Jack Hidley said:
Hi Jack.
Thanks for the explanation.
I was a bit surprised by the 899 litre calculated box capacity for a Qtc of 0.707 though.
How much would all these factors increase the effective volume of the NHT-3.3's 80L subwoofer enclosure?
Unless I am missing something major, considering the huge contrast between 899L and 80L, I still can't see how the NHT-3.3's 1259 subwoofer could be particularly highly damped.
Cheers,
Glen
G.Kleinschmidt said:
It also begs the question "how highly damped must it be to better a ported enclosure?" Probably not very.
Simon
I'll try to post some comparitive numbers tomorrow if I make it in to work.
I think you are being a little misled by comparing 900l to 80l. I think the comparison that makes sense is the Qtc of the driver in the box as you've modeled it to the driver in the box with a more accurate model.
Remember as the Qts of the driver approaches 0.707 the required enclosure volume goes to infinity to achieve a Qcb of 0.707.
I think you are being a little misled by comparing 900l to 80l. I think the comparison that makes sense is the Qtc of the driver in the box as you've modeled it to the driver in the box with a more accurate model.
Remember as the Qts of the driver approaches 0.707 the required enclosure volume goes to infinity to achieve a Qcb of 0.707.
Hi Glen,
These simulation tools are taking constructors' eyes off one of the balls of technical information which must be juggled as part of LS design.
Design is the *art* of juggling these ball with correct weighting in order to achieve a specific reproduction performance.
!!!!! However; software packages tend to based upon amplitude measurements where steady sines are already established at t=0 !!!!!
!!!!! None give an indication from t=0 of fundamental driver response during a first cycle of waveform reproduction !!!!!
*An initial wavefront transduced by a LF LS driver is naturally limited by its own mass/field etc, any open air path dimension, the mass of air it moves in any enclosure or vent for energy storage during a first half cycle, and any wave generated pressure differentials developed due to any of these or cabinet/room reflections.*
The current drawn by a dynamic loudspeaker during a first cycle of voltage drive is modified by driver motion generated back-EMF, which becomes modified by all of *-* above.
As soon as a choke/resistor is inserted in series between a voltage output amplifier and a driver in a cabinet (especially one which stores energy during a first half cycle in order that it might subsequently be released to increase SPL for all following half cycles) then both the amplitude and phase of back-EMF become deleteriously modified in time with respect to amplifier output, and driver reproduction dynamics become unnaturally softened !
Computer 'tools' are leading many users down a 'scenic' route and far away from hearing realistic dynamic 'music' reproduction back on the main road which constructors could be on.
Of course folks are free to enjoy the delayed energy-storage methods for inducing a flat SPL, but they are not being warned about constructions which might lead to inadequate satisfaction of recorded music reproduction and/or a committment of funds which could lead to driver/component redundancy.
The more driver energy is stored and re-released in order to increase SPL - the less the LF dynamic amplitudes are able to create a coherent illusion of reality.
Cheers ........ Graham.
These simulation tools are taking constructors' eyes off one of the balls of technical information which must be juggled as part of LS design.
Design is the *art* of juggling these ball with correct weighting in order to achieve a specific reproduction performance.
!!!!! However; software packages tend to based upon amplitude measurements where steady sines are already established at t=0 !!!!!
!!!!! None give an indication from t=0 of fundamental driver response during a first cycle of waveform reproduction !!!!!
*An initial wavefront transduced by a LF LS driver is naturally limited by its own mass/field etc, any open air path dimension, the mass of air it moves in any enclosure or vent for energy storage during a first half cycle, and any wave generated pressure differentials developed due to any of these or cabinet/room reflections.*
The current drawn by a dynamic loudspeaker during a first cycle of voltage drive is modified by driver motion generated back-EMF, which becomes modified by all of *-* above.
As soon as a choke/resistor is inserted in series between a voltage output amplifier and a driver in a cabinet (especially one which stores energy during a first half cycle in order that it might subsequently be released to increase SPL for all following half cycles) then both the amplitude and phase of back-EMF become deleteriously modified in time with respect to amplifier output, and driver reproduction dynamics become unnaturally softened !
