OB project with 2x LO15, B&C DE250 & Autotech SEOS 15"


  • Member
    dbo on #1104

    Hello to all,
    I finally managed to bring my project to materialization. OB speakers with 2x LO15, B&C DE250 & Autotech SEOS 15″ each. To exactly “finish” it will require some more time with EQ, but results were promising even the first day.

    I’d like to share my experiences and ask a couple of questions. For the first, all four drivers seem to be flawlessly made and untouched from overseas shipping.

    A week after I’m much closer to complete satisfaction, but this is only first week of a process that may take months. Overall these speakers are showing significantly greater potential than my previous Emerald Physics CS2, which in my room were quite excellent and among those best I heard anywhere. CS2 completely cured any audiophilia nervosa I might had being more than 20 years in audio, and if I was not plain curios enough while at the same time being able to stretch for 4x AE LO15s during last year, I could live with CS2s very happily for many more years. Briefly for those who are unfamiliar with EP CS2, this is the same concept and size like my new AEs, but with 2x Eminence Alpha 15A, Selenium D220Ti and Dayton 12″ round waveguide. In my opinion CS2 is still the most significant recent consumer speaker, clearly showing a strength of a clever concept and skill over just throwing more money in a wrong direction.

    As before, speaker is powered with >100 W per channel MOSFET for lows and 2A3 SET for highs, crossover is Behringer DCX2496 with full Selectronic mods. Signal is kept digital all the way to DCX, then after it gets out as analog I have 4-channel step-attenuator before amplifiers. I connected LO15s 8 Ohm coils in series at each driver, than two drivers parallel to the amplifier. That should give 8 Ohm total load.

    From the front, new speaker looks almost exactly like this rendering, I’ll post real photos some other time. Tweeters also have open backs so it is a dipole all the way.

    As a quick description of a sound I get now it is somewhat of a mixture of CS2 sound I had earlier, and of my impressions from auditioning JBL Everests.

    In comparison with CS2, new one gives even clearer sound throughout the spectrum, sound fills the room more fully and evenly, dynamics are also a bit better, and low end has more energy. I can always turn bass higher on CS2 too, but this one sounds more free from compression down low regardless of bass quantity. Also, I seem to get less (or none…) “structural” annoyances. This new baffle should be better than CS2 – new one is 2″ baltic birch plywood baffle plus quite heavy steel backbone. In the beggining, even with suboptimal EQ scheme and some clearly present peaks, sound was still quite listenable and easily flowing. While at CS2 I really needed to nail the EQ right, with new speakers there seem to be a larger margin for error while still hearing a good music.

    I started with XO and EQ scheme exactly as with CS2. Surprisingly, it was quite good. Later I decided to go from scratch and it is now better, but some corrections are still very close to ones with CS2. I guess those have room related cause. I left XO as 48db/oct Butterworth. CS2 have LP point at 1.24k, and HP point at 1k. For new ones I lowered both to 970 Hz.

    That is my first question. Should I also put new LP a bit higher than HP (say LP at 1.1 k while HP is at 970 Hz), and why is that so at CS2? I know I can experiment, and I had and I will, but number of combinations of XO points with tracks at where I may actually hear the difference is vast, so I would like to narrow it down theoretically first.

    Also is there any scheme of a LO15 motor structure? I tell people LO15 is a dual-coil design, but cannot really describe how it looks like inside.

    Thanks


    Member
    simon5 on #10598

    For the crossover, if we were in an ideal world, you would want the same frequency with the high pass as the low pass, but depending on the design or the speakers or the room used, you can have a peak or a dip at the crossover frequency when doing that. That means you either want an overlap or you want them separated further. You can use software to model that, but the best would be to measure them. Not that easy in a home environment, but feasible if you are determined. That’s not to mention the baffle step compensation…

    Dual coil is only two sets of windings around the voice coil former, so you can choose between two impedances by choosing how you wire them. You can think of it as an electromagnet with two separate windings… A bit like a transformer with a center tap…


    Member
    dbo on #10599

    Thank you for your answers Simon.

    @simon5 wrote:

    For the crossover, if we were in an ideal world, you would want the same frequency with the high pass as the low pass, but depending on the design or the speakers or the room used, you can have a peak or a dip at the crossover frequency when doing that. That means you either want an overlap or you want them separated further. You can use software to model that, but the best would be to measure them. Not that easy in a home environment, but feasible if you are determined. That’s not to mention the baffle step compensation…

    If it’s only about the peak or dip, then I don’t worry that much. Listening and simple measurements with handheld SPL meter are telling me this area is fine. But to know exactly if this point is the best for matching directivities of bass and horn is much more difficult task I guess. However, it should be around 1k.

