Hi Rob
Sound cards have no problem with CD quality reproduction (unless they are junk) so there is no reason to believe that they could not simulate anything to the desired degree. This would include linear or nonlinear effects. IF Soundeasy cannot do an adequite job, thats too bad, but there is no reason why it could not be done. DCR is a linear efect, even as it reacts with the amps output impedance, so just get its frequency response effect modeled and you have the effect.
If I believed in listening as a design tool then I would pursue this and get it to work, but alas, if you know me then you know that I do not design this way. To me there is no point in listening until you have a design that measures precisely the way that you intended and then a aural simulation is not necessary because you have the real thing. Even then, I don't put much emphasis on "my" listening. You just fool yourself that way.
SoundEasy can do it, certainly for the linear effects. The full system (box, drivers and crossover) can be modeled and then auditioned. I do it all the time. John could elaborate on any non-linear capabilities.
I usually still use CALSOD for final optimization partly because it is still the only software I've found that has the option to specify inductor gauge that is then used to calculate inductor DCR during the optimization process. This, as you know, is more important for traps that must be tuned. I even found that the results for common laminated-core inductors I use from Madisound come very close by specifying 12 gauge for the conductor. No non-linear effects, of course, but as you point out, they are generally linear in the operating currents found in crossovers.
If anyone wants to hear the difference in DCR values of inductors, it's far faster to make a change in an emulator than to do anything with a physical crossover short of a switch box of some kind.
And of course any decent software can have the output impedance of the amplifier specified to examine the interaction with the crossover.
Dave
Facts of freedom!
Oh WOW dantheman- I complete agree and dis-agree! Almost everything there was to know about audio was discovered and documented by 1950 and fully forgotten by 1965. 1947-1948 is when the AES journal started in an attempt to bring audio into a centralized periodical of peer review. I can say from first hand experience peer review is vastly over rated being more like "member of the clubbed" review. Clubbed into submission to participate in the popular knowledge. Am not saying there is not a lot of good stuff in AES but will say beyond any shadow of doubt there is more junk there. So many of the "qualified research" papers have so many important factors which have been completely ignored in favor of the popular knowledge of the moment that I quit reading the journal in 1978. You and geddes bring up Toole over and over again and my take on that effort is it so sloppy in the parameters I find important as to make the work a curiosity only.
Many here have thought there would be agreement on audio principles found within these threads but disagreement is the rule. I find this the most useful of all audio forums because people will speak their minds and flaming is not permitted. In this way the persistent individual can explore and discuss many ideas and concepts through reading these threads. This is a GREAT thing. Unlike the journals, the conclusion is never made and the factoid of the moment is not crammed down our throats by the peer review process. It truly is a forum of ideas and experiences and not a monument to the ivory tower league.
I will once again point out DeForrest invented the audion tube and had no idea how it worked. Armstrong came along and discovered its' intimate details. Often it is the work of but one person (like a Tesla or Maxwell) that bring the pieces so many have struggled to discover and document together into a coherent and sensible explanation. It is very clear here in this forum when certain members talk it is very well worth listening and likewise, it just as well to completely ignore others.
Live with it. Love it or hate it or anywhere in between. It is one of the few places I have seen where ideas may be presented and not be summarily rejected by peer or other writer review because that group or an individual said "I know better than you" though this does happen often with cetain writers here. We all know who they are always saying my way is the only way and I am always right. Personally, this is my favorite reading place in all audio of anything since 1975. I have finally found a place where I can actually learn something new! I believe others can also.
🙂🙂=SUM
Oh WOW dantheman- I complete agree and dis-agree! Almost everything there was to know about audio was discovered and documented by 1950 and fully forgotten by 1965. 1947-1948 is when the AES journal started in an attempt to bring audio into a centralized periodical of peer review. I can say from first hand experience peer review is vastly over rated being more like "member of the clubbed" review. Clubbed into submission to participate in the popular knowledge. Am not saying there is not a lot of good stuff in AES but will say beyond any shadow of doubt there is more junk there. So many of the "qualified research" papers have so many important factors which have been completely ignored in favor of the popular knowledge of the moment that I quit reading the journal in 1978. You and geddes bring up Toole over and over again and my take on that effort is it so sloppy in the parameters I find important as to make the work a curiosity only.
