No announcement yet.

T/S Parameters in ARTA LIMP vs REW - preferences?

  • Filter
  • Time
  • Show
Clear All
new posts

  • T/S Parameters in ARTA LIMP vs REW - preferences?

    When the DATS v3 was out of stock I built a jig with the 100 ohm resistance and have tried it in the same conditions using the added mass method in both REW and ARTA LIMP. Of course there are differences between the programs; some individual parameters are very close and others are not. For those who have tried, have you found one to be more accurate than the other? Or does it even matter as long as I pick one because the goal is to help with simulations in design before moving to next steps in the process? And BTW, I see what folks mean about manufacturer claims. I've been measuring 4 W130s-4s and the Fs is between 62 and 65 Hz. Manufacturer claims 50 Hz.

  • #2
    If you are correctly calibrated using LIMP, then you should see almost no differences between the two measurement systems. What specification is showing different?
    Don't listen to me - I have not sold any $150,000 speakers.


    • #3
      I'll paste them in below.


      fs 64.5 Hz
      Qms 3.912
      Qes 0.656
      Qts 0.562
      Fts 114.8
      Mms 6.74 g
      Cms 0.903 mm/N
      Rms 0.699 kg/s
      Vas 6.71 litres
      Bl 3.903 Tm
      Eta 0.26 %
      Lp (1W/1m) 86.35 dB
      Dd 9.60 cm
      Sd 72.4 cm^2
      Added mass measurement: Unit 3 - mm - same ART
      Added mass 30.000 g
      Air temperature 23.0 C
      Air pressure 1013.25 mB
      Air density 1.1919 kg/m^3
      Speed of sound 345.0 m/s

      Fs = 64.45 Hz
      Re = 3.70 ohm[dc]
      Le = 119.09 uH
      L2 = 202.79 uH
      R2 = 5.83 ohm
      Qt = 0.53
      Qes = 0.61
      Qms = 3.79
      Mms = 6.54 grams
      Rms = 0.690370 kg/s
      Cms = 0.932975 mm/N
      Vas = 6.87 liters
      Sd= 72.38 cm^2
      Bl = 3.998868 Tm
      ETA = 0.29 %
      Lp(2.83V/1m) = 90.06 dB

      Added Mass - Constant Bl Method:
      Driver unbaffled
      Added mass = 30.00 grams
      Membrane Diameter= 9.60 cm


      • Colonel7
        Colonel7 commented
        Editing a comment
        And the rest of the parameters in REW if these are of interest. I just cut and pasted the "simplified" above

        Dated: Jul 24, 2021 10:26:41 AM
        From measurement Unit 3 - free air
        Zmin 3.86 ohm
        fmin 296 Hz
        f3 1328 Hz
        Le(f3) 0.277 mH
        Motional impedance parameters
        R0 21.81 ohm
        CMES 442.6 uF
        L0 13.753 mH
        beta 0.0000
        omega0 601.0
        Blocked impedance parameters
        RDC 3.70 ohm
        dR -0.04 ohm
        Re 3.66 ohm
        Leb 2.2 uH
        Le 0.558 mH
        Rss 100000.0 ohm
        Ke 0.0528 S-H

    • #4
      It's Le that looks very dissimilar. Now that I look at the others they seem pretty close. I've taken so many measurements over the last 2 days that they started out much wider. I first used about 44g, then 20, then 30. The 30g ones are the closest, probably because it was most uniform in coverage and more importantly I I woke up to the fact that I needed to switch between the programs for each free air and then added mass measurement, so as not to reposition the mass at all for each unit.


      • #5
        I've got ARTA (and use if for FR) but never used the LIMP module for T/S parms (or Z-data).
        That's because I've had DATS(and WT3) for many, many years now.

        I CAN tell you that DATS is set up to make a "standard" Le measurement at either 1kHz (used for woofers, mids, and inductors) OR 1O kHz (which is used almost exclusively on tweeters). I'd research to see if YOUR programs are using the same "standard", as that COULD account for your variation.
        Your "main 3" parms (: Qts, Fs, and Vas) look awfully close to me - easily within the realm of "non-repeatability".


        • #6
          Thanks. I'll look more into the manuals and post info on Le if there's anything to report.


          • #7
            I always figure +- 10% is close enough for sim work. Did you break the drivers in before measuring them?


            • #8
              I did for 3 hours each but only with music. We had guests staying over and they did not appreciate pink noise when I started. Now that I know the Fs for each unit (and no guests!) I'm tempted to give them another few hours of sine waves close to the Fs, then remeasure.


              • #9
                I took a more in-depth look at the manuals and how Le is treated differently by REW and ARTA. Bottom-line is that I'm not going to worry about the differences because they are arrived at differently. I'll summarize below but bear in mind any errors are due to my own interpretation, so more knowledgeable folks feel free to correct me.

                REW's "simplified model components" (not frequency dependent) are for users that will be using circuit simulators that don't incorporate frequency dependent values for blocked impedance. ARTA's optional bells and whistles are REW's simplified model. ARTA does incorporate non-linear least squares estimation to lessen the error magnitude and what he describes as "lossy inductance", while it's not clear what REW does to reduce error in its simplified model. Without the extra options of checking the radio buttons for LSE and calculating lossy inductance, ARTA uses linear regression to minimize error.

