Announcement

Collapse
No announcement yet.

Measuring effect of port length on box tuning - odd results

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

  • #16
    Originally posted by datinker View Post
    killa said "So your measurements are rolling off faster than the model?"
    YES, that is what I am saying. Your suggestion that one has to measure output from the port is the sort of thing I am suspecting. As I said, I am measuring from the front of the speaker and the port is on the back.
    I think that answers isaeagle's question as well, but I will provide some more detail about the setup. The speaker is sitting on the edge of a table so it has free air around it except for the surface it is sitting on. The area is a bit cluttered so I get quite a few reflection artifacts in my graphs, but that does not affect the low end roll off.

    These are intended to be bookshelf speakers so I suppose it would be more realistic to measure them placed near a wall. That should return most of the port output back towards the listener, I would think.

    Now you have me thinking about the greater distance the bass frequencies have to travel to reach the listener in front. It makes me think ports should really be on the baffle. But then, there is the phase shift business...


    Ports can be on front, back, or side or bottom. They may interact more with a wall behind, them but phase is not a reason to put a port on any surface....

    You need to measure the port output directly, as most of the deeper bass from your speakers will be coming FROM the port and not the woofer. The low point in your impedance sweep, is where the port is putting out the most sound and the woofer should be barely moving.(minimum motion frequency range)

    Comment


    • #17
      killa, thanks for the link to the Jeff Bagby paper. That is just what I need to understand what is going on with the measurements,

      Kevintomb, you said that the low point in the impedance sweep is where the port is putting out the most sound and the cone is barely moving. By low point you must mean the trough between the two impedance peaks, right? I would have guessed that the port was doing all the work at the lower of the two impedance peaks.

      As for my somewhat vague thought about phase, what I was trying to say was two things really. One was just that with a port on the back the sound has to travel an extra 6 or 8 inches to reach the listener in front, a significant offset in time alignment. The sounds have to be out of phase, but I don't know what the practical limit of perceptibility is at low frequencies. It just seems like if the port were on the front this would be a non-issue.
      However, and I think killa may have been suggesting this yesterday, I don't know if the port is out of phase with the cone in any case just due to the fundamental physics of the interaction between the port and the box. I don't understand how that works well enough to know.

      Comment


      • #18
        Originally posted by datinker View Post
        killa, thanks for the link to the Jeff Bagby paper. That is just what I need to understand what is going on with the measurements,

        Kevintomb, you said that the low point in the impedance sweep is where the port is putting out the most sound and the cone is barely moving. By low point you must mean the trough between the two impedance peaks, right? I would have guessed that the port was doing all the work at the lower of the two impedance peaks.

        As for my somewhat vague thought about phase, what I was trying to say was two things really. One was just that with a port on the back the sound has to travel an extra 6 or 8 inches to reach the listener in front, a significant offset in time alignment. The sounds have to be out of phase, but I don't know what the practical limit of perceptibility is at low frequencies. It just seems like if the port were on the front this would be a non-issue.
        However, and I think killa may have been suggesting this yesterday, I don't know if the port is out of phase with the cone in any case just due to the fundamental physics of the interaction between the port and the box. I don't understand how that works well enough to know.
        Not Kevin but yes he he meant the trough in between the two peaks is your tuning frequency.
        As for what I was trying to say about ports and phase. I didn't mean it would affect the output in a negative way. I was trying to explain the port/driver interaction and why the driver response drops off the way it does before tuning. This doesn't really matter for what you are trying to do. It was for a better understanding of what is happening.

        Comment


        • #19
          Originally posted by datinker View Post
          Thanks everyone for the interesting and informative responses. I learned that Fb in a ported box is the midpoint between the two impedance peaks. Which means I have seen pretty much just what djkest said is predicted by model, Fb of ~53Hz. What software are you using to model this? Especially curious since you noted that there is an nasty port resonance to be aware of. I assume that was a modelling prediction?

