|4th January 2002, 04:48 PM||#11|
Wizard of Kelts
diyAudio Moderator Emeritus
I think you have a good idea in building the thing first and fine-tuning it later.
The only reason people say not to cross a subwoofer over above 150 Hz or 100 Hz is that they figure it will be in a separate unit several feet from the main speakers. If the main box contains the subwoofer, that is obviously not the case here.
There seems to be only octave in question-100 Hz to 200 Hz-and surely you can hit on the right combination of crossover point and slope to fill it in correctly.
|6th January 2002, 05:47 PM||#12|
Join Date: Dec 2001
Location: Milano, Italy
Pre-scriptum (!) I've realized that the attachment is too big for this list, so I've sent it to the e-mail address you have provided.
as promised (..or threatened..) I did some calculations on your project. I guess the outcome
can help you in the setup of your system. All graphs mentioned are in the attached 'zipped' word file.
1 - Panels and box volumes are simulated according to the dimensions you have supplied (8x44x30).
2 - Woofer: no parameters available, I've estimated a 'generically good 4-Ohm woofer'. If you want to send me the data of your unit we'll complete the simulation.
Anyway, since it is used only as a sub, we should be reasonably accurate.
3 - Drivers are placed on panel with tweeter on the top edge and mid as close as possible.
This assumption is important for all response simulations.
4 - As general rule, the purpose of any Simulation is just to put the project on the right path. All fine tuning will always be up to your ear!
Now, in Pic 1 and 2 we have the simulation of the Scan 8546 in free air and in 10 lt sealed box.
First note: having a lin X-max of 6.5 and a X-max of 10, with 200W you should not fear any damage.
Pic 3 is the free air/infinite baffle simulation of the tweeter
In pic 4 we see the response of the drivers on infinite baffle and on the actual panel that you use.
I think this graph answers your first question about panel diffraction, this being the difference between continuous and dotted lines.
Note that the woofer is simulated as mounted in the lateral panel, that is much bigger than the front, hence the smaller effect on his curve.
Note also that at certain freq. we have the effect of the reflection of the emission on front panel.
The bigger the panel, the more appreciable are the reflections, that's why many hi-end designs tends to have a small front
panel or put a lot of attention on dumping the reflections
With pic 5 and 6 we have the system behaviour with the 'best combination' of the filters you
want to test.
This is : 7mh coil on woofer, 500microF + 1,4 mH/6.8micro on mid and 2Ohm/6,8micro/1 mH on tweeter.
The output really looks not bad, but I think there's a couple of things that is possible to improve.
A - the big 500 cap on the mid. For sure a big electrolitic before the mid is not a reccommended practice for
a good sound. And anyway this big cap is doing very little to protect your speaker, see Pic 9. It reduces the output of
a couple of db in the 60-250 Hz range, but brings 2 resonances of 3 db at 40Hz and of 0.5 db in the 400-800 range.
I would simply get rid of it, it's giving more troubles than it's worth. The Scan in 10 lt has a really smooth roll-off, why
change something that is already very good?
Maybe you could put a smaller cap with a switch and activate it when in HT mode, for normal music you don't need any
protection on the Scan.
B - Look at the response in pic 6 : at 0 degrees is good, at -5 is almost perfect but at +5 there is a dip of 6-7 db, most
important the depression is about 1-1/2 octaves wide (see the 85 db spl line as ref.)
At 10 degrees the behaviour is nearly horrible.
This means that the energy (emission in a solid angle) delivered by the system is not optimized.
(or, in other words, there is a destructive interaction between the 2 Scan drivers)
Normally the effect on the musical reproduction is a non good 'reconstruction' of the ambience.
Pic 7 and 8 shows the system with the x-over changes I would suggest.