Many here have thought there would be agreement on audio principles found within these threads but disagreement is the rule. I find this the most useful of all audio forums because people will speak their minds and flaming is not permitted. In this way the persistent individual can explore and discuss many ideas and concepts through reading these threads. This is a GREAT thing. Unlike the journals, the conclusion is never made and the factoid of the moment is not crammed down our throats by the peer review process. It truly is a forum of ideas and experiences and not a monument to the ivory tower league.
I will once again point out DeForrest invented the audion tube and had no idea how it worked. Armstrong came along and discovered its' intimate details. Often it is the work of but one person (like a Tesla or Maxwell) that bring the pieces so many have struggled to discover and document together into a coherent and sensible explanation. It is very clear here in this forum when certain members talk it is very well worth listening and likewise, it just as well to completely ignore others.
Live with it. Love it or hate it or anywhere in between. It is one of the few places I have seen where ideas may be presented and not be summarily rejected by peer or other writer review because that group or an individual said "I know better than you" though this does happen often with cetain writers here. We all know who they are always saying my way is the only way and I am always right. Personally, this is my favorite reading place in all audio of anything since 1975. I have finally found a place where I can actually learn something new! I believe others can also.
🙂🙂=SUM
Last edited:
Software is statistical
Simulators and all that are fine but they are not the real thing! There is no substitute for direct measurement and direct experience. I suggested the simplest experiment with a small resistor on a cheap speaker to convince anyone of the effects of series resistance. This is a 45 cent test which will show a very solid result. Why bother with a bunch of software when such easy direct results are available? Is it the draw of "the computer has the answer?"
This reminds me so much of the earlier minimum phase discussions. How about qualifying/validating the software by creating a know test system using the two tone/multi-tone method and then testing the software to see if it conforms to a known value? No one was interested. Why? Conjecture is it is a lot of trouble and the results would likely conclusively show the software does not work. I have done this probably 45 times for 45 pieces of software. 2 work, the other 43 do not.
I really am a fan of direct measurement over anything of computer. This is true even though I was one of the earliest computer model makers back in 1975. Computers are quite useful but software is another story. Modern software is statistical in nature and the same action can result in a different outcome. How is one supposed to discover with that variable thrown in? Validation is required or software is useless.
SUM
Simulators and all that are fine but they are not the real thing! There is no substitute for direct measurement and direct experience. I suggested the simplest experiment with a small resistor on a cheap speaker to convince anyone of the effects of series resistance. This is a 45 cent test which will show a very solid result. Why bother with a bunch of software when such easy direct results are available? Is it the draw of "the computer has the answer?"
This reminds me so much of the earlier minimum phase discussions. How about qualifying/validating the software by creating a know test system using the two tone/multi-tone method and then testing the software to see if it conforms to a known value? No one was interested. Why? Conjecture is it is a lot of trouble and the results would likely conclusively show the software does not work. I have done this probably 45 times for 45 pieces of software. 2 work, the other 43 do not.
I really am a fan of direct measurement over anything of computer. This is true even though I was one of the earliest computer model makers back in 1975. Computers are quite useful but software is another story. Modern software is statistical in nature and the same action can result in a different outcome. How is one supposed to discover with that variable thrown in? Validation is required or software is useless.
SUM
This is nothing new, we know the effects of DCR. The fact remains that one can examine the impact before ever building anything. One can also hear the difference by auditioning the changes. None of that is counter to your method, just another one. It's just easier to do with a DF package. I and others already have it, it's becoming more common. The point is that it is just as valid as your 45 cent test and far more flexible.
