DIY Synergy/Unity spreadsheet

Discussion in 'Waveguides and Horns' started by BillWaslo, Apr 29, 2012.

  1. Hi Maxphan,
    Yeah, it was the same placement as before. But only because the ports had been made for a three-way version and I only used those as they were. It's not easy to make these with the ports movable. Probably the woofer ports should be moved forward (maybe a lot) but I don't feel up to making entirely new horns just to test that. It would be best to figure out where they could go physically, and then model them in HornResponse in the range of possible positions to see what works best (so it gives ok bass, but still reaches high enough for decent overlap with the tweeter). As the port gets closer to the tweeter, it wants to get smaller, but too small a port will strangle bass as will a position too close to the throat.
  2. Hi JaCro/James,

    I'd be a little worried about the perforated metal vibrating and resonating with all the pressure from the woofers. But it might work out. I'm also considering using some of the Geddes style reticulated foam in just the radial corners of the waveguide, to attenuate the highs going there there and so reduce the effect of the discontinuity. I have a big block of the stuff in my workshop.

    But really, the on-axis irregularity around 4kHz doesn't seem to bother the sound any that I can tell. Probably because the radiation pattern is already so much more consistent than anything else I've heard that the relative effect is quite minor. But maybe it would be better with the slight 4kHz stuff fixed. Maybe someday when I have lots of time (if ever!).

  3. Hi bwaslo/Bill,
    Fascinating project and great results! Congratulations! I have to admit I've read through this thread at least a couple of times now.
    Can I ask a question about the linear phase crossover for this posted on page 5?
    I get most of it (even with my admittedly rudimentary understanding of crossovers ;) ) but I am curious about what led you to the values and arrangement for the low pass section to the mid drivers?

    As it seems unusual compared with what I am generally used to - or am I just being naive?

    Cheers Paul.
  4. The lowpass section to the midranges is done that way to increase delay to the midrange, to better bring it into time alignment with the tweeter (more lowpass filter sections delays the wave going through). On a linear phase Synergy the difficult part is getting the output of the drivers to attach in the proper delay with the output coming out the horn from the further-in (higher frequency) drivers. You also have to do the electronic crossover slopes with the proper phase/delay characteristics (which usually means rather gentle slopes, used with plenty of driver overlap regions). The mid and woofer drivers have acoustical lowpass filters that form from the compression chambers below the ports.

    The most important thing (for me anyway) is to be way lucky! This one just happened to work out with a quite simple crossover for a three-way design; that doesn't happen all that much... I'd like to find non-buyout drivers that I can get to work as well, but so far haven't come across them (some to try, but not as promising).

    This was all done using measurements from each driver mounted in the horn and simulated/tweaked in Jeff Bagby's PCD program. Came out pretty much dead-on as modeled.
  5. Many thanks for the reply and the suggestion toward Jeff Bagby's PCD program, I'll have a look at that later and see what it can teach me. :)

    I get the idea behind aligning the output from all the sources on the synergy horn,
    [SIZE=small]But I'm still a little puzzled by the exact form and values used on the LP on the mids? (looks like 4th order but not like what I'm used to) Sorry, just me wanting to fully understand everything properly.[/SIZE] ;)

    Did that all come from the Jeff Bagby programme? Or did the unusual arrangement come from another source?

    Cheers again,
  6. For passive crossover design it's best to not think in terms of "classic" filters (butterworth, cheby etc) There are a world of betweener types and some sections.might be only for phasing or eq or avoiding a bump. Even crossover.order isn't any thing fixed Ls and Cs can have resistors across them or in series making them not cause an entire rolloff in the drivers' ranges. And the entire response shape that matters is of the driver and crossover so looking at a well-done crossover circuit alone isn't likely to.make obvious sense.
  7. Unfortunately you still haven't answered my question there! :(
    If you are unwilling (or unable) to answer could you point me to somewhere which can supply the information? So I can work it out for myself.

