Regarding the THD chart are you telling me that that speaker is capable of producing 120 dB at 10Hz with only 17% THD because I don't believe that speaker is physically capable of doing that SPL ?
According to WinISD with the same speaker mounted in a 100 liter cabinet with 2000 watts drive I don't get more than 99dB SPL and that is with 33mm peak diaphragm displacement so those distortion charts are a bit misleading or maybe I have overlooked something and not reading them properly. Simple test is to drive the speaker at 15 Hz until you reach the max power or displacement and then measure its output spectrum and waveform fidelity which I suppose is what that next table attempts to show.
And like I have said before if you want 18 inch performance in the bottom end why not use two 12 inch drivers which will give you the same surface area as one 18 inch whilst giving you much better high frequency performance and less directivity !!
regards
david
Last edited:
Sure you have plenty to smile about with the right tools for the job and 1.6 G-flops 32 bit floating point DSP in control 😉
Attachments
Last edited:
David,
No, the speaker does not do 120 dB at 10 Hz 🙄.
Josh Ricci has rigorously tested dozens of speaker systems for output and distortion:
Data-Bass
There are several charts in my post #99, Josh Ricci's CEA2010 Max Passing SPL (dB) chart indicates the LMS Ultra 5400 Sealed does 94.7 dB at 10 Hz with 18.8% distortion.
The drive level was probably in excess of 2000 watts, though he was careful not to burn the borrowed driver. He has destroyed a few of his own 😀.
The CEA2010 test is a metric of comparison for determining "clean" output of a sub.
Although you may think the CEA2010 acceptable distortion is too high, using the CEA2010 to test a subwoofer using your processing vs. an unprocessed sub would provide a fair comparison as to the efficacy and improvement it provides.
Your distortion reduction examples thus far lack any absolute SPL figures, so provide us no "real world" actual low distortion output capability information.
High output, accurate, low distortion speakers do tend to be expensive, the (still) unanswered question is if your processing with the motional sensor, loudspeaker, control board with "goodies" and the processor to run them can match the output at a given distortion level of a high quality speaker and cost less.
If you can prove it can, I'd be interested in purchasing the system, but until you provide actual SPL in your reports, and a price for the system we have no way to compare the cost-effectiveness of your system.
I do applaud your work, proper integration of processing and transducers can make great sonic improvements.
Art
Tranquility Bass,
The sine wave sweep testing done by me and presented at my website to produce the distortion graphs, output compression magnitude, long term output graphs, etc...Are all referenced to output measured at 50Hz at 2 meters outdoors ground plane and for DIY or passive systems the voltage drive level is also recorded. What this means is that a 115dB level on the THD chart does not mean that the system was driven at 115dB over the entire frequency bandwidth but that it "should" be producing that output referenced to 50Hz barring significant compression of the output occuring at 50Hz. In order to see the amount of output a system is generating at any specific frequency over a 10-120Hz bandwidth while producing the distortion levels shown on the distortion charts one has to also look at the corresponding measurement from the "long term output compression" graph from which the distortion measurements are derived.
I hope this clears up any confusion there. No the LMS driver was not producing 120dB at 10Hz...More like 95dB. 😉
Concerning MF control of subwoofers...I have 4 systems tested at my website that feature MF of various flavors. The Velodyne CT-150, Velodyne DD18+, Paradigm Sub2 and Rythmik FV15HP. 2 are vented alignments and 2 are sealed. None of these appear to offer dramatically superior sonic performance either subjectively or as measured or distortion characteristics versus a number of other non MF controlled systems. Which is not to say that it does not offer verifiable improvement. In fact one of the subwoofer systems the Velodyne DD18+ allows user control of the gain setting of the MF system. It was measured and compared with the gain at minimum and at maximum settings and with the feedback gain at maximum it did improve the distortion performance somewhat and flattened the frequency response quite a bit as well demonstrating a verifiable improvement. The driver in this subwoofer is a heavy duty long throw 18" which is quite expensive and the amplifier is a robust 1250w rated plate unit, so it already likely has rather good performance without the MF system. Additional to testing these 4 systems I have also owned 2 more subwoofers that had MF and I am currently working on developing a system for personal use that incorporates a MF system as well.
