Serious' cheap OB

Discussion in 'Speakers' started by Serious, Oct 18, 2016.

  1. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Continued here: https://www.superbestaudiofriends.org/index.php?threads/my-real-ob-speaker-project.4083/


    (Similar to what OJ did in the CS days: link)

    Hinted at it before, but never made a thread because I didn't make a crossover until tonight. Had to pick up the parts that were at a friends' place. 2-way design, Eminence Beta 15A and Fostex FE83En. FE83En is a little offset from the baffle center for better FR. Drivers are as close as possible. For now, 1st order low pass for the woofer, no parts in front of the wide-bander. The idea is to use the driver and baffle roll off. I use a super simple impedance correction circuit that actually works well with the coil (I measured the impedance and did a few simulations). About 6mH in series and 6.8 Ohm and 37uF in parallel. Coil is quite big, but that's the price you have to pay. Still have to see how much that takes away. It should be possible to get an air core inductor for it, but the one I'm using right now isn't. DC resistance doesn't have to be super low, less than 2 Ohms should be fine.
    I have to redo the tweeter baffle mounting to align the acoustical centers of the drivers. This means placing the tweeter 12cm back (When I do that I will also make both drivers point at the listening position.). Right now there's a slight suckout at the crossover region (at about 300Hz). I know, the FE83En does start to distort at such low frequencies, but from my measurements it's actually not that bad. Inverting the bass does fix the crossover but murks up the 500Hz-2kHz balance because of the 1st order crossover. It's late at night so I can't crank it up. I will see how bad the distortion really is when I listen to it more tomorrow.
    Aligning the acoustical centers will probably give me a better response (simulation looks promising compared to the same simulation with the 120mm offset that I have right now) and better, more coherent imaging and better driver integration from "sloping the baffle" experiments that I did before I installed the crossover. I'm a fan of time-coherent 1st order crossover designs.

    Eventually I will replace the FE83En with something better. No sidegrades here. Will probably not go Fostex for my 2nd driver. The Beta15 will most likely stay for a while. Pretty happy with it.

    (Forgot to mention: Both baffles are 18mm thick. The FE83En benefits from chamfering or using a thinner baffle. I've not chamfered the baffle I'm using right now which hurts the airiness and openness, but I want to go for a thinner baffle later because I'm lazy. The dimensions right now are 60cm x 1m. Pics later.)

    TBC
     
    Last edited: Aug 16, 2022
  2. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Update:

    First up, here's a crappy pic of what it looks like right now. The whole baffle is at a 7° angle (which isn't enough). The Beta 15A is 60cm from the floor and the FE83En is at 85cm height.

    CrappyPicSeriousOB.jpg

    Here's what the FR looks like at the listening position, calibrated with the B&K target. I also attached an uncalibrated one for better comparison against other speakers graphs. I prefer to look at the calibrated ones. These are 1/12th octave smoothed. I also attached the L channel step response. Drivers aren't aligned yet.
    LR OB.png
    Turns out that there is no 350Hz suckout, it's just that my left wall is different from the right. There's an evident 2kHz bump. The narrow dip after it isn't that bad in this graph, certainly not as bad as the spec sheets suggest. The 2kHz bump I think is both the bass driver and the Fostex, but it's mostly the Fostex. This isn't caused by my baffle dimensions, but could very well be because the back of my baffle isn't chamfered. Both this bump and the treble rolloff will most likely get slightly better on a thinner baffle, but it won't make a world of a difference. The same tendencies will still be there. It does sound that rolled off (this is almost on-axis). The 4" driver, which has a slight rising response, will most likely measure with better extension to 16k but drop off more after that. The 5" is probably too peaky for me. There's some ringing around 9kHz, but otherwise the driver sounds overly smooth.
    The bass extension is actually pretty good, it's just that I sit almost in the middle of the room which gives me a big dip around 20Hz and a peak at about 40Hz. Bass extension seems decent even to 20Hz. Still, it's almost impossible to get rid of the room modes, even with an open baffle. I tried many positions and this seemed like a pretty good compromise. I have more than 2m space to the back wall. Listening distance is about 2m and the left and right wide banders are about 1m80 apart.

    The midrange is pretty good, especially considering that the crossover is right at about 500Hz. Sounds pretty neutral overall. The extra lower mid/upper bass distortion from running the FE83En without a crossover isn't nice, but it's not horrible either. I would say the 2kHz peak and accompanying D2/D3 distortion spike is worse when listening at higher volumes.

