No announcement yet.

REW Impedence measurement help

  • Filter
  • Time
  • Show
Clear All
new posts

  • REW Impedence measurement help

    Well after my failure using a MiniDSP and 2 TDA7492s, I figured, well I can just do a passive design.

    So, I built an impedence jig using my own sketch.

    It is based on Hong's excellent guide for using ARTA. The REW manual has the same schematic so I figured it should work. Green is the ground wire.
    Here's my jig and sketch.

    The dotted line is the ground. I've since re-wired with a separate ground from each RCA plug going to the black speaker post just to make sure all the connections worked as well as soldered all the connections instead of the test hot glue.

    I do the impedence calibration and get the 100% of reference level and a reading of 100 ohms from 0hz to 20khz, a straight, flat line.

    When I plug in the speaker, I get readings in the hundreds of thousands of ohms in the lower registers and a few thousand ohms as it goes towards 20khz.

    I think the wiring is good, can anyone shed some light on this? I can share some screenshots of REW if that will help.

  • #2
    Did you drizzle glue over your solder joints? Your wires appear to be going to the right places, apart from the white wire at the top which I can't see that it is soldered to the resistor with the other white wire. Apart from that I would question the use of RCA connectors, are you not putting an amplifier in between your soundcard and the jig?
    "I just use off the shelf textbook filters designed for a resistor of 8 ohms with
    exactly a Fc 3K for both drivers, anybody can do it." -Xmax


    • #3
      In this photo i just hot glued the joints after twisting them together tightly. I have since soldered everything together after removing the hot glue. The white at the top is connected to the resistor with the other white wire.

      No amplifier, using a Behringer UCA202 usb soundcard line in/outs and a 100R sense resistor. The notation in Hong's sticky thread and in the REW manual says if you use an amp that you need a different resistor, but just using the line outs would be ok.

      I've tried using the headphone out on the UCA202 as well and get the same results.


      • #4
        I had problems with the UCA202 also. I gave up and got a cheap(ish) Focusrite 2i2. I still have to use the heaphone out with LIMP.
        Last edited by tom_s; 09-07-2017, 09:22 PM. Reason: Missed the last line in his last post
        Co-conspirator in the development of the "CR Gnarly Fidelity Reduction Unit" - Registered Trademark, Patent Pending.


        • #5
          Thanks Tom - that's too bad as I bought the UCA specifically to do these measurements! I am able to hear the sweep through the speaker, so I think it is driving the load fine if that is an indication. Is it? I really have no idea what I'm doing with the electronics side of things.


          • #6
            Afaik the uca has known low frequency crosstalk issues which affects the lf response for both fr and z measurements, e.g. or

            That won't cause readings like you describe though


            • #7
              I have succeeded in taking impedence measurements!!

              The reason it wasn't working was the UCA 222 (not 202) has a specific ASIO driver that needs to be used. I found a setup thing on youtube that mentioned this. This allowed the line-outs to be properly used by REW. So, if you are using a UCA222 or 202, get the ASIO driver file.

              Impedence jig worked fantastically.

              Here's the Peerless/SVS buyout woofer impedence in a sealed box of just shy of 15 litres (slightly over .5 cubic feet), the SB26 STAC tweeter, and the soundcard calibration file. It shows a rolloff below 20 hz but pretty flat through 20khz.


              • #8
                "Ordinarily" (meaning: this is just my op. - not backed by any theory that I know of, only by about 6000 observations over the last 15 yrs or more) any "dynamic" loudspeaker (voice-coil driven, so does NOT apply to ribbons, planars, piezos (or is it piezoes?), etc.) will show an inductive "rise" above Fs (as YOUR measurements do), with the curve below Fs basically "mirroring" that above Fs (for the most part), "flatlining" at some point close to the driver's "Re" value.

                So (not sure what it is about your setup that's NOT properly shaping the Z-curve down low - below 200Hz maybe) I think it would be safe (and prudent) to shape your tweeter's curve so it's flat (around 4ohms) below 200Hz, and to shape your woofer so it mirrors down to about 6.5ohms at the bottom of your scale. In practice, the woofer's Z-curve (below, say, 200Hz) will (should) be shaped more by your box design, and won't really alter any passive filter (unless you're trying to passively roll off the bottom end - which is basically never done). Even using your data "as is" on the tweeter MIGHT NOT have any sig. effect - but probably WILL, IF you'd shoot for just an electrical 1st order high pass.


                • #9
                  The rising low end is the crosstalk issue in the audio device, see the data-bass link for another example (as well as what happens if you use 2 UCAs, one each for input and output, to avoid that crosstalk problem)


                  • #10
                    Chris Roemer and 3ll3d00d , thanks for you observations.

                    Regarding crosstalk, I don't see that in the calibration of the soundcard, is this an issue only with impedence measuring?

                    I won't be shooting for a first order on these tweeters, my aim will be around 1600 4th order LR, as that's what sounded the best to me when running different slopes and crossover points through the MiniDSP. As I don't think I've ever turned up anything past 90db in actual listening, I think the tweeters should be fine. When I crossed higher, I thought I could hear the woofer have some ringing even with a big notch on its resonant peak.

                    On the woofer, would some of that rise below FS be a sign that my box isn't sealed properly? I just have it in a sealed alignment, unibox predicting a roll off around 60hz. I do have a removable baffle on the front and have been struggling to get it to seal properly while looking ok with the seam.

                    I'm hoping to get some measurements done this morning, so I will share those in a new thread later if it all works out.


                    • #11
                      NOPE (on the woofer "rise"). For a closed box, (IIRC) you should see a peak VERY similar to free air (Fs), but somewhat higer in Freq., and not quite as tall (I THINK?).
                      You can model the (boxed) Z-curve in something like WinISD ( ) and watch the Fs peak slide higher (Hz) as the box gets smaller.
                      Your measured Z-curves should pretty much coincide w/the box model.