Yes, Ben..
That's the intend of the feedback circuit, to cancel all (most of) the distortion. But since most of the distortion (percentage-wise) is in the first 3 harmonics, al limited feedback-bandwidth could also do.. however it would not perform optimal.
And if in case of a Sub the Fmax would be rather low, the BW of the feedback-loop could also be limited. So IPAL could perform rather well, at least canceling the first harmonics out of the distortion and reducing overall distortion greatly.
On the other hand a good working MF feedback-loop ought to capture most of the produced harmonics to at least 5kHz.
Greets,
Edwin
His esgigt,
Post #99: "...optical feedback...appears less complicated than I imagined."
Yes, but, the devil is always in the details, the paper also shows the difficulties arising from secondary effects like cone breakup (P.: 39). So cone integrity is another thing to keep in mind. I seem to remember that was a problem in using capacitive feedback as implemented by Backes & Mueller in the early 1970s. So you have to keep things in the pistonic range of the driver.
Regards,
Post #99: "...optical feedback...appears less complicated than I imagined."
Yes, but, the devil is always in the details, the paper also shows the difficulties arising from secondary effects like cone breakup (P.: 39). So cone integrity is another thing to keep in mind. I seem to remember that was a problem in using capacitive feedback as implemented by Backes & Mueller in the early 1970s. So you have to keep things in the pistonic range of the driver.
Regards,
Assuming the priority goal of MF is to reduce distortion, I would think that investigating other methods to achieve this same goal would be useful.
As I have proposed several times, I would think that adding more drivers (and maybe more power if necessary) to the passband of interest would reduce distortion much more than MF (assuming that the system is pushed beyond a few watts). This allows for both lower distortion at low power and the ability to provide a lot more max spl. The only conceivable way that MF could possibly compete is if you sized the MF system so that it would never exceed a fraction of it's thermal or mechanical capacity, which is almost never the case.
As we know from literally decades of experience, using a system without MF well within it's thermal and mechanical limits can be quite pleasant despite all the "evils" or resonances in the passband or imperfect impulse response.
So now that a few MF proponents that actually know what they are talking about have joined the discussion, what is the consensus on using more drivers and keeping the system well within it's thermal and mechanical limits to reduce distortion vs MF?
While I clearly do realize that the same system design logic could be applied to a MF system, this is rarely the case. Other than the possible exception of the IPAL (depending on how it's used), ALL of the MF usage I've seen has been making vastly undersized systems perform at a mediocre standard - the Rhythmic subs, the Sony speakers, etc.
If you look at Rhythmic's list of reasons we need direct servo, this list actually makes my point for me.
Bullet points 1,2 and 6 all point to using the system beyond it's natural linear thermal and mechanical range. The remaining "advantages" are fully achievable with a well designed system that does not utilize MF.
Since fully half of the Rhythmic talking points have to do with pushing a system beyond it's limits and keeping it relatively linear (and the rest are achievable without using MF at all), let's look at what is actually required to do that. To compensate for loss of force due to Bl and Cms non linearity at the driver's limits you need to add MORE power. To compensate for loss of force due to thermal stress (power compression) you need to add MORE power. None of this makes any sense. You can achieve the same goals by adding more drivers and not pushing the system to it's limits in the first place.
As for the "articulate, tight, transparent, and well-defined bass" (or lack thereof), this is usually dictated by the enclosure design and room induced effects (room gain, room modes, reflections and cancellations), not the driver's ability to accurately reproduce the signal.
When the room produces "ringing, boxy" effects that are an order of magnitude worse than the "problems" MF is addressing, and these room effects occur AFTER the sound leaves the speaker (so MF can't do anything about it), how valuable is MF in practical terms?
As I have proposed several times, I would think that adding more drivers (and maybe more power if necessary) to the passband of interest would reduce distortion much more than MF (assuming that the system is pushed beyond a few watts). This allows for both lower distortion at low power and the ability to provide a lot more max spl. The only conceivable way that MF could possibly compete is if you sized the MF system so that it would never exceed a fraction of it's thermal or mechanical capacity, which is almost never the case.
As we know from literally decades of experience, using a system without MF well within it's thermal and mechanical limits can be quite pleasant despite all the "evils" or resonances in the passband or imperfect impulse response.
So now that a few MF proponents that actually know what they are talking about have joined the discussion, what is the consensus on using more drivers and keeping the system well within it's thermal and mechanical limits to reduce distortion vs MF?
