No announcement yet.

Help with 3-way design - Klipsch Heresy III clones!

  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    OK, I'll try something w/the PC83 ...

    Your 3n(ohm) "shunt" resistor (runnning to gnd) in your woofer section is technically shunting about 60% of your amp's power away from ALL 3 drivers. OUCH! You should be able to remove it w/out changing any of the inter-driver relationships (balance). I think ALL your curves should pop up about +4dB w/that out of there. (You can temp. change its value to 300 or 3000 ohms to get essentially the same effect.)

    Here are the 1st (of many) "tips".

    There are more benefits (than not) to putting the "L-pad" (resistor network) over next to the tweeter (at the "end" - past the filter) rather than out in front.

    I SEE the "DCR" next to your coil values (they're ALL 0.28n(ohms) - probably a default?). Kinda important to have this accurate to within +/- 1/2n (0.1 is better).

    The software you're using wants XYZ coordinates for all your drivers (default looks like it's ZER0s). This is the horizontal, vertical, and "depth" offsets of (basically) the front of all the voice coils (w/the tweeter usually the reference - the "design axis"). And it's in meters (looks like). Some guys use the dust cap for ref.
    SOME softwares: if your tweeter is at 0,0,0 (and your 3 drivers run vertically down the center of your baffle), the all X-coords could be "0", but the mid will be below the tweeter (and behind - a bit) so maybe a "Y" of -4" (-100mm, or -0.10m), and a "Z" of -0.006m (1/4" behind the tweeter?).
    The woofer is even farther down, and farther back (since the driver is physically larger). Maybe -0.25m (-10") w/a Z of -0.025 (-1"?).

    Your filters on the woofer LP and mid HP have "peaky" "shoulders" (high "Q"), not generally a good idea. To keep the same cross pt. try doubling the series component and halving the shunt (OR it's the other way around). Your mid almost looks like you could flip the polarity on it (and then also the tweeter). How do I know this? 'Cause the summation line (above where the two drivers' curves cross) should be +3dB to +6dB higher than the cross pt. - yours looks like only +1 - +2.

    Now you've got to get a handle on baffle-step compensation. It's an effect that (generally) causes SPL to drop off on bass freqs. (maybe from 1kHz on down). To compensate, we typically use larger series coils on the woofer to attenuate its level at 1k -6dB compared to 100Hz. (This depends on MANY things, including speaker placement in the room). Then the mid and tweeter would also need -6dB more attenuation. SOME prefer only 3dB of BSC (baffle-step comp.). If you don't account for any at all, you'll probably find the bass lacking in your design. The free Tolvan's "Edge" software (on the net) lets you create your baffle and place drivers on it and plots the resulting response curve so you can see what I'm talkin' about.

    Do NOT get addicted to this stuff. Do NOT get addicted to this stuff. Do NOT get addicted to this stuff. Do NOT get addicted to this stuff. Do NOT get addicted to this stuff.


    • #17
      Excellent tips! I'll adjust some of the things you mentioned above. The software I'm using has a diffraction simulator so I'll input the driver locations and try to simulate it now that I have some direction...

      I'm thinking of off-setting the tweeter and mid away from the center vertical because it will allow me to build a "mini" sealed enclosure by using 3 of the existing outer walls of the larger chamber... If that makes any sense. I also thought about using a capped 3" PVC tube to seal the mid woofer... This would be the easiest way to build an enclosure inside an enclosure in my opinion and maximize the free volume. I'm not sure if the tube shape would affect the sounds, however. If I can use the tube, then I canter all drivers vertically, which I think will look better.


      • #18
        I ran a sim w/your 3n shunt resistor, and when I remove it your SPL did NOT jump up, but your system impedance DID rise from below 2n up to 4 or better.

        Try this(component "string" left to right (amp > driver).
        Tweeter HP: 3uF series cap and 0.10mH series coil (to roll off top end), 0.70mH shunt coil (to ground), L-pad: SeriesResistor=4n(ohms) / PR=8n
        Mid BandPass: 2n series resistor (attenuation), 0.50mH series coil, 25uF series cap, then 2 shunts; a 10uF cap, and a 4.0mH coil (
        Woofer LP: 3.0mH series coil (low DCR), 30uF shunt cap, a Zobel (next to woofer) using an 8n resistor + a 40uF cap.

        Crosses ended up near 700Hz and 3.8k. Runs around 84dB +4 to +5dB of BSC.


        • #19
          thanks, i tweaked some of your capacitor values slightly because some of those caps are only available in film and foil which would've cost $10 / piece. Also, the tweeter is the only one with inverted polarity in the larger screenshot below. If i invert the woofer, I get a pretty big bump in the 100-500hz region -- is this okay?

          mid-woofer polarity inverted:
          Click image for larger version

Name:	woofer inv..JPG
Views:	92
Size:	37.0 KB
ID:	1429207

          I input additional info such as baffle diffraction & enclosure info. I updated all the dimensions as well. I'm not sure if i'm properly modeling the toe-kick (wedge) angle of the cabinet. The bottom wedge sets the front baffle at about -10 deg. in relation to the vertical plane. I put "-10deg" for the vertical axis in the diffraction tool (is this correct)? See screenshots below. Do i need to account for this in any other locations? I.e. driver offsets below in relation to tweeter...with the angled baffle, the driver distances from the listening position will be pretty close to even since the tweeter would be the furthest back and woofer would be closest to the listening position.

