Announcement

Collapse
No announcement yet.

I could really use some help designing 3 way Crossovers

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    Might want to double check the mid-range portion. I can't sim that with PCD, but to the extent I can, it doesn't look right to me, and you have quite a dip in the response.

    Comment


    • #32
      Originally posted by DeZZar View Post

      Doesn't really help the rest of the community learn anything though does it? Perhaps they are just afraid of having their suggestions mercilessly shot down in a barrage of personal attacks! lol....
      Big surprise.
      BSME
      Kannapolis NC

      Comment


      • #33
        What if you flip the mid?

        Comment


        • #34
          Originally posted by Chris Roemer View Post
          What if you flip the mid?
          Translate that mid filter for me. I've seen you lay x-overs out this way before.

          Is that a ( 22uf and 4mH high-pass ), and a ( 1.5mh and 4uf low-pass ), with an added 1uf cap to make a notch (tank type notch) for the low-pass?

          Comment


          • #35
            Well, THAT's not MY suggested XO (BION), but it DOES look like one of mine.

            The mid uses an L-pad for attenuation (EYE almost always prefer an "L" on a mid or tweeter because the PR helps flatten the driver's Fs, in or out of box, which helps the HP filter (part) roll the bottom end off).
            Aside from that, it's just a 2nd order HP & LP. HP being the 22uF series and 4mH shunt (I used 55uF and 4mH !). The LP is a (notched/tanked) 2nd order w/the 1.5mH series and 4uF shunt (I used 0.9mH and 6.8uF - w/OUT a notch). The small (1uF) cap across the 1.5mH is the notch. Wolf (and others) rec. the 4n(ohms) resistance in the 1uF "bypass" because w/out it (AND - in this case, the 2n "R3" for attenuation) you COULD have a nearly direct short (for the amp to deal with) of high freqs. which would otherwise pass (unabated) through C2(22uF) > C5(1uF - which in reality would only pass VERY high freqs. - but SOME good power amps can be linear up to 100kHz - possibly even higher?) > and > C3(4uF) to GROUND.

            I've found (over many, many, many sims) that having the series components (C2 & L3) grouped together ahead of the shunts (L2 & C3), OFTEN helps hold up the system's minimum impedance (which is OFTEN a problem on 3-ways, 'cause you've basically got 3 drivers all in parallel (overlapping - the problem occurs when they overlap too much!) each other. It's (obviously) worse the more you use 4n drivers (over 8n).
            I'm not really sure why this^ is so, but I have observed it often. SOME of the sims aren't flexible enough to let you "mesh" the HP & LP components like this, meaning you might have to go: HP seriesCap > HP shuntCoil, then LP seriesCoil > LP shuntCap (or vice versa).

            You DO get (slightly) diff. responses (using the exact same 2 caps & 2 coils) with the 2 diff. topologies, so the impedance profiles DO end up different (obviously).

            (I wonder if I explained all of that properly?)
            Last edited by Chris Roemer; 08-30-2020, 04:57 PM. Reason: THIS XO (not mine) DOES use an L-pad. MINE also did, kinda (I moved the series resistor "out front" (near the amp end) to "tilt" the FR a bit.)

            Comment


            • #36
              Originally posted by Chris Roemer View Post
              Well, THAT's not MY suggested XO (BION), but it DOES look like one of mine.

              The mid uses a series resistor for attenuation (EYE almost always prefer an "L" on a mid or tweeter because the PR helps flatten the driver's Fs, in or out of box, which helps the HP filter (part) roll the bottom end off).
              Aside from that, it's just a 2nd order HP & LP. HP being the 22uF series and 4mH shunt (I used 55uF and 4mH !). The LP is a (notched/tanked) 2nd order w/the 1.5mH series and 4uF shunt (I used 0.9mH and 6.8uF - w/OUT a notch). The small (1uF) cap across the 1.5mH is the notch. Wolf (and others) rec. the 4n(ohms) resistance in the 1uF "bypass" because w/out it (AND - in this case, the 2n "R3" for attenuation) you COULD have a nearly direct short (for the amp to deal with) of high freqs. which would otherwise pass (unabated) through C2(22uF) > C5(1uF - which in reality would only pass VERY high freqs. - but SOME good power amps can be linear up to 100kHz - possibly even higher?) > and > C3(4uF) to GROUND.

              I've found (over many, many, many sims) that having the series components (C2 & L3) grouped together ahead of the shunts (L2 & C3), OFTEN helps hold up the system's minimum impedance (which is OFTEN a problem on 3-ways, 'cause you've basically got 3 drivers all in parallel (overlapping - the problem occurs when they overlap too much!) each other. It's (obviously) worse the more you use 4n drivers (over 8n).
              I'm not really sure why this^ is so, but I have observed it often. SOME of the sims aren't flexible enough to let you "mesh" the HP & LP components like this, meaning you might have to go: HP seriesCap > HP shuntCoil, then LP seriesCoil > LP shuntCap (or vice versa).

