We're talking about MF woofers... so a very limited bandwidth.
Square - waves or something close will never be fed into those speakers in reality. The HF components in such a signal are also absolutely irrelevant. Besides, phase relations and rise times, in my opinion are better established with alternative measuring methods.
Squarewaves are nice for rise time fetishists building amplifiers.
No particular relevance other than to illustrate the effect of phase on a simple waveform. The OP had stated that MFB would lead to perfect waveform reproduction. The plots were simply to illustrate this is not the case, and that linear phase is required if that is a goal. Some people who have experimented with linear phase say that it improves the impact and perception of bass transients. There are several threads on this topic in the Multi-Way forum.
Exactly, this was the point of the second attachment. Send a square wave through a bandwidth limited device like a subwoofer and you get mostly the fundamental tone sine wave.
Ok. I'll get a cleaned up version posted later this week.
Considering discussions here, maybe I'll add a plot of group delay as well.
You may believe this, but that doesn’t make it so. This is likely a similar situation to when you insisted that the Kenwood Sigma circuit was acting as a bridge based MFB system rather than just electrical feedback around the amplifier using remote sensing to reduce output impedance of the amplifier seen by the loudspeaker. After 20 posts and a half-dozen or so efforts at explanation, you still respectfully disagreed.
http://www.diyaudio.com/forums/subwoofers/239941-analog-servo-sub-17.html#post3999755
If flat response and distortion reduction could be had with current feedback signal alone, why would anybody(including you) bother with VC or accelerometer feedback? If it was possible it would certainly make MFB much easier to implement, rather than messing with extra VC or accelerometer mounting difficulties.
A thought experiment…what If you epoxied the voice coil to the pole piece, so no motion. Measuring the signal from VC representing cone velocity(ignore mutual coupling), you would get zero at all frequencies. Measuring the current and you would get a signal inversely proportional to the blocked impedance. (at frequencies below resonance, it will actually be no different than if the VC had not been epoxied). The VC MF system would crank up the gain at all frequencies to try and get the cone motion to match the input, the current feedback system would not.
An attempt to reframe the issue…current thru the VC provides a signal proportional to the INPUT forcing function applied the woofer cone. VC velocity provides a signal proportional to the OUTPUT of the woofer-box system, encompassing all their mechanical and acoustical properties. If we could somehow determine the OUTPUT (including any non-linearities) just by knowing the INPUT, then we wouldn’t need MFB.
This concludes my clarification efforts.
Last edited:
Yes, and both approaches would be working correctly. In one case, correcting the erroneous velocity and the second, not correcting the non-erroneous acceleration. Simple enough circuitwise to work with either.
Perhaps not quite as puzzling a gedankt experiment as the one with the follow in the falling elevator and the easy to read watch. Maybe speed of light is more mysterious.
Our house renovations are nearly done and our 10th grandchild arrived. Still only one great grandchild (that we know of). Soon I'll try to provide data to augment the mostly abstract and sometimes oddly nasty posts. Or data from anybody else too. Can't be soon enough.
B.
Last edited:
I think we need to clear the air when it comes to square waves and tone bursts. Some of us do dirt-under-the-fingernails research (when not renovating the house). As a diagnostic tool, can't beat pushing square waves through an amp or tone bursts through a speaker and watching in real-time on an oscilloscope. That's research.
Many of us are pretty sure that phase and strange convolutions like comb filtering don't play a big part in the delivery of music. So when a HiFi passes the frequencies needed to produce a nice square wave OK, the sound probably will be great even if the phases are screwed-around such that the image one the 'scope is not square and pretty at all. EVEN SO, square waves make a great test bench tool. Abstract talk about transfer functions is not meaningfully observable at the end of my alligator clips.
As I've said a bunch of times in this thread already, MF (and other corners of music reproduction) are crying for quantitative measures that reflect performance. Same is true of room performance. Being able to quantify your efforts is essential to progress.
Ben
Many of us are pretty sure that phase and strange convolutions like comb filtering don't play a big part in the delivery of music. So when a HiFi passes the frequencies needed to produce a nice square wave OK, the sound probably will be great even if the phases are screwed-around such that the image one the 'scope is not square and pretty at all. EVEN SO, square waves make a great test bench tool. Abstract talk about transfer functions is not meaningfully observable at the end of my alligator clips.