    I find my dad's speakers a little bright for my tastes. Those are almost flat with no applied speaker target compensation. My dad finds these too warm.


    I will get distortion measurements up later. By far not as clean as HD800/other good headphones, but that was expeceted. This whole project cost about as much as a new HD6X0.
     

    Attached Files:

    Last edited: Oct 19, 2016
  3. sphinxvc

    sphinxvc Gear Master (retired)

    Staff Member Pyrate BWC
    Joined:
    Sep 28, 2015
    Likes Received:
    3,320
    Trophy Points:
    113
    @Serious. Looks like a cool project. What effect does chamfering have, and what driver are you considering next?
     
  4. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    I haven't really tried it yet, but I measured it on a piece of cardboard. Maybe slightly less of that 2kHz bump and slightly more above 10kHz. I'm pretty sure the difference in sound will be bigger than the measured differences suggest. Removing the grill on the HD800 and HD600 makes a difference even if the measurements don't show anything.

    The drivers... I decided that I need to do more listening to the higher end widebanders before I make a decision. Considering the Lowther PM6A or DX3, Not sure how good the Tang Band W8-1772 or 1808 is. I want to give the Voxativ's another listen, but I think they only make 1 driver that's in my budget. If none of the widebanders convince me, I will most likely go for Accutons.
    Especially the widebanders are much more sensitive than the FE83En, which means I would have to double up on the Beta 15As, but I'm not sure if the Rag likes a 3 Ohm impedance in the bass. On the other hand you could probably make a real D'Appolito design this way.
     
  5. spwath

    spwath Hijinks master cum laudle

    Pyrate BWC
    Joined:
    Dec 13, 2015
    Likes Received:
    7,894
    Trophy Points:
    113
    Location:
    Madison, WI
    How do you know what speakers will perform well on an open baffle?
    I have a pair of jbl 2226h's, would they be any good?
     
  6. sphinxvc

    sphinxvc Gear Master (retired)

    Staff Member Pyrate BWC
    Joined:
    Sep 28, 2015
    Likes Received:
    3,320
    Trophy Points:
    113
    Found a good write-up on chamfering, sharing: http://www.troelsgravesen.dk/chamfer.htm
     
  7. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
  8. spwath

    spwath Hijinks master cum laudle

    Pyrate BWC
    Joined:
    Dec 13, 2015
    Likes Received:
    7,894
    Trophy Points:
    113
    Location:
    Madison, WI
  9. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Qts seems rather low and Fs is pretty high. Will most likely not get you good bass extension, although the room plays a major role here. The response to 1.5kHz looks very nice - much better than the Beta 15 which has some nasty cone breakup just below 2kHz. I do wonder what it sounds like. - Higher rated power handling, lower Qms, more damped cone breakup - might not sound better.
    I wonder why they all place the drivers so far apart. I don't think the floor will get you so much more bass and the big distance would probably give horrible off-axis and phase responses, at least with the crossover that I'm running at the moment. Also drivers that are far apart always sound far apart. (Image shifts with frequency, which I hate. I don't like it when the tweeter is above the ear level, when the rest of the drivers are at or below ear level. Unfortunately that's almost always the case with tower speakers.) I guess that's just different priorities. I can time-align my two drivers and get very close to a single driver phase response, but in an open baffle and with much better bass.


    Another thing that I want to try just for the heck of it: Use the HD800 driver as a tweeter. The HD800 driver is very good and has a super smooth FR. I feel the HD800 driver is very uncolored sounding, possibly more so than some of the exotic tweeter designs. Because of its comparatively large size (versus normal tweeters) you get more beaming from the mid-treble on, but that's not a bad thing IMO. Of course the problem is the sensitivity. My estimate puts it at about 80-82db at 2.83V/1m. The FE83En is about 88-90db efficient. That's a big difference.
     
    Last edited: Oct 20, 2016
  10. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    JBL 2226s will be absolutely horrible (massive rolloff) in an OB unless you design a compensation circuit. The 2226s will certainly be efficient and large enough in a home environment to take bass boost to offset the OB losses. The issue with the 2226 is the low Qts of approximately 0.3 (from what I can remember). The Eminence Beta's have a Qts of 0.6. I suspect it might be higher than that based on real world performance. OJ used the Alphas with their > 1.00 Qts, which in theory would have even better OB bass extension, but he mentioned they were craptastic - too uncontrolled.