I don't know what software you've found lacking, but I have been measuring and designing for years and have found exceedingly tight correlation between design and subsequent measurements. The software I use is not statistical. The same action yields the same result. Every time. My experience is apparently quite different from yours. It does not mean that yours invalidates mine and vice versa. You seem to have a different opinion on that. I do not accept your opinion on this matter.
Dave
P.S. The discussion on M-P is certainly not over as I'd say that there is still not yet unquestionable and convincing evidence to counter the generally accepted theory as most other evidence is indicative of M-P behaviour. The posts that question the validity leave it very much in doubt.
Last edited:
Since there is some discussion of active vs. passive I'll toss in a few cents worth.
We put a lot of study into active systems while I was at KEF, primarily for trying to design better studio monitors.
A couple of points to ponder: First a triamped system with 3 x 33 watt amps does not equal 100 watts. As you subdivide the spectrum you need similarly large amps each section unless you know your signal will always be equally distributed. (It won't.)
With passive networks you deal with the interaction of crossover and driver impedance curve. Not a difficult task with todays modeling software (or KEF's in 1985!) Also with the power handling of passive components. Issues of woofer damping aren't really a factor. You can easily achieve a target system Q while accounting for moderate inductor resistance.
With active crossovers you need to be comfortable with S domain polynomial fits, factoring into biquads and dealing with op-amp circuits. Also not difficult these days, but a different "domain". You can play with non-standard amplifier output impedances but I think there are more benefits for the full range fans than the multiway system fans.
I don't know of any good speaker designers that couldn't make either approach work.
One thing that really bothered us at KEF was the amount of power wasted in the equalization process (assuming a passive network) while flattening the midrange response of the typical woofer. It is easy to loose 10dB of midrange with a moderate sensitivity woofer. Furthermore, while making response flat you are making the maximum output curve (at clipping) flat, when it could have been the higher and more the natural shape of the peak output curve of typical music.
I ended up designing a system for broadcast monitoring called a P60 that had just enough crossover network to blend the drivers together but did all of the EQ actively: a hybrid network. Today, with DSP, I think this is a very valid approach: let the passive network blend the drivers with regard to the best possible polar performance and use DSP EQ to achieve really flat response. Drive it with a large, flat, low Z amp. Add music. Enjoy.
David
We put a lot of study into active systems while I was at KEF, primarily for trying to design better studio monitors.
A couple of points to ponder: First a triamped system with 3 x 33 watt amps does not equal 100 watts. As you subdivide the spectrum you need similarly large amps each section unless you know your signal will always be equally distributed. (It won't.)
With passive networks you deal with the interaction of crossover and driver impedance curve. Not a difficult task with todays modeling software (or KEF's in 1985!) Also with the power handling of passive components. Issues of woofer damping aren't really a factor. You can easily achieve a target system Q while accounting for moderate inductor resistance.
With active crossovers you need to be comfortable with S domain polynomial fits, factoring into biquads and dealing with op-amp circuits. Also not difficult these days, but a different "domain". You can play with non-standard amplifier output impedances but I think there are more benefits for the full range fans than the multiway system fans.
I don't know of any good speaker designers that couldn't make either approach work.
One thing that really bothered us at KEF was the amount of power wasted in the equalization process (assuming a passive network) while flattening the midrange response of the typical woofer. It is easy to loose 10dB of midrange with a moderate sensitivity woofer. Furthermore, while making response flat you are making the maximum output curve (at clipping) flat, when it could have been the higher and more the natural shape of the peak output curve of typical music.
I ended up designing a system for broadcast monitoring called a P60 that had just enough crossover network to blend the drivers together but did all of the EQ actively: a hybrid network. Today, with DSP, I think this is a very valid approach: let the passive network blend the drivers with regard to the best possible polar performance and use DSP EQ to achieve really flat response. Drive it with a large, flat, low Z amp. Add music. Enjoy.
