Tannoy CPA 12 Crossover

Discussion in 'DIY Discussion' started by Tenson, Apr 30, 2011.

  1. Tenson

    Tenson Moderator

    Joined:
    Nov 12, 2003
    Messages:
    5,947
    Likes Received:
    1
    Location:
    Kent, UK
    I went to visit Dev yesterday and take a look at his Tannoy CPA 12 based speakers which he reported some hardness from using the standard crossover.

    Edit: Please note measured frequency response below ~100Hz is not 100% accurate due to noise from a road outside.

    Below is the plot of the speaker with the standard crossover.

    [​IMG]


    Here is the schematic of the standard Xover.

    [​IMG]


    Here is the tweeter frequency response with no crossover.

    [​IMG]


    Here is the CPA-12 tweeter impedance.

    [​IMG]


    Here is the woofer frequency response.

    [​IMG]


    Here is the CPA-12 woofer impedance.

    [​IMG]
     
    Tenson, Apr 30, 2011
    #1
    1. Advertisements

  2. Tenson

    Tenson Moderator

    Joined:
    Nov 12, 2003
    Messages:
    5,947
    Likes Received:
    1
    Location:
    Kent, UK
    I then experimented with a completely new design of crossover and have come up with two options.

    Option 1:

    Minimum of components, this was the first design. It uses a notch filter on the tweeter and the woofer to obtain a smooth response as both drivers show a nasty bump, and I think this is the cause of the hard sound Dev talked out.

    While it achieves a much smoother frequency response than the Tannoy one, it does so by making the woofer and tweeter about 60deg out of phase at the point where they would otherwise peak, so they help to cancel each other out. In a normal speaker with separate tweeter and woofer this is not such a great idea as it relies on acoustical summing of the drivers, and normally this is not ideal and changes off-axis. The dual concentric design of the Tannoy however makes this a much more reasonable option as the drivers should sum almost perfectly.

    Make sure you use good low DCR inductors on the woofer, not just on the first inductor but also in the notch filter.


    Here is the frequency response of the new 'AudioSmile' design.

    [​IMG]


    Here is the same design with the tweeter reversed so you can see how well the phase aligns. If the phase alignment were perfect there would be a sharp notch where they cross. As you can see the drivers are partly out of phase with each other, on purpose though.

    [​IMG]


    Here is the crossover schematic.

    [​IMG]



    Option 2:

    Option 2 is designed to maintain phase alignment over the drivers crossover as well as the smooth frequency response. This is done by the addition of an all-pass filter to the tweeter network. The other component values have also changed slightly compared with option 1.


    Here is the frequency response of the new option 2 crossover network.

    [​IMG]


    Here is the same design with the tweeter reversed so you can see how well the phase aligns. If the phase alignment is good there would be a sharp notch where they cross. As you can see this is the case.

    [​IMG]


    Here is the crossover schematic.

    [​IMG]
     
    Tenson, Apr 30, 2011
    #2
    1. Advertisements

  3. Tenson

    cooky1257

    Joined:
    Aug 30, 2007
    Messages:
    797
    Likes Received:
    0
    That's great stuff Simon.

    In contrast here's the 215 DMT xover, same tweeter but loaded by larger horn/bass driver.
    [​IMG]
     
    cooky1257, May 1, 2011
    #3
  4. Tenson

    lunatek

    Joined:
    Jul 28, 2011
    Messages:
    5
    Likes Received:
    0
    Thank you for a very informative thread. Exactly what I was looking for!

    For some odd reason I didn't find these messages when I searched audiosmile forums for the first time. Strange. But now I have seen the light, eventually!


    Thanks

    Janne
    Turku, Finland
     
    lunatek, Jul 28, 2011
    #4
  5. Tenson

    cooky1257

    Joined:
    Aug 30, 2007
    Messages:
    797
    Likes Received:
    0
    Simon, Just to be clear, are these notch filters to correct the 'faulty' driver response ?
    I only ask as I understand it one of Devs drivers was misbehaving?