As previously mentioned it is highly likely that a good MF system will make much larger improvements to a lesser quality driver than higher end more capable unit. Specifically MF can make improvement in lowering of distortion, damping of the driver movement and tracking the signal shape, plus it can improve tracking of the input signal amplitude by counteracting compression and losses in the driver due to thermal compression, loss of motor force and increased suspension stiffness at higher excursions and other non linearities in tracking the signal. However there are definite limitations. It is not going to make a system more powerful than it already would have been, only more accurate. The limitations of driver displacement, thermal handling and amplifier reserves are unchanged and the MF control system will only improve things up to the point that these limits start to be encountered.
MF will not enable the Lanzer 15" to perform beyond its inherent physical capabilities and have the displacement, power handling, headroom and low distortion capabilities of an LMS or LMSR driver but it will improve its performance up to the point that those limitations are encountered.
As far as the audibility of the improvements gained by MF in distortion and signal tracking I have to agree with Art. I have used, heard and tested so many bass systems over the years and done far more real world testing and measuring than most and bass distortion is for the most part subjectively benign until it reaches truly appalling levels even in a reflection free environment with test signals and no upper frequency content masking the bass. With actual music or full bandwidth material masking the bass it is surprising the amount of harmonic distortion that it takes to be audible. There are a few studies that have been conducted on this. I find that typically the point where bass systems start to sound bad is more related to non harmonically related mechanical noises in the driver itself, amp clipping or resonances or buzzing/rattling in the cabinet or hardware, or air noises from vents that occur when the system is driven hard. Still I'm of the opinion that lower distortion is better always, better signal tracking is good and better dynamic reproduction with lower output compression at the system limits is also, so for some it will be worth looking into. This is why I am going to employ MF with drivers that are already incredibly low distortion and excellent performing to begin with even though I am skeptical that there will be any obvious audible improvement.
We know that MF can provide verifiable performance improvements so regardless of how audible these improvements in the bass are subjectively...If it can be made reasonably affordable, reliable and easily implemented...Why not use it?
The sine wave sweep testing done by me and presented at my website to produce the distortion graphs, output compression magnitude, long term output graphs, etc...Are all referenced to output measured at 50Hz at 2 meters outdoors ground plane and for DIY or passive systems the voltage drive level is also recorded. What this means is that a 115dB level on the THD chart does not mean that the system was driven at 115dB over the entire frequency bandwidth but that it "should" be producing that output referenced to 50Hz barring significant compression of the output occuring at 50Hz. In order to see the amount of output a system is generating at any specific frequency over a 10-120Hz bandwidth while producing the distortion levels shown on the distortion charts one has to also look at the corresponding measurement from the "long term output compression" graph from which the distortion measurements are derived.
I hope this clears up any confusion there. No the LMS driver was not producing 120dB at 10Hz...More like 95dB. 😉
Concerning MF control of subwoofers...I have 4 systems tested at my website that feature MF of various flavors. The Velodyne CT-150, Velodyne DD18+, Paradigm Sub2 and Rythmik FV15HP. 2 are vented alignments and 2 are sealed. None of these appear to offer dramatically superior sonic performance either subjectively or as measured or distortion characteristics versus a number of other non MF controlled systems. Which is not to say that it does not offer verifiable improvement. In fact one of the subwoofer systems the Velodyne DD18+ allows user control of the gain setting of the MF system. It was measured and compared with the gain at minimum and at maximum settings and with the feedback gain at maximum it did improve the distortion performance somewhat and flattened the frequency response quite a bit as well demonstrating a verifiable improvement. The driver in this subwoofer is a heavy duty long throw 18" which is quite expensive and the amplifier is a robust 1250w rated plate unit, so it already likely has rather good performance without the MF system. Additional to testing these 4 systems I have also owned 2 more subwoofers that had MF and I am currently working on developing a system for personal use that incorporates a MF system as well.
As previously mentioned it is highly likely that a good MF system will make much larger improvements to a lesser quality driver than higher end more capable unit. Specifically MF can make improvement in lowering of distortion, damping of the driver movement and tracking the signal shape, plus it can improve tracking of the input signal amplitude by counteracting compression and losses in the driver due to thermal compression, loss of motor force and increased suspension stiffness at higher excursions and other non linearities in tracking the signal. However there are definite limitations. It is not going to make a system more powerful than it already would have been, only more accurate. The limitations of driver displacement, thermal handling and amplifier reserves are unchanged and the MF control system will only improve things up to the point that these limits start to be encountered.