David
Last edited:
Hello Earl
Thanks for your response. It got me thinking and I was missing something the amps output impedance. I forgot I could input it into the software.
Rob🙂
Speaker dave has some interesting conclusions. Has anybody done an ABX with a DSP? If you fret about too many transistors, you'd scream about a DSP.
By most accounts, you are wrong about amp power. The stats on brief peak power demand goes straight skyward as you enlarge the passband and dividing it up by bi-amping reduces it far more than in proportion. It is the occasional peak clipping that drives people to go big power. I am conservative about the value of going big power (and ESLs survive the destructive power of clipping products) but the statistical reality of occasional super high peaks is true... esp. women's choral music.
When I get to heaven, all the simulations will be perfect. Just because all the mathematically tractable variables, no matter how inaudible, are "under control" does not mean that all the important variables are accounted for when your half-inch mdf speaker ends up between your coffee table and your curtains.
The low output impedance of the amp is there to damp back EMF from the voice coil arising from unexpected (error) motions of the cone... which aren't in the simulation because they don't exist in the ideal world of the model speaker.
By most accounts, you are wrong about amp power. The stats on brief peak power demand goes straight skyward as you enlarge the passband and dividing it up by bi-amping reduces it far more than in proportion. It is the occasional peak clipping that drives people to go big power. I am conservative about the value of going big power (and ESLs survive the destructive power of clipping products) but the statistical reality of occasional super high peaks is true... esp. women's choral music.
When I get to heaven, all the simulations will be perfect. Just because all the mathematically tractable variables, no matter how inaudible, are "under control" does not mean that all the important variables are accounted for when your half-inch mdf speaker ends up between your coffee table and your curtains.
The low output impedance of the amp is there to damp back EMF from the voice coil arising from unexpected (error) motions of the cone... which aren't in the simulation because they don't exist in the ideal world of the model speaker.
Last edited:
Thanks Ben and SUM.
Speaker Dave, I just completed a system similar to what you just said but in reverse order. I used the Behringer CX3400 3-way crossover(in 2-way mode) and the passively EQed the drivers to better fit the desired response. It worked rather well. Since all drivers measured a bit different, I was able to change the EQ circuit just a bit to suit each more perfectly. Of course this didn't need to take place after amplification.
That's what kills me about the commercially produced speakers I'm listening to now--their measurements are so close. Irks me in a way....... I'd think they'd be further off from each other since there's no way they are actually measuring each speaker they sell as "matched pairs" let alone slightly modifying the crossover to make it work more perfectly or pick 2 out of a batch of many that measure so closely. It would seem that the expense of that would be outrageous. Maybe there just using successive units from the assembly line, but I don't know that it would necessarily lead to this tight of tolerances. IOW, I don't now how they've done it, but they've done it just the same to paraphrase Tampa Red
My limited experience matches Ben's last post.
Anyway, many ways to skin the cat.
Dan
Speaker Dave, I just completed a system similar to what you just said but in reverse order. I used the Behringer CX3400 3-way crossover(in 2-way mode) and the passively EQed the drivers to better fit the desired response. It worked rather well. Since all drivers measured a bit different, I was able to change the EQ circuit just a bit to suit each more perfectly. Of course this didn't need to take place after amplification.
That's what kills me about the commercially produced speakers I'm listening to now--their measurements are so close. Irks me in a way....... I'd think they'd be further off from each other since there's no way they are actually measuring each speaker they sell as "matched pairs" let alone slightly modifying the crossover to make it work more perfectly or pick 2 out of a batch of many that measure so closely. It would seem that the expense of that would be outrageous. Maybe there just using successive units from the assembly line, but I don't know that it would necessarily lead to this tight of tolerances. IOW, I don't now how they've done it, but they've done it just the same to paraphrase Tampa Red
My limited experience matches Ben's last post.