While I clearly do realize that the same system design logic could be applied to a MF system, this is rarely the case. Other than the possible exception of the IPAL (depending on how it's used), ALL of the MF usage I've seen has been making vastly undersized systems perform at a mediocre standard - the Rhythmic subs, the Sony speakers, etc.
If you look at Rhythmic's list of reasons we need direct servo, this list actually makes my point for me.
Rythmik Audio • Servo subwoofer products
Bullet points 1,2 and 6 all point to using the system beyond it's natural linear thermal and mechanical range. The remaining "advantages" are fully achievable with a well designed system that does not utilize MF.
Since fully half of the Rhythmic talking points have to do with pushing a system beyond it's limits and keeping it relatively linear (and the rest are achievable without using MF at all), let's look at what is actually required to do that. To compensate for loss of force due to Bl and Cms non linearity at the driver's limits you need to add MORE power. To compensate for loss of force due to thermal stress (power compression) you need to add MORE power. None of this makes any sense. You can achieve the same goals by adding more drivers and not pushing the system to it's limits in the first place.
As for the "articulate, tight, transparent, and well-defined bass" (or lack thereof), this is usually dictated by the enclosure design and room induced effects (room gain, room modes, reflections and cancellations), not the driver's ability to accurately reproduce the signal.
When the room produces "ringing, boxy" effects that are an order of magnitude worse than the "problems" MF is addressing, and these room effects occur AFTER the sound leaves the speaker (so MF can't do anything about it), how valuable is MF in practical terms?
Last edited:
Hi JAG,
As e.g.: the dual driver ACE Bass systems show that distortion reducing enclosure design, and motion feedback based electronic design can be used together. As to a specific case-and quantifying the respective merit-that would require building, and measuring otherwise it's just guesswork.
Regards,
As e.g.: the dual driver ACE Bass systems show that distortion reducing enclosure design, and motion feedback based electronic design can be used together. As to a specific case-and quantifying the respective merit-that would require building, and measuring otherwise it's just guesswork.
Regards,
The cone-brakeup issue has never been solved by a MF setup to my knowledge, because it occurs in the upper regions of the units FR and is of pure mechanical nature. Besides, avoiding cone breakup is good engineering practice.
A good point.
But it seems natural to use stiffer cones (KEF B139?) for MF systems because more control is shifted to the feedback circuit rather than the kind of accidental motion of ordinary Rice-Kellogg drivers.
I wonder how much cone break-up is reflected back to an accelerometer or VC sensor? Can't be totally invisible to those methods.
Ben
Sure, good design is good design, and MF can be applied to a good enclosure design. The point was that a lot of the time "boxy, ringing" sound has to do with bad design and/or room effects (things that MF can't fix), but it's usually blamed on other things.
As far as quantifying respective merit, that's the heart of the matter. Personally I don't see any merit in MF. First the system has to have well designed enclosures. Then the system has to be properly sized to keep well within the thermal and mechanical limitations of the drivers. (Remember, at xmax we are already at 10 -20 percent distortion, and possibly even more, so you have to keep WELL under xmax if low distortion is a number one priority. And thermal compression usually starts at a fraction of the driver's rated power handling so you have to keep the power to a very low level.) Next the room has to be tamed.
That's 3 factors that each individually produce a lot more benefit than MF. And until all of that is done the benefits of MF will be dubious at best. Otherwise you are attempting to fix real problems with the wrong tool. AFTER those things are done MF might have some small amount of benefit but realistically it would be a very small benefit because the ACTUAL problems have already been identified and fixed.
Last edited:
MFB is such an intellectually satisfying technique that I find it surprising more people haven’t experimented with it. That being said, in the past 10 years or so I think MFB has become a solution in search of a problem. Modern FEM modeling techniques for mechanical /magnet/electrical systems have made it possible to improve woofer performance to the point that there is little to be gained except when it is desired to push a woofer beyond its inherent linear range in search of higher SPL (without increasing size or number of woofer). Looking at most any recent distortion vs. frequency curves for a well designed driver and you can see that the last octave of a woofers bandwidth is the main area of potential benefit from MFB. Yet, this is exactly the range where our ears seem to be “deaf” to the predominately 2nd and 3rd order harmonics generated.
So, ignoring for the moment the possibility that reducing distortion in the bottom octave is unnecessary leaves open three areas where MFB could be of use.
1) woofer systems that use cheap drivers due to cost constraints.
2) woofer systems that have size constraints imposed by design priorities other than performance.
