Lynn Olson said:
Lynn,
On a serious note, both the open baffle and Acoustic Resistance Box (quasi-cardioid, first introduced in 1950) yield minimum phase response and when the respective crossover filters are applied they will have some band pass response which will define the stored energy. So just what is it in regards to stored energy that you want to minimize?
The goals implied by Lynn was that minimal stored energy is amongst the highest of priorities. DEQX can do nothing for subtracting stored energy AFAIK. It can eq the on axis reponse and do all the crossovers for you, but it can't correct the time domain (other than driver to driver offset).
Don't get me wrong, I am not bad mouthing the DEQX, I have one. It is a great tool if you use it appropriately, but it is not a cure-all for more fundamental problems. In this case it can't remove stored energy, so one must still plough through the hard work of testing to minimize this from the start.
Don't get me wrong, I am not bad mouthing the DEQX, I have one. It is a great tool if you use it appropriately, but it is not a cure-all for more fundamental problems. In this case it can't remove stored energy, so one must still plough through the hard work of testing to minimize this from the start.
JoshK said:
Hi Josh
That is not precisely correct. If the resonance - a better word for energy storage IMO - is minimum phase then correcting the amplitude response will correct the time response. But many things like structural resonances, acoustic resonances and diffraction are not minimum phase in that their effects on the response differs at different field locations and these cannot be so corrected as you state. But one can't make a blanket statement like this since its not always true. Some things are correctable with EQ and some are not. Thats why its important to sort out which are which and attack each with that solution that works in each particular case. You cannot simply lump together all FR aberations and expect to correct them with electronic EQ.
brucemck2 said:
Having no knowledge of the AH425/288 or how it performs I really could say. About all I can say is that the NaO U-frame is a good woofer system if you want quasi-cardioid bass from about 100 Hz down to the lower cut off of the woofer. Other issues to be addressed are sensitivity, max SPL capability and blending the to the polar/power response of the 100Hz up source, etc. See my brief discussion on crossocers between different source types.
DEQX & U frame
Hi Brucemck2 and John K,
I hope this is of interest and helps answer your DEQX question.
I have been experimenting with the DEQX since 2004 and have found it a very useful device indeed.
I now use my own design of battery power supply with the DEQX boards and external DACS (NOS or 32 bit ESS DAC's) and have built both Alloy and Acrylic casework (Acrylic is best!) to address the recognised weakness' (necessary commercial compromises) in the DEQX at their retail price point.
In 2005 I used it with great success as the centre piece of a system to showcase the potential of my
"Variable Path length cancellation" U frame open baffle bass / low mids (up to 350 Hz in this case) combined with the sublime Manger driver housed in a cast teardrop head unit of my design.
The Variable Path length ( VPL…. I never was any good at marketing! ) address's the damping issues inherent in U frame design. I.e. take John K's example of the 18 inch U frame with its 170 Hz peak.
Try taking a large "shark bite" out of each of the side cheeks of the U frame back box and you instantly (and cheaply!) loose 80% 0f the resonance problem but keep 80% of the bass extension.
You may need to "bite" the top as well if the frame is less than 70cm (approx 32 inch's) tall.
I found that by evenly spreading the path length cancellation I got the best of both worlds. Further experimenting led to a series of golden ratios of baffle width, maximum and minimum side cheek depth, baffle height and of course how to best match these to the bass or bass / mid drivers I wanted to use.
With the benefit of the DEQX and the VPL U frame I could use really superb drivers that would not normally be able to work in an open baffle design.
As far as drivers go I favour paper cone, linen surround very low Mms, very high Bl, oversized copper voice coils and highest possible sensitivity. I am fan of Wave guides and certain horn shapes, Earl Geddes is very talented and I have learned a lot from reading his white papers and forum posts.
My latest design is a custom wave guided Manger driver above 400Hz, a 10 inch or 12 inch paper cone in a shallow wave guide from 400Hz down to 100Hz and then a choice (still experimenting…..) of sealed 15 inch or 18 inch VPL U frame.
The Manger sounds astonishing in a shallow oval wave guide.
Here is a section of product description from the DEQX web site :
" The DEQX Calibrated™ PDC preamp/processor measures the performance of your loudspeaker system, then processes the signal going to the speaker so that it compensates for errors in both the time and frequency domain.
DEQX’s proprietary time domain correction measures the extent that some frequencies lag behind others, a phenomenon known as ‘group delay'. DEQX can then delay, and phase align, the early frequencies so that the later ones can catch up. This results in the sound we finally hear having the same natural time-coherence as is present in the audio signal. Simultaneously, the frequency response is corrected so that the volume of frequencies that the speaker has made louder or softer is also corrected to match the original signal. Again this occurs at hundreds of different frequencies while also correcting phase at all frequencies. "
Could John K please elaborate on why he thinks that the DEQX can not do the above?
