the speaker, the resulting step, being broad-band, excites the natural frequencies of the system. This results in a kind of step response superimposed on the desired exponentially decaying sine wave envelope, which seriously confounds the measurement. In the example below the speaker was excited at about 10 V rms at 100 and 60Hz. The scope was triggered off the switch drive which was not synchronized with the sine wave. I put the scope into infinite persistence mode so as to generate an envelope as the speaker's voltage waveform "crawled" across the screen. Clearly, these are anything but exponential decays.
Simply windowing the tone-burst prior to the switch opening, while narrowing the bandwidth, will damp the motion of the speaker as long as it's still driven from a voltage source. Perhaps a smooth increase in the source resistance, a sort of "source impedance window" would have the desired effect. The question is, does there exist a window that will not excessively damp the system while not excessively exciting it with out-of-band energy?
I seriously doubt I'm the first to try to go down this rabbit hole, and not sure I'll pursue it further at this point. Thanks for indulging me and ideas are welcome.
--Frank
(http://40Hz.jpg)
I used other frequencies, too, but can't get the forum to swallow the images. I'll post separately if anyone wants. -F
I too am a little confused about what you are asking...
I was introduced to step functions(?) back in the 70s a technician... we were splicing together pitch shifted audio samples and hard blanking or muting to 0V between samples could cause unexpected perturbations. The energy introduced could be larger if there was DC immediately before or after the blanked segment. You might reduce the energy introduced into your speaker by coordinating the disconnect with zero crossings, but if you have ever listened to tone bursts there is still audio perturbations related to the on/off step function even if started and stopped at zero crossings. Also with tone bursts you want to present even number of cycles otherwise you will introduce a DC component from the odd half cycle.
In addition driving and then disconnecting an amp from a speaker, while in motion, will turn that speaker motor into a microphone of sorts generating voltage output related to the continued motion.
I doubt this answers your question, but some stuff to chew on...
JR
-
It's an odd one. I feel like, having a Physics background, I ought to be able to make a reasonable go of this.
So.
Let's consider the free-air driver case for starters.
If you excite it at a fixed frequency, there are three possible places you can "let go" of the driver:
- Maximum displacement, zero velocity
- 0 displacement, maximum velocity
- Some displacement and some velocity. Okay, that's many places.
Bear in mind that, at low frequencies, loudspeakers are a damped spring-mass system. When connected to an amplifier, Qts is low, ~0.3. Disconnected, Qts = Qms, which tends to be high, ~10.
ie, there'll be ringing when disconnected. That's according to Frank's plan.
We're going to assume there's no hysteresis of any kind. ie, what happens to the cone is a function of the usual equations of motion. That assumption does usually hold, more or less. I've heard of cases where it might not be true, but the effects are fairly small so we can put that to one side for now. If someone would like to go into that, I'd be interested.
Let's consider the three cases I've outlined above:
- With maximum displacement and zero velocity, the driver will ring at Fs. That should be obvious - think about a pendulum.
- With maximum velocity and zero displacement, you've effectively given the cone a "kick". It has kinetic energy. My bet would be that over the time of a cycle or two, the output frequency will shift to Fs and then the energy will dissipate.
When it comes to damped oscillators, I like to imagine a pendulum in custard. If you gave the pendulum a really hard kick, it'll set off quickly but then tend towards its natural resonance. Probably as it reaches the excursion maximum.
The third case is just a convolution of the above cases - it'll set off quickly, from a non-zero displacement. So long as the speaker it within it's linear region, I don't think there's any additional complexity to consider.
Now, let's slap that driver into a ported cabinet. Sealed cabinets are trivial - the spring of the air inside the cabinet contributes to that of the suspension, and nothing really changes from the above.
So. Ported cabinets.
Simple enough: you've got a mass (slug of air in the port) and a spring (compliance of air in the cabinet).
... Which is all being excited by another mass (cone) that's on another spring (suspension) which mostly combines with the first spring (air).
At this point, it's complicated.
However, we have a secret weapon: impedance curves.
Having thought a while, I'm convinced that the "let go" output will simply tend towards that of the speaker being driven by a high impedance. For a ported box, there are two impedance peaks. One's below the port tuning, and one's above.
I suspect that what would happen when you "let go" of the cone is that the output will head towards those two frequencies.
On the 'scope, I'd expect to see them modulating each other. It kind-of looks like that on the screen captures, but it's difficult to tell for sure.