    I expect to measure speakers with more sophisticated equipment and educated help within a month, then I’ll get back with results.

    @simon5 wrote:

    Dual coil is only two sets of windings around the voice coil former, so you can choose between two impedances by choosing how you wire them. You can think of it as an electromagnet with two separate windings… A bit like a transformer with a center tap…

    So these coils are winded one over the other? Not one coil in in front and the other on the back? So if I connect only one coil, driver will still behave properly? Only with less power handling and warmer coil?

    Regards


    Participant
    stryke on #10600

    I would suggest investing in some basic measurement system. The omnimic from parts express is the most simple and easy to use. The crossover is not only about selecting the crossover frequency to get flat response but also keeping phase coherent throughout the crossover region. A crossover can measure flat but not sound good due to poor phase integration. One of the easiest ways to get close is to invert the polarity of the driver. Adjust the phase and/or time alignment until you get the deepest null possible at the crossover point. Then when flipping the polarity back around they will be in phase very close. If you don’t have ability to adjust phase or time alignment all you can do is make slight shifts to the crossover point.

    John


    Member
    dbo on #10601

    Thanks for the answer John, and my respect for the drivers, they are really great. I get some fantastic music out of these.

    I have Behringer microphone and can adjust time delay, crossover point and slope at digital crossover. I hope to get good help in measuring in a month or sooner. When I see exactly how it’s done I’ll invest into something of my own so I can fiddle with it and learn how to improve speakers over time.

    Could you please comment on this below and help me understand it; regarding LO15 being dual coil driver.

    So these coils are winded one over the other? Not one coil in in front and the other on the back?

    If I for some reason connect only one coil, driver will still behave properly? Only with less power handling and warmer coil?

    Regards


    Participant
    stryke on #10602

    The coil in the LO15 is a standard dual VC configuration. It is essentially a 4 layer coil. The first two layers wound down and backup the former connect to the one set of terminals. The second layers down and up connect to the second set. Each set of windings is 8ohm nominal. They can be wired either in series for 16ohm or in parallel for 4ohm. The typical wiring would be to wire the VC in series for 16ohm woofers and wire the 2 woofers in parallel to get back to 8ohm for the pair. You do NOT want to use only one set of terminals on each woofer. This cuts the motor strength of the woofer in half. It cuts efficiency and raises the Q of the driver. You would end up with a woofer Q that is far too high to be useful. Additionally this is a very easy way to burn up the VC. The LO and Dipole woofers have comparably a very small VC vs other TD woofers. Powering one VC and leaving the other unpowered will trap heat in the VC and allow it tu burn up easily.

    What makes the LO and Dipole Series unique is the underhung geometry and of course the Lambda design with full copper sleeve on the pole. This underhung geometry keeps the BL very linear throughout the Xmax of the driver. The Bl curve is extremely flat right up until the woofer hits the Xmax point. This is very critical. In an open baffle, the low end decreases due to the baffle rolloff. This has to be compensated for by higher excursion and more output from the woofers. They operate at higher excursions for the same SPL levels than a woofer in a sealed or vented enclosure. Because they operate at higher excursions, more Bl linearity is required at these excursions. The underhung provides this linearity.

    Then of course is the linearity of inductance. With high excursions, inductance changes as well. A typical overhung woofer has an S shaped inductance curve. As the VC moves inward, Le goes up. As it goes outward, Le goes down. This is clear to see in the impedance curve of the woofer. It is critical because as impedance changes, the response curve changes as well. You can see an extreme example of this here with the Altec 414. Look at the curves at 1KHz. On the outward stroke, the resistance at 1KHz is about 15ohm. At rest it is about 21ohm. At the inward stroke it has gone up to about 26ohm. There is a difference of 9ohm in resistance when the driver is moving to 5mm Xmax!

    https://sites.google.com/site/drivervault/driver-measurements/tang-band-75-1558se/gpa-altec-414/le-x

    Now contrast that to the LO15. A properly designed woofer with full copper sleeve to eliminate this effect. At the same 5mm Xmax there is NO change to the impedance curve whether at rest, on outward or inward stroke. The following note was made by Brandon Thill who did the measurements:

    “whippersnapper02@yahoo.com – Jun 26, 2009 9:30 AM
    I know, why only the 5mm plot? Well this was the last TD driver I tested Le(x) for after seeing all the others what’s the point? Put a well designed full copper sleeve on the polepiece and Le(x) will be excellent. In the case of ALL the TD drivers it’s flawless.”

    https://sites.google.com/site/drivervault/driver-measurements/15/ae-speakers-lo15/le-x

    The changes in the impedance curve greatly affect the response of the woofer. Especially in a situation where the woofer is being called on to regularly move to high excursions in an open baffle. If the response curve is continually changing, it is not possible to design a crossover that is accurate. The measurements taken to optimize the system with the woofer at the rest position are no longer valid once the driver starts moving. This is true for both active and passive crossovers. However with passive crossovers it is even more of an issue when the component values of the crossover are determined by the impedance curve itself. When the impedance curve changes, none of the passive components are then accurate.