As I've said a bunch of times in this thread already, MF (and other corners of music reproduction) are crying for quantitative measures that reflect performance. Same is true of room performance. Being able to quantify your efforts is essential to progress.
Ben
Last edited:
If you actually care how things sound you are measuring the wrong things and using the wrong tools. Speakers can sound FANTASTIC regardless of whether they can produce clean square waves or tone bursts. And with today's other (more useful in many cases) products there's not much need for an oscilloscope for speaker design or analysis. For example, devices that measure impedance and a variety of different simulators, and a measurement mic are vastly more useful than a 'scope.
Phase is everything and if you don't believe that try enjoying your music on speakers that are 180 degrees out of phase and within 1/4 wavelength of each other. You won't get the theoretical perfect cancellation but common sense is all you need for this thought experiment.
Comb filtering can provide deep notches in response at various places in space. This means different frequency response in different places. This is bad and if you don't get that I don't think there's any way to make it more clear. This leads nicely into discussions about dispersion patterns and off axis behavior too, and an investigation into what's really important when trying to make good sound.
As far as Toole is concerned, on axis frequency response is number 1. If you get frequency response wrong there's no hope, you might as well just quit. This is so obvious I think he only bothers to mention it once in his book because this is absolute basics.
Dispersion patterns and off axis behavior comes in somewhere behind on axis frequency response but it vital for good sound as judged by blind audiences at Harmon.
(Note that phase problems and comb filtering can mess up both your on axis AND off axis response so how could these things possibly be unimportant?)
Tying in somewhat with off axis behavior is reflections, what frequency range is reflected, how strong the reflection is and how long it arrives after the on axis wave. These reflections cause out of phase issues with the on axis wave at the listening position and cause peaks and notches similar to comb filtering.
Toole spends a lot of time and a huge amount of pages in his book talking about this stuff.
I don't recall him mentioning square waves or tone bursts, although it's been awhile since I've read it straight through. If you could point me to where he says this stuff is even remotely important I'd appreciate it, and in return I'll let you know which entire chapters say that 1 meter square esl panels playing well up into the high midrange (or higher) can't possibly be used in a serious quality system.
On this we can agree, more data is better than less. The problem is that there is already lots of data, you just chose to ignore the stuff that doesn't confirm your biases.
But at this point, after talking about this stuff on this forum for 8 years I think it's well past high time for you to put together a MF system and measure it. So far you've shown nothing technical, only ideas, many of which have been repeatedly shown to be inaccurate and incorrect.
Last edited:
Ben,
Congrats on grandchild #10.
Yes, square waves do make for a great (but nasty sounding) test signal.
Once you get around to measuring phase, you will find that for a loudspeaker to reasonably reproduce a square wave, it must have near flat phase response in the pass-band tested. That is not an abstract concept, it is a direct result of the transfer function.
If you (correctly) employ FIR filters, pretty much any speaker's phase response can be made to be flat.
Using FIR filters plus MF (like the IPAL system does) can result in flat phase and reduced distortion at high drive levels.
Using good drivers (like the IPAL system does) can shake your house up at quite high (for domestic comparisons) SPL with no audible distortion, and employing FIR filters can flatten phase response (thereby making a speaker reproduce square waves) for free, other than the time it takes to learn how to implement them to achieve your goals.
Art
Last edited:
My friend says he got square waves from his Quads* - at some frequency(s)... maybe, sort of. If I sound envious, it is because I have never had much luck getting a mic to pick anything resembling a square wave from a single ESL cell except by playing around with mic positions on my bench. Certainly not forthcoming from in-service speakers where the phase does somersaults Hz by Hz and inch by inch into the room.
But if you have sq wave mic pix to show, post 'em. And I will be very envious.
Sq waves great for amp R&D, but not possible for my speakers. (For Loathers?) Tone bursts are trivial for amps to reproduce, but a revealing test for speakers... but hopeless if they have "group delay".