Computer 'tools' are leading many users down a 'scenic' route and far away from hearing realistic dynamic 'music' reproduction back on the main road which constructors could be on.
Of course folks are free to enjoy the delayed energy-storage methods for inducing a flat SPL, but they are not being warned about constructions which might lead to inadequate satisfaction of recorded music reproduction and/or a committment of funds which could lead to driver/component redundancy.
The more driver energy is stored and re-released in order to increase SPL - the less the LF dynamic amplitudes are able to create a coherent illusion of reality.
Cheers ........ Graham.
Graham,
is that your way of saying : "don't aim for a theoretical flat Butterworth response" and possibly "never use High Q to boost the bass response"?
If my interpretation of your post is close to correct, then what speaker Q would you aim for to give a reasonable chance of reproducing realistic transients?
is that your way of saying : "don't aim for a theoretical flat Butterworth response" and possibly "never use High Q to boost the bass response"?
If my interpretation of your post is close to correct, then what speaker Q would you aim for to give a reasonable chance of reproducing realistic transients?
Jack Hidley said:
Ok, Thanks!
I’m pretty sure I understand what you are saying here, but I don’t see how it is out of line with my observations – namely that the design compromises within the Foster 1259 driver simply prohibit it from producing (in any practical sealed enclosure - not just that of the NHT-3.3) both a low F3 and a highly damped Qtc.
For example, with 0.9 ohms series resistance and in an 80L sealed enclosure, an F3 of 28Hz is produced, with a Qtc of 1.13.
The sealed box volume required to get the Qtc down to even 0.8 simply isn’t practical.
The reason I am studying the NHT-3.3 and the performance of various subwoofer drivers is because I’ve been repeatedly told throughout this thread that a Qtc even as low as 0.8 will not be sufficient to avoid “boomy” bass – some have even suggested aiming close to 0.5!
But here were have the NHT-3.3, a speaker with a reputation for very solid low bass, which has a high Qts subwoofer driver that apparently makes a Qtc of even 0.8 an impossibility, let alone 0.707 or lower.
I would dearly like to know the actual subwoofer driver Qtc value for the NHT-3.3 if anyone out there has actually measured it!
Cheers,
Glen
SimontY said:
No, for me it begs this question - "how much better would it sound with a Qtc of 0.707 as well".
Why? Because with only a slightly higher F3 (by 5Hz) that can be done with the Dayon RSS315HF-4.
This is the driver I currently intend to use 4 of for my (now 180L total volume each) speaker pairs.
Cheers,
Glen
G.Kleinschmidt said:
Glen, I did not even read your entire post here, but I know this driver well, and something does not seem right in your sims. Did you use walls lined, or heavy fill in your sims? The 3.3 used a steel laminate inductor, the .43 resistor was probably the DCR of the inductor. The driver Qms and thus Qts is reduced when there is stuffing right up against the back of the driver.
You've pointed out other BS in this thread, the suggestion of a low .5 Qtc is just nonsense. Room mode resonances are much higher Q than the low end HP, the issue is probably more of exciting them when there is strong LF extension and people claim that they hear poor bass. Also, an interesting characteristic of 2nd order HP functions is that the output at Fc is 20 log of Qtc. For a Qtc of 1 the output is flat at Fc, of course with peaking above, .707 is - 3dB, .5 is - 6 dB.
The same driver with a different box size and stuffing will require 4 times the power obviously at Fc with a Qtc of .5 as compared to a Qtc of 1. Most will run into serious thermal issues (thermal compression or burnout) when 4 times the power and high output is required. Also, Small's parameter K is optimal with a Qtc slightly above 1. A Qtc between .7 and 1 is usually optimal in my opinion.
Note also that the peaking above Fc with higher Qtc helps with the baffle step issue, especially for smaller bookshelf systems with higher Fc values.
I run my 1259s vented, yeah I know everyone says this is a sealed box driver only. But if you know what you're doing, it's not a problem. I favor vented systems for the excursion, and distortion reduction. I generally tune Fb low for a more gradual rolloff.
You're post concerning Qtc with the quote from the LDC is correct.