MF will not enable the Lanzer 15" to perform beyond its inherent physical capabilities and have the displacement, power handling, headroom and low distortion capabilities of an LMS or LMSR driver but it will improve its performance up to the point that those limitations are encountered.
As far as the audibility of the improvements gained by MF in distortion and signal tracking I have to agree with Art. I have used, heard and tested so many bass systems over the years and done far more real world testing and measuring than most and bass distortion is for the most part subjectively benign until it reaches truly appalling levels even in a reflection free environment with test signals and no upper frequency content masking the bass. With actual music or full bandwidth material masking the bass it is surprising the amount of harmonic distortion that it takes to be audible. There are a few studies that have been conducted on this. I find that typically the point where bass systems start to sound bad is more related to non harmonically related mechanical noises in the driver itself, amp clipping or resonances or buzzing/rattling in the cabinet or hardware, or air noises from vents that occur when the system is driven hard. Still I'm of the opinion that lower distortion is better always, better signal tracking is good and better dynamic reproduction with lower output compression at the system limits is also, so for some it will be worth looking into. This is why I am going to employ MF with drivers that are already incredibly low distortion and excellent performing to begin with even though I am skeptical that there will be any obvious audible improvement.
We know that MF can provide verifiable performance improvements so regardless of how audible these improvements in the bass are subjectively...If it can be made reasonably affordable, reliable and easily implemented...Why not use it?
Last edited:
Josh Ricci's post is a model of beneficial information. Thanks.
Question: you mention MF systems using vented boxes. But I thought it is an inherent principle that MF systems are not applicable to vented boxes because the cone motion is not closely correlated with sound output across the broad resonant region?
The question of creating physical measures that properly characterize speaker performance is an old and important discussion. I do not find it puzzling that MF systems have extended low bass, and that is easily measured. But I am puzzled that MF systems have perceptibly "tighter" bass, yet "tight" is related to sounds above the bass speaker range.... yet MF is controlling the bass speaker in a way that leads to a sense of "tight."
Ben
Question: you mention MF systems using vented boxes. But I thought it is an inherent principle that MF systems are not applicable to vented boxes because the cone motion is not closely correlated with sound output across the broad resonant region?
The question of creating physical measures that properly characterize speaker performance is an old and important discussion. I do not find it puzzling that MF systems have extended low bass, and that is easily measured. But I am puzzled that MF systems have perceptibly "tighter" bass, yet "tight" is related to sounds above the bass speaker range.... yet MF is controlling the bass speaker in a way that leads to a sense of "tight."
Ben
Last edited:
In my opinion bass distortion is audiable and we all need MF subs!
My MF sub is using 2x10" Peerless XXLS woofers with the accelerometer attached to only one of them(!). They share the same volume.
Even though these are expensive decent drivers THD was 26% @ 23Hz (12mm X-max) and 8% @ 35Hz (8mm X-Max) without MF. With MF this came down to 5% and 1%. I am sure I could hear the difference when I disconnected the feedback loop when playing music. More punch and a cleaner sound.
About cost, imagine you want to build two bass towers with a total of 12 woofers. With MF we can attach the acc. to one of them and bring the distortion down well into single digit for all of them even with cheap drivers.
I will for sure buy your system when you finish.
Pictures below is sound output from my simple analogue MF system at 30V RMS across the terminals. 23Hz (12mm X-Max) THD from 26% to 5%
This IS audiable
My MF sub is using 2x10" Peerless XXLS woofers with the accelerometer attached to only one of them(!). They share the same volume.
Even though these are expensive decent drivers THD was 26% @ 23Hz (12mm X-max) and 8% @ 35Hz (8mm X-Max) without MF. With MF this came down to 5% and 1%. I am sure I could hear the difference when I disconnected the feedback loop when playing music. More punch and a cleaner sound.
About cost, imagine you want to build two bass towers with a total of 12 woofers. With MF we can attach the acc. to one of them and bring the distortion down well into single digit for all of them even with cheap drivers.
I will for sure buy your system when you finish.