The woofer is cut with a bigger coil, the mid is free on the high pass side, while his low-pass and the tw high-pass are
modified (see draft in the attached, note the position of the 1.8 resistor that must be between cap and coil to properly equalize
the tweeter response)
In pic 8 you see that the emission at +/- 5 is within +/- 2 db, and the response @ 10 degr. is well acceptable,
showing only a narrow dip in the 3.5-4.5 Khz range (less than half octave wide, and so not critical)
One last tip: in both systems the woofer must be connected out of phase, see pic 10 if you don't believe me!
|7th January 2002, 10:28 AM||#13|
How can I ever thank you? You have spent so much time energy and intellect over this problem.
I know the 8546 has a Xmax of 6.5mm but the 12" I am using (Audio Concepts dual voice coil version of the SV12 (www.audioc.com) has a linear one way Xmax of about 11mm.
Instinctively I guess quantity of bass is a function of volume of air moved. This means that Volume displace of te 6846 is 150cm2 x 0.65cm = 97.5cm3. The volume displacemetn of the DV12 is 420cm2 x 1.1cm = 462cm3. That means that the DV12 can move about 5 times the air of a 8546.
Hence I was worried that when the 12" is really moving the 8546 will have reach it's bass limits. That is what I am worried about.
Also the reason I was loooking to cross the woofers a little higher (9mh is about 80Hz I think) is that it would have afforded me a bit more senitivity. WIth a 9mh Inductor the system sensitivity will be about 85db which awful low. even 2db more would be much more reasonable.
Do the simulations you have done include difraction loss. From fig 4 it seems the 6" (solid black line) has a 10db loss from about 400Hz to 40Hz.
Lastly the 12" is about 90db at 4 ohms (3.1 ohms DC with both coils in parallel), the 6" is 88db at 8ohms (5.5 ohms DC). Given that we are compensating for difraction loss and further reducing the level of the 6" (and also padding down the tweeter to match) how does one match the 12" which is more sensitive to he 6" or (padded down) 1".
I'll see if I can find specs of the 12" and send them to you.
|7th January 2002, 11:49 AM||#14|
|7th January 2002, 02:14 PM||#15|
Join Date: Dec 2001
Location: Milano, Italy
first you don't need to thank me at all: I'm playing with this hobby for fun. If this means that my brain is deranged, well, this is another problem!
Ok, let me try to answer your questions
1 & 2 - You're right about you calculations on qty of air moved, but again don't worry too much about it.
Excluding the effects of reflections in the room, the SPL that you listen is the emission of DV12 plus 8546 minus (destructive interaction between the two).
In simple words, it is exactly the yellow curve in the graph. This curve is simulated with the mic at 1mt from center of tweeter @ 2.83V (= 1W/8Ohm) input, and you see that is in the range of 87-88 db.
All theoretical 'point calculations' of the sensitivity are superseeded by this global SPL curve, that is a (still approximated but more accurate) picture of your system.
So, in reality, the limiting factor is the mechanical and termical handling of the 'weakest' unit. This is the Scan 8546 and, in my opinion, you will never reach this limits with a 200w amp and a musical program. Of course if you plan to play a 32 Hz tone at 0db for 5 minutes, there is more than a chance to kill the driver, but first you'll have some other interesting side effects, like cabinets opening, glasses broken and your wife/fiancée or neighbour that will try to set fire to your house or directly to your pants.
The point is that you can pull more SPL on the bass side by the DV12, but the linearity of the system will be affected (emphasized on the bass), unless you don't filter the 8546 in order to match the higher level delivered by the DV12.
9mH filters the woofer nominally at 55 Hz, that is it also equalizes his level. The idea is to use the DV12 to complete the 8546 emission at lowest frequencies, providing the 'punch' but leaving to the 8546 the responsability of the sound.
Again, this is a simulation: you may start this way and try with different coils until you reach the best sound balance.
3 - Yes, all global simulations are on actual panel. Normally these simulations are 'conservative' at low freq, that is you should expect a low freq level a bit higher in the real world.