  8. Paul,

    Bill, kinda did answer your question there. Don't think of passive crossovers as having a specific response, Butterworth, L-R, etc... The values and even schematic diagram of the passive components don't matter a lick. The overall acoustic response, time domain response, polar pattern, resonance control and stored energy of the crossover network, driver, speaker baffle and whatever else contributes to the final response are all that matters.

    I've made thousands and thousands of crossovers for probably 70 or 80 speakers. Only 8-10 of which saw the commercial light of day. People don't always believe it's true, but with programs like LSPCad and SoundEasy it's possible to have a flat response +/- 1-3dB within minutes of measuring a speaker. I have never, ever, once, ever liked the sound of the speaker with the first, computer solved "perfect" response. There has always been improvements with playing with tweeter and woofer rolloff rates, 10dB per octave? Sure. No problem. Moving the tweeter slightly in time through all pass networks? Yep. In LSPCad I once designed a great sounding crossover for a Vifa XT woofer, Seas Excel tweeter two way....then I looked at the final impedance. It would have been 1.2 Ohms at -60 degrees from about 500Hz to 800Hz. I would have put almost every amplifier made into self protection mode or the smoke pile. It took me 3 weeks to get that sound at a "reasonable" impedance of 3.8 Ohms and -30 degrees.

    That speaker never made it out of the lab.

    I've had speakers with 2'nd order electrical slopes have a final acoustical slope of somewhere on the order of 8'th order. (I also had a very interesting conversation with the folks from Joseph Audio at CES, infinite slope my a&$.)

    All Bill is saying, all I'm saying is, by limiting yourself to textbook or "traditional" circuit variations you'll be eliminating the vast majority of where you need to be in your crossover design. I rarely used a standard circuit schematic, I almost always had notch filters, all-pass sections, impedance correction, and more. Modern crossover simulators are an amazing aid. SoundEasy and LSPCad are invaluable, learn them, use them, they will make your speakers sound much...much better.

  9. Nope he didn't.

    I asked "what led you to the values and arrangement for the low pass section to the mid drivers?"
    and followed up with "Did that all come from the Jeff Bagby programme? Or did the unusual arrangement come from another source?"

    I never got an answer, all I got was it was "done that way to increase delay to the midrange" and that it's "best not to think in terms of classic filters".
    Does that tell me anything about how such an arrangement was come up with? No, not at all!
    That's what I really want to know.

    I'm not limiting myself to traditional circuit variations, that's the whole point of me asking about it! To learn what's going on - but when I ask I just seem to get vague answers which don't explain it properly or even slightly in fact. Only stating what was the desired end result.
    So is the only way for me to find out about such things to get Soundeasy or LSPCad? As nobody can explain?
    Will they have an explanation of what this section is achieving? and more importantly How?
  10. Designers are led to specific crossover configurations by a kind of six sense or intuition developed over doing lots of crossover designs and then confirmed by rigorous testing and optimized by a trial and error process. Bill's first crossover on page 3 of the thread is indeed difficult to understand but he got the results he was looking for and that speaks well for his process. I'll guarantee you he didn't just draw the schematic based on some simulations and then build it. The 2nd version, developed based on suggestions from the master himself, Tom Danley, is elegantly simple, a goal we all strive for and applaud when we see. He added the magic of carefully placed and tweaked crossover filter Q and pole/zero frequency locations to the placement of the drivers in the horn and driver frequency range overlap inherent in Bill's original design.
    I think the message is you would have to be there in the lab with them to understand how they got where they got. But we've got sufficient clues to follow their paths if we choose.
  11. Hi Paul,

    Your question was kind of hard to answer without knowing where you're coming from. I tried to respond with the function of the odd "soft lowpass" on the midrange (which is: delay). It is also a crossover filter, too, of course, but that I'd have thought was obvious.

    There are no formulae to present nor a how-to-do-it manual. For loudspeakers, particularly with passive crossovers, everything (impedance, delays, magnitude, resonances, radiation pattern, power handling) interacts. You basically need a good modeler (such as PCD) and some intuition on what the various components are likely to do when connected different ways and in different values. Then try varying the parts and configurations in the modeler to see what they do and work with them until you either get where you want to go - or decide it can't be done. Used correctly, PCD will give very accurate results - though you might find you want to tweak by ear afterwards from the original curve shape you went after.

