What to look for when measuring speakers

Collapse
This topic is closed.
X
X
 
  • Time
  • Show
Clear All
new posts
  • CraigJ
    Senior Member
    • Feb 2006
    • 518

    What to look for when measuring speakers

    O.K., I don't know where this post belongs, but I'll post here since it's my Phoenix clone. I'm hoping to turn it into a "What to look for when measuring speakers" discussion. I recently picked up a calibrated mic from Herb, and now I'm really dangerous. Here is my current frequency response (below 250 hz turned off)

    Click image for larger version

Name:	Picture1.webp
Views:	16
Size:	56.3 KB
ID:	943281

    At the risk of sounding less than informed, other than "flat", what exactly should I be looking for? Should I put the speaker on a lazy susan and average the 0-30 degree responses? Speaker was placed in center of the room, with extended mic on one meter from center of speaker. Drivers are two SS 8545s and Jed's XT25 double magnet in a MCM waveguide with RS28 in back.

    Thanks for the assistance.

    Craig
    Last edited by theSven; 27 June 2023, 21:05 Tuesday. Reason: Update image location
  • TacoD
    Super Senior Member
    • Feb 2004
    • 1078

    #2
    Some things I look for when doing measurements.

    Frequency Response,
    Measure driver + x-over and compare with simulated response. This feedback loop will help to interpret the simulations of for example lspCAD.

    Another thing I do is measuring horizontal and vertical response on various angles, then you can see how good the x-over works (right x-over point with regard to beaming/ diffraction/ floor bounce). Also do not forget to measure on different loudspeaker distances, especially important for larger loudspeakers, this will tell you if the drivers are summing correctly (no large dips, other anomalies).

    Do not forget impedance measurements, look for wrinkles in the curve (possible resonances which are not suppressed).

    Comment

    • ThomasW
      Moderator Emeritus
      • Aug 2000
      • 10934

      #3
      I showed your post to Jon just before he climbed into his Element to drive back home. Jon said "he needs to read up on doing impulse response tests"...... :W

      IB subwoofer FAQ page


      "Complicated equipment and light reflectors and various other items of hardware are enough, to my mind, to prevent the birdie from coming out." ...... Henri Cartier-Bresson

      Comment

      • CraigJ
        Senior Member
        • Feb 2006
        • 518

        #4
        Thanks for the new thread Thomas. Pretty sure I'm not the first and only "beginner" that is using a dxc as crossover and implementing a mic and software to improve the speaker's sound. I also appreciate the heads up on researching more regarding impulse response tests.

        TacoD, advice much appreciated. From my short memory, JonW had an excellent thread regarding speaker design when he built his current finished speakers. When I measure from different points, how critical do I get when the response changes? Heck, I can fumble over the mic, reset everything, and get a different response....I can just hear CJD saying "don't move the mic until you are completely finished measuring".

        Again, thanks for the excellent advice. Perhaps I should head on over to the library and get a good book.

        Craig

        Comment

        • dsrviola
          Senior Member
          • Oct 2007
          • 119

          #5
          I hope this discussion goes further. I'd like to know how people are measuring off axis response, and how best to makes sure the work you do on the crossover is also optimizing vertical and horizontal off axis issues. Also wondering about how measuring and designing for an MTM (or in my case WMTMW) is different than a standard two way. In my particular listening set up, consistency of vertical response is a priority for me. I'm wondering what I need to be looking for as me and my bud start working on a new crossover design.

          Comment

          • JonW
            Super Senior Member
            • Jan 2006
            • 1582

            #6
            Hi Craig,

            Oh boy, you’re in measurement territory now. Look out! It took me a while to get up to speed on this topic. Many of my troubles- and all the great help people here provided me with- are shown in the Spassvogel thread:

            Probably no need to wade into that, though. Maybe we can help you here. I’ll help as much as I can. In the spirit of giving back (or was that to confuse you completely... :W

            For starters, I might recommend just trying to mimic the data from a manufacturer. Take one driver, put it in your cabinet (or a large baffle), set up the system, and see if you can get something familiar looking. It probably won’t look exactly the same but if it looks familiar, you know you’re on the right track.

