Hi there,
I am modeling speaker crossover for
bookshelf ~6L BR box with SICA 5.5 coax tuned ~67Hz, 40-50g of volume filling approx.
Already was stuck on faulty mic, faulty amp, several measurements with human (mine) error and so on. Such a simple task of measuring everything went extremely long and tedious. Maybe there are more my errors somewhere, but I cannot find them. Please correct my workflow.
Now I have everything in VituixCAD measurements done according to Mr. Kimmo Saunisto bible with REW and transferred to VituixCAD. Everything done more or less by the letter, only measurements made every 15 degrees, not 10, and only horizontal plane. It is a coax, so for these first simulations it will be enough.
The stuff which still gives me headache, from simple to more advanced:
1. Given the fact, that coax woofer part of driver which is positioned more for mids, not low notes - how much "ok-ish" is my BR volume? I want them to play as bookshelfs without subwoofer. Maybe some time I will add one if I build one I like. My taste is no subwoofer at all than cheap one. But this is topic for another conversation. My tactics was lowering the inside volume of the box from ~9,5L to ~6L, leaving everything the same. It sounds little better on smaller boxes, all the simulations and measurements made in 6L 68Hz BR box. So - is this acceptable BR tuning? It is far from perfect, but going for higher internal volume and lower tuning freq - it is outside this driver comfort zone and pretty ugly simulation curves. Doing opposite kinda defeats the purpose to have them as the only low freq sound source. Manufacturer recommends 8L 68Hz or 9L 58Hz boxes. What box size and tuning is your choice?
2. Impendance measurements and simulation. I am still building simple homemade rig for measurements with 100Ohm resistor, there is no DATS on my table, and probably there will be none in near future as my relationship with Dayton is bad ATM. So for now I am using manufacturer provided impendance curves, which probably are measured on big standard baffle with no box on the other side or any other random method. BR box adds another spike in impendance, in the tuning frequency, and, if I am right, not much in the upper range? Crossover in my simulations is ~2,5kHz, so is it safe to assume, that BR box raises impendance in the form on second spike a lot, and has very little impact in the higher frequencies, correct? Similarly with tweeter: it is closed box itself by construction, so no change depending on the woofer BR box, but does the baffle size/shape impact tweeter impendance curves? And finally: is there any tool which can MERGE manufacturer impendance curve with BR box curve? I haven't found in Vituixcad tools one, just the predicted one as in Winisd and other tools.
3. Harman "curve of preference", VituixCAD score. How acceptable Harman curve is for you personally? I have been experimenting with Vituixcad + EQ APO, soundcard as digital crossover, with pretty interesting results. I screenshoted them at 30 degree listening angle, as 20-40 degree shows smoothest SPL curve, highest score, also sounds most natural while listening
3.1) Manufacturer suggested crossover, for unknown baffle, 9L, 58Hz tuned box. Mine is ~6,2L, ~68Hz. Completely unbearable, too shouty, but not outright bad. You can find similarly tuned small speakers with "HiFi curve" in ASR reviews or in some showrooms. Probably with simple 3-way EQ you could make it even listenable. All the other crossover came from this one. Note not so quickly dropping woofer response up crossover frequency:
3.5) Crossover with best score (without notch filter). Very listenable Probably one to buy components for. 3-order LP on woofer too now. That extra 1Ohm resistor on tweeter makes response smoother, but I doubt that to the level you can hear difference. Pleasant sound, realistic vocals, bit too sharp on the 5-10kHz range on some sounds, don't know exactly where, but not annoying:
3.6) Also with series notch. At the cost of 3 extra components - even smoother response and +0,05 preference rating score.
The other crossover variants are pretty much the same story: even the pretty ugly one can easy score 5,5+, which is unrealistic, so I need to checkbox target curve to get realistic sore how good it is. Which ones to use:
Full space or Constant DI?
SL_ON? SL_LW? SL_PIR? SL_SP?
Which scoring ASR reviewer use? Want to compare my creation to objective tests of other speakers.
My goal is to have pleasant sounding speaker - this SICA woofer has a good timbre, so it is a good building block, want to squeeze most from it. Also it has to be psychoacoustically correct, it will be for some time my main speaker on my desk, my head has to be clear after hour, two, three hours listening in background levels, also several songs per day at louder levels. Probably Harman curve is a good place to start, correct?
4. Tried to simulate all orders of crossovers, the ones that produce workable curves are: 3 order on tweeter and 2 or 3 on woofer. Only two combinations. The question is simple - if we have VituixCAD, most of the answers about the correctness of slopes etc. are answered in what simulation shows? There is no rule to make it symetric-assymetric? Whatever works, right? Have no idea how to read impedance bumps and that they mean and where they are ok, and where they are not. Only understand, that there is no BR bump in my simulation, already asked about it. So, besides that, are my simulations more or less correct?
Thank you for your patience! All the opinions and facts are welcome. I am now in the level where the more I am going into DIY audio, the more I understand that I suck.