Thanks and all the best.
Derek
Hi Brucemck2 and John K,
I hope this is of interest and helps answer your DEQX question.
I have been experimenting with the DEQX since 2004 and have found it a very useful device indeed.
I now use my own design of battery power supply with the DEQX boards and external DACS (NOS or 32 bit ESS DAC's) and have built both Alloy and Acrylic casework (Acrylic is best!) to address the recognised weakness' (necessary commercial compromises) in the DEQX at their retail price point.
In 2005 I used it with great success as the centre piece of a system to showcase the potential of my
"Variable Path length cancellation" U frame open baffle bass / low mids (up to 350 Hz in this case) combined with the sublime Manger driver housed in a cast teardrop head unit of my design.
The Variable Path length ( VPL…. I never was any good at marketing! ) address's the damping issues inherent in U frame design. I.e. take John K's example of the 18 inch U frame with its 170 Hz peak.
Try taking a large "shark bite" out of each of the side cheeks of the U frame back box and you instantly (and cheaply!) loose 80% 0f the resonance problem but keep 80% of the bass extension.
You may need to "bite" the top as well if the frame is less than 70cm (approx 32 inch's) tall.
I found that by evenly spreading the path length cancellation I got the best of both worlds. Further experimenting led to a series of golden ratios of baffle width, maximum and minimum side cheek depth, baffle height and of course how to best match these to the bass or bass / mid drivers I wanted to use.
With the benefit of the DEQX and the VPL U frame I could use really superb drivers that would not normally be able to work in an open baffle design.
As far as drivers go I favour paper cone, linen surround very low Mms, very high Bl, oversized copper voice coils and highest possible sensitivity. I am fan of Wave guides and certain horn shapes, Earl Geddes is very talented and I have learned a lot from reading his white papers and forum posts.
My latest design is a custom wave guided Manger driver above 400Hz, a 10 inch or 12 inch paper cone in a shallow wave guide from 400Hz down to 100Hz and then a choice (still experimenting…..) of sealed 15 inch or 18 inch VPL U frame.
The Manger sounds astonishing in a shallow oval wave guide.
Here is a section of product description from the DEQX web site :
" The DEQX Calibrated™ PDC preamp/processor measures the performance of your loudspeaker system, then processes the signal going to the speaker so that it compensates for errors in both the time and frequency domain.
DEQX’s proprietary time domain correction measures the extent that some frequencies lag behind others, a phenomenon known as ‘group delay'. DEQX can then delay, and phase align, the early frequencies so that the later ones can catch up. This results in the sound we finally hear having the same natural time-coherence as is present in the audio signal. Simultaneously, the frequency response is corrected so that the volume of frequencies that the speaker has made louder or softer is also corrected to match the original signal. Again this occurs at hundreds of different frequencies while also correcting phase at all frequencies. "
Could John K please elaborate on why he thinks that the DEQX can not do the above?
Thanks and all the best.
Derek
Re: DEQX & U frame
Hi Derek,
Very interesting post.
I'm experimenting with U frames. Could you post some pictures of these cabinets?
If it's the speakers featured in 6moons, i think they dont show these details. Or i was too sleepy last night when i read it
/me takes note.
As I posted also in the Manger thread, what would be your opinion of a Bipolar setup with two W05?
Also, they made a one-way speaker, the Swing, with an interesting cabinet curved design with response down to 100hz, which is amazing considering the tender driver.
Manger Swing
It must be hard to find a mid-woofer to keep the pace of the Mangers!
Hi Derek,
Very interesting post.
Overkill Audio said:
I'm experimenting with U frames. Could you post some pictures of these cabinets?
If it's the speakers featured in 6moons, i think they dont show these details. Or i was too sleepy last night when i read it
/me takes note.
As I posted also in the Manger thread, what would be your opinion of a Bipolar setup with two W05?
Also, they made a one-way speaker, the Swing, with an interesting cabinet curved design with response down to 100hz, which is amazing considering the tender driver.
Manger Swing
It must be hard to find a mid-woofer to keep the pace of the Mangers!
I think you may have meant JoshK not JohnK as John K never mentioned anything about DEQX, but I did.
Just have a look at Gedlee's correction of my post. DEQX's blurb you quote talks of driver offset correction, not time-based resonance correction (for which it could correct for min-phase resonances only per Dr. Geddes).
Just have a look at Gedlee's correction of my post. DEQX's blurb you quote talks of driver offset correction, not time-based resonance correction (for which it could correct for min-phase resonances only per Dr. Geddes).
Manger and OB bass
Sorry I got your name wrong Josh K, I have read so many of John K's posts and a lot of great stuff on his web site that I must have had him in the back of mind!
I will take some pics of my latest experimental cabinets tomorrow and post a few, be warned they aint pretty! They are MDF test cabinets with black felt coverings not my previous commercial piano black and Walnut veneers...