Pictures below is sound output from my simple analogue MF system at 30V RMS across the terminals. 23Hz (12mm X-Max) THD from 26% to 5%
This IS audiable
The problem is low distortion bass is the exception to the rule rather than the rule. Most woofers produce prestigious amounts of distortion and could be improved by some sort of motional feedback in the same way an amplifier is improved by negative feedback. And you are right, motional feedback won't make a speaker perform beyond its limits in much the same way feedback won't make an amplifier drive beyond its rail voltages. However it does offer numerous benefits below these limits that you don't get with an open loop system. Also distortion is not necessarily characterized by one measurement metric such as THD. The mix of distortion components is equally important as the ear is more sensitive to higher order distortion components which I have also addressed in this design.
My focus in providing this design is in music reproduction rather than home theater where more distortion is tolerated and sometimes encouraged. With music reproduction less is better !!
Reduction of distortion is only one of the benefits of the application of motional feedback. I have just taken it to the extreme to show what the hardware is capable of doing. The other benefits are increased bandwidth, improved transient response, immunity to changes and aging of the speaker as well as immunity to changes in voice coil resistance with temperature and inductance modulation.
And if it is as simple as what you say it is then for all of those systems that you tested on your website and operated within their working limits should sound identical to each other if distortion wasn't an issue but surely they don't !! Also why do a lot of loudspeaker vendors such as tcsounds spend a lot of time trying to reduce motor distortion if distortion is not an issue ? I do however agree that mechanical noise is an issue on some speakers more than others and even if you correct the distortion in the motor you may have to contend with mechanical noise from the woofer suspension etc.
I'm not here to debate the subjective merits of low distortion bass reproduction in the same way that sub ppm low distortion amplifiers and dacs are debated here. I have to agree with Bruno Putzeys on this one when he says the following:-
In the end it is a personal choice as to what distortion levels are tolerated by the listener and should not be dictated by making excuses for what is otherwise the popular acceptance of inferior bass reproduction systems just because that's how it always has been done. I bet if an amplifier or DAC produced as much distortion as a woofer then somehow people would notice it and certainly wouldn't buy it or use it. Would you ?
Regarding the four speakers you site only one employs motional feedback the way I have used it. I don't think the Paradigm sub 2 is a mfb system because the website certainly doesn't mention anything about it and I think they scrapped mfb after some legal issues a few years ago. As for applying mfb to vented boxes this would have to go into the April fools basket. Have a look if there is a sensor in the vent 😉
Also you should put that disclaimer on your distortion charts otherwise they don't make any sense and are misleading.
regards
david
Ben,
I can't speak for the OP's motional feedback system's ability to be applicable to a vented box system.
The IPAL system can control any enclosure type.
It uses an integrated amplifier which the output impedance can also be controlled from the model.
From their white paper:
"The amount of feedback loop that may be acceptable depends on the specific application and the specific design where the IPAL system is being set.
The upper limit of the amount of feedback will be determined by the system Phase Margin Limits and it must be carefully set in order to let the system operate in stable and reliable mode under any condition.
From the Pressure Model control panel it is also possible to set the amplifier output impedance. This feature also permits the system to adapt the loudspeaker transducer to any specific acoustical design. The output impedance could have a bandwidth of operation as well. The optimal setting of these parameters must still be carefully set under the respect of phase margin requirements in order to get the necessary stability of operation."
As you may now recall, the IPAL corrected waveforms you went ga ga over shown in the paper were using a ported 36Hz Fb box using an 18".
Art
In these initial tests no SPL measurements were given for two reasons. Firstly the measurement is taken from the output of the accelerometer which should be close to the sound pressure but of course the levels will be totally different. Secondly the speaker was operated in free air so there would be a lot of bass cancellation.
You will also note that I also stated that it did not take much drive level to get this particular speaker to produce a large amount of distortion. Of course there are other factors that contribute to the distortion such as mechanical noises from the suspension and flexure of the cone. These are all issues with the speaker itself and not the electronics. Provided the amplifier is not being overdriven all of the electronics will add very little if no distortion and be able to reduce distortion from the speaker according to classic negative feedback theory.