I would have prefered a bit more magnet strength in the 1259, it is a very inefficient driver, and that would have lowered Qts, but at the time I purchased them they were about the only 12" driver with a 12-13 mm Xmax.
Pete B.
G.Kleinschmidt said:
Glen, the way to determine the NHT3.3's Qtc would be to include the crossover inductor value and RDC in the Unibox sim, then look at the input imedance peak at Fc simulated. Choose heavy fill in Unibox, if the impedance peak is still higher than actual, lower the driver Qms until it matches to account for fill up against the back of the driver. See what you come up with.
You can find a plot of the input impedance here:
http://www.stereophile.com/floorloudspeakers/1293nht/index3.html
You'd be surprised at what you find in top performing commercial designs that do not follow many of the DIY "expert" opinions. A higher Qtc system provides more ouput at Fc and in the passband than with very low Qtc, this helps when using a low efficiency driver.
Pete B.
If one single 8" woofer were used, I might use Qtc of 0.7, but with four 8" I wouldnt dare go above 0.6....I would say the same if big 15-18" woofers were used...too big risk of softy bass with too much "volume", and nothing to do about it...apart from RCL to remove system ressonance, but hard to get right without measurements
Glen, the simulators are based on lumped models obviously and they work very well for small enclosures or more cubic enclosures that are a good fit for the lumped model. However, even a 3 or 4' tower with the woofers at one end will include some transmisison line effects. This was covered as far back as Beranek, see Ch 8 Figure 8.9.
He shows that while we normally think of a lumped box as an air spring, the input impedance can become a mass reactance, or at least show a higher compliance that the lumped model would predict when transmision line effects are taken into account.
A good real world example of this is the AR-9 tower. I believe that the Fc is lower than would be predicted by the lumped model.
Comparing the Unibox results with the measured impedance should give you an idea of the magnitude of these effects and how simple adjustments to your sims can be used to account for the differences. Sims still work just fine.
Pete B.
He shows that while we normally think of a lumped box as an air spring, the input impedance can become a mass reactance, or at least show a higher compliance that the lumped model would predict when transmision line effects are taken into account.
A good real world example of this is the AR-9 tower. I believe that the Fc is lower than would be predicted by the lumped model.
Comparing the Unibox results with the measured impedance should give you an idea of the magnitude of these effects and how simple adjustments to your sims can be used to account for the differences. Sims still work just fine.
Pete B.
Rooms Are Cavity Resonators
Art Ludwig does a nice job with this:
http://www.silcom.com/~aludwig/images/hbl_opt.gif
http://www.silcom.com/~aludwig/Room_acoustics.html
Pete B.
Art Ludwig does a nice job with this:
http://www.silcom.com/~aludwig/images/hbl_opt.gif
http://www.silcom.com/~aludwig/Room_acoustics.html
Pete B.
In my experience, exactly!
I did once build a 3 way (actually, the first speaker I built) using a 12" Blaupunkt Overdrive subwoofer, a Seas MP14rcy/p, and a Vifa Vifa D25AG-35. The bass Qtc ended up around .85. (I based on box size on the published paramaters, which were a bit off). But--the bass from that speaker is really pleasant sounding. It perhaps has the edges rounded off a bit, so a wallop from a kick drum doesn't ring your ears quite as much as with other speakers I've built with lower bass Qtc, but I do like those speakers a lot. I didn't find it particularly difficult to blend the bass into the midrange. They are sealed box 3 ways using the BSC method we have been discussing here.
JJ
Jack--very interesting info. This sounds sort of similar to the principal used in the Hegemen subwoofer. (AX, Dec '03). I would like to hear more about that technique for woofer loading.
JJ
PB2 said:
PB2 said:
PB2 said:
Hi Pete.
I have not read up on how to calculate Qtc from the speaker impedance yet and I have not run any sims as such - just basic box size sims (no fill) for comparisons between different drivers.
I don't think there is anything wrong with the Unibox sim that I've run.
The schematic diagram I have of the NHT-3.3 crossover shows a 0.43 ohm resistor in series with the 12mH inductor.