3) systems meant to push the envelope of what is possible. (ie no matter how big or how many woofers you add to reduce bottom octave distortion, properly implemented MFB could reduce it)
I think the majority if not all MFB systems ever produced fall into categories 1) & 2).
My personal experimentation a decade ago mainly focused on 2).
For an example of what was/is possible, I measured a Velodyne ULD-15 20Hz distortion at 20mm P-to-P excursion with and without servo. Distortion dropped by roughly a factor of 30 from 15% to 0.5%. This result is in reasonable agreement with the 30dB loop gain and distortion level claimed in the manual. I’m a bit confused as to why recent Velodyne products have not shown the measureable reduction in distortion of past products.
just a guy’s recommendation that you can get the same performance by simply increasing the number woofers is perfectly valid. It’s brick simple, and is going to work every time. Well, every time you have enough money and room for all those woofers. It would take a lot of woofers to reduce 20Hz distortion by a factor of 30. And, as suggested above by category 3), whatever number of woofers you wind up with, MFB would allow them to play louder with lower distortion in the bottom octave. The question really becomes, what distortion threshold do you want to achieve, and how do you want to get there. In other words, I don't think there is anything inherent in MFB that improves sound quality in some unquantifiable/unmeasurable way. BTW, is 0.5% @ 20Hz really necessary? Even listening to test tones I didn’t think so. But it is possible with MFB(accelerometer based, NOT voice coil).
Concerning the Sony SA-W2500 subwoofer that started this thread:
A pitiful bit of engineering. They use a bridge network(partially/poorly compensated for woofer inductance) to extract the woofer cone velocity. As Bentoronto correctly pointed at, with traditional analog velocity feedback you need to avoid vented enclosures where the velocity of the cone is not a simple function of the SPL. In the case of the SA-W2500 it doesn’t really matter since the loop gain is a “whopping” 2.5 – 3.0dB and the response is actually flatter without the feedback signal. No distortion reduction either. I may still have measurements on this one laying around…if anybody is interested, I will see if I can dig it up.
Concerning the IPAL differential pressure sensor:
Looking at the equivalent model for any loudspeaker enclosure, you will see that sensing the pressure from one side of the woofer cone to the other provides a signal proportional to the force applied to the cone. This would encompass all forces…from the motor, the suspension, radiation impedance, the box, the port(if present), etc. Traditional analog feedback systems would have a tough time making proper use of this signal if the enclosure is anything other than a closed box. But with a DSP approach that includes virtual models where every signal in the model is available for comparison with the feedback it could certainly be used as advertised for vented enclosures.
So, ignoring for the moment the possibility that reducing distortion in the bottom octave is unnecessary leaves open three areas where MFB could be of use.
1) woofer systems that use cheap drivers due to cost constraints.
2) woofer systems that have size constraints imposed by design priorities other than performance.
3) systems meant to push the envelope of what is possible. (ie no matter how big or how many woofers you add to reduce bottom octave distortion, properly implemented MFB could reduce it)
I think the majority if not all MFB systems ever produced fall into categories 1) & 2).
My personal experimentation a decade ago mainly focused on 2).
For an example of what was/is possible, I measured a Velodyne ULD-15 20Hz distortion at 20mm P-to-P excursion with and without servo. Distortion dropped by roughly a factor of 30 from 15% to 0.5%. This result is in reasonable agreement with the 30dB loop gain and distortion level claimed in the manual. I’m a bit confused as to why recent Velodyne products have not shown the measureable reduction in distortion of past products.
just a guy’s recommendation that you can get the same performance by simply increasing the number woofers is perfectly valid. It’s brick simple, and is going to work every time. Well, every time you have enough money and room for all those woofers. It would take a lot of woofers to reduce 20Hz distortion by a factor of 30. And, as suggested above by category 3), whatever number of woofers you wind up with, MFB would allow them to play louder with lower distortion in the bottom octave. The question really becomes, what distortion threshold do you want to achieve, and how do you want to get there. In other words, I don't think there is anything inherent in MFB that improves sound quality in some unquantifiable/unmeasurable way. BTW, is 0.5% @ 20Hz really necessary? Even listening to test tones I didn’t think so. But it is possible with MFB(accelerometer based, NOT voice coil).
Concerning the Sony SA-W2500 subwoofer that started this thread:
A pitiful bit of engineering. They use a bridge network(partially/poorly compensated for woofer inductance) to extract the woofer cone velocity. As Bentoronto correctly pointed at, with traditional analog velocity feedback you need to avoid vented enclosures where the velocity of the cone is not a simple function of the SPL. In the case of the SA-W2500 it doesn’t really matter since the loop gain is a “whopping” 2.5 – 3.0dB and the response is actually flatter without the feedback signal. No distortion reduction either. I may still have measurements on this one laying around…if anybody is interested, I will see if I can dig it up.