Re the Manger, there is a lot to talk about with that driver and I don’t want to go off on a Manger tangent from Lynn's thread. I am happy to post here about my Manger experiments with open baffle / wave guide / matching bass & low midrange if Lynn and the guys don’t object. Lynn, please advise as I am wary of upsetting the master!
Very briefly or until otherwise advised here are my top ten points of view on the Manger driver and the best way to use it in its current guise:
(1) If you have listened to ( or tested with) the Manger using frequencies below 300Hz to 400 Hz you have not actually heard what the driver is really capable of. Sub 350Hz ish kills the Manger.
(2) When used with an shallow oval wave guide from 300Hz / 400Hz up (depending on crossover slope) the driver can reproduce sounds in like no other, technically and by subjective listening.
(3) The Manger must be used in a three way system as no driver capable of high SPL / low distortion sub bass can simultaneously have the required rise and settling time in the critical 100Hz to 400Hz band. The 100Hz to 400Hz band makes or breaks the Manger speaker system as a whole.
(4) Only large 10 inch to 12inch but still low Mms 20g to 50g drivers conforming to my previous posts spec (huge BL in particular) and used VPL U frame or OB with or without wave guides can attain the required speed. Large Sd covering very small linear movement (not pistonic, as all " mass on a spring drivers" oscillate way beyond their start stop points with every compression / rarefaction to such a degree that they can not be accurately described as pistonic) can accelerate and decelerate much faster than the usual 6 inch or 8 inch drivers covering long linear travel to match the Vd figure of the large Sd drivers in this range.
(5) Low inductance voice coils are a must.
(6) In order to minimise unwanted energy bouncing around the driver / cabinet structure you must maximise driver efficiency. A typical high end Hi Fi driver is approx 0.8% to 1.5% efficient at converting electrical energy into sound energy. Be careful to look at the frequency band you require the driver to cover, not just the optimum band the manufacturer quotes sensitivity at!
So where is 98% plus of you amplifier power going? It's being dissipated as voice coil heat and bouncing around the mechanical driver / cabinet system as Potential ( big rubber surrounds are soooo bad!) and Kinetic energy and then slowly (all things are relative!) being converted into heat through frictional loss.
Therefore if we double the efficiency to a massive, but attainable, 1.6% to 3.0% we can reduce by 50% the amount of unwanted and sonically destructive system resonance. Ie 50% reduction in distortion of the worst type. Much more involved here but I hope you can see the direction I am going with this.
(7) For the sub100Hz band, sealed is best for Home Cinema and most musicians prefer OB, U frame or VAL U frame for music. Always one big bass driver is better than two small ones if OB, U or VPL. If sealed subs, use 2, 3 or 4 corner / room boundary subs.
(8) Capacitors and inductors are "frequency dependant energy storage devices" i.e. they release different frequencies at different times, and some frequencies don’t get through at all…! Great for designing crossovers on paper but in the real world, time domain accuracy is the be all and end all, the real holly grail. Big passive caps and inductors store and release energy on a frequency dependant basis, they destroy time coherence and time coherence is one of the Mangers great strengths. Look at the great research into energy storage / release effects of passive filters on speaker systems John K's site, really interesting stuff.
So if you love what the Manger can do when you use passive crossovers imagine what it can do with a battery powered DEQX DSP and great DAC's…!
The standard DEQX DAC's can not begin to do justice to the Manger driver.
(9) The existing rear alloy basket on the Manger is a real weak point, much to be done here OB 1….
(10) The existing magnet system is far to weak. A doubling of the Bl is required to get the efficiency up to where it will really be the one to beat.
Again apologies for all the Manger stuff here but I hope it is also of some interest to the wider driver market as used by most of the DIY (and all of the commercial) designers.
All the best
Derek.
PS Josh, The Manger swing is a beautiful and very cool life style product, but way short of what the Manger driver can do in a fully developed 3 way system.
Sorry I got your name wrong Josh K, I have read so many of John K's posts and a lot of great stuff on his web site that I must have had him in the back of mind!
I will take some pics of my latest experimental cabinets tomorrow and post a few, be warned they aint pretty! They are MDF test cabinets with black felt coverings not my previous commercial piano black and Walnut veneers...
Re the Manger, there is a lot to talk about with that driver and I don’t want to go off on a Manger tangent from Lynn's thread. I am happy to post here about my Manger experiments with open baffle / wave guide / matching bass & low midrange if Lynn and the guys don’t object. Lynn, please advise as I am wary of upsetting the master!
Very briefly or until otherwise advised here are my top ten points of view on the Manger driver and the best way to use it in its current guise:
(1) If you have listened to ( or tested with) the Manger using frequencies below 300Hz to 400 Hz you have not actually heard what the driver is really capable of. Sub 350Hz ish kills the Manger.