          Click image for larger version

Name:	3-way FR_rev1.JPG
Views:	92
Size:	273.8 KB
ID:	1429205

          baffle diffraction - let me know if anything jumps out. I must admit, I don't fully understand what is going on here.

          Click image for larger version

Name:	DC250-8_DIFF.JPG
Views:	86
Size:	147.6 KB
ID:	1429208

          "full space" checked:

          Click image for larger version

Name:	DC250-8_DIFF_FS.JPG
Views:	86
Size:	146.0 KB
ID:	1429209

          Attached Files


          • #20
            anyways, here are some progress photos of the cabinets. book-matched walnut plywood panels are cut and edge-banded. toe-kick wedge is cut but I have yet to router the "arch" out. should be ready for glue later today !!!
            Click image for larger version

Name:	IMG_20200111_163404.jpg
Views:	99
Size:	399.3 KB
ID:	1429214 Click image for larger version

Name:	IMG_20200111_142054.jpg
Views:	97
Size:	436.4 KB
ID:	1429212 Click image for larger version

Name:	IMG_20200111_170536.jpg
Views:	96
Size:	386.3 KB
ID:	1429213
            Attached Files


            • #21
              Not fam. w/THAT sim software. Didn't know you were going to tilt them back (10* I guess)?
              Usually the woofer is the "reference" for phase (connected "normal").
              What If you try reversing the mid and running the tweeter normal? (So, flipping both mid and tweet from how you have them.)
              On MY sim (on the mid) I've got the 25uF series cap out front (w/the 2n resistor and small coil), NOT between the other 2 shunts. Can your XO sim do that?
              If not, I'll prob. tweak some stuff. (Running all the series parts "out front" helps to hold up the impedance on a 3-way.)

              I'm guessin' that you're sticking to about 1.5cf (instead of 2.5cf). You're limiting that woofer's bottom end that way. Mid to upper 40s instead of mid to low 30s. :-(
              Hopefully you're still venting them (about 3"id x 7" port)?
              EYE would probably still want to run a sub w/those (in the small box). YMMV


              • #22
                hey chris, i don't think the baffle diffraction effect was getting fed to the overall frequency response -- i *think* i fixed that. I tried all sorts of different polarity combinations and if I invert the mid only i get something like this:

                Click image for larger version

Name:	3-way_SPL_mid inverted.png
Views:	74
Size:	42.3 KB
ID:	1429261

                mid + tweeter inverted:

                Click image for larger version

Name:	3-way_SPL_tweeter_mid inverted.png
Views:	68
Size:	45.2 KB
ID:	1429262

                tweeter only inverted:

                Click image for larger version

Name:	3-way_SPL_tweeter inverted.png
Views:	68
Size:	41.7 KB
ID:	1429263

                looks okay to me with just tweeter inverted, any issues with this? in regards to the cabinet size: i don't think we have enough space for a 60L enclosure...frankly im not sure how my wife is even allowing me to put the 41L enclosures in our living room. however, my simulation shows the following for the 41L cabinet with a 3"dia x 8"L port:

                f3 40.6 Hz
                f6 35.1 Hz
                f10 29.5 Hz
                Zmin 8.1 Ohm @ 5 Hz
                Zmax 42.8 Ohm @ 15.4 Hz
                GDmax 11.4 ms @ 39.4 Hz
                XmaxC 4.5 mm @ 5 Hz
                VmaxR 2.7 m/s @ 24.8 Hz
                Pmax 1 VA @ 5 Hz
                DRIVER: Dayton Audio DC250-8, 1 pcs in series
                n0 0.35 % Reference efficiency
                SPL 87.6 dB/W Sensitivity
                USPL 88.5 dB/2.83 Sensitivity
                EBP 57.4 Efficiency bandwidth product
                Dd 20.9 cm Effective diameter of driver
                Vd 154.4 cm^3 Maximum linear volume of displacement
                Cas 7.36E-7 m^5/N Acoustic equivalent of Cms
                Mas 5.73E1 kg/m^4 Acoustic equivalent of Mms+Mme
                Ras 2.92E3 Ns/m^5 Acoustic equivalent of Rms
                Rae 2.06E4 Ns/m^5 Acoustic equivalent of Re+Rg
                BOX REAR 1: Vb=41.0 l, Ql=7.0
                Fb 35.5 Hz System resonance frequency
                Cab 2.9E-7 m^5/N Acoustic compliance of air in enclosure
                Rab 1.55E2 Ns/m^5 Acoustic resistance due to absorption
                Ral 1.08E5 Ns/m^5 Acoustic resistance due to leakage
                VENT REAR 1: Dv=7.6 cm, Lv=20.3 cm
                Sp 45.4 cm^2 Effective area of port
                Map 6.94E1 kg/m^4 Acoustic mass of air in port
                Rap 1.55E2 Ns/m^5 Acoustic resistance of port losses
                Click image for larger version

Name:	enclosure.png
Views:	68
Size:	32.1 KB
ID:	1429265

                I think F3 of 40hz is good enough for me...if my simulation is correct. I have a pair of overnight sensation MTMs in my home office and their F3 is in the same range from what i can gather...i don't feel the need for a sub with those. I don't know if these will perform the same but if the simulation is correct, I think the bass should be sufficient for me.
                Attached Files