Announcement

Collapse
No announcement yet.

1st order inductor doing F all.

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

  • 1st order inductor doing F all.

    Hi, just got back to finishing an old project with Alpair 12PW- I made a new crossover and noticed the coil wasn't rolling off enough, so I stripped it down to test and added another coil in series, up to a value approx 1.15mh. This should mean a crossover point around 1000 hz or so. Still seems to be going strong up to 5 kHz and more..realise it's only a 6db roll off but it shd be doing more than it is. What am I doing wrong 😅 There's a little peak around 3khz- when I do a youtube test sweep, it's still pretty audible.
    Last edited by bill poster; 12-26-2016, 12:12 AM.

  • #2
    Hi Bill

    That's about the size of it.

    Nothing unexpected here.

    ​A resistive load fed by an inductor would indeed exhibit a first order roll off with a corner frequency determined by the inductance and load resistance.

    ​Your speaker driver however does not have an impedance vs. frequency function that is either resistive or flat and an inductor in series with it will interact with it's complex impedance and the result you will have to measure or else compute with software from measurements of the drivers impedance and SPL vs. frequency transfer functions.

    ​What you're experiencing is expected because the drivers impedance rises with frequency just like your inductor.

    ​Imagine for a moment you fed an inductor from another inductor and measured the result.

    ​The signal upon the inductor being fed by an inductor wouldn't fall with frequency.

    ​In fact it would be reduced by a constant ratio determined by the ratio of the two inductances.

    ​That's what's going on with your inductor and driver.

    ​A driver isn't a full inductance at high frequency but it is pretty inductive causing what you are observing.
    Last edited by daryl; 12-14-2017, 10:52 PM.

    Comment


    • #3
      Thanks Daryl. Confusingly the first crossover I put together seemed to have a large dip right in the upper mids. V similar but different values. ie 1st on Alpair and 2nd on tweeter with L pad

      Comment


      • #4
        Cd it be the original L Pad was attenuating the impedance (and the lesser padded new one isn't?)

        Comment


        • #5
          Ok decided to go for zobel. The impedance at x/o is up by 10 ohm so shd I put 8 or 10 ohm in the calculator? The Ve is UH (49.3)so I converted it to MH (0.05)and stuck it through a zobel calculator. So it gives me a 0.5uf cap and 10 ohm resistor. Not sure if I got the value correct on the cap or not.

          Comment


          • #6
            Is this a 2-way w/a desired Fc around 1000Hz then?

            Comment


            • #7
              Sorry Chris, not v clear by me. Wanted to stick to a simple x/o and roll off the woofer 1st order to avoid bsc and the tweet 2nd. I know it's probably not ideal. The tweet would cross around 3khz. Estimated a zobel circuit of approx 7.5 ohm resistor and 8.5 uf cap

              Comment


              • #8
                Bill, here are the traced files and 3K x-o inductor. Fprawn tracer is idea for this this work. Give us the tweeter and baffle and we can get you something even closer.

                Alpair-12PW-freq.frd

                The zma file is bad...
                Alpair-12PW-imp.zma
                Attached Files
                Last edited by jhollander; 12-27-2016, 08:14 AM. Reason: Bad impedance file
                John H

                Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

                Comment


                • #9
                  Hey John, I did the same thing . . . BUT

                  something doesn't jive between the linked Z-plot and Mark Audio's T/S specs for this driver.

                  You show a Zmax around 32n @ 36Hz, and a Zmin (should be close to Re) just over 10n at 200Hz.
                  The specs show an Fs @ 38Hz, and an Re (should be close to Zmin) around 5.5n.

                  The plot's Y scale looks a bit wonky, so I set the Y-axis to "log", and put markers @ 10 and 40 ohms (w/FPTrace). With the distance the same between 80/40n, and 40/20n, AND 20/10n, THEN the bottom of the scale (equivalently) SHOULD be 5n (not zero). There's kind of a "loose" marker of "6" hanging out there??

                  Anyway, MY plot shows a Zmax of 25n @ 37Hz, and a Zmin of 9n @ 200Hz. But . . .

                  when I get some unexpected (oddball) data from a Z-plot like this, I'll dump the parms into WinISD and look at a Z-plot THERE.
                  WinISD shows a Zmax of 20n @ 39Hz, and a Zmin near 5.5n @ 1.3kHz.

                  (So, we need more reliable data, at least data that's not in conflict w other supposedly good data.)

                  In any case, this driver has a very low Le value, which means a Zobel won't do much 'cause the inductive rise just isn't that pronounced.
                  By sim (using my STILL BAD z-trace) I get a flattening Zobel of 30n w/about 5uF of capacitance.
                  W/NO Zobel (just a 1.15mH series coil, 0.9n DCR) I'm seeing about -3dB down near 1kHz, and -6dB around 2k (-8dB @ 3k).
                  W/MY Zobel, there's no change at 1kHz, and -6 is around 1.8kHz (-10dB @ 3k).
                  W/bill's Zobel (7.5n 8.5uF) and his 1.15mH coil set, response is down about -12dB @ 3k.

                  To know where we stand, I think that driver needs a DATS sweep run on it (free air OK, but in-box would be even better).
                  Last edited by Chris Roemer; 12-26-2016, 07:54 PM.

                  Comment


                  • #10
                    Chris agreed, my trace is obviously bad. The Mark audio PDF impedance graph is also bad considering the Re value of 5.4. Unibox and response modeler agree with a lower Zmax, 20 ohms
                    John H

                    Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

                    Comment


                    • #11
                      Originally posted by jhollander View Post
                      Bill, here are the traced files and 3K x-o inductor. Fprawn tracer is idea for this this work. Give us the tweeter and baffle and we can get you something even closer.

                      [ATTACH]n1311529[/ATTACH]

                      [ATTACH]n1311530[/ATTACH]
                      Thanks for this. I'm using the viawave RT850 tweets for this build, baffle is 98x26cm.

                      http://viawave.ru/en/
                      Seeing as the tweet is flat to 1500hz, wd a 3khz LR2 be an alternative?

                      Comment


                      • #12
                        Not likely you could do an LR2. Via Wave recomended second order electrical plus driver roll off is near 4th order.
                        John H

                        Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

                        Comment


                        • #13
                          Playing around this morning, added baffle in, used derived woofer impedance file. Not sure about the viawave impedance. The mark audio seemed lumpy on the 10 inch baffle. I guess they are optimized for full range. I notched out the 450 Hz hump. This seems like it should be measured as well.

                          John H

                          Synergy Horn, SLS-85, BMR-3L, Mini-TL, BR-2, Titan OB, B452, Udique, Vultus, Latus1, Seriatim, Aperivox,Pencil Tower

                          Comment


                          • #14
                            Ok got a zobel hooked up but didn't see the reply- thanks so much, will look now and digest!

                            Comment


                            • #15
                              The baffle width of 10" seemed the minimum - is the hump at 450 is a mix of baffle step or diffraction and driver response? Looks like it really needs that notch. Guessing the bumpy response from 2-4khz doesn't make it any easier.

                              Thanks so much for taking the time to do this, really appreciate it, so the least I can do is finish it! Looks like my design execution is flawed but I'll stay positive

                              Comment

                              Working...
                              X