(2) When used with an shallow oval wave guide from 300Hz / 400Hz up (depending on crossover slope) the driver can reproduce sounds in like no other, technically and by subjective listening.
(3) The Manger must be used in a three way system as no driver capable of high SPL / low distortion sub bass can simultaneously have the required rise and settling time in the critical 100Hz to 400Hz band. The 100Hz to 400Hz band makes or breaks the Manger speaker system as a whole.
(4) Only large 10 inch to 12inch but still low Mms 20g to 50g drivers conforming to my previous posts spec (huge BL in particular) and used VPL U frame or OB with or without wave guides can attain the required speed. Large Sd covering very small linear movement (not pistonic, as all " mass on a spring drivers" oscillate way beyond their start stop points with every compression / rarefaction to such a degree that they can not be accurately described as pistonic) can accelerate and decelerate much faster than the usual 6 inch or 8 inch drivers covering long linear travel to match the Vd figure of the large Sd drivers in this range.
(5) Low inductance voice coils are a must.
(6) In order to minimise unwanted energy bouncing around the driver / cabinet structure you must maximise driver efficiency. A typical high end Hi Fi driver is approx 0.8% to 1.5% efficient at converting electrical energy into sound energy. Be careful to look at the frequency band you require the driver to cover, not just the optimum band the manufacturer quotes sensitivity at!
So where is 98% plus of you amplifier power going? It's being dissipated as voice coil heat and bouncing around the mechanical driver / cabinet system as Potential ( big rubber surrounds are soooo bad!) and Kinetic energy and then slowly (all things are relative!) being converted into heat through frictional loss.
Therefore if we double the efficiency to a massive, but attainable, 1.6% to 3.0% we can reduce by 50% the amount of unwanted and sonically destructive system resonance. Ie 50% reduction in distortion of the worst type. Much more involved here but I hope you can see the direction I am going with this.
(7) For the sub100Hz band, sealed is best for Home Cinema and most musicians prefer OB, U frame or VAL U frame for music. Always one big bass driver is better than two small ones if OB, U or VPL. If sealed subs, use 2, 3 or 4 corner / room boundary subs.
(8) Capacitors and inductors are "frequency dependant energy storage devices" i.e. they release different frequencies at different times, and some frequencies don’t get through at all…! Great for designing crossovers on paper but in the real world, time domain accuracy is the be all and end all, the real holly grail. Big passive caps and inductors store and release energy on a frequency dependant basis, they destroy time coherence and time coherence is one of the Mangers great strengths. Look at the great research into energy storage / release effects of passive filters on speaker systems John K's site, really interesting stuff.
So if you love what the Manger can do when you use passive crossovers imagine what it can do with a battery powered DEQX DSP and great DAC's…!
The standard DEQX DAC's can not begin to do justice to the Manger driver.
(9) The existing rear alloy basket on the Manger is a real weak point, much to be done here OB 1….
(10) The existing magnet system is far to weak. A doubling of the Bl is required to get the efficiency up to where it will really be the one to beat.
Again apologies for all the Manger stuff here but I hope it is also of some interest to the wider driver market as used by most of the DIY (and all of the commercial) designers.
All the best
Derek.
PS Josh, The Manger swing is a beautiful and very cool life style product, but way short of what the Manger driver can do in a fully developed 3 way system.
Re: Manger and OB bass
Looking for them!
Ahem, It was me not Josh
You can post freely anything regarding the Mangers in this thread
Overkill Audio said:
Looking for them!
Ahem, It was me not Josh
You can post freely anything regarding the Mangers in this thread
My remarks about what I'm contemplating for the bass section have been a little opaque, partly because I'm still working it out with Gary Pimm.
Let's move several steps back and look at the issue of stored energy in the bass region - particularly the 100 to 800 Hz mid-bass to upper-bass region. Cone drivers are inherently flat and in the piston band, unless there is a serious problem with suspension or cone design (yes, I know about spider resonance - but that is much less problematic than the cone breakups at higher frequencies). By "stored energy" I mean not crossover artifacts, nor the rise-time of the impulse, but the "clutter" and resonant tails after the impulse has passed, particularly in the critical 0~5 mSec interval.
Although the driver is well-behaved, closed or vented boxes are not. With no filling, long-lasting standing waves dominate the decay picture. The effect of box filling is to damp the standing waves more quickly, but no filling can actually remove them, and the designer has to be careful not to get the filling too close to the cone, or it will start to mass-load the cone and depress efficiency and HF response.
Open-baffles are free of high-Q standing waves in the sense of closed or vented boxes, but diffraction around the edges is a real problem. Instead of a group of high-Q resonances, there's a single delayed backwave that diffracts around to the front after a delay of a millisecond or so, depending on baffle size. Making the baffle asymmetric and moving the driver off-center usefully blurs the backwave in the time domain, but it still emerges as a single delayed pulse - just more spread out.
