Hi all
I am looking to create an MTM speaker using the following drivers:
http://www.falconacoustics.co.uk/downloads/Morel/Specs sheet CAT308.pdf
http://www.falconacoustics.co.uk/downloads/Vifa/P21WO-20-08.pdf
I have built 2 way speakers in the past designing the xo myself with good results but never an MTM. I have picked up a few things from various project websites so this will be an opportunity for me to confirm my knowledge and learn a few things on the way. I am no expert so please chip in.
The reason I've picked these drivers is their nice smooth response curves, would blend nicely at ~2khz and both fs are far from the xo freq.
They are both 8ohm, maybe would be better if the tweeter was 4ohm to match the woofers once paralleled. Anyhow let's consider the drivers a given and see if we can make them work.
Again I have chosen 2khz as the xo freq. I read somewhere that the drivers need to have a 90deg phase difference at the xo freq. so that it's truly d'appolito. This is achieved by deploying a 3rd order butterworth xo. We will confirm that in simulation later.
2 woofers in parallel will give a +6db gain, so since the woofer sensitivity is 91db and tweeter 90db, an -7db loss pad is needed on the woofers.
Finally we need to equalise the woofer impedance with a zobel network. I have calculated one for each woofer, maybe someone can help me consolidate them into a single CR leg.
Using the various calculators from lalena I have ended up with the attached initial design. I have also plotted the overall output from all drivers (equalised for sensitivity) and overall impedance.
Wavelength at 2khz xo freq is ~165mm so the driver centres need to be within that. That is certainly possible tweeter-to-woofer but not possible woofer-to-woofer. We'll place them as close as possible or even offset the tweeter to the inside of the axis between woofer to woofer centres if wood thickness allows - we'll come to this later.
Also the 165mm dimension or multiples of it needs to be avoided anywhere on the speaker (width/height/length/driver standoff distances etc.) to avoid any standing waves which will highlight any imperfections of the xo region.
Questions:
1. The db curve is flat enough and the overall impedance does not fall below ~4ohm. However the impedance seems to peak to ~10ohm at ~3khz then fall again and then creep up at >20khz. Does that impedance curve pose a problem for an amplifier capable of driving a 4ohm load? If so can the z curve be remedied / ironed out a bit? They look like a temperamental load to me with z going up and down but maybe is not as bad as i think...?
2. The xo at the moment is a little messy in terms of high part count. Is there any way of consolidating it by combining parts? e.g. 2 zobels into 1 etc.?
3. How do I go about calculating BSC? Using the formulas in the ESP website let us assume the baffle will have a 250mm width. Thus 115/0.250 => 460hz so we need say 3db bsc at that frequency. I have seen L||R circuits in front of the xo which act as bsc but not sure how to calculate the values. Googling "bsc calculator" seems to wield driver specific results; Does that imply the rest of the xo parts/interactions need to be taken into account?
Any other comments / general improvements to the xo please shout. Also feel free to correct if any of the above statements / logic are not right.
I am looking to create an MTM speaker using the following drivers:
http://www.falconacoustics.co.uk/downloads/Morel/Specs sheet CAT308.pdf
http://www.falconacoustics.co.uk/downloads/Vifa/P21WO-20-08.pdf
I have built 2 way speakers in the past designing the xo myself with good results but never an MTM. I have picked up a few things from various project websites so this will be an opportunity for me to confirm my knowledge and learn a few things on the way. I am no expert so please chip in.
The reason I've picked these drivers is their nice smooth response curves, would blend nicely at ~2khz and both fs are far from the xo freq.
They are both 8ohm, maybe would be better if the tweeter was 4ohm to match the woofers once paralleled. Anyhow let's consider the drivers a given and see if we can make them work.
Again I have chosen 2khz as the xo freq. I read somewhere that the drivers need to have a 90deg phase difference at the xo freq. so that it's truly d'appolito. This is achieved by deploying a 3rd order butterworth xo. We will confirm that in simulation later.
2 woofers in parallel will give a +6db gain, so since the woofer sensitivity is 91db and tweeter 90db, an -7db loss pad is needed on the woofers.
Finally we need to equalise the woofer impedance with a zobel network. I have calculated one for each woofer, maybe someone can help me consolidate them into a single CR leg.
Using the various calculators from lalena I have ended up with the attached initial design. I have also plotted the overall output from all drivers (equalised for sensitivity) and overall impedance.
Wavelength at 2khz xo freq is ~165mm so the driver centres need to be within that. That is certainly possible tweeter-to-woofer but not possible woofer-to-woofer. We'll place them as close as possible or even offset the tweeter to the inside of the axis between woofer to woofer centres if wood thickness allows - we'll come to this later.
Also the 165mm dimension or multiples of it needs to be avoided anywhere on the speaker (width/height/length/driver standoff distances etc.) to avoid any standing waves which will highlight any imperfections of the xo region.
