Analog Servo Sub

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Yes, one of the channels is a servo controlled sub channel.
It is a new version of the still popular Philips mfb or motional feedback speakers from the 1970's.
some diy people from Holland are giving the concept new life.
more info here:http://mfb.piratelogic.nl/data/docs/537.usermanual.pdf
there are even scanned original servicemanuals which explain the workings on the homepage of the forums I just posted.
This should help a lot.
They are currently sourcing customized accelerometers for use with any woofer. The originals are not availeble anymore.
I have 8 of the 8" mfb woofers laying here for my next project :)
I also have 2 pairs of working mfb 544 speakers.
I am now typing from my phone but if I read your posts tomorrow maybe I can help with translating some stuff that Google messes up in their transl.

greetings
 
Here we go again (in Robbintip's link... as far as I'm able to read it), the usual MFB story: how to make poor and small drivers/enclosures produce OK bass.

That is a worthy challenge for Philips and audio businesses. I wish them luck.

But at DIYaudio, we're trying to take good speakers and make them better with MFB. So the techniques that can improve poor and small bass boxes may or may not be directly applicable here, except as inspirations.

Ben
 
When did they appoint you a director of R&D here?

Is there anything substantive and direct in the mildly-stated quote of mine that you'd like to comment on? Was I too harsh - even judgmental - in saying, "may or may not" or maybe you took serious offense when I suggested the work could be an inspiration to others?

Or do you just have a great need to surly ventilate today on this forum?

Well, I DEFINITELY and TRULY believe that work may or may not be useful and I DEFINITELY feel it might be inspirational. And I'll stick with those views.

Ben
 
Last edited:
Is there anything substantive and direct in the mildly-stated quote of mine that you'd like to comment on? Was I too harsh - even judgmental - in saying, "may or may not" or maybe you took serious offense when I suggested the work could be an inspiration to others?

Or do you just have a great need to surly ventilate today on this forum?

Well, I DEFINITELY and TRULY believe that work may or may not be useful and I DEFINITELY feel it might be inspirational. And I'll stick with those views.

Ben

A totally pointless comment.

regards
Trev
 
I understand that you are not able to completely understand the dutch text completely.
By saying MFB only makes small woofers sound OK, you are wrong.

MFB does a couple of things.
The first thing is making the woofer follow the exact input signal, and thus making it sound better than only the woofer being fed by an amplifier, in a normal way.
This is the part you want to use I understand, make the woofer replicate the exact input signal.

Then, to achieve monstrous bass from a small woofer, the input signal is corrected (before amplification) to compensate for the woofer's roll-of. Every woofer in a box has a roll-of at the low end. It doesn't matter if the box is too small or not, you will simply compensate more or less for this.
If a woofer has a -3db roll-off from 60 hz and down, you will essentially make a filter that adds +3db from 60Hz and down. This is the clever part that makes a small woofer sound big. To go more in detail, a rumble filter is needed, because it will try to produce even 2Hz signal at audible levels, frying your woofer by doing so.

The first part of the MFB workings, the correction circuit for the woofer making it duplicate the input signal rather than just having a go at it, is the part you would most likely use.

The second part however, is also interesting. When you modify the rumble filter of the newer 587 MFB's, you will get distortion-less (thus color-less) f3 25Hz sound. You see the 8" woofer in a 25L box (bigger than the 8L box of the first types) work very hard.
The result is astonishing. You can hear frequencies normally only produced by big 12" subs, played effortless, clean, and non-colorated. When you hear these speakers play, YOU WILL **** BRICKS!!! Really, if you haven't heard them, you don't know what I mean.
The limit is the Xmax and amp power.
If you for instances have high enough Xmax and power on a 12" subwoofer, you will be able to get the sound of a 21" sub. I am not joking, this is for real!
The sound produced is fast and clean because you will have a small mass, but still can reach deeper and cleaner then you would ever imagine coming out of your trusted speakers.
If used right, and not playing at extreme levels, you will get enough deep bass from 8" tower speakers to eliminate your 12" subs.

Then there is more.
MFB speakers are actively crossed over (full 3-way in the later series) and using a subtractive filter. They are fase corrected as well.

There is more to this speaker than just making a small woofer sound OK.
MFB 545 12" studio monitors for example produced the deepest sound from a 12" at that time without using ports/TL.

The deep bass then is not affected by the small box, because the correction circuitry simply makes the woofer follow input signal rather than be affected by box dimensions. Normally you would have the least colored sound at the tuning frequency of the box. This is not so with MFB, it doesn't have colored sound at low frequencies at all.
The advantage of this is you can have correct sounding bass at low frequencies, without having to use a port or a TL with their disadvantages such as steep roll-of, muddied, less tighter bass, coloration, name it all.
The disadvantage is that you will at some time reach your Xmax when tuning them lower and lower.