One of the odd things about the heavily-filled Gary Pimm boxes - which are open at the back - is that the box modes are much less severe than you might expect (this is by measurement in the time domain). They're almost acting as if they were very short transmission lines, instead of a closed box with the back removed.
I've been discussing what would happen if the GP box were stripped down even more, becoming more like a lossy framework than a box. Bear with me here - visualize an ordinary slant-baffle, with vestigial side panels, just enough so the slant baffle doesn't fall over. Extend the side panels further back, but make them of perf board with carpeting glued on to each side. Make more perf boards parallel to the side panels, directly behind the driver, and at right angles to the sound coming from the back of the driver. If you stand directly behind the baffle, you will be able to see part of driver between the array of vertical carpet-covered perf boards.
The intent is to attenuate the backwave 6~12 dB over a broad band while minimizing reflections - which is why the perf boards are at right angles to the backwave, unlike the Hartley "Boffle" of the early Fifties. The operating principle is similar to a low-restriction car muffler or a silencer on a gun - a lossy mesh is used to attenuate a shockwave without adding restriction.
Now, I could be all wet, so I've asked Gary to measure something like this and see what it actually does in the time domain. It's not a closed box, it's not a GP box, and it's not really an open-baffle, either. It's a transitional structure, and the goal isn't a certain polar-pattern, but minimization of stored energy in the time domain.
The only other enclosure I can think that puts first priority on minimization of stored energy is the B&W Nautilus tapered-pipe design. I'm not looking for 100% absorption of the backwave - instead, I'd be happy with a modest 6~12 dB attenuation and diffusion in the time domain (instead of the backwave coming around the front all at once, it takes multiple paths that are attenuated and spread out in time).
If the attenuation really is broadband, the polar pattern will move away from a dipole and towards a cardioid pattern - again, transitional between a dipole and a closed-box monopole. The phase and magnitude of the backwave will control the exact pattern. Do I expect it to be stable with frequency? Only measurements will tell, although I expect plenty of variation.
Sorry about the vague description - I only mentioned this to Gary a couple of weeks ago, since he has the combination of instrumentation and intellectual curiosity to see if there is any merit to this idea.
Let's move several steps back and look at the issue of stored energy in the bass region - particularly the 100 to 800 Hz mid-bass to upper-bass region. Cone drivers are inherently flat and in the piston band, unless there is a serious problem with suspension or cone design (yes, I know about spider resonance - but that is much less problematic than the cone breakups at higher frequencies). By "stored energy" I mean not crossover artifacts, nor the rise-time of the impulse, but the "clutter" and resonant tails after the impulse has passed, particularly in the critical 0~5 mSec interval.
Although the driver is well-behaved, closed or vented boxes are not. With no filling, long-lasting standing waves dominate the decay picture. The effect of box filling is to damp the standing waves more quickly, but no filling can actually remove them, and the designer has to be careful not to get the filling too close to the cone, or it will start to mass-load the cone and depress efficiency and HF response.
Open-baffles are free of high-Q standing waves in the sense of closed or vented boxes, but diffraction around the edges is a real problem. Instead of a group of high-Q resonances, there's a single delayed backwave that diffracts around to the front after a delay of a millisecond or so, depending on baffle size. Making the baffle asymmetric and moving the driver off-center usefully blurs the backwave in the time domain, but it still emerges as a single delayed pulse - just more spread out.
One of the odd things about the heavily-filled Gary Pimm boxes - which are open at the back - is that the box modes are much less severe than you might expect (this is by measurement in the time domain). They're almost acting as if they were very short transmission lines, instead of a closed box with the back removed.
I've been discussing what would happen if the GP box were stripped down even more, becoming more like a lossy framework than a box. Bear with me here - visualize an ordinary slant-baffle, with vestigial side panels, just enough so the slant baffle doesn't fall over. Extend the side panels further back, but make them of perf board with carpeting glued on to each side. Make more perf boards parallel to the side panels, directly behind the driver, and at right angles to the sound coming from the back of the driver. If you stand directly behind the baffle, you will be able to see part of driver between the array of vertical carpet-covered perf boards.
The intent is to attenuate the backwave 6~12 dB over a broad band while minimizing reflections - which is why the perf boards are at right angles to the backwave, unlike the Hartley "Boffle" of the early Fifties. The operating principle is similar to a low-restriction car muffler or a silencer on a gun - a lossy mesh is used to attenuate a shockwave without adding restriction.
Now, I could be all wet, so I've asked Gary to measure something like this and see what it actually does in the time domain. It's not a closed box, it's not a GP box, and it's not really an open-baffle, either. It's a transitional structure, and the goal isn't a certain polar-pattern, but minimization of stored energy in the time domain.
The only other enclosure I can think that puts first priority on minimization of stored energy is the B&W Nautilus tapered-pipe design. I'm not looking for 100% absorption of the backwave - instead, I'd be happy with a modest 6~12 dB attenuation and diffusion in the time domain (instead of the backwave coming around the front all at once, it takes multiple paths that are attenuated and spread out in time).