            Start out with frequency response only. It’s easier and intuitive with regard to what it should look like if all is well. When you get something looking good, try impedance. You’ll want impedance for the final crossover design. I’d probably say skip distortion for the moment.

            Make sure your levels are all OK- that you’re not overloading the sound card or that you’re not having too low an MLS chirp.

            JustMLS and the other measurement programs that I know of require you to build a little cable rig with a resistor and such. You don’t just wire up a driver straight and then measure.

            To protect tweeters from low frequency thumps when you turn on amps, etc. it’s recommended to put a large cap in the wiring. A really large value (300+ uF) may not influence the tweeter output where you are concerned with but should protect the tweeter from blowing should something bad happen. Mids and woofers probably don’t need that.

            Start with your microphone 1 meter from the driver, pointing straight at it. Gate the measurements such that you only collect data for around 7 milliseconds after the MLS chirp. Otherwise you’ll get the room bounce modes which you don’t want at this point. Maybe adjust the microphone-to-driver distance in the program to get the phase to look OK. If that all looks good, disconnect the driver and connect the next one. (Take out the big tweeter cap if you had one.) Do not touch anything else (microphone, sound card levels, any parameters in the software, nothing). Then take a measurement of the next driver. Most people start with the tweeter and then move down from there.

            When you want to get a little fancy, try a measurement with all the drivers connected at once. If you have a preliminary crossover (or when you’re using your DCX) you can also put that in the loop. Actually, I might do it both with and without the crossover to get a comparison. If you’re all flat with the crossover and all the drivers going, that’s where you want to be- for starters. Voicing may come next.

            If you want off-axis information (it never hurts to have it), rotate the speaker maybe 30 degrees. Or move the microphone to be 1 meter from the driver(s) and 30 degrees off axis. Measure as you did already. You will likely see the high frequency tweeter information drop off, say, above 10,000 Hz. As you tweak your crossover you may want to input the off-axis measurements now and then, if it’s important for your goals.

            I haven’t yet had a chance to try Fuzzmeasure (even though I’m a Mac guy), which it looks like you’re using. Maybe download the JustMLS section of the LspCAD manual/tutorial and try to follow along with that. It’s not a great tutorial, but some of the numbers there may help you out.

            Does any of that help?

            -Jon
            Last edited by theSven; 27 June 2023, 21:05 Tuesday. Reason: Update htguide url

            Comment

            • CraigJ
              Senior Member
              • Feb 2006
              • 518

              #7
              Hi JonW,

              I was hoping you'd pop your head out of the lab and give some advice. Thanks for the very informative info. As mentioned, I think we learned more about speaker design through your Spassvogel project than any other project I can remember. At this point, I'd like to stick with the dcx and am "fine tuning" three of my current projects. I'm a true believer that my speaker system has dramatically improved with the assistance of measurement equipment thus far. With that being said, and making the assumption that my projects have no major faults, I want to get better at measuring speakers with the goal of improving sound quality.

              Regarding the actual measuring, speaker is placed one meter from mic, with mic extended on a diy extension. Mic is at tweeter level for mtm, and midrange level for the wwmt. O.K. so far? From here, I can record the measurement as seen above; fairly flat with added eq. When I start turning the speaker on a lazy susan, the FR is going to change and how do I average the results, or perhaps what do I average? 7 milliseconds for data chirp, got it, thanks, think I used 347.83 milliseconds, good for an auditorium. My masters is not in math, and I know there is an equation around here somewhere for reflections so that saved hours of research.

              Regarding voicing, I take that as matching the volume of the drivers...at this point I'm going with a fairly straight line and can adjust output level with more listening time.