Concerning the IPAL differential pressure sensor:
Looking at the equivalent model for any loudspeaker enclosure, you will see that sensing the pressure from one side of the woofer cone to the other provides a signal proportional to the force applied to the cone. This would encompass all forces…from the motor, the suspension, radiation impedance, the box, the port(if present), etc. Traditional analog feedback systems would have a tough time making proper use of this signal if the enclosure is anything other than a closed box. But with a DSP approach that includes virtual models where every signal in the model is available for comparison with the feedback it could certainly be used as advertised for vented enclosures.
Wonderful to have bolserst contribute his valuable knowledge*.
Yes, despite the obvious theoretical obstacles, nothing preventing you from using the IPAL item with vented enclosures as per his constraints. Providing that:
1. you are installing it where drunk dancers are the audience since it may do strange things to the freq response
2. (a) you know the math model of the vented box, (b) you can program your DSP with that model accurately enough, and (c) the box has the goodness not with depart from the DSP model from time to time
3. the microprocessor can handle all that stuff within the phase timing needed for the highest frequencies of interest, ummm, gotta be a lot fast than my Behringer 2496 with its old CPU.
Ben
*I wish Steve would pop into the passive radiator thread.
Well, you nailed it exactly. I figured I would tell the truth for a change.
I can see why. It's been most of 20 years since I last listened to an MF system (the one I built), and I still remember how dramatic the improvements were when feedback was added.
One of the goals of my employer at the time, was to squeeze a quart out of a pint pot, to a degree. I was using an 8" woofer in a pretty small box, and the close-miked response was ruler-flat down to well below 20 Hz; I think the -3dB frequency was actually around 10 Hz.
Whether an 8" woofer could generate enough SPL down there for human ears to actually detect, during normal in-room listening, was another matter!
But I did discover that, personally, I found the experience of listening to a loud, steady, 20 Hz sine-wave quite unpleasant. It felt more like a moth fluttering in my ears than an actual sound. Ugh!
Agreed. At one stroke, you get light weight, precise and repeatable characteristics, extended frequency response, resistance to normal environmental conditions (humidity, temperature, etc). We can all thank the push to put air-bags in every car for producing these wonderful little MEMS accelerometer chips.
I don't want to play Chicken Little, endlessly warning about the sky falling. But, as far as I understand it, the series resistor sensing approach basically raises the output impedance of the power amplifier, effectively driving the speaker with a constant-current source.
This dramatically reduces the damping factor of the amplifier, meaning the woofer Q at its fundamental resonance rises far above its Qts value - and you get boomy bass.
The same problem exists with valve power amps with pentode output devices, which also have rather high output impedance. The problem of poorly controlled bass due to high output impedance has been bothering the valve Hi-Fi crowd for a long time.
In general, then, constant-current drive of a woofer makes the behaviour around the fundamental resonance worse, not better.
If I have somehow misunderstood the principle behind current-sensing approach to motional feedback, please correct me.
I agree, a comparison of the measured acoustic performance of motional-feedback done properly (accelerometer) vs conveniently (various emf-sensing schemes) would be interesting to see.
I'd like to touch on the cardboard dust-cap issue. To get good benefits from MF, you need a substantial amount of negative feedback. To maintain stability with a substantial amount of negative feedback, you need well controlled phase behaviour out to fairly high frequencies - you need a decent phase margin at the unity-gain frequency, just as you do with op-amp feedback, or any other servo system.
All of this means you cannot tolerate any additional mechanical resonances in the speaker-accelerometer system. Mounting an accelerometer to a soft and floppy cardboard dust-cap may not be a good idea; the dust-cap, particularly when carrying the extra weight of an accelerometer, will almost certainly have mechanical resonances at relatively low frequencies compared to the actual cone-break up modes, along with the 180-degree phase shift that accompanies each resonance. There goes your chances of good loop gain along with loop stability.
Instead, the goal is to have the sensor as light as possible, and the connection to the voice coil as mechanically stiff as possible. In this way, the sensor itself won't add any troublesome phase shifts or resonance peaks to the combined frequency response of the (speaker + accelerometer). That way, we end up limited only by the mechanical qualities of the speaker itself, not by the sensor we added on.