The TL;DR is that I'm not convinced that watching a speaker after it's been disconnected will tell you much information about the previous excursion state.
Chris
-
It's an odd one. I feel like, having a Physics background, I ought to be able to make a reasonable go of this.
So.
Let's consider the free-air driver case for starters.
If you excite it at a fixed frequency, there are three possible places you can "let go" of the driver:
- Maximum displacement, zero velocity
- 0 displacement, maximum velocity
- Some displacement and some velocity. Okay, that's many places.
Bear in mind that, at low frequencies, loudspeakers are a damped spring-mass system. When connected to an amplifier, Qts is low, ~0.3. Disconnected, Qts = Qms, which tends to be high, ~10.
ie, there'll be ringing when disconnected. That's according to Frank's plan.
We're going to assume there's no hysteresis of any kind. ie, what happens to the cone is a function of the usual equations of motion. That assumption does usually hold, more or less. I've heard of cases where it might not be true, but the effects are fairly small so we can put that to one side for now. If someone would like to go into that, I'd be interested.
Let's consider the three cases I've outlined above:
- With maximum displacement and zero velocity, the driver will ring at Fs. That should be obvious - think about a pendulum.
- With maximum velocity and zero displacement, you've effectively given the cone a "kick". It has kinetic energy. My bet would be that over the time of a cycle or two, the output frequency will shift to Fs and then the energy will dissipate.
When it comes to damped oscillators, I like to imagine a pendulum in custard. If you gave the pendulum a really hard kick, it'll set off quickly but then tend towards its natural resonance. Probably as it reaches the excursion maximum.
The third case is just a convolution of the above cases - it'll set off quickly, from a non-zero displacement. So long as the speaker it within it's linear region, I don't think there's any additional complexity to consider.
Now, let's slap that driver into a ported cabinet. Sealed cabinets are trivial - the spring of the air inside the cabinet contributes to that of the suspension, and nothing really changes from the above.
So. Ported cabinets.
Simple enough: you've got a mass (slug of air in the port) and a spring (compliance of air in the cabinet).
... Which is all being excited by another mass (cone) that's on another spring (suspension) which mostly combines with the first spring (air).
At this point, it's complicated.
However, we have a secret weapon: impedance curves.
Having thought a while, I'm convinced that the "let go" output will simply tend towards that of the speaker being driven by a high impedance. For a ported box, there are two impedance peaks. One's below the port tuning, and one's above.
I suspect that what would happen when you "let go" of the cone is that the output will head towards those two frequencies.
On the 'scope, I'd expect to see them modulating each other. It kind-of looks like that on the screen captures, but it's difficult to tell for sure.
The TL;DR is that I'm not convinced that watching a speaker after it's been disconnected will tell you much information about the previous excursion state.
Chris
Chris, I appreciate your detailed reply. Thanks. You understand what I was trying to do and why it doesn't work, at least in this form. I was reaching way back to my days of motor speed controllers that use the motor's own back EMF as the tachometer. Speakers and permanent-magnet (or shunt-wound) DC motors have a lot in common but they are not the same.
I've been looking at input impedance as a way of getting to excursion but the model is not so simple, requiring values of parameters that we'd rather not have to know a priori. This might be what the better powered speaker manufacturers do for their excursion limiters. I think what folks here would like is a fairly direct way to measure excursion that can be used as a post-design check and for DIY side chain filter equipped peak limiter settings.
Here's an article from the good folks at TI about one approach. They're concerned with fingernail-size phone speakers but the physical model looks to me as though it would apply to 21 in. subs as well.
http://www.ti.com/lit/an/slaa857/slaa857.pdf (http://www.ti.com/lit/an/slaa857/slaa857.pdf)
All for now. Guess I better stew awhile and save up for that interferometer ::)
--Frank
-
Chris, I appreciate your detailed reply. Thanks. You understand what I was trying to do and why it doesn't work, at least in this form. I was reaching way back to my days of motor speed controllers that use the motor's own back EMF as the tachometer. Speakers and permanent-magnet (or shunt-wound) DC motors have a lot in common but they are not the same.
I've been looking at input impedance as a way of getting to excursion but the model is not so simple, requiring values of parameters that we'd rather not have to know a priori. This might be what the better powered speaker manufacturers do for their excursion limiters. I think what folks here would like is a fairly direct way to measure excursion that can be used as a post-design check and for DIY side chain filter equipped peak limiter settings.