              The manual for JustMLS will be downloaded as soon as I get on my Intel Mac. And yes, you've been a big help!

              Craig

              Comment

              • Jed
                Ultra Senior Member
                • Apr 2005
                • 3617

                #8
                Thomas and Jon are right, you need to set your gate from where the signal starts to just before the reflections in the room are to minimize interference in the measurements. The length of the gate should be around 6ms after the signal begins-- in most conditions. Then, if you want to get fancy you can do a splice using a long gate nearfield response spliced with the top end of another farfield short gated measurement: so you can get more low end information to come close to anechoic like conditions, yet still have accurate phase and top end response. JustMLS allows you to apply BSC to that measurement to get an accurate farfield representation because the nearfield measurement will NOT show the baffle step do to the close proximity of the microphone to the cone. Full gates are still good to do without a splice, so that you can compare the general amount of BSC that is needed; however, a long/full gate at 1+ meters will have too much room interference in the midrange and bass because of the room boundary reflections at those frequencies to be useful as final data. Therefore, farfield with short gate, farfield with a long gate, and splice of nearfield/farfield with BSC are all usefull in the measurements before deciding if the overall measurements used in the crossover software are representative of what the drivers are doing in reality.

                Or, one could go to a costly anechoic chamber.

                Jed

                Comment

                • Dennis H
                  Ultra Senior Member
                  • Aug 2002
                  • 3791

                  #9
                  The deal with gating is you look at the impulse response and you can see the first reflection as a blip a few msec after the main impulse. Put your gate before that blip and you have just the speaker's response with no reflections. The catch is the LF response is meaningless at frequencies below 1/gate-time. So you need to look at both the gated response for the high frequencies and the ungated (or long-gated) response for the low frequencies, understanding that the LF is mucked up either way. It's mucked up by the math in the gated response and it's mucked up by reflections in the ungated response (shrug) .

                  About measuring off axis, you don't want to see sudden changes in the response when you move the mic a little bit. If you do, something is wrong, most likely in the crossover.

                  Edit: Jed and I were typing at the same time and said about the same thing.

                  Comment

                  • Saurav
                    Super Senior Member
                    • Dec 2004
                    • 1166

                    #10
                    How do you arrive at the 6ms value? Assuming the mic is 1m away from the driver, and the mic/driver are 1m above the floor, and the floor is the closest boundary... the reflected sound travels ~2.2m, so it's traveling 1.2m further than the direct sound, which comes to about 3ms.

                    Either I did the math wrong, or some of my assumptions aren't valid.

                    Comment

                    • Jed
                      Ultra Senior Member
                      • Apr 2005
                      • 3617

                      #11
                      Originally posted by Saurav
                      How do you arrive at the 6ms value? Assuming the mic is 1m away from the driver, and the mic/driver are 1m above the floor, and the floor is the closest boundary... the reflected sound travels ~2.2m, so it's traveling 1.2m further than the direct sound, which comes to about 3ms.

                      Either I did the math wrong, or some of my assumptions aren't valid.
                      One would have to look at the impulse graph to determine the exact figure, as I mentioned in the other part of my writeup. It is also good practice to elevate the speaker more than 1m off the ground, preferably outside in quiet conditions. How much reflection one determines acceptable in the gated window is another matter, in which case an anechoic chamber would be nice to avoid any limitations in the measurement conditions available versus how much low end response is needed to get enough measurement resolution below the baffle step.

                      Comment

                      • Saurav
                        Super Senior Member
                        • Dec 2004
                        • 1166

                        #12
                        Thanks. I always got really short gate times when measuring in-room. Outside and elevated, I could get 6ms.