For the 8" woofer I worked with, there was a nasty mechanical breakup at around 1 kHz. It showed up as a huge spike in the frequency response. I couldn't make that go away, so it would set the upper limits of what could be achieved with MF.
So my goal was to get my actual piezo accelerometer (with its mounting) light and stiff enough to have no mechanical resonances below, say, 5 kHz, preferably even higher; high enough so the servo feedback would be limited by the 1 kHz speaker breakup mode, and not by the accelerometer.
In the end, I managed to implement 18 dB of loop gain (negative feedback). My goal had been 20 dB, but I never managed to get beyond 18 dB.
With 18 dB of loop gain, the close-miked acoustic response of the woofer was virtually ruler-flat from just above 10 Hz to a little below 1 kHz, with smooth roll-offs at each end. It had to be used in a 3-way speaker system, since there aren't too many good tweeters that can reach down to 1 kHz.
It took some electrical trickery to achieve that good a result; I struggled for a while to find a way to manage that obnoxious 1 kHz resonant peak in the driver!
-Gnobuddy
Ben,
1. The IPAL is an adaptive system, it follows the designer's intended frequency response regardless of surroundings or drive level.
2. The math model of the box is not needed to program the DSP, but accurate measurements at various drive levels and voice coil heating are.
3. The IPAL microprocessors certainly are a good bit faster than the much older (and less expensive) Behringer 2496. That said, the latency inherent in any DSP can be virtually eliminated by delaying the pick up transducer by the same amount as the DSP latency.
Not that gender has anything to do with this discussion, but your previous referral to me as "he/she" seems to suggest you may have forgot that I have been a "man" since reaching adulthood in 1974, having started my professional audio career designing speaker systems and mixing live bands years before that, and that I am not some nameless ghost haunting your posts, I am Art Welter, owner of Welter Systems, Inc. (since 1985) designer of the sound (and lighting) systems used by Eclipse Concert Systems and Southern Thunder Sound from 1978-1992, and have used the "weltersys" name since the days that AOL limited names to a maximum of 9 letters.
Cheers,
Art
I believe that Art has accurately transcribed the opinions of the manufacturer for all our benefits to read in Pts 1 and 2. If Art had added any argument to the manufacturer's claims, then I would hope somebody would comment on his value-added comments.
Earlier, I presented some specifics in connection with their AES publication about why I have misgivings about IPAL's trustworthiness.
For Pt 3, seems to violate feedback theory. Does Art really mean that using quantum computers, you might be able to pull what he describes?
Ben
Last edited:
This and other narrative and quantitative parts of bolserst post consider small perturbations and ill-linearities of expensive precisely designed and manufactured drivers.
As bolserst says, a small amount of feedback would result in a small but welcome improvement. The benefit would be on the order of improvement of using multiple drivers (if that gross and expensive solution appeals to you and to your spouse). In any case, maybe no casual listener or polite guest would notice.
But as others who have lived with MF have repeatedly pointed out, the biggest payoff is in the control, tightness, extended bass, and - yes, I will say it - quickness of the woofer.
The dramatic improvements shown in the IPAL tone burst pictures* - that's what MF users are talking about. The error is gigantic, unlike the familiar hard-to-hear errors from spiders and surrounds and bad magnets. A small amount of feedback has a big effect on these perturbations.
Ben
*granted, I think those IPAL tone bursts are done with a vented box which gives truly wretched tone bursts which sound as lousy at the o'scope pictures look - unless you use MF
Last edited:
I've read similar comments concerning MFB over the years, but I am not sure that it is something inherent or only possible with MFB. Have you ever compared "control, tightness, extended bass, quickness" of a sealed box woofer with MFB against the same sealed box woofer without MFB but Eq'd to the same response shape(magnitude & phase)? When comparing apples-to-apples like this using tone bursts I did not notice much if any difference in the starting and stopping of the waveforms until testing beyond the woofer's Xmax. (ie beyond where the voice coil starts to leave the gap). Note that the LP crossover should be in place and identical for this comparison.
Yes, the IPAL tone burst comparisons are with a vented box. Pretty dramatic, and shows what is possible when you move beyond traditional MFB techniques. The price for entry of course is high-power DSP and the knowledge how to use it.
Yes, vented box... which shows that introducing MF even to unsuitable boxes like vented ones, can be pretty beneficial.
Likely that today you'd want lots of processing (requiring DSP) somewhere in the system or in a few places like inside the loop and for general EQ. For sure, it has been known forever that VC methods need to be corrected to acceleration driver operation (meaning bass boost), as expected (or do I have that backwards).