          As for the question of fill ....
          I prefer Unibox for modeling, it's also freeware. Play around with the fill options and you can see the difference fill has on vented vs sealed. http://audio.claub.net/software/kougaard/ubdwnld.html

          http://techtalk.parts-express.com/fo...khanspires-but
          http://techtalk.parts-express.com/fo...pico-neo-build
          http://techtalk.parts-express.com/fo...ensation-build

          Comment


          • #20
            What's Fill Got to do With It

            I now have the speakers I was asking questions about above back in October all put together. All that is left at this point is deciding how much, if any, stuffing is optimal inside the enclosures, and maybe a last little bit of X/O tweaking. These are intended to be a Christmas present for my son, which is why I am trying to decide about this last question this week. Once the fill is in I can close them up and be done with the enclosures and construction.

            I am attaching 3 pictures to this post so you can see what I am talking about when I say I am working with elliptical enclosures of 7 liter volume. The volume was directly measured by filling the boxes with rice. As you can see in the pictures the ports are on the outside of the box so as not to lose any precious internal volume. The hole in the back is radiused on the inside all the way around and on the outer surface it is radiused only over about 1/4 of the circle on the arc section facing up the port. The front and back walls diverge at about 7 degrees so there are no parallel surfaces. There is also a horizontal brace inside so the top and bottom parts of the ellipse are of unequal size. The SB13 woofer is surface mounted and the BC25 tweeter is flush mounted which in combination with the backward leaning baffle makes the acoustic center depths of the two drivers almost perfectly even relative to a listener on-axis. The cross-overs are mounted outside of the enclosures so they can be tweaked all I want without wearing out any enclosure screw holes.

            Now for the question of fill - I made some measurements with my DATS 2 of the woofer in the fully stuffed enclosure and the empty enclosure. I made impedance sweeps for both cases with the ports open and with the ports blocked with a wad of paper towel tightly jammed in the throat. I have attached a 4 panel figure showing the DATS sweeps of those 4 tests.

            What I am seeing in these data is that the stuffing does what is expected in the blocked port case, it lowers the resonance peak by about 10Hz. There is only a tiny change in the height of the peak in the stuffed vs. unstuffed case with the port blocked.
            With the port open you see the typical twin-peak impedance curves. In the stuffed case the higher fq peak is shifted to the left by about 5Hz, which is what I would hope for. What I have read in various articles is that stuffing a ported box does not have the effect of increasing the apparent volume of the box. That does not seem to be true in this case, but the effect is smaller in the ported vs. closed box, so the trend is in the right direction.
            In the port open case the two peaks are between about 0.5 and maybe 1 ohm taller with the port open vs closed. That is consistent with what I have read - which is that stuffing the box lowers the Q.

            I have tried measuring the output from the port by doing a frequency sweep while holding the mic about 3" above the port opening and I get a curve with a peak at about 110HZ that rolls off by 10 db at 60Hz. The rolloff is the same in the stuffed and unstuffed cases. There is a difference between the curves up in the midrange. There are a couple of bumps about 6 db tall at ~1100 and ~1600 Hz in the open port case which are barely visible in the stuffed case. I don't know what the best physical set-up is for the direct port measurement so I did not attach those graphs. I don't like the fact that there seems to be no low frequency boost at the port so I am suspicious about these sweep results.
            I am open to suggestions about how to make this measurement.

            Comment


            • #21
              Originally posted by datinker View Post
              The speaker is sitting on the edge of a table so it has free air around it except for the surface it is sitting on...
              The modeled response is half-space. Placed like that is not half space, and you may be dealing with a floor bounce cancellation as well. To measure half-space below the baffle step frequency you measure ground plane, away from boundaries, preferably outdoors, unless you happen to have a anechoic chamber handy.
              These are intended to be bookshelf speakers so I suppose it would be more realistic to measure them placed near a wall. That should return most of the port output back towards the listener, I would think.
              Port radiation is omni-directional.
              www.billfitzmaurice.com
              www.billfitzmaurice.info/forum

              Comment


              • #22
                I've used and tested this woofer extensively and the box is a bit small for a ported alignment. Not impossible though. Are you able to measure outside? If not I would pull the speaker out into the room and do the same measurement with the mic less than 3" from the port. Even try half and inch. Then go half an inch from the woofer dust cap. Those two measurements need to be considered together to determine the overall response. If you want to compare results I have a testing video on youtube. A simple search should turn it up. The speaker is called Life S5. Sent from my SGH-I337M using Tapatalk
                https://www.youtube.com/channel/UCm2...oSKdB448TTVEnQ

                Comment


                • #23
                  Originally posted by datinker View Post
                  In the port open case the two peaks are between about 0.5 and maybe 1 ohm taller with the port open vs closed. That is consistent with what I have read - which is that stuffing the box lowers the Q.
                  ​That measurement with the title "no stuffing in the box" seems a bit suspect. That dip in the impedance between the two peaks should be a lot lower if the box was airtight (except for the vent). It looks like the impedance curve for a stuffed and vented box.
                  Brian Steele
                  www.diysubwoofers.org

                  Comment


                  • #24
                    Thanks again everyone for the helpful comments!