I have seen that one set of MFB's, old style (8" woofer in 8L cabinet) can vibrate my wall and dropping my pictures/clocks/paintings from it, while the bass was to low to actually hear. You really have to hear them to understand what I mean.
By saying it is just sounding OK, you are undermining it. I have never heard ANY sub (and I heard a lot) producing cleaner bass than controlled woofers like MFB.
If you refer to the whole speaker as sounding just OK, you are right. Over 150Hz, MFB woofers can sound colorated (hey, 1976...) also, the mids, tweets and the high amp are not top notch. This has several causes, first, it was 1976, second, they were made to not be expensive speakers that no one could afford. They were made for men that wanted big speakers sound, but had wifes that only would take small speakers in their living room. The speakers are therefore full of compromises.
By eliminating WAF, going past the compromises and use todays drivers and technology, you can create something that simply no company offers.

Philips is no small fish in the audio world. The only thing is that they made their stuff for regular consumers, not for audiophile, piles of money spending people. They had B&O to do that for them back then, also Marantz originated a lot of equipment from them back than, as were lots of others.

Phlips first mass produced the light bulb, but also made the first Hi-Fi speaker. They invented dome tweeters, dome midranges, they were first at producing a speaker with correction circuit for the woofer at mass production, they invented the compact cassette, the CD, the DVD, and lots of other stuff you use on daily basis.

I'm not saying Philips is the best, I only highly recommend to dive in a little deeper on these MFB speakers. They do exactly what you want (and a little more as well). I see you are open minded about this, that is why I regret it so much that all this information is in Dutch, as it is difficult to translate.

When you hear a corrected woofer play for the first time, you will be blown away! I sincerely hope you will finish your project and come to a nice result.

I would strongly advice you to buy a set of these speakers second hand (revised ones!!) and hear for yourself. If you even remotely like it, I would suggest going on with your project, and believe me, you will.

Greetings.
 
1. No one is a bigger or more frequent advocate for MFB than me, as you may have noticed over the past few years. I experimented with it first around 1967 and had MFB in my woofers for many decades although not recently. But thanks for your sermon anyway.

2. Yes, Philips are swell folks.

3. I am glad your results with little drivers are so encouraging. Of course, that kind of "progress" making small speakers better with MFB has been published for half a century or more, if you will check your history. I hope will all will live to hear rooms shaking from an 8-inch driver. It would be wonderful to have more excellent small speakers of the sort you advocate and the sort that experimenters on this forum typically use for initial testing. I actually believe MFB R&D should be especially focused on extra-small drivers (making 2-inch drivers listenable).

4. Judging as best as I can from close attention to purposes of posts at DIYaudio, I continue to think people here want to see some R&D making large good-quality woofers even better, not small speakers tolerable (as has been proven for half a century already....). Please.

Ben
 
4. Judging as best as I can from close attention to purposes of posts at DIYaudio, I continue to think people here want to see some R&D making large good-quality woofers even better, not small speakers tolerable (as has been proven for half a century already....). Please.
I think MFB is great for BOTH small 6 inch woofers AND >10 inch (sub)woofer usage.
If you push a 6 inch woofer to the limit to get some 30-40 Hz SPL, this will cause distortion. What does MFB do? Attempt to lower that distortion, because it is a disturbance compared to the reference.
If you push a 12 inch sub to the limit to get some 15-20 Hz SPL, the same holds.

Anyway; Back on-topic!
Armand, any news?
 
Last edited:
Hi Armand,

This is Jac Brown who you referred to in the first post regarding the audioexpress article. I have been hanging around diyAudio forums as lehmanhill.

I stumbled across your thread while searching for something else. I'm no expert, but maybe I can contribute some of my experience with servo subwoofers to this thread. I am just back in town from a trip, so I need to do some catching up with life and your thread, but I will check back in when I catch up.
 
I think MFB is great for BOTH small 6 inch woofers AND >10 inch (sub)woofer usage.
If you push a 6 inch woofer to the limit to get some 30-40 Hz SPL, this will cause distortion. What does MFB do? Attempt to lower that distortion, because it is a disturbance compared to the reference.
If you push a 12 inch sub to the limit to get some 15-20 Hz SPL, the same holds.

Anyway; Back on-topic!
Armand, any news?
Yes, back on topic: contrary to what you pose as a definitive rhetorical challenge, the same does NOT necessarily hold because other factors are involved.

