Building Biamp (Biwire) Speakers

[fpitas]

A primer for those just getting interested in bi-amping:

https://sound-au.com/biamp-vs-passive.htm

As has been said above, bi-wiring is generally feel-good stuff. Harmless but pointless.

And my obligatory statement about active vs passive crossovers. Passive crossovers very often are not "conventional" filters like Butterworth, Bessel etc. They are shaped to optimize the driver acoustic response, taking into account the raw driver response. Essentially, EQ is built into them for the particular driver. With active you are generally stuck choosing from an assortment of conventional filters, so EQ is necessary to flatten driver response (generally an octave beyond the crossover point) and take care of obnoxious peaks etc. I've used active for many years, and I wouldn't dream of going passive, but it's not the simple slam-dunk that internet articles would have you believe.

[Chris Roemer]

On the woofer, you size the series coil for the proper baffle-step compensation. You'll likely need a notch (tank) filter on that one for that peak above 2kHz.
After you see where that falls, you can design an L-pad to cut the tweeter output to match the woofer's level above 1k or so.

Here's what I get -
2nd order LP w/tank (across coil) and Zobel(like-thingy): 2.5mH (low DCR) series coil, then a 12uF shunt cap (to gnd), lastly (next to - or "across" the woofer) a RC using an 8n(ohm) resistor and a 10uF cap. For the notch filter (across the big coil) use a 4n resistor and a tiny 1.5uF cap.

2nd order HP (tweeter) w/an L-pad (about -5dB of attenuation): a 8uF series cap, then a 1.0mH shunt coil (can be #18 or even #20). On the "L" SR/PR = 4n / and 12n (series and parallel resistors).

Runs around 84dB (+/-2) w/Fc near 1.5kHz. Full BSC. Solid 8ohms. The "Silky" (-070) dome is -18dB down at resonance, so it MIGHT survive!
A better choice (to cross that low) would be an RS dome.

[zinger084]

So I landed on the following: I'm sticking with the Dayton Audio DS270-8 10" Designer Series Woofer Speaker but went with buying a couple Peerless DA32TX00-08. The low Fs I think will help me. I just pulled the trigger on a 4th order LR crossover set of components to try and axe them down each quickly. I'm going to RLC the woofer and the tweeter (first time i'll be attempting a crossover and RLC from bare components, but I do have a reasonable amount of circuit board and component experience).

So, another few questions.

Since I went with a 4th LR (and not a 2nd LR where i needed to reverse tweeter polarity) do I need to reverse on a 4th LR to avoid that nasty dip at the crossover frequency?

Also, with the RLC networks, does it matter which order they are placed in or what positive or negative they "start" from?

Thanks.

[Chris Roemer]

You'll answer your phase ?s in your sim.
My XO still looks pretty good, but change the HP filter to a 6.8uF cap and a 0.90mH coil. Raises Fc slightly to 1.6k.

I'm lookin' at mid 30s in 2.5cf w/a 3" "Precision Port" that's 7" long.

[zinger084]

So another thing just hit me… on another set of speakers I’m designing, since I’m using two woofers instead of one (again, on another separate set of speakers separate from this 10” and 1-1/4”), does that change any values on the crossovers or RLCs?

[buggers]

Amazing would be a bit over stated. Perhaps adequate for a non audiophile application, and for starting out in the designing rabbit hole.

[fpitas]

Here we go again.

[zinger084]

Here are my 4th LR values…

[fpitas]

You really can't design a crossover w/out using .frd (freq. response) and .zma (impedance) data files in a simulation program.
Most here would not run a 10" up to 1.5k, or even try using it in a 2-way (not using a dome tweeter, anyway).
Care to share your drivers?

OP: Please read that carefully. Online calculators are a joke.

[zinger084]

Sorry guys. Just trying to learn more here. If I’m wrong and troubling you too much, I’ll move onto elsewhere.

[DeZZar]

Sorry guys. Just trying to learn more here. If I’m wrong and troubling you too much, I’ll move onto elsewhere.

Trouble away...that's what the forum is for. Just take the advice you get on board given the sizeable knowledge and experience gap between yourself and those that have taken the time to contribute to your thread thus far.