    @Serious: You can try a U-frame for the woofer (but it looks like you don't need it), and put weird uneven jagged wings in a U-frame arrangement behind the Fostex.
     
    Last edited: Oct 20, 2016
  11. spwath

    spwath Hijinks master cum laudle

    Pyrate BWC
    Joined:
    Dec 13, 2015
    Likes Received:
    7,894
    Trophy Points:
    113
    Location:
    Madison, WI
    Hmm, ok.
    Trying to find a good use for them
    And you are right, .33
     
  12. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Stick your JBLs in big ass boxes. They are pretty amazing. Big drivers + big boxes = quality bass.
     
  13. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Bass extension would be much better if it wasn't for my room rolling everything below 40Hz off where I'm sitting. I don't think a U-Frame would lower the OB bass rolloff point enough where I would get more bass extension. More bass? Yes. Better extension? Nope.
    I want to try the speakers a little closer to the back walls. That wall is at an angle which avoids direct backwall reflections, so close distance shouldn't be a big problem.

    For the Fostex: Ideally I'd want to leave the space at the sides as open as possible to get the best elimination of edge reflections. Why did you go for the jagged U-Frame? As a sort of a diffusor?
    When I put the driver behind the woofer mounting baffle it will sit on a much smaller baffle anyway. 20x60cm instead of 100x60. Maybe this helps reduce the audibility of the distortion as this will roll off below 400Hz or so, or maybe it will make the distortion even worse as the fundamentals get rolled off more than the harmonics. I suspect it's the former as the disortion itself when using the harmonic frequency as a reference wouldn't change and the woofer takes over below that. The driver does have pretty high distortion across the band.
     
  14. spwath

    spwath Hijinks master cum laudle

    Pyrate BWC
    Joined:
    Dec 13, 2015
    Likes Received:
    7,894
    Trophy Points:
    113
    Location:
    Madison, WI
    That seems like the best option, but not at the moment because big driver + big boxes = big speakers. And I dont have room.
    Either A: I convince my family to let me upgrade their vizio soundbar, or B: I stash the JBL's somwhere safe for a few years until after college/I get my own place.

    Either way, I got them vitrually for free. $50 for the two, plus yamaha horn loaded cabs for them, plus yamaha horn tweeters, plus yamaha pa amp, plus shure sm58 mic. and $50 for all that, still cant beleive it/
     
  15. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,777
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Uneven and/or asymmetrical edges will result in less of the OB up and down response. You could always play with cardboard and see what happens. I actually had stuff like this behind my OBs.
     
  16. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
  17. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Still doesn't seem to explain why the drivers are so far apart, except he probably wanted to use the floor reinforcement for the bass and avoid floor bounce lower in frequency. Would like to see a system step response at 2 or 3m listening distance at his preferred listening height. Probably won't look anywhere close to a textbook 2nd order xover, even when not taking the driver response into account. The woofer has its acoustical center at the crossover frequency about 12cm behind the full range driver. Because it's mounted so far away, the difference only becomes larger.
    I can angle my baffle so that the step response looks very good. That's at a 23.4° angle. The other option (for placing both on a straight baffle) would be using the woofer backwards with its magnet facing the listening position. That might not be a bad option.
     
  18. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Made some progress: I did two things yesterday:
    I moved the FE83En 9cm behind where it was before, which should align the acoustical centers without the crossover. This means that the drivers don't add perfectly at the crossover region (which also wasn't the case before), but the phase should be very similar above and below the crossover region. Even if it's not perfect, the changes it made to the sound are astonishing. There's a certain type of imaging and immediacy that only time-aligned speakers seem to have.
    Here's what it looks like now:
    Box2.jpg

    The other thing that I did is to chamfer the baffle. Some of you know how much of a difference removing the small, almost acoustically transparent, piece of foam behind the HD6X0 drivers makes. Now imagine removing a big, solid chunk of wood that obstructed most of the drivers' back and probably acted as a small transmission line.
    Back.jpg
    There's still a lot of shoutiness when listening from behind. Have to avoid direct reflections with the shout as much as possible.

    Of course this also means that the baffle dimensions changed. I actually planned it more like this instead of having one big baffle. I attached what a farfield simulation using "The Edge" looks like now (2 Both) and what it looked like before (1 Both) (Bass is the upper graph).