                        Comment

                        • Face
                          Senior Member
                          • Mar 2007
                          • 995

                          #13
                          How about laying layers of insulation or bass traps in front of the speaker? My floor bounce goes away completely after laying a GIK 244 on the floor between my speaker and mic.
                          SEOS 12/AE TD10M Front Stage in Progress

                          Comment

                          • Jed
                            Ultra Senior Member
                            • Apr 2005
                            • 3617

                            #14
                            Originally posted by Face
                            How about laying layers of insulation or bass traps in front of the speaker? My floor bounce goes away completely after laying a GIK 244 on the floor between my speaker and mic.

                            I do the same for in room measurements. Sort of a pseudo anechoic solution.

                            Comment

                            • JonW
                              Super Senior Member
                              • Jan 2006
                              • 1582

                              #15
                              Hi Craig,

                              You're getting good help here from Jed and Dennis. They know more about this than I do so listen to them first. I found it a little rough getting to my first successful measurements. So with some patience and the folks here you'll soon be on your way.

                              Originally posted by CraigJ
                              As mentioned, I think we learned more about speaker design through your Spassvogel project than any other project I can remember.
                              Nice to hear that the thread and project was a help to someone besides myself. Thanks. The learning process is posted there, as is the final design. Although I doubt anyone else will ever build the speakers (expensive, difficult cabinets to make, Zaph has a speaker with the same drivers, etc.).

                              Yes, I think you will find that the measurements will yield a much better speaker. You still have to use your ears, though. (See voicing, below.)

                              Originally posted by CraigJ
                              Regarding the actual measuring, speaker is placed one meter from mic, with mic extended on a diy extension. Mic is at tweeter level for mtm, and midrange level for the wwmt. O.K. so far?
                              Yes. :T The only thing I might do different is put the mic at the tweeter level for the WWMT as well. Assuming that you are building them to have the tweeter at your ear level. I figure you might as well put the mic as close to where your ear will be and match all the phases, etc. to that. But we could also argue that it's not a big deal if your mic is at 1 meter and a 7 ms gate (see what Dennis and Jed said) when you really listen to the speakers 3 meters away and without a gate.

                              The 7 ms (or 6, etc.) is a rough place to start that seems to work for most people. You may need to shorten it a little, even down to 2 or 3 ms. See what Saurav, Jed, and Dennis are saying.


                              Originally posted by CraigJ
                              From here, I can record the measurement as seen above; fairly flat with added eq. When I start turning the speaker on a lazy susan, the FR is going to change and how do I average the results, or perhaps what do I average?
                              Keep the on-axis and off-axis measurements separate. No averages. When you build your xover model start with the on-axis data. As things come together, now and then plug in the off-axis data and see how it looks. If things are going well, the off-axis data and xover will still look OK (except for the high frequency loss). If the on- and off-axis models look dramatically different it might indicate a problem. But if you're making a speaker for mostly on-axis performance, the off-axis data are really of secondary importance, as I understand it


                              Originally posted by CraigJ
                              Regarding voicing, I take that as matching the volume of the drivers...at this point I'm going with a fairly straight line and can adjust output level with more listening time.
                              I'm not totally sure what voicing is to everyone else but I can tell you what I did... But I was making a passive xover so it may be pretty different for you. I got my measurements, modeled a crossover in LspCAD and then clipped (not soldered) together the various caps, resistors, and inductors to have a listen. It was decent. Then I went about changing the value of each component, one at a time. And listened to see if it got better or worse. For example, if I had a 10 uF cap I tried a 12 and an 8. If, say, the 12 was a little better I then tried a 14, etc. It took a while but it definitely made for a much better sounding speaker. Then I plugged all the new values back into the crossover model and it still looked good. Slightly different looking in the models (neither better not worse) but definitely better sounding.

                              Here was a big thing for me: Getting the phase to overlap between the 2 drivers. When I got a nice phase overlap, that's when the speaker really came together. Phase is from the crossover (not sure how it works when active) but it's data you need to be sure to collect at this stage.