Next tests will be done on a properly sealed enclosure and measurements taken from a mic 😉
regards
david
You will also note that I also stated that it did not take much drive level to get this particular speaker to produce a large amount of distortion. Of course there are other factors that contribute to the distortion such as mechanical noises from the suspension and flexure of the cone. These are all issues with the speaker itself and not the electronics. Provided the amplifier is not being overdriven all of the electronics will add very little if no distortion and be able to reduce distortion from the speaker according to classic negative feedback theory.
Next tests will be done on a properly sealed enclosure and measurements taken from a mic 😉
regards
david
Those systems that Ricci mentioned won't have anyway of sensing what is going on in the vent and yet they claim servo or motional feedback !! Go figure !!
regards
david
Interesting that Armand found one accelerometer "tames" two drivers.
Earlier someone asked about the signal in the feedback loop. That is quite interesting to eyeball. Also very interesting is the inverse-distorted signal sent to the drivers! Long ago, in days of crude instruments, I seem to recall finding X-Y plots evocative.
Minor quibble with David: ALL that MF does is reducing distortion. But doesn't that include bass extension!!! And all the tightening effects are really just making the driver more veridical.
Anybody have an explanation for MF working in bass-reflex boxes?
Ben
Earlier someone asked about the signal in the feedback loop. That is quite interesting to eyeball. Also very interesting is the inverse-distorted signal sent to the drivers! Long ago, in days of crude instruments, I seem to recall finding X-Y plots evocative.
Minor quibble with David: ALL that MF does is reducing distortion. But doesn't that include bass extension!!! And all the tightening effects are really just making the driver more veridical.
Anybody have an explanation for MF working in bass-reflex boxes?
Ben
yes of course it includes bass extension which is another benefit 😉
Regarding mfb in bass reflex boxes I have seen one case where it was applied to a passive radiator system with accelerometers attached to both the driver and radiator but the reduction in distortion was at best marginal. The problem with a vented box system is that it is essentially a 4th order system and the phase shift limits the amount of loop gain that can be applied. Even if you apply compensation you have to be careful that you don't drive it very hard below its box resonance frequency where there is minimal loading on the driver and the net output falls off rapidly with frequency.
regards
david
Regarding mfb in bass reflex boxes I have seen one case where it was applied to a passive radiator system with accelerometers attached to both the driver and radiator but the reduction in distortion was at best marginal. The problem with a vented box system is that it is essentially a 4th order system and the phase shift limits the amount of loop gain that can be applied. Even if you apply compensation you have to be careful that you don't drive it very hard below its box resonance frequency where there is minimal loading on the driver and the net output falls off rapidly with frequency.
regards
david
A gentle reminder: There is no need to quite the entire post just above yours.I have removed much of the over-quoting.
Thanks
The problem is that the total sound coming out of the box - vent plus driver - is the net output. It is the net output we want to control in order for the level to remain flat across the speaker's range.
Unfortunately, the amount of movement of the cone is way different than that net output (1) at resonance and (2) at points where the vent is out of phase (below resonance) and (3) in phase (above resonance).
Therefore, controlling the motion of the dust cap with MF is quite different than controlling the output for a bass-reflex, TH, transmission line, and such.
That's how things are: MF harms the freq response*.
But it IS coherently related for IB (if truly infinite, eh), sealed boxes, and sort-of for horns esp. with rear-wave sealed boxes like Klipschorns.
As I posted earlier, MF does not provide bass extension per se. It "undistorts" the driver so that is produces extended bass... just getting more like the musical signal which is which is a lessening of distortion albeit an odd use of the term. That's negative feedback!
Ben
*It is almost inconceivable to finely hand-tune a very, very complex correction into the feedback loop for those unruly ups-and-downs of cone motion in BR boxes.
Last edited:
It works because multiple identical drivers, sharing the same volume and operated in electrical parallel or series or a combination of both behaves like one larger speaker with the same surface area of all of the drivers combined. The key point is that all drivers have to share the same volume, radiate into the same environment and be fed with the same signal otherwise there will be an imbalance.
regards
david
Last edited:
In other words mfb is not practical for vented boxes or any other box type with a resonant secondary radiator of some sort.
Regarding bass extension with motional feedback you get an extension plus reduction in distortion. Without mfb you can use an eq but distortion will increase with the boost and any changes in the speaker parameters will see a change in the frequency response.
regards
david
First of all, I think we need to ask for hard data before we conclude it "works." It may help a bit in practice and may address certain grosser distortions. But it is quite screwy in theory. Maybe it is my origins at Bell Labs where the basics of negative feedback were elaborated a long time ago by a chap named Nyquist, but the thought is totally unorthodox.