    In the case of the CPA15 xover FWIW C1 is 20uF, C2 is 6,8uF, L1 is 1.5mH, L2 is .33mH, R1 is 8.2 ohm, R2 is 2.2ohm

    Cooky
     
    cooky1257, Jul 29, 2011
    #5
  6. Tenson

    Dev Moderator

    Joined:
    Jun 19, 2003
    Messages:
    5,764
    Likes Received:
    3
    Location:
    Ilford, Essex, UK
    Frank, yes, the bass cone of the measured driver has a small tear in it, which I assume is due to abuse at some stage which may also have affected the HF unit. The tear has been glued, but when I get around to finishing these speakers (distracted by a lot of other stuff :() and listening again, I'll decide whether to have them reconed or not.

    Current state of the speakers:-

    [​IMG][​IMG]
     
    Dev, Jul 29, 2011
    #6
  7. Tenson

    Tenson Moderator

    Joined:
    Nov 12, 2003
    Messages:
    5,947
    Likes Received:
    1
    Location:
    Kent, UK
    I'm not sure to be honest, since I only measured that one speaker. It could be a bad compression driver making the peak, or it could be how they all are.

    I have a hunch that is how they are meant to be, because the response is generally swinging around a flat line. If the compression driver did have that peak taken out the response would still not be flat because of the woofers peak in a similar place.
     
    Tenson, Jul 29, 2011
    #7
  8. Tenson

    Tenson Moderator

    Joined:
    Nov 12, 2003
    Messages:
    5,947
    Likes Received:
    1
    Location:
    Kent, UK
    Nice Walnut Dev.
     
    Tenson, Jul 29, 2011
    #8
  9. Tenson

    Dev Moderator

    Joined:
    Jun 19, 2003
    Messages:
    5,764
    Likes Received:
    3
    Location:
    Ilford, Essex, UK
    It'll look redder when I've finished:).
     
    Dev, Jul 29, 2011
    #9
  10. Tenson

    cooky1257

    Joined:
    Aug 30, 2007
    Messages:
    797
    Likes Received:
    0
    Cheers Simon,

    I suspect that peak could be a combination of an undamped resonance due to burnt off ferrofluid and a function of the smaller 12'cone/horn-this peak isn't present on the 15".....I'm going to root out the PEQ settings for both cabs..

    Nice cabs Dev.

    Update; The peq is different, the 15 has a smoother response through 800-2k range. HF Ferrofluid may be ok too but I've yet to see one that doesn't need a recharge.
     
    Last edited by a moderator: Jul 29, 2011
    cooky1257, Jul 29, 2011
    #10
  11. Tenson

    lunatek

    Joined:
    Jul 28, 2011
    Messages:
    5
    Likes Received:
    0
    Image links?

    Hello!

    I returned to check the measurements and crossover designs but it seems that all of them are now unavailable. Any chance finding the pics somewhere?

    I would be happy to find them here onnthe forum or perhaps someone could email those to me?


    *email address removed


    Thanks for any help!



    Janne
     
    lunatek, Mar 8, 2013
    #11
  12. Tenson

    Tenson Moderator

    Joined:
    Nov 12, 2003
    Messages:
    5,947
    Likes Received:
    1
    Location:
    Kent, UK
    Hi Janne,

    It seems they were lost when the forum had a big crash about a year ago. I've re-uploaded them for you, so they should appear in the original posting.

    Simon
     
    Tenson, Mar 9, 2013
    #12
  13. Tenson

    lunatek

    Joined:
    Jul 28, 2011
    Messages:
    5
    Likes Received:
    0
    Thank you!

    Fantastic. This was fast!


    Janne
     
    lunatek, Mar 12, 2013
    #13
  14. Tenson

    Mimic82

    Joined:
    Dec 20, 2015
    Messages:
    1
    Likes Received:
    0
    Is there any change some one could reupload the pictures. I bought some cpa12 to nut they sound real nasty :)
     
    Mimic82, Dec 20, 2015
    #14
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.