                              Originally posted by CraigJ
                              And yes, you've been a big help!
                              That's too bad. Everything I said was wrong and intentionally misleading. :W

                              Comment

                              • penngray
                                Senior Member
                                • Sep 2007
                                • 341

                                #16
                                Originally posted by Face
                                How about laying layers of insulation or bass traps in front of the speaker? My floor bounce goes away completely after laying a GIK 244 on the floor between my speaker and mic.
                                Thanks for the suggestion!

                                Comment

                                • dsrviola
                                  Senior Member
                                  • Oct 2007
                                  • 119

                                  #17
                                  Measurements cont'd

                                  Are there any specific concerns to watch out for with an WMTMW design when taking measurements. My (center to center) mid to mid spacing is 9.5" and my woofer spacing is 32.25". Wondering what the correct distance should be for making measurements without the dreaded ground bounce ruining everything while still being able to determine driver integration. How low is necessary to cross from the woofer to the mid and not run into spacing issues. There's 11.5" c to c between each mid and woofer. I'm hoping to cross over to the mids no lower 300Hz-400Hz.

                                  Drivers: Accuton C12, C79, and C95. (old part numbers)

                                  Also, what should I expect as far as vertical lobing issues and they're effects on the sound? And do different types of crossover topologies deal with this particular effect better than others, and what are their trade offs.

                                  I should be able to measure the speakers outside with the bottom of the cabinet approx. 36" off the ground. That would make the tweeter about 57" above the ground.

                                  Too many questions for one post. Sorry.

                                  Comment

                                  • JonMarsh
                                    Mad Max Moderator
                                    • Aug 2000
                                    • 15209

                                    #18
                                    Typically a WMTMW would have the primary design axis on the tweeter axis. The behavior off this axis depends greatly on the chosen crossover points, the crossover slopes and phase integration in the crossover (for example, LR4 networks (standard and slightly asymmetric) will usually produce good response plots on axis and a deep null, but have issues with vertical lobing and total power response that some designers will prefer to address with something like a B3 alignment that has phase quadrature in the crossover region and a much taller vertical lobe if the crossover frequency and driver spacing are chosen appropriately).

                                    As to ground bounce, this is a factor that should/must be taken into account in the basic system design; unless you're going to listen to a speaker outdoors jacked up off the ground or with special sound absorbers on the foor, it might not be a bad idea to give this some thought in developing your system concept, especially the work of Roy Allison.

                                    Generally, if you can't follow Allison's recommendations explicitly because of your preferred/desired configuration, a taller source in the affected frequency ranges helps distribute the effects so that you don't get as large a null in one frequency area at a specific listening distance.
                                    the AudioWorx
                                    Natalie P
                                    M8ta
                                    Modula Neo DCC
                                    Modula MT XE
                                    Modula Xtreme
                                    Isiris
                                    Wavecor Ardent

                                    SMJ
                                    Minerva Monitor
                                    Calliope
                                    Ardent D

                                    In Development...
                                    Isiris Mk II updates- in final test stage!
                                    Obi-Wan
                                    Saint-Saëns Symphonique/AKA SMJ-40
                                    Modula PWB
                                    Calliope CC Supreme
                                    Natalie P Ultra
                                    Natalie P Supreme
                                    Janus BP1 Sub


                                    Resistance is not futile, it is Volts divided by Amperes...
                                    Just ask Mr. Ohm....

                                    Comment

                                    Working...
                                    😀
                                    😂
                                    🥰
                                    😘
                                    🤢
                                    😎
                                    😞
                                    😡
                                    👍
                                    👎
                                    Searching...Please wait.
                                    An unexpected error was returned: 'Your submission could not be processed because you have logged in since the previous page was loaded.

                                    Please push the back button and reload the previous window.'
                                    An unexpected error was returned: 'Your submission could not be processed because the token has expired.

                                    Please push the back button and reload the previous window.'
                                    An internal error has occurred and the module cannot be displayed.
                                    There are no results that meet this criteria.
                                    Search Result for "|||"