It depends on the source of the distortion. That is an important point (in more detail below) because it is one of those things that is absent and therefore gets ignored in all the talk about models that is deep in this thread and related threads.

When distortion arises from non-linearities of spiders and motors, MFB with an accelerometer can fix that pretty good. But when it arises from a driver (little or big) running out of moxie*, it can not be fixed with MFB no matter what your feedback fraction.

See the difference? See why modeling gets you only so far?

Ben
*or an accelerometer and related pieces that aren't as good as your woofer
 
Last edited:
Yes, back on topic: contrary to what you pose as a definitive rhetorical challenge, the same does NOT necessarily hold because other factors are involved.

It depends on the source of the distortion. That is an important point (in more detail below) because it is one of those things that is absent and therefore gets ignored in all the talk about models that is deep in this thread and related threads.

When distortion arises from non-linearities of spiders and motors, MFB with an accelerometer can fix that pretty good. But when it arises from a driver (little or big) running out of moxie*, it can not be fixed with MFB no matter what your feedback fraction.

See the difference? See why modeling gets you only so far?

Ben
*or an accelerometer and related pieces that aren't as good as your woofer

No different than an amplifier hitting the rail. Everything has limits.

regards
trev
 
Bentoronto does have a point. I am currently measuring the improvement on the acoustics and the influence of the MFB caused my THD to drop to 1/3 of it's non-MFB value. The controller however, should push back harmonics by a factor 10 (20 dB) so indeed not all is related to the actuator distortion.
Question remains; why would the distortion introduced in a subwoofer be greater, spider/actuator-wise?
 
I am currently measuring the improvement on the acoustics and the influence of the MFB caused my THD to drop to 1/3 of it's non-MFB value. The controller however, should push back harmonics by a factor 10 (20 dB) so indeed not all is related to the actuator distortion.

A few questions on your setup:
1) What frequency were you measuring distortion at?
2) What sensor are you using for measuring acceleration? You mentioned in another thread that the sensor distortion may be an issue.
http://www.diyaudio.com/forums/subw...trolled-woofer-controller-27.html#post3573140

3) Are you comparing distortion(with & without feedback) using a microphone? or the output of the accelerometer. If accelerometer has linearity issues, you would need to be comparing its output to see distortion reduction close to theory.
4) Were all harmonics reduced by roughly the same amount? The distortion reduction capability for each harmonic is related to the loop gain at the frequency of the harmonic, not the maximum loop gain.
http://www.diyaudio.com/forums/subw...trolled-woofer-controller-23.html#post3561894

As with amplifiers, I would expect that only the 2nd harmonic(and perhaps 3d) would be reduced nearly exactly by its associated loop gain factor. Higher order harmonic reduction doesn’t necessarily follow the slope of increasing loop gain as the driver generated distortion is fed back to the input and generates harmonic content not present when woofer is driven open loop.
5) Any chance you are reaching limits(running out of moxie) on your woofer(excursion), sensor(G-capability), or driving electronics(clipping)?
 
1) What frequency were you measuring distortion at?
Fundamentals at 30,40 or 60 Hz.
2) What sensor are you using for measuring acceleration? You mentioned in another thread that the sensor distortion may be an issue.
What I meant in that post was that in order to reach 15% distortion on your accelerometer, you really need to be vandalizing the woofer. At least mine. The accelerometer is the ACH-01. On the microphone, reaching 15% THD indeed is easier.
3) Are you comparing distortion(with & without feedback) using a microphone? or the output of the accelerometer. If accelerometer has linearity issues, you would need to be comparing its output to see distortion reduction close to theory.
My reduction in harmonics on the accelerometer is a 95% match to the predicted/calculated reduction. That part is good, but I was hoping for a acoustic reduction that would by in the same magnitude. However, that seems to be less.
4) Were all harmonics reduced by roughly the same amount? The distortion reduction capability for each harmonic is related to the loop gain at the frequency of the harmonic, not the maximum loop gain.
The harmonics on the accelerometer were suppressed by the predicted magnitude of the sensitivity function. No surprises there. That is how I designed the openloop.
Higher order harmonic reduction doesn’t necessarily follow the slope of increasing loop gain as the driver generated distortion is fed back to the input and generates harmonic content not present when woofer is driven open loop.
I am having a hard time trying the find out what you mean in the above. It is not the amplifier which is the dominant cause of the measured harmonics, but the woofer itself. 2nd-6th harmonics on the sensor output caused by the woofer, are WAY more dominant than the harmonics of the amplifier. The goal is not to reduce harmonics introduced by the amp, but introduced by the woofer.
5) Any chance you are reaching limits(running out of moxie) on your woofer(excursion), sensor(G-capability), or driving electronics(clipping)?
Nope, I monitor about each and every single signal in order to prevent such events.
 