                    Brian noticed what may be evidence of a leaky box - that is very possible. The back of the enclosure was not fully screwed down and not caulked yet.

                    I don't have a convenient way to measure outdoors at this time, or even a large empty room with enough floor space for a good ground plane test, so I tried the nearfield measurements from 1/2" in front of the woofer and 1/2" over the port mouth as ryanbouma suggested. What I found then is that measured very near the diaphragm the woofer rolls off steeply below 100Hz, which agrees with what I find measured at 1M. Stuffing makes no difference to this measurement.
                    When I measure up close to the port mouth I see a gentle slope down below 100Hz with a 2db hump at about 110Hz. That hump is similar to what I saw at 3" above the port, but the roll-off is much more gentle measured right up in the mouth. This looks a lot like the model prediction now.
                    The comparison between the very stuffed and the unstuffed box at the port mouth is interesting. The curve from the unstuffed box has a pretty big peak just above 1kHZ and a strong dip at about 350Hz. The curve from the stuffed box is much smoother and is 6-10 db lower through the crossover region than the unstuffed. Better still, it has the same slope below 100 Hz but it is about 2 db higher between 70 and 30 Hz. It is 10db down at 30 Hz.
                    I think I have to conclude that stuffing in this ported enclosure is a good thing. It seems to lower the box tune a hair and suppress some midrange resonances. The latter effect looks quite beneficial.

                    I tried another experiment to verify what I measured with my ears. I ran an online tone generator at 150 Hz on the stuffed speaker and then jammed the paper towels in the port. I could hear very little difference. I repeated this every 10 Hz down to 30 Hz and the plug to no-plug difference is very noticeable from 80 Hz down to 50 Hz, less so above and below that range. The port is doing it's thing. Stuffing is not preventing it from working as assessed by ear and by mic.

                    Comment


                    • #25
                      If you parallel the woofer's terminals w/a series 5ohm resistor and 15uF cap, you'll flatten the impedance rise, making your LP filter work easier.

                      Comment


                      • #26
                        Originally posted by datinker View Post
                        Brian noticed what may be evidence of a leaky box - that is very possible. The back of the enclosure was not fully screwed down and not caulked yet.
                        ​I've attached a screenshot of the impedance curve for my 27-year old Mordaunt-Short MS3.2s, which I'm currently "restoring" at the moment. I've illustrated on the image the difference between where the impedance null is and where it ideally should be. Both left and right speakers show the same impedance curve which suggests a lossy vented enclosure. And guess what - when I ran a few pure tones through the speakers at the vent frequency, I heard definite signs of a leak - in BOTH speakers. Looks like some sort of "design flaw" with the MS3.2s that I never picked up on before (they were consigned to surround speaker duty many years ago, before I started doing detailed measurements). The leaks are near the tweeters, and I'm pretty sure of what the cause likely is, and I will address it when the replacement tweeters arrive.

                        ​I suggest using weatherstripping tape along that removable rear panel to ensure a proper seal. I also suggest ensuring that the speaker is properly sealed up before performing other tests, as if there are any leaks, those leaks can skew your measured results a bit.



                        Brian Steele
                        www.diysubwoofers.org

                        Comment


                        • #27
                          There is a zobel in the cross over circuit. The DATS sweeps were made directly on the speaker terminals so you see the impedance rise there.

                          The way the front and back panels fit on these enclosures is not a straight interface between flat surfaces, more of a stepped edge. I don't have any CNC tech to make shapes like this so there are a lot of hand made imperfections. The front is permanently attached with fairly hard caulk. The rear panel will go on with silicone so I can cut it open later if I have to, but it isn't convenient for testing purposes. I just have to make my choices here and stick with them. Whatever issues are left after final assembly will have to be dealt with in the cross overs or lived with.

                          Comment

                          Working...
                          X