                For frequency dependent values, REW incorporates additional parallel models of Le and Ke derived from a 2008 JAES paper (Thorborg and Unruh) on modeling semi-inductance, with the goal of accounting for the transition of the voice coil from a conventional inductor at low frequencies to a semi-inductor at higher ones (compared to what Chris cited for DATS above taken at 2 individual frequencies). It also uses another Thorborg and Futtrup JAES paper model the effect of electrically conductive material in the magnet system.

                While REW is more complex in its modeling beyond its simplified model and ARTA's extra options, there's not a way to gauge whether it is any more accurate. I don't have knowledge of how the calculations are done in DATS so have no comment on that.

                Separately re: T/S parameters in general, In the most recent update of the ARTA manual the author states T/S parameter values such as Vas and MMS can have an error rate of 20-30% due to the limitations of T/S being a simple linear model and transducer systems are non-linear.

                Hope this helps.


                • #10
                  At any rate, Le is basically not a factor in "box design" (bottom-end rolloff), and while it DOES play a part in middle and upper-range driver response, it's affects are typically overshadowed by other aspects of cone, suspension, and motor design. THAT's why we make use of .frd files when we sim driver systems (and don't just use a simplified Le value to try to predict the top-end rolloff of a driver).


                  • #11
                    First, you've used an added mass that is rather large. Aim for an added mass about equal to the Driver Mms, so about 6 grams by your information provided. You should see a change in Fs by about 25%. This will give you a greater degree of accuracy for the important Vas value.

                    Le on it's own is a pretty useless specification, especially since you have a sweep of the driver impedance already. Le is just too simple of a value, speaker motors are more complex than simple inductance. I can't speak for REW, but ARTA/LIMP includes a few different "lossy inductor" models to characterize the "Le" in a more complex and accurate manner. You can read about them in the manual under section 5.4.

                    Once you've run the T/S parameter calculation, close the window and press F3 to compare the result of the lossy inductor model against the measured impedance. That should give you all you need to know how accurate it is.
                    "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


                    • Colonel7
                      Colonel7 commented
                      Editing a comment
                      Thanks - I will do it again with the lighter added mass

                  • #12
                    This document better describes the lossy inductance models in ARTA.

                    "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


                    • #13
                      I re-did the measurements this morning with 6.25g and on ARTA the F3 tracks the measured impedance. Yesterday I ran the drivers an additional 75 minutes with a sine at 62 Hz and remeasured at 30g, before I realized that 20-30% would put me at approx 6g for added mass (~Mms). The cones are much more compliant to pushing and flex. The FS really hasn't changed but Vas has risen with REW. I've cut and pasted the differences in Fs-Qt-Vas with both ARTA and REW. Haven't had an opportunity to look at the other parameters. Thanks to everyone for the continuing advice as I'm new to this but expect to learn from my errors. Edit: "compliant" was a poor choice of words considering its relation to Vas - I meant the cone isn't as stiff. For lack of a better description it is easier to move up and down by hand and feels rubbery.
                      180 mins music + 75 mins 62 Hz Sine Same break-in music and sine
                      Visaton W130s-4 ARTA 30g ARTA 30g ARTA 6.25g
                      Spec Unit 1 Unit 2 Unit 3 Unit 4 Unit 1 Unit 2 Unit 3 Unit 4 Unit 1 Unit 2 Unit 3 Unit 4
                      Fs 50 Hz 63.96 64.86 64.45 63.78 63.6 64.67 63.88 63.19 63.83 64.96 64.27 63.58
                      Qts 0.45 0.52 0.53 0.53 0.54 0.52 0.53 0.53 0.54 0.53 0.54 0.54 0.53
                      Vas ft3 0.49 0.255 0.273 0.243 0.244 0.255 0.24 0.248 0.245 0.224 0.219 0.225 0.22
                      + 75 mins 62 Hz Sine Same break-in music and sine
                      REW 30g REW 30g REW 6.25g
                      Fs 50 Hz 63.4 65.3 64.5 64.5 63.5 64.3 63.9 63.2 63.6 64.6 64 63.3
                      Qts 0.45 0.569 0.566 0.562 0.562 0.597 0.609 0.598 0.623 0.612 0.611 0.612 0.645
                      Vas ft3 0.49 0.209 0.249 0.249 0.237 0.225 0.233 0.236 0.227 0.243 0.255 0.249 0.247
                      Last edited by Colonel7; 07-28-2021, 02:46 PM.


                      • #14
                        Take both your results and model them in a cabinet to see a real comparison.

                        T/S is sensitive to measurement voltage difference, ambient temperature and himidity difference...but you will find that in general one number goes down and another goes up, and the net change in a cabinet is pretty much zilch.
                        "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


                        • #15
                          Yes, on to the next steps. Thanks again for pointing out correct added mass. The drivers seem to be very consistent among units although not necessarily to spec. I bought these drivers to learn as a first experience with design and building my own cabinets, building a number of prototypes such as TM, MTM, floor-stander 2 and maybe 2.5 way, and an eventual 3-way center. The Visatons were available and on clearance which fit the bill, and the published specs seemed to support sealed or ported for experimentation. Based on the measurements thus far the cabinet will need to be considerably bigger for sealed compared to Visaton's suggested alignment. I'll still try it for one of the TM or MTMs. This will keep me busy for many months. I know I'll mess up a lot but the intention is to learn along the way, avoiding the frustration of messing up with nicer drivers and screwing up cabinet finishing. I plan on experimenting with that as well. Really enjoying the process of it.