Anybody know a web link of free tone bursts? One that produces precise cycle counts, not millisecond durations. The tone bursts on REW are Hann-shaped (they taper gently); dunno why. I don't have intuitions about how a good or bad speaker would handle these.
Ben
*Quad ESLs have multiple cells but the freq workload is narrowed in the cell array as the freq rise
Last edited:
Revealing what exactly? As weltersys just said, a square wave will show you that your speaker has flat phase and adequate frequency response. Since you don't care about phase you won't ever achieve a square wave.
Unity/Synergy horns are said to be able to produce square waves well enough but I doubt you care as they are completely contrary to your design philosophy in just about every way.
Tone bursts just reveal how accurate a speaker is at producing a tone burst. Can you tell the difference between a regular low distortion speaker playing well within it's linear region vs a theoretically perfect speaker that can produce clean tone bursts in a blind test listening to music? I seriously doubt it, your ears are not nearly as sensitive as your 'scope, especially at your age.
You quote "group delay" as though it's a phrase that doesn't have a clear meaning. It's already been explained to you that group delay is an actual delay caused by phase. But again, since you don't care about phase, you can't do anything about group delay and you can't ever achieve the goals that you value so highly.
Last edited:
But you have Dayton-Wright 1 meter square esl panels IIRC, not Quads. As far as I could tell when I looked them up yours did not work like you say the Quads work. Yours seem to have a bunch of "full range" cells, not different frequency cells.
There are of course probably dozens of models of Dayton-Wright esl and it's quite hard to find any tech information other than marketing specs, but as far as I know your esl panels are a full range dispersion nightmare.
I still regularly use my Tektronix 465B purchased at an electronics swap-meet in non-working condition shortly after college graduation. For repairing my antique radios, or troubleshooting amplifiers, etc. I find an analog O-scope superior to digital when tracking down instabilities and odd high-Q resonant circuit behavior as well as troubleshooting modulation problems.
Looking at square waves is a quick way of getting a feel for the overall response(ie transfer function) of the device you are testing. However, it requires knowledge and experience to interpret the meaning of different waveform shapes that show up. If you look at the attached guidance(vintage but still applicable), you will see in every case(except Fig.1 - J) the interpretation of the particular alteration to square wave shape is given in terms of what the frequency response or phase response must be. (ie transfer function). Figure 1 – J could have equally been describes as have a high-Q peak in the response. Note that a high-Q dip or notch would also show ringing, but the leading excursion of the ringing is of opposite phase.(See attachment #2). Implicit in the interpretations of the square wave shape is that the response is minimum phase. For example, you could have a high-Q all pass filter having flat magnitude response and you would be hard pressed to distinguish from a notch based on just looking at the square wave response. (see attatchment #3)
For loudspeaker design, I generally swap my O-scope alligator clips for REW alligator clips and skip the middle-man. With measurement tools available to us today, like REW, you can measure the frequency response and phase directly and avoid the need for interpretation. I’ll be posting the square wave response spreadsheet in a few days and you can play with generating each of the waveform shapes shown by entering the response/phase description. In short, square wave response and transfer function are two sides of the same coin. Whichever side you or anybody else are most comfortable with, is fine with me.
I still find tone bursts to be useful diagnostic signals for MFB systems when evaluate stability of the feedback loop, providing a quick thumbs-up/down. For example, you may recall the instability at 13Hz eventually tracked determined to be caused by ACH-01 mounting issue. There looked to be something amiss based on the frequency/phase response plot, but the tone burst made it really obvious. (see Fig. 2) http://www.diyaudio.com/forums/subwoofers/239941-analog-servo-sub-5.html#post3588625
In my experience, the MFB woofer system by itself can be fully quantified with near-field frequency/distortion response & phase measurements. Add the effect of the room and you’ve entered an area of ongoing investigation/research by many in the field. Any and all ideas are welcome.
Attachments
Last edited:
OK. Since a few folks around here are busy trying to establish who has the most lab experience going back to the "brass instrument" days and who knows the most cute tricks to "teach" the others....
When I was at Bell Labs* fooling with MF (my hobby time and encouraged by Bell, but not part of my professional role at The Labs) I used a Spectrograph and a storage scope with a Polaroid camera.