If the attenuation really is broadband, the polar pattern will move away from a dipole and towards a cardioid pattern - again, transitional between a dipole and a closed-box monopole. The phase and magnitude of the backwave will control the exact pattern. Do I expect it to be stable with frequency? Only measurements will tell, although I expect plenty of variation.
Sorry about the vague description - I only mentioned this to Gary a couple of weeks ago, since he has the combination of instrumentation and intellectual curiosity to see if there is any merit to this idea.
Definative time correction with DEQX
Hi Josh K ,
Below I have copied a straight quote from Kim Ryrie the DEQX MD. I asked him to please explain once and for all exactly what are the true time / phase correction abilities the DEQX. Also how can the DEQX be used to help reduce the unwnted stored energy in the driver / cabinet system. He very promptly (over the weekend ) answered my email with this reply:
" Yes the DEQX is often used only as a traditional active crossover, but it also has Group Delay correction features that are unique to DEQX. DEQX Group Delay correction goes well beyond the driver to driver time correction implied by “time alignment” requirements. The DEQX can, if set-up to do so, correct the timing errors WITHIN individual drivers, i.e. group delay errors. So in addition to allowing timing adjustment between drivers (time alignment), it can also compensate for the timing/phase of hundreds of frequency groups independently and differently depending on frequency! So that correction will surely impact the effects of stored energy timing, although as you say, even DEQX is not a magic bullet to stopping stored energy .
Provided you are using the DEQX’s own anechoic measurement software and using DEQX’s correction and linear-phase crossovers, the measurement technique determines how low in frequency the actual time domain correction will go. You can set the Group-delay correction ‘window’ during which ALL timing errors will be corrected in response to the actual measurement – so getting the measurement right is critical because this cannot be done manually due to the massive amount of data involved per driver or speaker. According to their issued and pending patents, the DEQX essentially pulls apart the original audio stereo source into its thousands of component sine waves, so that individual frequencies (not just a driver) can be independently compensated in gain AND phase (time) before the audio signal is re-assembled. So this means that the DEQX goes beyond minimum phase EQ such as multi-band parametric EQ, although all DEQX processors also provide that for final real-time EQ tweaking to taste.
The default Group-delay correction window in DEQX’s software is 3ms, which allows all frequency groups travelling for up to a metre to be made coherent again (i.e. as per the original source signal). So in effect, the earliest ‘on-time’ frequencies (typically highs) are variously delayed by up to three milliseconds to allow the later frequency groups (mainly lower mids and bass) to catch up. This can occurs to thousands of frequency groups simultaneously. This is also why the DEQX uses massive amounts of floating point DSP power that needs external SDRAM memory to cope with the lengths of waveform being dealt with over the several milliseconds – much of their processing power is also required to allow the whole FIR processing to be completed within a few more milliseconds so that the audio is not delayed to an extent that it won’t synchronise with video – DEQX has a patent on this low-latency process (traditional useful lengths of FIR DSP processing take about 100ms to 400ms just to process!) Traditional DSP crossovers only use a fraction of this processing power and only need relatively tiny amounts of internal memory – apart from DEQX’s excellent ADC (x2) and DAC (x6) converters followed by tracking analogue volume controls, this is why the DEQX is more expensive than other active crossovers such as Behringer or DBX. "
I hope the above clarification is of interest to any and all speaker builders as I believe very few of us have fully understood the potential of the DEQX software. Even in standard form its an awesome piece of kit and with a battery power supply and external DAC'S is downright scary good!
Cheers
Derek.
Hi Josh K ,
Below I have copied a straight quote from Kim Ryrie the DEQX MD. I asked him to please explain once and for all exactly what are the true time / phase correction abilities the DEQX. Also how can the DEQX be used to help reduce the unwnted stored energy in the driver / cabinet system. He very promptly (over the weekend ) answered my email with this reply:
" Yes the DEQX is often used only as a traditional active crossover, but it also has Group Delay correction features that are unique to DEQX. DEQX Group Delay correction goes well beyond the driver to driver time correction implied by “time alignment” requirements. The DEQX can, if set-up to do so, correct the timing errors WITHIN individual drivers, i.e. group delay errors. So in addition to allowing timing adjustment between drivers (time alignment), it can also compensate for the timing/phase of hundreds of frequency groups independently and differently depending on frequency! So that correction will surely impact the effects of stored energy timing, although as you say, even DEQX is not a magic bullet to stopping stored energy .