    All of these issues are solved with the Lambda motor. http://www.aespeakers.com/Lambda001-1.php


    Member
    dbo on #10603

    Thank you very much for this all in one place explanation.

    Drivers are great, in my listening situation, and only “eye measured”, longest excursion is around 5 mm, that would be loud music with much low bass. In most situations, loudest I listen (meters say around 95 dBA at the seat), excursion goes to some 2 mm or so.

    Still waiting to measure the speakers…


    Member
    dbo on #10604

    Still not measured them properly but don’t care any more. I have my own methods to adjust them. Speakers are fantastic. This is as close to perfect for speaker this size as I ever heard at any price or technology used, of course with proper EQ.

    I experimented with two other configurations – only one bass driver instead of two, and 2.5-way with 15 mH coil attached in series with lower driver.

    I had to readjust EQ and crossover for evaluation of both of these of course.

    My impressions are that first one does not bring anything better to overall presentation, I wanted to see if it is somewhat “clearer”, but driver only have to move more for desired volume. At first, 2.5-way seemed interesting, sounded somewhat smoother, more relaxed or something like that. But over time I noticed that a part of that “electrostatic speed” is missing, relaxed was in fact a bit lazy, and bass was a tiny bit blurred, midbass lost a bit of a kick. These were all very subtle effects, I’ll be fine with it if I could not compare it to something better than this. However, I needed only a moment after I returned it into normal 2-way operation to realize that this is clearly better. Although both cones are working in 2-way and 2.5-way, 2-way gives more realistic size of everything that is going on. I know from before that I prefer larger midrange. In this case two 15″s are also better than one. As for 2.5 way, simple measurement shows that 15mH coils affects response from 100 Hz upwards. Perhaps I should also experiment with smaller coils to see what happens then.

    When I’ll measure, I’ll be most interesting to see how is horn time aligned with cones. I the meantime, my thinking is that drivers are time aligned if coils of tweeter and bass are in the same plane (both xover cuts are digital 48dB Butterworth). I can easily see where is tweeter voice coil, and I guess that woofer has it between back edge of the basket and magnet. If I’m right, then bass voice coil is some 12 mm in front of the tweeter coil. So I delayed bass drivers at digital crossover for 12 mm distance.

    Am I very wrong thinking like that?


    Member
    dbo on #10605

    Just a bump for my question regarding time alignment in last paragraph above, perhaps someone can comment. Thanks.


    Member
    simon5 on #10606

    You can mechanically align them, that’s for sure, it won’t hurt.

    Problem is, a crossover will change the phase, that means it changes the time alignment also…

    It depends what kind of digital crossovers you have… are those active crossovers or real digital ones that do the crossover step before the DAC step ? You can align them in the digital domain instead of mechanically… then you send everything to the DAC and be done with it.

    I would just tweak the crossover until when it’s plugged reverse phase that you have the deepest null at x-over frequency… again mechanically aligned won’t hurt, but it’s not a perfect solution, you can run into other problems trying to do that. 8)


    Member
    dbo on #10607

    Crossover is digital, Behringer DCX2496 with full Selectronic mods http://www.selectronic.fr/dcx2496-modifie-par-selectronic.html#.

    Slopes are 48 dB and it is all done in digital domain before D/A conversion. Drivers are in position which aligns horns and basses fronts at the baffle. In this position it seems to me that bass coils are approximately 12 mm (possibly +/- some) in front of tweeter coil. I measure positions in which I’m guessing that coils really are. During design phase, I had no intention to physically align coils, their current position is a result of overall construction.

    So, until I could measure them, I digitally delayed basses for 0.03 ms, I hope I’m not too far off. It would help further if I’d know exact LO15 coil width and placement.


    Member
    simon5 on #10608

    I don’t know if you have a good microphone, but the best way to do that I think would be to measure, tweak the delay, measure again… etc.

Viewing 12 posts - 1 through 12 (of 12 total)

You must be logged in to reply to this topic.