Last edited:
If higher order harmonics are caused by cone break up, how are you going to correct for it? The woofer is not functioning as a piston anymore, so you would need more accelerometers on a single cone? Could you explain this?

I've read that it is possible to use lasers as accelerometers for woofers. Its expensive and not small at all, but might be interesting to use for lab use, if you could borrow one.
 
My reduction in harmonics on the accelerometer is a 95% match to the predicted/calculated reduction. That part is good, but I was hoping for acoustic reduction that would by in the same magnitude. However, that seems to be less.
Excellent :up:
With your accelerometer distortion reductions matching predicted reductions, your MFB system is working properly. If your mic measured distortion trends at 30Hz(no cone break up issues) don’t match the ACH-01 results, then perhaps your mic is operating in a non-linear fashion and contaminating the results. You mention using the Panasonic WM61A capsule…had you modified the capsule per Linkwitz to handle higher SPL with reduced distortion? If not, it is probably not suitable for low distortion measurements with mic placed in the high SPL zone close to a woofer. Do you have access to another mic to compare with?


I am having a hard time trying the find out what you mean in the above. It is not the amplifier which is the dominant cause of the measured harmonics, but the woofer itself.
My apologies for being unclear when switching topics from woofer distortion to amplifier distortion. I was not talking about the amplifier in your MFB system which probably has >40dB feedback and very low distortion. I was trying to point out that when you have a non-linear system that you are wrapping a feedback loop around, the higher order harmonics don’t necessarily reduce in direct proportion to the amount of loop gain the way lower harmonics do as loop gain is increased from 0 to 20dB. It depends on the exact nature of the non-linearity. I mentioned amplifier distortion since this behavior had been analyzed in the past by Peter Baxandall, Nelson Pass, and most recently Bruno Putzeys. Obviously woofer distortion would not trend in the same manner as a square-law amplifying device, but it would not be unreasonable to expect trends in reduction of higher order harmonics to differ from lower order.

More info here if interested.
http://www.linearaudio.nl/linearaudio.nl/images/pdf/Volume_1_BP.pdf (see page 14)
https://passlabs.com//download.php?...ges/uploads/mypdf/distortion_and_feedback.pdf (see page 9)
 
Last edited:
There seems to be continuing confusion on one point that should be obvious. We aren't aiming to make the accelerometer output clean, but the speaker output. With good feedback, you should have tiny distortion at the accelerometer because the accelerometer is the criterion.

In practice, any much distortion from the mic confuses matters although a before-and-after test design teases that out some.

About cone break-up, that is a reality for many cones and a crucial issue for MFB. Anybody have a parameter for that in their beloved models? No surprise "yes's"? As far as I know, none of feasible feedback sensors do anything except make the dust cap linear (and mic MFB is a non-starter). So, you just have to buy a driver with a stiff cone and hope for the best. My impression is that most distortion luckily arises from grosser dys-motions of the cone.

Ben
 
Last edited:
There seems to be continuing confusion on one point that should be obvious. We aren't aiming to make the accelerometer output clean, but the speaker output. With good feedback, you should have tiny distortion at the accelerometer because the accelerometer is the criterion.

In practice, any much distortion from the mic confuses matters although a before-and-after test design teases that out some.

Hmmmm...I guess I didn't notice any confusion on this point. It really only comes into play if the accelerometer is not linear. But you are right, the measured acoustic performance is what really counts. Unfortunately many(most?) readily available mics are not capable of handling high SPL at low frequencies without self-generated distortion, so measuring the accelerometer output is the next best thing. I was spoiled by having access to a B&K 4133 mic when doing my testing.

As I have mentioned before, the ACH-01 is a rather linear accelerometer on paper and comparison with the B&K mic proved it in practice. My experience with the ACH-01 was that if a particular mic measured distortion significantly higher than the accelerometer distortion it was due to non-linearities in the mic capsule, not the ACH-01.

BTW, I did not experiment with any other accelerometers, so these comments pertain ONLY to the ACH-01.
 
Isn't reducing harmonics with feedback just simulating a hard cone? In other words, why not use an stiff cone with high-frequency break-up? Then you'd only have to correct it's nonlinearity..
Also, when correcting higher order harmonics, you'd have to cover lot's of different frequencies at the same time, am I right? That sounds like a lot of work.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.