Provided you are using the DEQX’s own anechoic measurement software and using DEQX’s correction and linear-phase crossovers, the measurement technique determines how low in frequency the actual time domain correction will go. You can set the Group-delay correction ‘window’ during which ALL timing errors will be corrected in response to the actual measurement – so getting the measurement right is critical because this cannot be done manually due to the massive amount of data involved per driver or speaker. According to their issued and pending patents, the DEQX essentially pulls apart the original audio stereo source into its thousands of component sine waves, so that individual frequencies (not just a driver) can be independently compensated in gain AND phase (time) before the audio signal is re-assembled. So this means that the DEQX goes beyond minimum phase EQ such as multi-band parametric EQ, although all DEQX processors also provide that for final real-time EQ tweaking to taste.
The default Group-delay correction window in DEQX’s software is 3ms, which allows all frequency groups travelling for up to a metre to be made coherent again (i.e. as per the original source signal). So in effect, the earliest ‘on-time’ frequencies (typically highs) are variously delayed by up to three milliseconds to allow the later frequency groups (mainly lower mids and bass) to catch up. This can occurs to thousands of frequency groups simultaneously. This is also why the DEQX uses massive amounts of floating point DSP power that needs external SDRAM memory to cope with the lengths of waveform being dealt with over the several milliseconds – much of their processing power is also required to allow the whole FIR processing to be completed within a few more milliseconds so that the audio is not delayed to an extent that it won’t synchronise with video – DEQX has a patent on this low-latency process (traditional useful lengths of FIR DSP processing take about 100ms to 400ms just to process!) Traditional DSP crossovers only use a fraction of this processing power and only need relatively tiny amounts of internal memory – apart from DEQX’s excellent ADC (x2) and DAC (x6) converters followed by tracking analogue volume controls, this is why the DEQX is more expensive than other active crossovers such as Behringer or DBX. "
I hope the above clarification is of interest to any and all speaker builders as I believe very few of us have fully understood the potential of the DEQX software. Even in standard form its an awesome piece of kit and with a battery power supply and external DAC'S is downright scary good!
Cheers
Derek.
As Earl knows we have different positions on drivers and MP response. First let's distinguish between a minimum phase response and a minimum phase system. A minimum phase response, to me, simply means that the amplitude and the phase associated with that amplitude can be related by the Hilbert-Bode transformation. A minimum phase system would then be a system with output characterized by the same H-B relationship between amplitude and phase. In a simple electrical or mechanical system where there is only a single output this is straight forward. However, for a loudspeaker driver the output can be a function of the observation point. That the response varies with position doesn't preclude that the response at any observation point is MP. The problem with a driver is complicated because, even in the case of a perfectly rigid piston, we can think of the radiated sound as being the sum of the radiated energy form an infinite number of sources distributed over the surface of the piston. Each of these sources moves with the mechanical velocity of the piston and in so-far as the mechanical system is linear and MP, then so is the motion of each of these sources.
When we are far from the piston and on axis, all the sources sum in phase and thus the on axis response is a direct indication of the mechanical motion of the piston. However, when we move off axis the distance from each source (on the piston surface) to the observation point varies. As the wave length become shorter we start to see the development of a nonuniform polar response due to these path length differences. The system remains linear and the response is typically still MP relative to any observation point even though it may be different from the on axis response.
Since the on axis response is indicative of the piston motion, any anomalies in the on axis response are indicative stored electromechanical energy in the driver. Thus is we eq the on axis response to correct these anomalies we correct for the stored energy. We will still suffer the change in the off axis response due to the distributed nature of the sources over the piston surface. But the applied EQ will correct for the stored energy in the electromechanical system.
The off axis anomalies generally have little to do with stored electromechanical energy in the driver. They are just the consequences of the finite size of the driver radiation at frequencies with wave length on the order of, or smaller than the driver circumference. The only way to deal with this is to limit the use of the driver to a frequency range where is has constant directivity.
I presume that some of you will say, "what about breakup?" That is another issue, but not so relevant as good design would limit the use of a driver to well below any region of breakup.
When we are far from the piston and on axis, all the sources sum in phase and thus the on axis response is a direct indication of the mechanical motion of the piston. However, when we move off axis the distance from each source (on the piston surface) to the observation point varies. As the wave length become shorter we start to see the development of a nonuniform polar response due to these path length differences. The system remains linear and the response is typically still MP relative to any observation point even though it may be different from the on axis response.
Since the on axis response is indicative of the piston motion, any anomalies in the on axis response are indicative stored electromechanical energy in the driver. Thus is we eq the on axis response to correct these anomalies we correct for the stored energy. We will still suffer the change in the off axis response due to the distributed nature of the sources over the piston surface. But the applied EQ will correct for the stored energy in the electromechanical system.
The off axis anomalies generally have little to do with stored electromechanical energy in the driver. They are just the consequences of the finite size of the driver radiation at frequencies with wave length on the order of, or smaller than the driver circumference. The only way to deal with this is to limit the use of the driver to a frequency range where is has constant directivity.
I presume that some of you will say, "what about breakup?" That is another issue, but not so relevant as good design would limit the use of a driver to well below any region of breakup.