So, bolserst wants a shoot-out between sealed boxes and MF. At Bell Labs, I did my testing with (don't laugh) a Karlson 15. The difference on test tones such as impulses was dramatic. Not as dramatic for music and I didn't EQ much (which is essential with VC MF).
With modest feedback ratios, used it for years with a Klipschorn bass. Pretty modest improvement. Likewise for a 6 cu ft sealed box sub. Made no difference with my 1956 AR-1 but that has a unique speaker design. All these tests were in the days before room EQ and so very hard to say if the FR made any sense.
B.
Last edited:
Tone burts rendition depends on fc and Qtc.
If MFB or equalised drivers in sealed box have their "apparent" parameters identical, the tone bursts they deliver are identical.
The difference between the two is that MFB should deliver less non linear distortion.
I’ve performed the “shoot-out” several times in the past, I was asking if you had.
Take your Karlson 15, or whatever modern woofer you have handy, put it in a suitably sized well-built box and add MFB and EQ until you like the way your test tones sound and impulses look on a scope. Now, measure frequency response(mag & phase) and save it. Then, remove MFB and re-adjust EQ to match the response and phase you just measured. Repeat listening to test tones and looking at impulses. When operating within X-max, I noticed no obvious differences. Based on vintage, I’d guess Xmax for the Karlson 15 was likely < 2mm.
Exactly my point.
If you aren’t pushing your sealed box woofers to the point that distortion is a concern, then you can achieve the same response flattening/extending benefits of MFB with EQ alone. If distortion is noticeable, then MFB can provide improvement over EQ alone.
Last edited:
Yeah he also said "I think MFB has become a solution in search of a problem."
If you want to selectively quote to make your case, I can do that too, and my quote is better.
WHAT!?!
The Sony MF speaker has a five inch woofer (I think). How is doubling that up to 2 five inch woofers gross and expensive? Your current sub is a 12 inch woofer (IIRC) in a tiny box. If your "polite guests" have a problem with a couple of 12 inch woofers in maybe 2 cu ft total you have really weird friends.
Also the 2 woofers will have lower distortion than a single MF woofer AND the two woofers will go louder, so there's a double benefit to that approach.
Do we have to read the Adire paper again? Quickness is not an issue - the motors on any reasonable driver have plenty of force to push the cone as quickly as it needs to go to play back the signal at all frequencies in the passband.
MF is not steroids, it can't provide extended bass any better than eq can. All it can do is marginally reduce distortion, which is better done in other ways, like using more drivers.
Have fun listening to your tone bursts. As a few people have mentioned, a clean tone burst is not required for pleasant listening, arguably it's pretty unimportant in the grand scheme.
...
Are you being serious here?
An externally hosted image should be here but it was not working when we last tested it.
That's what you used to test MF? Probably the most muli resonant device of the time? And you say a (theoretically) single resonance ported box is unsuitable?
That K15 is a 6th order bandpass AT LEAST and maybe more if you count the upper chamber separate from the driver chamber, as there is a divider board. That's 2 (or 3) chambers in that box depending on how you look at it, and a BUNCH of impedance peaks inside the passband.
How did you account for all those impedance peaks and phase shifts inside the passband for that testing? How did you account for all the impedance peaks inside the Klipschhorn's passband when you applied MF? If you didn't account for the several impedance peaks in the passband of these speakers (and you didn't because you clearly don't know how), then you did not apply MF correctly and it's hard to say what the MF was actually doing - certainly not what you think it was doing anyway. And you can't quantify the "modest improvement" you experienced in the Khorn and 6 cu ft sealed box since you didn't actually measure anything.
And the one speaker that you might arguably have set up with MF correctly - the small sealed box AR-1 - the MF made no difference even to your completely subjective and totally biased ears.
All this testing is only mildly interesting from a completely subjectively and obviously biased perspective, but as a scientific analysis it's grossly incompetent and really means nothing.
Last edited:
Exactly my point.
Exactly the problem. Easy to say, "identical speakers sound identical". But. at least for Back-EMF feedback, wholly different world when the amp has a negative output impedance vis a vis the driver.
The system no longer resembles a T/S passive driver in a sealed volume because - as with op amps that would self-destruct without feedback - the behaviour with sufficient gain is defined by the feedback parameters, not the driver.
The problem is trying see the output as a variant of T/S behaviour instead of feedback theory.
Ben
The problem is that the guy you are trying to teach here has done the tests and the measurements. You haven't.