In particular...Abandon the online calculators and never ever visit them again. They are literally worthless. Seek out crossover simulation software like VituixCAD or xSim in order to even come close to successfully designing your speakers.

[zinger084]

OP: Please read that carefully. Online calculators are a joke.

Interesting... maybe a little more background would help. I've referenced the following websites to build my own crossover design google sheet:

https://calsci.com/audio/X-Overs.html

Chapter 91 in http://www.bcae1.com/

as well as http://users.on.net/~isaacmcn/audio/impcomp/impcomp.htm

I was feeling pretty good about my spreadsheet that using T/S's from 2-way systems, along with Fc and Q values yields Crossover network reqs, zobel reqs, RLC reqs, and l-pad reqs for my network. Should i not be referencing these online formulates and calculations?




Also, does this help shed some light on my experience level?

I'm soliciting recommendations for software to use and will look into those DeZZar mentioned...

[fpitas]

Yes, if you add Zobel networks the online calculators become vaguely useful, but with lots of unnecessary parts.

[wogg]

Interesting... maybe a little more background would help. I've referenced the following websites to build my own crossover design google sheet:

...snip for size...

Also, does this help shed some light on my experience level?

I'm soliciting recommendations for software to use and will look into those DeZZar mentioned...

Very interesting! I'd wager you'll have an easier time than most. So are you using the T/S parameters to calculate the impedance per the Fs, Q, and Le specs? If so, that's way better than calculators that presume 8 or 4 ohms across the whole spectrum. It's still missing the acoustic data though, so the electrical roll off may be pretty close but the final acoustic crossover will be off.

I've done write ups on my site about modeling. They're getting a bit old, using Bagby's Excel tools and WinPCD which are still quite good. I've been moving toward VirtuixCAD myself though for a one stop application that covers many of the required tools like diffraction simulation and near field / far field response blending.

Sim from manufacture data / traces: http://woggmusic.com/advanced-speake...gn-simulation/
Real life build with detail on making measurements: http://woggmusic.com/supernova-minimus-speaker-build/

[zinger084]

Very interesting! I'd wager you'll have an easier time than most. So are you using the T/S parameters to calculate the impedance per the Fs, Q, and Le specs? If so, that's way better than calculators that presume 8 or 4 ohms across the whole spectrum. It's still missing the acoustic data though, so the electrical roll off may be pretty close but the final acoustic crossover will be off.

I've done write ups on my site about modeling. They're getting a bit old, using Bagby's Excel tools and WinPCD which are still quite good. I've been moving toward VirtuixCAD myself though for a one stop application that covers many of the required tools like diffraction simulation and near field / far field response blending.

Sim from manufacture data / traces: http://woggmusic.com/advanced-speake...gn-simulation/
Real life build with detail on making measurements: http://woggmusic.com/supernova-minimus-speaker-build/

Excellent links, thanks for sharing. I'm going to look into SPL Copy and WinPCD to start. I assume these will be the tools I need to get the proper driver files into digital format and then be able to run sims on these files.

I also will look at VirtuixCAD, as it appears it doesn't need Excel and can do it all (design crossovers, model plots, etc)?

[Chris Roemer]

Wogg knows his stuff.

To "see" (understand) "baffle-step", you can use the (free) Tolvan's "Edge" software. You can make up a baffle and position your drivers on it and SEE the (approx.) -6dB of "lost" bass that you end up with when not taking it into account.

Here're your 4th order filters on your woofer (green) and tweeter (blue), along w/MY filters (orange & yellow) and their (theoretical) "sum" (the white line).
The "apparent" rise of the white line (below about 1000 Hz, typically) - (and the orange line, the DS270) is making "baffle-step compensation" (BSC). Once the speakers are placed up off the floor (on stands), and away from walls, that "rise" will flatten out (to essentially a flat line).


Here - v - the green curve is your 2nd order filter (woofer) and 2nd on tweeter (blue). The 2.5kHz breakup is actually quite horrific here.
(Note - MY suggestion is also only 2nd order, but with an added Zobel and a notch filter (across the woofer coil).)