Re: Definative time correction with DEQX
Derek, can you either post or PM me with a source for a battery supply for the DEQX? (I've already heavily modded mine.)
Apologies for getting off topic vis-a-vis the Ariel, which has been very enlightening for me (thanks Lynn for the persistence)
My question might be better stated: is there any reason to expect that a DEQX would handle a 425 horn plus a U box woofer or a sealed box woofer any better/worse than it would a more "traditional" three way cone drivers in sealed boxes arrangement? Or, any reason to expect the passive crossover in those instances to outperform the DEQX in some material aspect?
Overkill Audio said:
Derek, can you either post or PM me with a source for a battery supply for the DEQX? (I've already heavily modded mine.)
Apologies for getting off topic vis-a-vis the Ariel, which has been very enlightening for me (thanks Lynn for the persistence)
My question might be better stated: is there any reason to expect that a DEQX would handle a 425 horn plus a U box woofer or a sealed box woofer any better/worse than it would a more "traditional" three way cone drivers in sealed boxes arrangement? Or, any reason to expect the passive crossover in those instances to outperform the DEQX in some material aspect?
Lynn Olson said:
Lynn,
It amazes me how you continue to misrepresent the dipole back-wave problem and the open backed, damped woofer. I posted some simulation results previously showing exactly that the back wave isn't an issue at all when the driver is correctly eq'ed and filtered. Perhaps you didn't get the idea or don't like sims. So here are some measurement and additional sims.
First, here is a simulation based on a measured response.
An externally hosted image should be here but it was not working when we last tested it.
The green trace is the measured response on axis of a small 5" driver mounted in a 8" circular baffle. The read trace is the simulated bandpass response which required the equalization curve shown in blue. It equalizes the response to a 150 Hz -2.5K Hz LR4 HP/LP bandpass.
Next, here is the simulated impulse response:
An externally hosted image should be here but it was not working when we last tested it.
Again, green is the raw, unequalized dipole response. The negative swing is clearly evident as are other factors resulting from the drivers response irregularity. It is not an inverse of the front impulse because the lack of high frequency content from the read side wraping around the baffle. The red trace is the simulated impulse of the equalized dipole. It is the impulse of the bandpass response. The same impulse would be obtained if the system were an infinite baffle bandpass with the same amplitude.
Finally, here is another simulated impulse. This one is identical to the previous one but the blue trace is the overlay of a 150 Hz HP, 2.5k Hz LP LR4 electrical filter.
An externally hosted image should be here but it was not working when we last tested it.
It overlays the eq'ed dipole response exactly showing that the only artifacts (or "ringing") in the response is that of the bandpass filter.
Well, I'm sure you will say there are only sims so here are some measured results of the same thing starting with the raw dipole frequency response and then a measurement of the equalized dipole response (performed using SoundEasy's digital filtering capability).
An externally hosted image should be here but it was not working when we last tested it.
Lastly, here is the measured impulse of the raw dipole (green) and eq'ed system (red).
An externally hosted image should be here but it was not working when we last tested it.
As you can see the measurement concur with the simulated results.
The point remains that when designed correctly the dipole impulse looks no different that an infinite baffle impulse.
[edit] And for what it's worth, here is a comparison of the impulse on axis (red) and at 30 degrees off axis (green):
An externally hosted image should be here but it was not working when we last tested it.
Also with regard to G.Primms opend backed, heavily damped woofer it would be nice if you gave credit where is is deserved. Nothing against G. P. here, just an observation that it isn't his invention. What the heck, I've been using it for almost 10 years. This type "acoustic resistance box" has been around for a long time, since the 50's and was analyized by Thomas Holmes in 1986 AES paper, and modeled by Juha Backman in a 1999 AES presentation.
Link for battery supply
http://www.armedforces-int.com is a link to a good supplier of rechargable Lithium Ion battery supplies.
What you need is a minimum of 8 Ah capacity with a steady 15volt
(plus or minus 0.5 volts) and the matching battery charger.
My supply was made for me by a guy who works for me so I dont know the exact details.
If you get stuck I can try and get some more details.
I have been trying to post some photos of my VPL Uframe bass / low mid speakers but the files are all too big. Can someone help by sending me a private email address I can send the files to so they can be reduced to fit this site?
Thanks
Derek.
http://www.armedforces-int.com is a link to a good supplier of rechargable Lithium Ion battery supplies.
What you need is a minimum of 8 Ah capacity with a steady 15volt
(plus or minus 0.5 volts) and the matching battery charger.
My supply was made for me by a guy who works for me so I dont know the exact details.
If you get stuck I can try and get some more details.
I have been trying to post some photos of my VPL Uframe bass / low mid speakers but the files are all too big. Can someone help by sending me a private email address I can send the files to so they can be reduced to fit this site?
Thanks
Derek.