    A post on a discussion forum isn't going to be sufficient to give the intuiution needed, though some basic study of passive filters and component impedances would help. Can't tell you where to get that though (other than Engineering School). Probably a Google search or three would help.

    Yeah, pretty much. Either that or build several hundreds of circuits and measure. Trial and error is a lot easier on a computer!

  12. I actually found the first one perfectly easy to understand, there's nothing unusual in it at all! It was very straight forward and easy for anyone to with a rudimentary crossover knowledge to understand.
    The second (linear phase) one is mostly pretty simple too and very logical when considering the slopes applied by the design.

    However personally I haven't experienced much in the way of crossovers which relate to the particular chosen Low pass element for the mids - so I thought I would ask about it from Bill as he was responsible for producing it and would no doubt know!

    Was that the part suggested by Mr. Tom Danley? Was it suggested by a crossover design program? Or was it the result of Bill's knowledge and experience?
    That's what I'd like to find out! And I'd like whoever is responsible for it to explain exactly why that form was chosen?
    It's a pretty simple question really.

    I really don't see that we "have to be there in the lab with them to understand" the thought behind it, certainly no other technology imposes such restrictions, education would be severely lacking if everyone worked that way!

    I thought that was the point of DIY forums, to share ideas and findings in the pursuit of better audio reproduction!
    Or have I done the audio equivalent of asking for the colonel's secret KFC recipe or something!? :p
  13. Hi Bill,
    So it appears that you are saying that it was just a rigorous trial and error arrangement using PCD until the desired results were achieved?
    So no outside input or it being based on anything else previously encountered to suggest it.
    That's basically all I wanted to know! :) Cheers!
  14. Like Jack and I said, though, it does require some insight -- you can't just combine raspberries, curling irons, toenails and - oh, ah - maybe some capacitors - or whatever you find at the grocery store - randomly until an ideal crossover appears! It's not monkeys at a typewriter (but perhaps experienced monkeys?).

    For example, when doing that crossover, the phase wasn't falling fast enough in the midrange than it should for a linear phase, so I knew I needed to get a little more overall delay in the midrange, so I bumped it by two more 2nd order sections damped till I got the delay as I wanted without too sharp of a phase-change in midrange drivers' band. BTW, the part I got from Tom on the passive crossover was to use the inductor shunting the tweeter (he pointed out that it could keep the diaphragm stiff at low frequencies). The active crossover that I'm using, though, was almost entirely Tom's design that he did for me (I did a slight bit of tweaking on that, but that was it). Delay is a LOT easier to do with a DSP crossover.
  15. Hi Scott:
    Yes, the original crossover isn't all that difficult to understand from a basis of experience; there are just a lot of pieces and a rationale to be understood for each one, and probably a been there/done that feeling for some of them. What I got out of it was justification for spending the extra bucks on DSP. I'm only doing one of these and its my time that needs to be optimized more than my wallet. What I got out of the 2nd circuit was respect for the master. If I can do that than I can use the pair of pro amps I just bought in another room of the house.
    I've been to engineering school but really all that did was give me the basis on which to build the learning needed to solve real world problems on the job. A good thing too because the technology I'm dealing with today didn't even exist when I went to school. And now I'm having the fun and adventure of learning acoustics. But sound reproduction has been around since the 1920's or so. Imagine being able to come up with so fundamental an innovation as the synergy horn in such a mature field after so many giants have had their go at it! And I think Tom has said else where he hasn't gone to engineering school. So hats off to him. But also let no one else be discouraged. Get a couple books, read some papers and get your hands dirty. This stuff may be arcane but it does yield to a concerted effort.

  16. There are two ways to learn how to come up with the variations, and in reality most folks who know how to do it, do both.

    1. Run thousands of simulations using the software of their choice to watch the impact of electrical elements on the acoustic response.
    2. Study the underlying math (signal analysis, transfer functions, Fourier Transform, Laplace Transform, linear system theory, non-linear system theory...depending on how fancy and detailed you want to get).