4 - as said above, by equalizing the emission level of the units in order to equilibrate the global response of the system.
|8th January 2002, 06:52 AM||#16|
Thanks again Sandro,
one more question came tome.
Why are the DCR of the inductors so high.
I have a 1mh indusctor that is only 0.2 ohms and a 5 mh Indctor that is only 1 ohm. And these are both air core. Using a lathe I can easily build a 9mh air core inductoer that is less than 1 ohm. It will big for sure but is not air core better?
One reason I find to use 9mh is if I find too little resonse in the 60-100Hz range I can wind off a few outer turns of the 9mh inductor and get a smaller inductoer. It is more difficult to make a small inductor larger.
|8th January 2002, 11:24 AM||#17|
Join Date: Dec 2001
Location: Milano, Italy
DC Resistance: just because I assumed a 'conservative' value!
With low DC coils you'll see some benefit for those in series (the wf and Mid), almost nothing for the tweeter one.
The simulation for 9mH/0.5Ohm, 1mH/0.2 Ohm gives an increase of about 0.5 to 1 db in the range 40Hz-4000Hz
If you can build a low res air core inductor of 9mH it is better than a ferrite core, I just estimated a ferrite because normally the commercial coils of that value are solid core.
|8th January 2002, 11:28 AM||#18|
Sandro! You have spent so much time energy and intellect over this problem. You are blown!
I know the 8546 has a Xmax of 6.5mm but the 12" I am using (Audio Concepts dual voice coil version of the SV12 (www.audioc.com) has a linear one way Xmax of about 10.5mm. Instinctively I guess quantity of bass is a function of volume of air moved. This means that Volume displacement of the 8546 is 150cm2 x 0.65cm = 97.5cm3. The volume displacement of the DV12 is 420cm2 x 1.05cm = 442cm3. That means that the DV12 can move about 4.5 times the air of a 8546. Hence I was worried that when the 12" is really moving the 8546 will have reach it's bass limits. That is what I am worried about.
Also the reason I was loooking to cross the woofers a little higher (9mh is about 60Hz I think) is that it would have afforded me a bit more senitivity. WIth a 9mh inductor the the 12" would not be able to help the 6" in the 150-200Hz range and the 6" would require more difraction compensation hence the system sensitivity will be lower. Right?
Do the simulations you have done include difraction loss. From fig 4 it seems the 6" (solid black line) has a 10db loss from about 500Hz to 70Hz. The simple circuit that you have sent does not have any visible difraction copensation circuit. Are you assuming tha the 12" will be able to compensate for difraction loss even after having a 9mh inductor in series? Also from fig. 4 I notice that the loss between 200hz and 500Hz is 6db. This is the worrisome range. I read your graphs wrong initally as I assumed the marking on x axis wer 10, 100, 100, 10,000Hz not 20, 200, 2000, 20,000Hz. Hence I got the post on the site wrong too. Please forgive me.
I think the 1ohm resistance of the 9mh inductor will reduce the level of the 12" to the level of the 6" and 1" 2 way. Right? However wont this affect the damping of the 12"? I am told that any resistance between a driver and a amp reduces the damping drastically. This damping is not noticeable for tweeters as it only affects low frequency performance. Right?
Also I wonder why you put the 1.8ohm resistor between the cap and inductor in the tweeter circuit. Is there a reason for this? I have never seen this topology before. Also the Q of the tweeter crossover will be rather high with a 5uf cap and 0.25mH inductor. Given the 9900 has no ferro fluid damping and hence a high resonance peak at 500Hz would this not make the tweeter suseptable to damage. The 10ohm resistor in parallel with the tweeter was a simple way to damp this peak. As the tweeter impedance went up (with lowering frequency) more current and hence power would be absorbed by the 10 ohm resistor.