Anyway, many ways to skin the cat.
Dan
Last edited:
Confusion again- All things being equal, in a multi amp system the energy delivered to each driver should be about the same as delivered to the drivers, without crossover loss, as a full range passive crossover system. In a multiple amplifier system each amplifier delivers less energy. That does not mean less voltage output is required from the amplifiers. Just less energy from each amplifier. Multiple domains- time, frequency, and energy- voltage, current, and reactance as another example.
I have done a lot of laboratory dynamic range testing as describe earlier in this thread. One of these days will post photo results from the effects of increased amplifier damping on dynamic range and distortion whenever I get around to it. Other test results suggest increases in dynamic range of +6dB over a .1 ohm series resistor. No way to achieve that with any passive network. So like many things, passive networks are a trade off of other things like- I only have one good amplifier so it must be a full range system. Very valid. Or like me- I have DSP and lots of really nice amps so one for each.
My next upgrade plan is to try out some FIR filters and see about that. Very attractive characteristics which are also impossible with passive components.
I am glad some peoples software does the same thing each time. That would make validation to a known standard much easier. Do you validate? I always do. Part of the scientific method- calibration and validation of method. It is the researchers job to do this.
🙂=SUM
I have done a lot of laboratory dynamic range testing as describe earlier in this thread. One of these days will post photo results from the effects of increased amplifier damping on dynamic range and distortion whenever I get around to it. Other test results suggest increases in dynamic range of +6dB over a .1 ohm series resistor. No way to achieve that with any passive network. So like many things, passive networks are a trade off of other things like- I only have one good amplifier so it must be a full range system. Very valid. Or like me- I have DSP and lots of really nice amps so one for each.
My next upgrade plan is to try out some FIR filters and see about that. Very attractive characteristics which are also impossible with passive components.
I am glad some peoples software does the same thing each time. That would make validation to a known standard much easier. Do you validate? I always do. Part of the scientific method- calibration and validation of method. It is the researchers job to do this.
🙂=SUM
Last edited:
But when less than an ohm in my passive network raises my Q from .4 to .7 (or similar dB variations in a vented box), I'm not going to lose sleep over cone motion error.
Read T.H. Wiik of SEAS on progressive spiders if the subject interests you. An easy solution to DC offset problems
David Smith
The software will model any change in the frequency response due to a change in series R. As Ben says, it won't model any distortion (loss of woofer control) caused by reduced damping factor. But the software is a perfect tool to find out if such distortion exists. Build an active system with the software doing the modeled crossover and build the same system with caps and coils. If the caps and coils are introducing any distortion that isn't being modeled, there should be a measurable and (maybe) audible difference.
My opinion, any difference you hear due to series R is because of the frequency response change. Simply putting an R in series and listening will of course change the sound but it doesn't prove anything. All that can be modeled perfectly. Any 'extra' distortion is inaudible IMO at least with normal drivers. If you have something exotic, say a very low impedance driver that requires a specialized amp, active may be the way to go. Active is also good if you want to use drivers with very different sensitivities.
Yes, whatever works. Each method will have theoretical as well as practical plusses and minuses. I knew an engineer that would put 5 or 6 little resonant notch circuits into each network. They were usually good for a dB or two of droop over an octave or so. When we pointed out he could have achieved the same thing with smarter optimization of the basic network...
David
I've measured woofer distortion with one of the old tube amps with the variable damping controls. There was no difference in distortion at low frequencies. There was a difference at mid frequencies where the nonlinear magnet B was distorting the current. A higher amplifier source impedance will tend to linearize that as the high source R dominates the nonlinear contribution of the woofer (i.e. linearizes the current).
JBL SFG and Mac LDHP, and other similar type magnet structures with flux shunting rings will be better in this regard. Alnico (and I think NEO) are better than straight Ferrite.
In the LF high excursion region non-linear B and suspension will dominate and changng source R seems to have no effect.