    Here's what the FR looks like now (B&K target compensated):
    LR 2 BK.png
    I don't think I will need a filter for the 2kHz shoutiness, but it does bother me. I might try to make a filter at one point.

    Here's the step response now: (I measured it slightly closer to the speakers to avoid the couch reflection, but I still get reflections. I want to measure them outdoors. I tried to keep the relative driver distances the same as at the listening position.)
    L 2 step.png
    I know, it's not perfect. But I think it looks much better than before.

    Sounds much more open now. More dynamic. I think it sounds pretty good now. For some reason the bass now subjectively extends deeper. (I also moved it a little. At one position I got a -5db point at 25Hz.) Treble is more present, mostly due to the more open back. Actually sounds more extended now. Midrange isn't perfect. This is where making the crossover better would improve things. A larger midrange driver to be able to lower the xover frequency, maybe. The tiny 3" widebander has a lot of distortion across the band (about 1% D2 and D3 at 90db). Fortunately it doesn't seem to rise a lot as we go lower in frequency. I feel like the smaller baffle helps to avoid some of the lower midrange congestion at higher volumes that I experienced before.
    I don't think I would take KEF LS50s over these. This is really starting to become something special. Drivers also have more burn in by now, which I'm pretty sure made it sound better.
     

    Attached Files:

    Last edited: Oct 23, 2016
  19. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Here's the right channel distortion at about 90db: Compare this to my HD800 95db* measurements (http://www.superbestaudiofriends.org/index.php?threads/hd800-measurements.2720/#post-80296)

    R 2 90db.png
    It says 1/12 octave smoothing, but there doesn't seem to be a feature to smooth distortion measurements. Looks really messy without smoothing. If there's a way to smooth these, tell me. These are at the listening position (about 2m) and using the harmonic frequency as a reference.** I know, it looks horrible. Most of the distortion is from the FE83En. The 4th order distortion around 200Hz and the D2 spike at 2kHz I would say are the worst. The consistently high 2nd and 3rd order throughout basically the whole range are easily audible. By far not as clean as the HD800. Not even close. 2nd and 3rd order in the midrange is at 1%. The HD800 probably has less than 0.02% D2 in the midrange and even less D3. The HD800 midrange distortion gets really hard to measure at sane volume levels with cheap, small omni mics. The 2nd order bass distortion rise actually sounds worse than it looks because this is using the harmonic frequency as a reference. The 25Hz spike is from the 50Hz hum that the mic seems to have with my laptop. Still, driving the FE83En without a high-pass filter seems to work out in terms of distortion. No nasty distortion rise in the lower mids (except the D4).

    To be fair these measurements aren't super horrible for speakers. The bass distortion, especially up to 30Hz is actually pretty low for speakers. Even the high distortion in the midrange isn't terrible by speaker standards. Ask Voxativ for distortion measurements of their $$$$ drivers.

    *90db on the coupler translates to 95db on the head.
    **You have to use the harmonic frequency as a reference for speaker measurements to eliminate the room effects from the measurements.
     
    Last edited: Oct 23, 2016
  20. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Small update: Been tweaking the crossover. Still no parts in front of the Fostex. I lowered the inductor to 3.9mH and changed the Zobel to 3.4Ohm / 32uF.
    The problem is that the Fostex essentially rolls off with a 12db/octave slope by itself on the smaller baffle. You know a textbook 1st order crossover is not gonna happen. With the right zobel circuit you could actually get extremely close to an almost perfect 2nd order xover with only the coil in front of the woofer, but I tried something like that and I think I still prefer the approach without reversing the polarity to one driver. The filter I have right now does give me pretty good phase and frequency response, even off-axis it seems. The step response still doesn't look as nice as I'd want it to. Proper phase response seems to be really important for the imaging.
    It's really hard to get measurements close to what you get at the listening position, but without reflections for the simulation. You sorta have to come up with some weird stuff. LOL, I actually windowed Fostex nearfield measurements so that the roll-off matched the 2m distance listening position roll-off. Even when measuring outdoors the floor bounce still makes it really hard to get useful measurements. The cold-rainy munich weather doesn't make it any easier, either.

    I found out that the Fostex doesn't have the 2kHz shout on-axis. All the extra energy is coming from reflections. My room isn't even the most reverberant. This also explains why I found it to be not as bad as shown in the FR measurements. Also, with no reflections the 2kHz suckout looks really bad, at least on axis. Super narrow and extremely deep, deeper than the spec sheets show, I think.
     

Share This Page