Spectrograph
That might settle the question of who is a fossil oscilloscope user (but certainly not who is sharpest).
Ben
*Murray Hill, of course, where the Nobel Prize winners worked and German guy named Manfred Schroeder worked a few doors from my office
When I was at Bell Labs* fooling with MF (my hobby time and encouraged by Bell, but not part of my professional role at The Labs) I used a Spectrograph and a storage scope with a Polaroid camera.
Spectrograph
That might settle the question of who is a fossil oscilloscope user (but certainly not who is sharpest).
Ben
*Murray Hill, of course, where the Nobel Prize winners worked and German guy named Manfred Schroeder worked a few doors from my office
Are you familiar with "Multi Sub Optimizer" program? I've only briefly glanced at it so far but it's supposed to take the guesswork and empirical fiddling out of the equation when setting up a multi sub system. Seems like the type of thing that might be right up your alley, although I can't vouch for it as I haven't downloaded it yet or even read much about it.
Optimizing subwoofers and integration with mains: multi sub optimizer - AVS Forum | Home Theater Discussions And Reviews
Indeed it is
I have been playing with it off and on since it was first offered; fairly good success...always interesting to see how things compared with what I came up with by manually tweaking setting in my simple spreadsheet "optimizer". The author was careful to point out and avoid Harman's patented SFM method. Of course that just begs for a comparison of the two approaches...which I have been slowly working on.
Attached is a PDF I made from the Help and Tutorial notes for the "MS Optimizer".
There was also a thread on the AVS forum you might remember seeing...
Optimizing subwoofers and integration with mains: multi sub optimizer - AVS Forum | Home Theater Discussions And Reviews
You got to use a mechanical spectrograph? Really? Very, very cool.
At some point I hope to acquire one for my antique instruments collection.
What was your role at Bell? I seem to recall you saying you were a psychologist?
Interesting to hear that they encouraged after-hours “play time” by their employees using company facilities….definitely seems to be a no-no in most current large companies.
Wonderful to be at Bell Labs "idea factory" during that great age:You got to use a mechanical spectrograph? Really? Very, very cool.
At some point I hope to acquire one for my antique instruments collection.
What was your role at Bell? I seem to recall you saying you were a psychologist?
Interesting to hear that they encouraged after-hours “play time” by their employees using company facilities….definitely seems to be a no-no in most current large companies.
https://en.wikipedia.org/wiki/The_Idea_Factory:_Bell_Labs_and_the_Great_Age_of_American_Innovation
Everybody called the stock room, "the gift shop" and I used the world's biggest anechoic chamber on Sundays... when they weren't testing a new thing called "electrets".
Bell had a bunch of Spectrographs, big like refrigerators. How else to freq analyze before FFT and computers? They'd spin and electrify the pages to mark them and make a big smell. Wasn't detailed enough for my MF purposes.
Our department was human information processing such as voice and instrument simulation including colour vision. Not sure which was Schroeder's department but he was respected even at Bell. A few doors down, a fellow biker buddy was developing some computer crap called Umis or Yunix or Linplus... long time ago. Or maybe it was C.
Ben
Last edited:
Indeed I have seen that avs thread, in fact I just posted a link to it in my last post.
Your attachment looks very interesting. I haven't downloaded the program or the instructions yet so this document looks like a good place to start.
I've got a calibrated Umik coming from Cross Spectrum Labs, I need to figure out which mini dsp I want to buy (which will probably require reading the manuals for ALL of them, including the manuals for all the plug ins, which could take some time) and I have get the latest REW and read the manual for that.
I have a lot of work to do.
Additionally I'm working on sourcing materials for some room treatments - I found some rigid Roxul on kijiji that's pretty local and a fantastic price (about 7x cheaper than retail with tax included) and $100 will get me more than I will need (fifteen 6 inch panels or thirty 3 inch panels). These will be for hanging treatments, I'll get some pink fluffy for the big corner traps. Having trouble making appointment with the seller though.
Time to put my money where my mouth is on these aspects of the system. My second annual busy season is ramping up so I might not get to actually do any of this work for a few months yet.