Here's an article from the good folks at TI about one approach. They're concerned with fingernail-size phone speakers but the physical model looks to me as though it would apply to 21 in. subs as well.
http://www.ti.com/lit/an/slaa857/slaa857.pdf (http://www.ti.com/lit/an/slaa857/slaa857.pdf)
All for now. Guess I better stew awhile and save up for that interferometer ::)
--Frank
Please excuse my simplicity but isn't the position sensor on servo drivers such as the Infinity Watkins woofers what you are after?
Also Apogee amps needed a sense lead right next to the voice coil. I know they were doing more than measuring voltage like the old Meyer processors. I am going to take a look and see what the Apogee circuit was doing or I am sure somebody will come along that knows.
-
I've been looking at input impedance as a way of getting to excursion but the model is not so simple, requiring values of parameters that we'd rather not have to know a priori. This might be what the better powered speaker manufacturers do for their excursion limiters. I think what folks here would like is a fairly direct way to measure excursion that can be used as a post-design check and for DIY side chain filter equipped peak limiter settings.
--Frank
I would be interesting in knowing what powered speakers might be doing this. In my experience people really give powered speakers a lot of credit for doing things they really don't.
-
I would be interesting in knowing what powered speakers might be doing this. In my experience people really give powered speakers a lot of credit for doing things they really don't.
I vaguely recall finding a patent back in the 70s from Phillips covering a feedback sensor on the LF driver. At the time this was probably used in modest bookshelf sized speakers.
JR
PS: Powered speakers are rich in potential for smart features (mostly protection in early days). Modern low cost DSP offer even more options. Cheap powered boxes are just like cheap non-powered, and at a minimum they probably use active EQ to win side by side comparisons at POS.
-
I vaguely recall finding a patent back in the 70s from Phillips covering a feedback sensor on the LF driver. At the time this was probably used in modest bookshelf sized speakers.
JR
PS: Powered speakers are rich in potential for smart features (mostly protection in early days). Modern low cost DSP offer even more options. Cheap powered boxes are just like cheap non-powered, and at a minimum they probably use active EQ to win side by side comparisons at POS.
In my early audio days (late 90's) I know Kenwood was pitching something similar in their car audio subwoofer line up. I found this link, which is surprisingly technical for the car audio market. Kenwood Sigma Drive (http://www.cieri.net/Documenti/Kenwood/Documenti%20tecnici/Kenwood%20-%20Sigma-Drive%20Technical%20Guide.pdf)
{EDIT} After rereading the above article that I posted, it looks like Kenwood borrowed this from their other markets, and not directly originating it from Car Audio{/EDIT}
It does appear that powersoft has the discussed feature set in its IPAL module. The marketing line states "One of the greatest limits on achieving outstanding acoustical performance is the lack of a global amplifier-loudspeaker feedback system, linking the electrical and the acoustical signal directly. The Differential Pressure Control" (DPC®) does exactly this: it allows a differential pressure signal, mechanically obtained from the two sides of the moving loudspeaker membrane, to be fed to the amplifier just as any other electrical feedback signal. This grants a complete real-time characterization of the transducer and its acoustical load conditions, allowing for an input signal-to-SPL closed loop design."
https://www.powersoft-audio.com/en/oem-solutions/ipalmod
Is anyone aware of commercial products with this module?
-
I would be interesting in knowing what powered speakers might be doing this.
I would, too. But I'm pretty sure no one is going to tell me. I can't help but to speculate.
-
I vaguely recall finding a patent back in the 70s from Phillips covering a feedback sensor on the LF driver. At the time this was probably used in modest bookshelf sized speakers.
It was Phillips. The Center for Computer Research in Music and Acoustics (CCRMA) at Stanford had a set of four around 1978. They thought they were the cat's meow back then. Some years later I asked someone about them and they said the speakers always sounded like they were underwater. People's perceptions...
There have been a number of attempts at position-feedback speakers over the years. Meyer had (has?) one that was sold as a large studio monitor. And then Powersoft has the integrated amplifier/actuator for large subwoofers that David mentioned. Has anyone seen one in the wild? No doubt there are others.
--Frank
-
I've been looking at input impedance as a way of getting to excursion but the model is not so simple, requiring values of parameters that we'd rather not have to know a priori. This might be what the better powered speaker manufacturers do for their excursion limiters. I think what folks here would like is a fairly direct way to measure excursion that can be used as a post-design check and for DIY side chain filter equipped peak limiter settings.