David Smith
Last edited:
An amp from Dave's museum? Love to see it!
Various odd old amps like that. Fisher? Cousin to all-electric motional feedback, one of the greatest ideas in speaker history (and the only realm in which feedback R&D has come to a cold stand-still... dunno why).
Can you tell us more about your test set-up, if the knob on the amp really changed anything in the output, driver enclosure, and what effect did the amp have on the low end of the frequencies.
I know one of the modeler's will ask, "Just why would I need feedback around my woofer? I can model the speaker perfectly, even heating up the voice coil."
Last edited:
Ben, I don't think you give us modelers enough credit. 🙂 Woofer servo feedback is alive and well. Some use an accelerometer attached to the cone and some use a second voice coil to drive the feedback loop. Velodyne subs are probably the best known. For DIYers, Rythmik has kits.
Servo subwoofers •Rythmik Audio Direct Servo subwoofers
All that said, motor design has improved so much in recent years, due in part to, yes, software, that most people don't find servo motors worth the hassle. There's little to no performance advantage if you start with a driver with symmetrical Bl, symmetrical suspension and low Le.
Servo subwoofers •Rythmik Audio Direct Servo subwoofers
All that said, motor design has improved so much in recent years, due in part to, yes, software, that most people don't find servo motors worth the hassle. There's little to no performance advantage if you start with a driver with symmetrical Bl, symmetrical suspension and low Le.
Interesting that you mention tube amps. Ever since I got a 500V shock, I decided It's play it in the save side. But I do understand it's another tipic that could go on and on.🙂
I was looking at a few other inductors, it seems how the impedance phase curves to zero differ with different. Hmm, it seems to get real complicated looking deeper. I even have impedance curves for different cored inductors shaped differently.
Seriously, I think there are lots of variables in speaker design. If 45cm of cable makes a difference, even with a series resistor, anything will make a difference. But golly, trying to sort all this out really is a pain in the neck.
What, your amps don't have the famous "Fisher Z matic" control?
This was quite a few years back (I still have the curves) and using a Fisher 80AZ. Fisher and a couple others of that era sold amps with variable damping knobs. A pot in the feedback loop gave a variable combination of positive(?) current feedback and negative voltage feedback. The end result was the ability to vary from a normal damping factor (say 20) down to a low damping/high Z output. The trick was doing it in a way that kept the volume roughly constant.
The measurement I took was a standard fundamental/second harmonic/third harmonic plot, but I used an old B&K oscillator that had a feedback loop to keep the woofer output constant. Otherwise the variable damping would give markedly different response curves and an apples and oranges comparison. The response curves are as expected: whatever the starting Q, the lesser damping positions progresively rasied level at resonance and high frequencies. As much as people get excited about the concept of "damping" in the end you are just allowing the output voltage to follow the shape of the impedance curve (with low damping) or forcing flat voltage (with high damping). Ain't nothin' in the amplifier "reaching out and grabbing the woofer cone".
By the way, I still have the amps but removed the Z-matic panel and circuit and converted to fixed bias via a zener and transistor regulator instead of a Cathode resistor.
David
Last edited:
I can't believe my ears!
Why that is exactly what single-sided class-A amp builders said a mere 80 years ago when some person suggested feedback around their amps. Besides, few people can hear 3% harmonic distortion on music... and we're so busy deciding just how much early wall reflection is best.
Mock horror aside, thanks for update on the precious few companies still interested in motional feedback. I experimented with it very crudely when I was at Bell Labs. Did pulse testing of a woofer (what does that reveal, measurement gurus?) in world's largest anechoic chamber and took Polaroid scope shots (long ago, eh).
Astonishing what a difference it makes when your electronics GRABS your woofer and makes it act right. You don't know the meaning of "tight bass" till you've worked with it...
...and destroyed some expensive drivers.... I add that to introduce some thrill-seeking into this thread.
- Status
- Not open for further replies.