Lastly the 12" is about 89db at 4 ohms (3.1 ohms DC with both coils in parallel), the 6" is 88db at 8ohms (5.5 ohms DC). Given that we are compensating for difraction loss and further reducing the level of the 6" (and also padding down the tweeter to match) how does one match the 12" which is more sensitive to he 6" or (padded down) 1". What I am trying to say is if the 6" is 88db then we add 3db of difraction compensation and the sens. is 85db. 12" is 89db nand hence there is a 4db gap in sensitivity.
I went to the audio concepts site. It seems that the do noy sell drivers anymore. Hence I will give you all the specs I have. Hope this helps.
Imp: 4 ohms (2 coils in parallel)
Xmax (one way linear): 10.54mm
Voice coil dia: 2"
Voce coil Height: 30.61mm
Air gap Height: 9.525mm
Voicel coil inductance: 2mH
Voice coil resistance: 3.11ohms
ACI (www.audioc.com) recomended sealed boxes of 2-3 cu. ft.
The specs of the 8546 and 9900 can be got from http://www.scan-speak.dk/
With reference to my above posts and all your posts and help I would like to suggest the following:
L1 = 8 mh (1 ohms) air core (this is the inductor for the 12" woofer)
C1 = 500uf (2 x 1000uf caps in series, 1 10uf polyester cap in parallel and 1 2uf wonder cap in parallel) the parallel caps will act like bypass caps. Maybe we can remove this cap and see. I only worry about excursion.
L2 = 1 mh (0.2 ohms) air core
C2 = 8.2 uf
R1 = 2 ohms (v/s 1.8 omhhs in your circuit) although I feel I would need 4ohms instead
C3 = 4.7 uf
L3 = 0.27mh
R2 = 10ohms (only to protect the tweeter).
Invert polarity on the tweeter since it's effective slope is 2nd order.
Does this make any sense?
|8th January 2002, 02:58 PM||#19|
Join Date: Dec 2001
Location: Milano, Italy
just a quick reply, I have just little time today
C1 : if you want the cap, use a 5x100micro good non-polaryzed electrolitic (ROE, Monacor or Intertechnik the brands I know, maybe your 'pusher' has a better alternative) + the poly & wonder cap.
It sounds better than the series, and easier to modify later if needed
R1 : I put a 1.8 because it is a standard value, if you have a 2 it's ok . Just stay on 10W dissipation.
C3 & L3 : Ok, nothing changes
R2 : don't use it, it is not necessary and brings some troubles. If you want to linearize the imp of the tweeter use a RLC vs ground with L = 1.2 mH, C = 86micro, R = 6.2 Ohm instead of the simple R2
With the current filter, the TW is already attenuated by 12 db at 2kHz, it should not burn.
DO NOT invert the polarity of the tweeter, theory and real world are a bit different! Or, if you can measure the output, try both connections and see the plot!
Will be back for the 'theory' and the bass calculations when less stressed, I am in year closing period now!
|9th January 2002, 07:08 AM||#20|
I thougth the R2 (parallel cap with tweeter) would be a cheap way to avoid using 3 components. I know the notch filter is better. Just trying to save money!
Ok will build notch filter. Since you have helped so far the least I can do is try out what you recommend.
I blew out 2 Morel MDT 33 tweeters before. How I dont know. I had a MTM using 2 Morel MDT33 tweeters and blew out both tweeters. I guess the SPLs I use are a bit more than normal.
|Thread Tools||Search this Thread|
|Thread||Thread Starter||Forum||Replies||Last Post|
|Baffle step difraction||JC951t||Multi-Way||7||18th March 2009 01:09 PM|
|Gunderson compensation..||mikeks||Solid State||155||25th September 2006 03:14 AM|
|Baffle Difraction &||ralph-bway||Multi-Way||36||9th January 2005 02:26 PM|
|Reduce difraction and distortion from the cabinet on DIY designs.||Bedroc||Multi-Way||1||24th February 2002 04:09 AM|
|New To Site?||Need Help?|