No announcement yet.

Troel W18 Crossover Tweak

  • Filter
  • Time
  • Show
Clear All
new posts

  • Troel W18 Crossover Tweak

    Hi Guys,
    I'm writing to show off my ignorance in crossover design...

    I'm working on the Discovery W18 by Troel, flat version with V2 crossover. Here the post: Link
    And here the Crossover:

    Click image for larger version

Name:	Troel 18W Crossover V2.JPG
Views:	326
Size:	64.3 KB
ID:	1480747

    Only issue is that I'm not going to use the Scan Speak R2604/832000, but the XT25BG60-04 (Dual magnet).
    From what I read around it should be a direct drop in if not for the sensitivity that's 1.3 db higher for the Vifa.

    Here 2 questions:

    Do you agree that I should be able to drop it in with the same crossover?
    How am I supposed to modify the Hi section to damp the tweeter of about 1.3db?


  • #2
    Here the link to Ray Tremblay's post where he shows the difference between Vifa XT25TG30-04 vs XT25BG60-04 and there is a discussion about the SS too.

    Thanks Ray!


    • #3
      Did you ask Troels?


      • fpitas
        fpitas commented
        Editing a comment
        Good point. He may have been down this road before.

    • #4
      If it were me, I'd drop that initial 1/2ohm resistor, and MAYbe increase the initial 6.8uF cap to 8uF (or even better, somewhere in-between). YMMV

      The .frd files EYE have DO show a +2-3dB "bump" above 12k on the Scan that the XT doesn't have (and you won't either).
      Also, the (filtered) XT's impedance runs 1/2 to 1 ohm lower than the Scan's (still stays well above 4ohms though).


      • #5
        djg : you are actually right, I'll try to ask!
        Chris Roemer : Why would you suggest those changes? what would be the effect? lower the crossover freq. instead of L-Padding?


        • #6
          Click image for larger version

Name:	1977.jpg
Views:	216
Size:	76.6 KB
ID:	1480794

          Red line is SS tweeter w/Troels filter you imaged.
          White is XT25 w/the initial resistor removed and the 6.8 cap raised to 7.4uF.
          Do as you wish (the XT FR is considerably flatter, but it's not what Troels used - but you already know that).


          • #7
            Thanks Chris,
            I'll go with your suggestion!
            Another question: Should I provision some more resistors to tune the output of the tweeter when built? If yes what values would you suggest?


            • #8
              On Troels orig. XO, the front (1st / left) resistor took the whole tweeter up/down. If it's too bright, adding 1n(ohm) back there should drop the XT by about -1dB (2n for 2dB).
              The rear (2nd / right) resistor ends up tilting the XT's response. If you dropped it to 1n, the Fc and shoulder don't move much, but the top end tilts "up" (possibly more like his orig. Scan tweeter). You could bump it up to 3n, but most don't care for a tweeter tilted down (or can hear it, if they're "mature", like me).


              • #9
                Where did you get the FRDs and ZMAs? Your own measures?
                I'm trying to simulate it in XSim and they behave quite differently! something is not right in my sim...


                • #10
                  here the snapshots of the results
                  Click image for larger version

Name:	XT25 vs SS Circuit.jpg
Views:	179
Size:	114.2 KB
ID:	1480888 Click image for larger version

Name:	XT25 vs SS.jpg
Views:	167
Size:	60.0 KB
ID:	1480889


                  • #11
                    other try:
                    Here the results of Troel's XO (same for both tweeters) using the TX25 frd and zma you sent to martyh a few years ago on this forum. And they match pretty well!

                    I'm kind of lost....

                    Click image for larger version

Name:	XT25 vs SS Circuit Matyh.jpg
Views:	170
Size:	57.8 KB
ID:	1480893


                    • #12
                      My files (just traces, pretty sure) of the Scan D2604 are from May 2016. The top end is NOT smooth.
                      It's about 92dB @ 7k, and 93 @ 14k, but is LOW at about 89.5dB @ 11k and again by 20k (so about 3dB worth of "wiggles" over 7kHz).

                      My files for 264-1012 (PE's #) are from 2013.
                      The FRD is most likely a trace.
                      I DO have 2 ZMA files (1 traced, and marty's was measured), but they're not significantly different.
                      Trace had Fs near 500Hz @ 19ohm Z-max. WT closer to 600Hz @ 21ohms.

                      On your schemo you list the Scan as "R2604" - but it's really D2604, right?
                      I think our SS files probably differ in .frds mostly.
                      Just sim both tweeters (with files YOU like) and fiddle w/the padding (and maybe that initial cap, if it helps).
                      By YOUR last post, it looks like you could just drop the XT right in there w/NO XO changes, right?

                      (OR, attach files if you want ME to try again.) These are ALL just sims.
                      The proper way is to do your own measurements "in box" and go from there.


                      • #13
                        Furthermore ... going through the (MY) archives, I see I've got 2 .PDFs from Tymphany for this driver.

                        Going by the dates that EYE downloaded them, the older one (2012) is called "Rev1" (revision 1, I assume).
                        It shows the double magnet on both the photo and the mechanical drawing.
                        It's datasheet "pretty much" matches my newer sheet, BUT: lists SPL as [email protected], and [email protected]

                        My 2017 sheet has a double magnet photo, but shows only a single magnet on the mech. drwg. (an error, I presume).
                        It lists SPL as 92.3dB @ 2.83v, and 88.6 @ 1w.

                        Since 3dB diff. is common between 2.83v and 1w (@ 4ohms), I'm rounding the specs to:
                        94dBv and 91dBw for the 2012, and 92dBv and 89dBw for 2017. That's a difference of about 2dB in sensitivity.
                        Funny thing is that I don't really see that on their plots though.
                        (This could be the main reason why designers ask you not to ask for advice on driver swaps - TOO many variables (errors?) when you don't measure in-box.)

                        If you want these PDFs I can try attaching them (or send them via email)?


                        • #14
                          The most sensible thing seems to be to measure them...
                          I'll proceed with the setup and let you know!
                          Have to build a test box first but for the tweeter should be able to proceed with only the baffle tough...