              You DO get (slightly) diff. responses (using the exact same 2 caps & 2 coils) with the 2 diff. topologies, so the impedance profiles DO end up different (obviously).

              (I wonder if I explained all of that properly?)
              Close enough for me!

              Comment


              • #37
                Here's one more:
                Click image for larger version

Name:	my 2nd xover design.jpg
Views:	153
Size:	37.8 KB
ID:	1450166

                frequency response doesn't look too bad

                Click image for larger version

Name:	my 2nd xover design fr.jpg
Views:	160
Size:	110.3 KB
ID:	1450167


                Comment


                • #38
                  Originally posted by clang View Post
                  Here's one more:
                  Click image for larger version  Name:	my 2nd xover design.jpg Views:	0 Size:	37.8 KB ID:	1450166

                  frequency response doesn't look too bad

                  Click image for larger version  Name:	my 2nd xover design fr.jpg Views:	0 Size:	110.3 KB ID:	1450167

                  I'd like to see the individual driver responses on this one, and the impedance if it's not too hard to post.

                  Comment


                  • #39
                    Not a problem:

                    Click image for larger version

Name:	my 2nd xover design fr.jpg
Views:	192
Size:	118.7 KB
ID:	1450225

                    Click image for larger version

Name:	my 2nd xover design im.jpg
Views:	175
Size:	118.9 KB
ID:	1450226

                    Comment


                    • #40
                      Between 1k-2k your summed resp. is higher than your mid. Never a good situation.
                      It means you've got cancellation going on (which is really a phasing issue).
                      The usual suspect (here) looks like the tweeter's extended response (below its Fc).
                      (Also, the top octaves look a couple dB "hot".)

                      Comment


                      • #41
                        Originally posted by clang View Post
                        Not a problem:

                        Click image for larger version

Name:	my 2nd xover design fr.jpg
Views:	192
Size:	118.7 KB
ID:	1450225

                        Click image for larger version

Name:	my 2nd xover design im.jpg
Views:	175
Size:	118.9 KB
ID:	1450226
                        Well, it's only 8 parts, and it has a decent impedance, and on axis response is in the ballpark. It might be optimized for a more upward axis. I'd try it, and see how it goes.

                        Comment


                        • #42
                          Originally posted by rpb View Post

                          Well, it's only 8 parts, and it has a decent impedance, and on axis response is in the ballpark. It might be optimized for a more upward axis. I'd try it, and see how it goes.
                          Yep, I'm thinking of trying it. I'll also pick up a few small resistors if I have to attenuate the tweeter. This one is my design so I might as well own it.

                          Comment


                          • #43
                            Originally posted by clang View Post

                            Yep, I'm thinking of trying it. I'll also pick up a few small resistors if I have to attenuate the tweeter. This one is my design so I might as well own it.
                            Since it's your design, I'll offer a suggestion. Try a smaller coil on the tweeter filter, and see how things look. If nothing else, it should protect the tweeter from damage. It will probably change the impedance at high frequency a little, so there's a trade-off. Might be minor though.

                            If you can add resistance of the coils in your sim, I'd ballpark them at .3 ohms. Don't worry about being exact. If it's a big coil, the values may be more important. The cheaper coils sometimes are higher resistance. That might be an advantage, or not. Depends on multiple factors.

                            Comment


                            • #44
                              Clang - there IS a problem.
                              You absolutely can NOT rely on an "out-of-phase" condition to achieve a "flat" FR curve.
                              That's exactly what your mid is up to. You (EYE) can SEE it right on your plots! Your mid (yellow) is higher than your summed response (the black line). This is BAD. I can only explain it to you, I can't understand it for you.

                              SOMEbody else please back me up here? ??

                              Comment


                              • #45
                                Originally posted by Chris Roemer View Post
                                Clang - there IS a problem.
                                You absolutely can NOT rely on an "out-of-phase" condition to achieve a "flat" FR curve.
                                That's exactly what your mid is up to. You (EYE) can SEE it right on your plots! Your mid (yellow) is higher than your summed response (the black line). This is BAD. I can only explain it to you, I can't understand it for you.

                                SOMEbody else please back me up here? ??
                                Chris... I pretty much agree that we could eliminate that, and it would sound better. I don't know what off-sets he used. My sim looked better, but was lower impedance, and more parts. This problem you point out is on one axis. What about 20 or 30 degrees vertical? (Yes, it might have a peak that is audible, but in my sim, it would likely have a null. ) These speaker will sit on the floor. There is no sweet spot listening position. Yes, you or I could do better with actual measurements, however all my sims were worse impedance wise than his. The problem you see may improve with the coil change I suggested. I wish that PCD could sim this xo. I tried to download this software, and my pc said it was a virus, so, I dropped the idea. I played with this sim yesterday for several hours, just for fun. Everything that made it measure better, reduced the impedance. Maybe that doesn't matter all that much.

                                Imagine walking around the room while these play. The off axis response of the woofer, and mid will be different. Summations, and AC will be out the window too. All that really remains is tonal balance, and not blowing anything up.

                                Comment

                                Working...
                                X