The best I can say for that multiple driver set-up is that it doesn't look like inherent instability, only potential instability.
Imagine this: you have two channels in your stereo amp playing the same signal. You use the feedback circuit of one channel to "correct" both. Make any sense? Hardly.
A bunch of drivers are in independent motion, not one big driver. In fact, they are "fighting" for the same interior air.
Ben
Last edited:
yes but all of the drivers see the same mechanical impedance and exhibit the same velocity and force provided they are identical.
you need to look at the equivalent electro-mechanical-acoustical analogous circuit and then you can see the bigger picture 😉
For N speakers connected in electrical parallel behaves like a single speaker with the following composite parameters:-
fs'=fs
Vas'=N.Vas
Qms'=Qms
Qes'=Qes
Qts'=Qts
no'=N.no
Sd'=N.Sd
Mms'=N.Mms
Cms'=Cms/N
Rms'=N.Rms
Bl'=Bl
Re'=Re/N
Ud'=N.Ud
For N speakers connected in electrical series behaves like a single speaker with the following composite parameters:-
fs'=fs
Vas'=N.Vas
Qms'=Qms
Qes'=Qes
Qts'=Qts
no'=N.no
Sd'=N.Sd
Mms'=N.Mms
Cms'=Cms/N
Rms'=N.Rms
Bl'=N.Bl
Re'=N.Re
Ud'=N.Ud
Alternatively you can experiment with Win ISD which has the ability to model an enclosure with multiple identical drivers.
regards
david
you need to look at the equivalent electro-mechanical-acoustical analogous circuit and then you can see the bigger picture 😉
For N speakers connected in electrical parallel behaves like a single speaker with the following composite parameters:-
fs'=fs
Vas'=N.Vas
Qms'=Qms
Qes'=Qes
Qts'=Qts
no'=N.no
Sd'=N.Sd
Mms'=N.Mms
Cms'=Cms/N
Rms'=N.Rms
Bl'=Bl
Re'=Re/N
Ud'=N.Ud
For N speakers connected in electrical series behaves like a single speaker with the following composite parameters:-
fs'=fs
Vas'=N.Vas
Qms'=Qms
Qes'=Qes
Qts'=Qts
no'=N.no
Sd'=N.Sd
Mms'=N.Mms
Cms'=Cms/N
Rms'=N.Rms
Bl'=N.Bl
Re'=N.Re
Ud'=N.Ud
Alternatively you can experiment with Win ISD which has the ability to model an enclosure with multiple identical drivers.
regards
david
"Perfect world" models aren't the least bit applicable here. We are talking about errors.
Unless all the drivers have identical errors (which would require they occupy the identical hole location on the panel so they have identical transfer curves*), you are sometimes increasing the distortion when you pump the feedback from one driver into the others.
But your math is just great 😉 - clearly your strong point.
All logical and math arguments aside, the whole enterprise of MF is fraught with risk of instability and burnt out expensive drivers (DAMHIK) because bass speakers are a mix of acoustic compromises and sleight-of-hand radiation resistance guesses. For sure, I would not add to the instability by using an utterly unconventional approach like using the feedback from one driver to power any other driver.
Ben
*or you could have 6 matched drivers in 6 identical boxes... but I still wouldn't wire them in series as you suggest.... even after consulting the wise old Win ISD
Unless all the drivers have identical errors (which would require they occupy the identical hole location on the panel so they have identical transfer curves*), you are sometimes increasing the distortion when you pump the feedback from one driver into the others.
But your math is just great 😉 - clearly your strong point.
All logical and math arguments aside, the whole enterprise of MF is fraught with risk of instability and burnt out expensive drivers (DAMHIK) because bass speakers are a mix of acoustic compromises and sleight-of-hand radiation resistance guesses. For sure, I would not add to the instability by using an utterly unconventional approach like using the feedback from one driver to power any other driver.
Ben
*or you could have 6 matched drivers in 6 identical boxes... but I still wouldn't wire them in series as you suggest.... even after consulting the wise old Win ISD
Last edited:
- Status
- Not open for further replies.