All for now. Guess I better stew awhile and save up for that interferometer ::)
--Frank
Frank,
Although measurement of excursion could be done with an interferometer, and you might find other uses for it, it's easy enough to simply see excursion, and can be easily recorded with video if you want to avoid proximity to the noise.
That said, since you seem to be on to the excursion measurement pursuit as it is related to distortion, since distortion is not specifically tied to excursion, excursion measurement in itself is of limited value.
The "better powered speaker manufacturers" do not employ "excursion limiters" per se, but a combination of high and low pass filters with frequency selective voltage limiting, and drivers designed to withstand the excursion within the voltage delivered. Drivers that can deliver high SPL output with low distortion require very high excursion, very high BL (motor strength) and very robust cones and suspensions, all of which are expensive.
A differential pressure sensing device feedback control loop based system like the Powersoft/B&C Ipal (Integrated Powered Adaptive Loudspeaker) can improve output linearity even more, although few will pay for the improvements that most can not detect in normal concert situations.
Art
-
Frank,
Although measurement of excursion could be done with an interferometer, and you might find other uses for it, it's easy enough to simply see excursion, and can be easily recorded with video if you want to avoid proximity to the noise.
That said, since you seem to be on to the excursion measurement pursuit as it is related to distortion, since distortion is not specifically tied to excursion, excursion measurement in itself is of limited value.
The "better powered speaker manufacturers" do not employ "excursion limiters" per se, but a combination of high and low pass filters with frequency selective voltage limiting, and drivers designed to withstand the excursion within the voltage delivered. Drivers that can deliver high SPL output with low distortion require very high excursion, very high BL (motor strength) and very robust cones and suspensions, all of which are expensive.
A differential pressure sensing device feedback control loop based system like the Powersoft/B&C Ipal (Integrated Powered Adaptive Loudspeaker) can improve output linearity even more, although few will pay for the improvements that most can not detect in normal concert situations.
Art
Art, I was pretty much kidding about the interferometer, although if I saw a deal on a nice used one I might be tempted 8)
My interest in measuring excursion is part just curiosity, but to the extent that there's a practical goal it's more to come up with good limiters for "all conditions" damage protection than controlling distortion in and of itself. In my personal use I rarely push things that hard anyway, but there's always the dropped mic or phantom power pop.
When it comes to what's inside powered speakers, I can only speculate based on what I might try were I the designer. Going in I didn't know how effective a peak voltage limiter using a simple side-chain equalizer is, but, based on what you, David S. and JR are saying, it sounds like it's good enough. On the other hand, amps such as Powersoft do have instantaneous output current measurement built-in, so an approach utilizing that might be a "simple matter of programming", as they say. But it may also fail to confer any advantage.
Thank you all for this discussion.
--Frank
-
In my early audio days (late 90's) I know Kenwood was pitching something similar in their car audio subwoofer line up. I found this link, which is surprisingly technical for the car audio market. Kenwood Sigma Drive (http://www.cieri.net/Documenti/Kenwood/Documenti%20tecnici/Kenwood%20-%20Sigma-Drive%20Technical%20Guide.pdf)
{EDIT} After rereading the above article that I posted, it looks like Kenwood borrowed this from their other markets, and not directly originating it from Car Audio{/EDIT}
It does appear that powersoft has the discussed feature set in its IPAL module. The marketing line states "One of the greatest limits on achieving outstanding acoustical performance is the lack of a global amplifier-loudspeaker feedback system, linking the electrical and the acoustical signal directly. The Differential Pressure Control" (DPC®) does exactly this: it allows a differential pressure signal, mechanically obtained from the two sides of the moving loudspeaker membrane, to be fed to the amplifier just as any other electrical feedback signal. This grants a complete real-time characterization of the transducer and its acoustical load conditions, allowing for an input signal-to-SPL closed loop design."
https://www.powersoft-audio.com/en/oem-solutions/ipalmod
Is anyone aware of commercial products with this module?
David,
The Martin Audio MLX subwoofer uses the ipalMod and associated B&C 18" drivers.
They used to have a 21" for clubs called the ASX but it is now discontinued.
I am not sure who else has an ipalMod system out in commercial production.
Lee
-
David,
The Martin Audio MLX subwoofer uses the ipalMod and associated B&C 18" drivers.
They used to have a 21" for clubs called the ASX but it is now discontinued.
I am not sure who else has an ipalMod system out in commercial production.
Lee
Power Sound Audio has a few subs for home theater. Not for small apartment use!