Questions:
1. The db curve is flat enough and the overall impedance does not fall below ~4ohm. However the impedance seems to peak to ~10ohm at ~3khz then fall again and then creep up at >20khz. Does that impedance curve pose a problem for an amplifier capable of driving a 4ohm load? If so can the z curve be remedied / ironed out a bit? They look like a temperamental load to me with z going up and down but maybe is not as bad as i think...?
2. The xo at the moment is a little messy in terms of high part count. Is there any way of consolidating it by combining parts? e.g. 2 zobels into 1 etc.?
3. How do I go about calculating BSC? Using the formulas in the ESP website let us assume the baffle will have a 250mm width. Thus 115/0.250 => 460hz so we need say 3db bsc at that frequency. I have seen L||R circuits in front of the xo which act as bsc but not sure how to calculate the values. Googling "bsc calculator" seems to wield driver specific results; Does that imply the rest of the xo parts/interactions need to be taken into account?
Any other comments / general improvements to the xo please shout. Also feel free to correct if any of the above statements / logic are not right.
Attachments
Is better if you use some kind of simulation software and input frd/zma's of the drivers. Use SPL trace if you want to input them from specs.
First thing, you shouldn't need to pad down the woofers, That is where you get your BSC from! you may roll them off a bit earlier (bigger coil) to get the desired effect, as Inductor said simulation is your friend here.
Second, you should NOT be looking at 3rd order from an electrical sense, you should be aiming for 3rd order acoustic rolloff. This will not likely be obtained with third order electrical filters!
Do a bit of reading on accoustic rolloff vs electrical. The electrical order of the filter, combines with the drivers natural rolloff to give you the final acoustic roll off. It is the accoustic that you need to worry about, though you do need to take the phase into consideration as well (which is affected by the electrical). Again simulation will help here.
If you haven't already, have a read of AllenB's sticky thread here on designing loudspeakers without measurements for a good primer.
Tony.
Second, you should NOT be looking at 3rd order from an electrical sense, you should be aiming for 3rd order acoustic rolloff. This will not likely be obtained with third order electrical filters!
Do a bit of reading on accoustic rolloff vs electrical. The electrical order of the filter, combines with the drivers natural rolloff to give you the final acoustic roll off. It is the accoustic that you need to worry about, though you do need to take the phase into consideration as well (which is affected by the electrical). Again simulation will help here.
If you haven't already, have a read of AllenB's sticky thread here on designing loudspeakers without measurements for a good primer.
Tony.
Baffle step makes you loose 6 dB at low frequencies and yields a rising frequency response. Baffle step correction usually simply is done by increasing the coil of the low-pass filter until the frequency response gets flat. Thus, you don't need to pad down the woofers.
Attachments
I am not aware of any software that runs on mac and is free. Any suggestions?
I used to use unibox in the past which was good enough but now it doesn't seem to work properly in open office.
I lately use the online version of winisd for vb and port calcs, which is limited but it does give an idea of driver+box response. I use the lalena calculator to overcheck.
Thanks for the clarification of acoustical vs electrical roll off order!! That explains in my mind why some zaph designs use 2nd order electrical x/o...!! I will do some more reading and redesign a 2nd order xo.
Clever way of adding bsc, if I understood you correctly I just need to adjust L2 ??
Hi Chatziva, Inductor was talking about crossover simulation software. I'm not aware of any mac specific versions. Jeff bagbies PCD is excell based, and there is speaker workshop. There is also a member here with a standalone speaker design program that I can't remember off the top of my head, but I don't know if there is a mac version.
edit: and yes increasing L2 will roll off the woofer earlier compensating for the rising response of baffle step. A crossover simulator helps a lot with working out this aspect.
Tony.
edit: and yes increasing L2 will roll off the woofer earlier compensating for the rising response of baffle step. A crossover simulator helps a lot with working out this aspect.
Tony.
Last edited:
Excellent thanks for that!! I like this software (but 100 euro is steep for a hobbyist!)
I can do crossover electrical simulation in ltspice, the only issue is to know what to change! I can do that with 2nd order xo but it gets complicated with higher order ones...
After a bit of reading and a lot of tweaking of the part values in ltspice I've come up with the attached xo.
First things first I've changed it to 2nd order.
For the zobel network I calculated R and C for woofer and then halved R and doubled C which gave a close enough single zobel leg for 2 woofers. Then I did some further tweaking later to flatten overall z more; no hard rules there but I increased R a bit and reduced C (which also reduces cost for a small impact on performance - especially true for >1mH coils and >20uf caps)
Also I have added a little resistor 2Ω behind the tweeter to attenuate it slightly and manipulate the overall curve as I could do it by just playing with c1 & L1. Might try to further optimise this later by trying to remove Rz1 altogether.
From the 1st graph the woofer phase at xo freq 2KHz is -200 deg (or +160 deg right?) and tweeter +50 deg. So I think that I am correct in saying that (electrically at least) the tweeter lags behind the woofer by 160-50=110deg which is close enough to the D'Appolito target of 90deg. Is it close enough? 😕
The 2nd graph now shows a flat enough overall response with about +6db bsc at 1KHz and below. Good.