    The circuit shown on Page 5 doesn't look all that un-orthodox except for Bill's previously mentioned shunt inductor. I'd heard about that before for the Synergy horns, there was someone on another forum complaining about wrecked tweeter diaphragms. Tom told him to fix it by using the shunt element which puts the back EMF brakes on the tweeter element.

  17. Aloha guys.
    Bill, thank you for sharing with us your project and all the help with it.
    I got all the drivers and crossover parts from PE. Build a horn but completely lost with driver exit location. I know that the inch is equal 2.54mm and when I convert all the numbers and trying to place every driver in panel it dosn't look right at all.
    Is anybody tried to build it in metric?
    For example 5.475 -how much is it in mm?

    thank you all who involved in this project and Tom D for his ideas and support for DIY community.
  18. Hi Sergey,

    Probably the confusion is because the diagram I drew doesn't really look much like the actual wood cuts. In actuality the four drivers get very closely clustered together -- midranges touch each other and nearly touch the woofers. Let me try to take a better photo to show what it should look like.
  19. Sergey,

    Nope, an inch is 25.4mm (or 2.54 cm)!
    5.475 inches is equal to 139.1mm.

    Here is a photograph showing of the side of one of the coSyne horns, so you can see how it is supposed to look. Does this help?


  20. I see.
    that a lot better!

    hope to finish in couple days.

    Thank you, Bill.

  21. Aloha Bill.

    Everything looks good now except one thing.
    Woofers enter - in your description clearly show that is 0.75. (19mm)
    When I put everything on place I get 0.985 or 25mm.
    Am I mistaken?
    How critical is it?

    there is an additional picture for those who are prefer metric.

    Attached Files:

  22. Do you mean the short dimension (0.75in) for the woofer ports? That should be 0.75inch, 0.985in would be much larger. The port does widen on the outer side of the panel, using the 45degree chamfer router bit. The hole size given is for what enters through to the inside surface of the panel- on the outside (woofer side) the opening is larger.

    Make each woofer port by drilling 0.75in diameter holes at the two cross points shown, then saw between each with a jigsaw to make the oblong opening. Then chamfer (or file) the openings so they are larger on the outside.
  23. I noticed the Celestion is out of stock at PE. Confess up you silent builders out there....
  24. Incoming: new Synergy spreadsheet, docs on the way ("soon")

    I was hoping to get the revised spreadsheet out, along with detailed and illustrated directions for making the horns, but some back problems (mostly gone now) and a good dose of flu (not gone) have slowed me down a lot. So I thought I'd at least get the spreadsheet posted for now. When I can get the instruction pdf done, I'll post it in a new thread here.


    Here's the link to the spreadsheet file:

    If you don't have Excel, you can instead use the free software GNUMERIC for Windows or Linux. Windows download version is at

    I've left this copy of the spreadsheet unlocked for users who wants to modify it for some reason (GNUMERIC doesn't allow unlocking). So-- when using, PLEASE BE CAREFUL TO ONLY TYPE IN THE CELLS WITH GREEN BACKGROUND, AS ABOVE.

    Some notes about what's different about this version:
    • · Dimensions are named with less complicated symbols (A, B, C… rather than stuff like “k1(h)”.
    • · Board junctions modified for easier assembly and to remove need for doing any “vertical” cuts. (No vertical sled or the like needed for any of the edge angles). All cuts are with the board parallel to a table saw.
    • · New methods of attaching panels before glue-up to avoid using duct tape, hot melt glue, or the like during assembly
    • · Allows for construction from pre-veneered boards, guiding orientation of the wood grain when cutting panels. Doing it this way requires very careful and precise cutting though – mdf or other materials that can be filed, ground down, filled with bondo or wood dough, and then painted, would be much, much easier. But wood grain does look cool, if you want to go to the trouble.
    I get the writeup out as soon as I can.
  25. Right in time!
    Thank you, Bill.
    Hope to finished t in weekend and start to build a crossover that Tom Danley made for us.
    Appreciate your help Tom.

    Mahalo from Hawaii,
    with all respect to you guys,

Share This Page