Scanspeak Project (first attemp)

[AndrewUK1990]

Hi everyone, I have just made my first ever attemp at modelling a crossover for my speaker cabinets which i have already waiting. The setup im going for is BBMT where the two mid/bass units are Scanspeak 18W/4531G00's the Midrange is a Scanspeak 12M/4631G00 and the tweeter is a Scanspeak D2905 950000.

I have followed the steps for processing the frequency response data for my specific cabinet & baffle using the FRD consortium tools. I then used these in speaker workshop to produce a crossover.

So im just asking any of you wise people if you could have a quick look at the pictures of my Xover design and the chart because I'm sure i will have done something wrong that ive overlooked seeing as though ive never done this before. Critical advise is appreciated - im sure there is plenty!

Quick CAD drawing of the baffle - it is not very accurate; just for a rough idea of what it looks like. Other two pics are freq response/phase and crossover design.






Thanks all,

Andrew.

[PeteMcK]

Andrew, It's not clear from the schematic, but the 10uF cap and 10 ohm resistor in the woofer section should be in series I think....

[AndrewUK1990]

Ok, thanks for the input, it was wired wrong there by mistake. I've had a play around with it some more - Im not good at drawing the schematics very well yet but i just thought id put up the design here incase anyone may spot something that was obviously wrong or incase i had done something that is known not to be a good idea - like putting an L-pad on a woofer for instance.

Hope thats not too confusing ! Ive also tried another crossover design 4th order to bring a -24dB slope to reduce bass drivers especially from receiving higher end frequencys. Ive got that crossover reasonably flat, but still tweaking. Ill upload it tomorrow and try give more details so its easier to understand. Off to bed now tho - midnight here and gotta be up for lectures at 9am

[wintermute]

yes those 10uF caps concerned me as well. I'd suggest turning on the option to generate the impedance plot for your filter network, it may save you some heartache!

Tony.

[nickf]

Hi Andrew,
the placement of the tweeter horizontally beside the midrange will place the crossover lobing effects in the horizontal plane, making for uneven dispersion across the room at the mid to tweeter crossover frequency.
Lobing effects are cause by the difference in distance between the listener and the tweeter and the listener and the midrange. Ie. if you are at 45 degrees to the speaker the distance between you and the tweeter is not the same as between you and the midrange. Thus frequencies which added perfectly when you were directly in front of the speaker may now subtract from each other.
To minimise lobing use higher order crossovers so that the frequency band where both midrange and tweeter produce nearly equal amounts of sound is smaller.
As you have drawn them, these speakers may work best on their side! or, is it possible to re-make the front baffle to place the tweeter above the woofer.

Nick

[AndrewUK1990]

Originally Posted by nickf

Hi Andrew,
the placement of the tweeter horizontally beside the midrange will place the crossover lobing effects in the horizontal plane, making for uneven dispersion across the room at the mid to tweeter crossover frequency.
Lobing effects are cause by the difference in distance between the listener and the tweeter and the listener and the midrange. Ie. if you are at 45 degrees to the speaker the distance between you and the tweeter is not the same as between you and the midrange. Thus frequencies which added perfectly when you were directly in front of the speaker may now subtract from each other.
To minimise lobing use higher order crossovers so that the frequency band where both midrange and tweeter produce nearly equal amounts of sound is smaller.
As you have drawn them, these speakers may work best on their side! or, is it possible to re-make the front baffle to place the tweeter above the woofer.

Nick

Thanks for the advice here - its really helping me out! I didnt realise about the lobing effect but ive just read some more about it now, seems like i would be better off designing a new baffle really putting the tweeter above the drivers just like you said. Although I can get around the problem of the tweeters/mid i think by making the speakers toe in towards the listening position if ive understood this correctly. Reason i dont want to change the baffle unless its obvious that it isnt going to be suitable is because they are old Wharfedale cabinets which i really like for whatever sentimentle reason :P (I will alter them if needs must though).

Secondly, ive been working on a 4th order crossover design and will be finishing it off tonight hopefully so that should help out with this lobing problem a bit too as well as cutting off the woofer a bit sooner which would atm be receiving too much high frequency.

Couple of questions too if i may - how do you turn on the impedance plot on the combined graph in speakerworkshop??? its been puzzling me a lot, just cant seem to find it! Seconly, i used FRD tools to get accurate modelled frequency response data for my drivers in this baffle/cabinet' but when i used unibox to model the cabinet size for each driver, i wasnt 100% sure weather to model it for one scanspeak bass driver and use that data for each, or to put 'both drivers in series' and model them at the same time. When i was doing it i just did each driver individually in seperate spreadsheets with the mid having the appropriate box volume of 1.6L and the bass drivers having a total box volume of 55L. The complete box volume is 55L so my question is: Was i right in modelling each bass driver using 55L as my box volume, or should i have halved the volume as there are two speakers, or should i have done both drivers in the same unibox modelling sheet and set the config as 'Two speakers in series'?

Many thanks,

Andrew.

[wintermute]

Couple of questions too if i may - how do you turn on the impedance plot on the combined graph in speakerworkshop??? its been puzzling me a lot, just cant seem to find it!

I don't think you can turn it on for a combined plot What I did was do the low and high pass filters separately (so I could use the optimizer) and then made a single filter network with all filters and drivers (with the same values as the optimized individual ones) and then calculated the response of that with the tick box for calculate network impedance ticked

so my question is: Was i right in modelling each bass driver using 55L as my box volume, or should i have halved the volume as there are two speakers, or should i have done both drivers in the same unibox modelling sheet and set the config as 'Two speakers in series'?

I would say the best approach would be to tell unibox you have two drivers (I assume they are 4 ohm if you are wiring in series) and model it with the 55L. If they are 8 ohm nominal woofers then definitely model them (and in your final implementation wire them) in paralell! wiring in series you get no real spl gain. Wiring in parallel you get a theoretical 6db gain.

Tony.

[AndrewUK1990]

Hi Wintermute, the woofers are indeed 4ohm hence why i have wired in series. They are 90dB sensitivity which is very high for a woofer so they dont need any extra gain either (unlike most 8ohm woofers).

I'll take your advice and model them in series and use that in FRC to get a more accurate freq response, but im guessing there wont be too much difference as they are in series.

Also thanks for clearing up the impedance plot! i have also done high pass, low pass and a bandpass on seperate networks then combined into one final network so ill hav a look tnyt for the impedance plot tick box.

Thanks again,

Andrew.

[dublin78]

You are braver than me. Congrats. All I can do is copy other designs.

You may as well cast your eye over these two projects with similar drivers:

Ekta Grande: Ekta GrandeScanSpeak 18W/
Vertigo Plus: Audio Components - Scan-Speak - Reference Line

Good luck

[AndrewUK1990]

Cheers for the links Dublin78, ive already had a good look at these though they are both very nice designs; gives me something to aim for