The impedance graph shows a minimum of 3.6Ω up to 200Hz and then it increases to a comfortable 6~10Ω for the amp. So the amp needs to be capable of 4Ω to get good bass.
I was thinking of increasing the xo freq slightly to allow for wider spacing between the drivers (but still within a wavelength) however the current xo is asymmetrical already so I am not quite sure what the xo freq is now...!!
Now I need to finalise enclosure dimensions and then somehow acoustically simulate the whole thing. 😕
Any other improvements at this stage ? 😎
First things first I've changed it to 2nd order.
For the zobel network I calculated R and C for woofer and then halved R and doubled C which gave a close enough single zobel leg for 2 woofers. Then I did some further tweaking later to flatten overall z more; no hard rules there but I increased R a bit and reduced C (which also reduces cost for a small impact on performance - especially true for >1mH coils and >20uf caps)
Also I have added a little resistor 2Ω behind the tweeter to attenuate it slightly and manipulate the overall curve as I could do it by just playing with c1 & L1. Might try to further optimise this later by trying to remove Rz1 altogether.
From the 1st graph the woofer phase at xo freq 2KHz is -200 deg (or +160 deg right?) and tweeter +50 deg. So I think that I am correct in saying that (electrically at least) the tweeter lags behind the woofer by 160-50=110deg which is close enough to the D'Appolito target of 90deg. Is it close enough? 😕
The 2nd graph now shows a flat enough overall response with about +6db bsc at 1KHz and below. Good.
The impedance graph shows a minimum of 3.6Ω up to 200Hz and then it increases to a comfortable 6~10Ω for the amp. So the amp needs to be capable of 4Ω to get good bass.
I was thinking of increasing the xo freq slightly to allow for wider spacing between the drivers (but still within a wavelength) however the current xo is asymmetrical already so I am not quite sure what the xo freq is now...!!
Now I need to finalise enclosure dimensions and then somehow acoustically simulate the whole thing. 😕
Any other improvements at this stage ? 😎
Attachments
Last edited:
chatziva, this isn't going to work. The Vifa P21W0-20-08 Woofer is going to be too loud in a Parallel D'appolito for the Morel MDT 30 tweeter.
I'd be looking at a regular two way using this circuit:
http://www.diyaudio.com/forums/multi-way/245917-diy-kef-celestion-5.html#post3714447
You possibly won't need ANY damping resistor in the bass shunt. You could, of course, do a series D'Appolito easily enough.
I'd be looking at a regular two way using this circuit:
http://www.diyaudio.com/forums/multi-way/245917-diy-kef-celestion-5.html#post3714447
You possibly won't need ANY damping resistor in the bass shunt. You could, of course, do a series D'Appolito easily enough.
Funny you should say this I was thinking the same. And it's going to be power hungry as well. If the tweeters could take it, I think it would realistically need a 400W amp.
That's fine I've got a hypex 400W amp but it's gonna demolish my house... or needs to be played at low volume settings. I wanted to make a high end MTM but perhaps that's way too high end...
The tweeters are actually CAT308 which is an MDT-30 replacement rated at 200Wrms so they might be able to take the beating in theory ??
Here's the overall output and power with 20Vac input.
Takes in about 100W and will be playing at about 120db (woofer sensitivity plus max db reading from my graph i.e. 30db if my calculations are right). That is quite loud...
Takes in about 100W and will be playing at about 120db (woofer sensitivity plus max db reading from my graph i.e. 30db if my calculations are right). That is quite loud...
Attachments
Last edited:
Anyway indeed it's not going to work... These drivers are a much better match as a 2 way... Back to the drawing board...!!!
I remember doing a MMT 3-Way project (or MTM type can't remember) for another member long time ago.
http://www.diyaudio.com/forums/multi-way/163796-some-design-help-please.html
http://www.diyaudio.com/forums/multi-way/163796-some-design-help-please.html
Yes looking again, I guess the bass will be hot by maybe 3 - 4 db in a typical room. I suspect you could get away with a morel CAT-378 Morel CAT 378 1-1/8" Soft Dome Horn Tweeter | 277-084 (newer version of my DMS37's) with it's 93db sensitivity, but see below.
The other thing I didn't look at is that the vifa's are 8" drivers. That is going to be a challenge if you are trying to do a true D'appolito MTM. The centre to centre distance is just going to be too large.
If you can run a pc simulation program on your mac, it may give you more options for crossover simulation.
Tony.
The other thing I didn't look at is that the vifa's are 8" drivers. That is going to be a challenge if you are trying to do a true D'appolito MTM. The centre to centre distance is just going to be too large.
If you can run a pc simulation program on your mac, it may give you more options for crossover simulation.
Tony.
- Status
- Not open for further replies.