Supposing that the 0.43R is there to represent the inductor DCR, I have re-run the sim with just this (optomistic) value and attached a screen shot below.
For a Qtc of 0.707, the sealed box volume calculates to a whopping 413L.
Now I can appreciate that fibre fill and other design tricks can increase the compliance of a sealed enclosure, but can they really make the 80L's of the NHT's subwoofer enclosure equivalent to +400L ?
It seems to me that solid bass performance can be had without abnormally low Qtc.
I totally agree about the claims made about the need for an ultra low Qtc. I do not find any support for it in the published texts that i have read - in fact the opposite. It is also a myth that continuing to increase the box volume will continue to lower F3. Beyond a certain point (depending entirely on the driver) the opposite is true.
For my current plan for a
W
MT
W
with the two Daytons, I have decided to design for an unfilled closed box Qtc of 0.707. This works out to 74L for each woofer with 0.5 ohms in series with each (for an F3 of 35Hz).
I can then increase the compliance further if desired with an amount of fibre fill. This will factor in the effects of voice coil heating. The LDC gives a chart showing a rather hefty rise in Qtc with voice coil heating, so it is good to err on the safe side WRT to box volume.
I'm currently planning a totally seperate sealed compartment for each woofer (this allows a much more rigid cabinet construction) and they are pretty much a cube in shape.
From my current level of understanding this should mitigate transmission line effects and produce real world results that pretty much match the simulations. ??
Cheers,
Glen
Attachments
Hi Glen,
You have to check "heavy fill" for the 1259 because the Qts is slightly high. Left side of the main spread sheet in Unibox.
You don't really have to understand how to get Qtc from the input impedance, just make the sim match the real world, then you're seeing the correct reduction in Qms due to box resistive losses and stuffing up against the back of the driver. It matters most at system resonance where the impedance peaks. One of the side tabs (bottom right) in Unibox has the impedance plot.
Unibox should also take into account the slight increase in effective volume due to the filling, however you can also adjust the box volume slightly to match the measured impedance for the NHT3.3. You might be compensating for the filling, or TL effects in the box, whatever it is they should match. I've found very good agreement without making any large adjustments.
About filling, not that you have to know the details:
http://users.ece.gatech.edu/~mleach/papers/Filling.pdf
Pete B.
You have to check "heavy fill" for the 1259 because the Qts is slightly high. Left side of the main spread sheet in Unibox.
You don't really have to understand how to get Qtc from the input impedance, just make the sim match the real world, then you're seeing the correct reduction in Qms due to box resistive losses and stuffing up against the back of the driver. It matters most at system resonance where the impedance peaks. One of the side tabs (bottom right) in Unibox has the impedance plot.
Unibox should also take into account the slight increase in effective volume due to the filling, however you can also adjust the box volume slightly to match the measured impedance for the NHT3.3. You might be compensating for the filling, or TL effects in the box, whatever it is they should match. I've found very good agreement without making any large adjustments.
About filling, not that you have to know the details:
http://users.ece.gatech.edu/~mleach/papers/Filling.pdf
Pete B.
Its my experience that polyfill is not very good, the result is blurred upper bass/midrange with loss of clarity and detail
the drivers needs free space/air to breathe freely
A completely stuffed box is something I would never do
I certainly wouldnt count on it to save a "flawed" design
the drivers needs free space/air to breathe freely
A completely stuffed box is something I would never do
I certainly wouldnt count on it to save a "flawed" design
tinitus said:
LOL!
Well since the polyfill didn't work for you, I guess that you have to look for something else.
I guess we can forget all about things such as internal reflections also.
Hi Glen,
IMHO you have just a incorrect QTs of the 1259
Qts=0.44 Vas=238 liter Fs= 16.5 Hz
It seems to me reasonable for the NHT 3.3 implementation
cheers,
IMHO you have just a incorrect QTs of the 1259
Qts=0.44 Vas=238 liter Fs= 16.5 Hz
It seems to me reasonable for the NHT 3.3 implementation
cheers,
inertial said:
Hi
I got all the parameters from here:
http://www.madisound.com/catalog/product_info.php?products_id=1402
Cheers,
Glen
- Status
- Not open for further replies.