We (sometimes) use a Zobel on a woofer to keep its impedance from climbing (over about 1k).
Here you can see the raw woofer (grey/white) climb to over 25n(ohms) at 10k (and beyond).
My Zobel is orange (8n + 10uF). Not only does it (basically) flatten the rise, but I'm also trying to maintain around 8ohms of impedance.
You can fiddle w/the resistor and cap in an XO sim and "watch" what your Zobel is doing!
Your "textbook" Zobel (in GREEN: 7n + 43uF - from the T/S parms) also keeps the rise down, BUT ... it's dropping the Z down into the 5-6n range (quite a "sag" near 1k).
No need for a Zobel on a tweeter (the raw tweeter is the red line w/a resonance around 550 Hz). Note how flat the red line stays (under 9n @ 10k).

SOMEtimes (on tweeters w/out FF - ferrofluid) you need to push down a tweeter's Zmax (@ Fs) 'cause it disrupts a normal rolling off of your highpass filter.
YOUR RLC (compensation) network is the blue line. It DOES push down the Z @ Fs, but (again), it sags quite low (5-1/2n @ 1.2k). Not only THAT, but these are both 8n drivers. If you'd cranked out "textbook" filters for a pair of 4n drivers, I'd wager that the Z would get dangerously low (near Fc), probably near 2-1/2ohms.
I don't even use a compensation network. Tweeters CAN be attenuated by using simple series resistance, BUT ... if you use an L-pad, it WILL push the Zmax @ Fs down some (HERE - the yellow curve, it's gone from about 300% (3x Znom - or 24ohms) down to half (150%, about 12ohms)). Not only that, but I've gotten about -5dB of attenuation out of my "L" (besides cutting the Zmax in half).



In short, YOUR ("textbook", or online calculated) filters would end up using about 14 (2nd order) or 18 (4th) parts for a 2-way (with 4 to 6 coils, including a 14mH on the woofer's RLC network - VERY expen$!ve - and unnecessary). EYE end up using only 10 (only TWO coil$ !), AND ... your "calculators" haven't compensated for baffle-step, or accounted for proper tweeter attenuation (BECAUSE of baffle-step), OR dealt with the woofer cone breakup near 2.5k.

Normally we don't care about (or compensate for) a woofer's Zmax @ Fs (shown here at 30Hz) because we're not trying to roll off its bottom end (using a passive XO filter). Not only that, but in a vented box you'll have a (smaller) PAIR of peaks, with a valley in-between.

(and yes, I AM retired !)

Late note: Most Dayton files COME w/.frd & .zma data files (they're in the zipped file pack under "Manuals and Resources"), THANKs PE !
(these files still need "prepping" for use - esp. when it comes to the "phase" data - the 3rd column in the F/Z files)

A "better" tracer (for drivers from, say, Peerless or Tang Band) is "FPTrace" (? not sure of the name - but user "FPrawn" (here on TT) made for all of us a few years back).

[zinger084]

Wogg knows his stuff.

To "see" (understand) "baffle-step...

(and yes, I AM retired !)

Wow, another great post. Thank you so much. I appreciate the multiple models that you ran and I understand (most) of what you are saying. I can see how it's important to use the software and ssee the results rather than just trust a single calculation. I'm going to spend some time looking into VirtuixCAD tonight and i'm going to buy a Dayton Audio iMM-6S microphone for some measurements.

What i've done in the past is build 2-way systems. I usually use WinISD to model the woofer and then I find a tweeter than crosses close to what i expect (not too low or too high). I purchase 2-way LR pre-built parts express crossovers. That all got stagnant and i want to learn and do more.

This is only one of my three latest builds, so i'll save those other two for another posts. But... here... i just want to see my order of events.

I have the cabinets built right now. I planned 3.62ft^3 net with a 9" port, 4" diameter.

I figured i needed a 4th order to drop the db quick to avoid distrortion on the tweeter and to avoid the nasty hump on the 10" woofer.

So... right now, if i begin to learn VirtuixCAD, can i use that to load my drivers TS and model them to get proper crossover component design? Can i also use it to measure once built and continue to tailor to my liking? etc?

[wogg]

Don't do the iMM-6S, it lacks the pass through 1/8" output to send the signal to your amp. Also, although I use the lowly iMM-6 for measurements, the UMM-6 mic is a better choice in the long run. You can use a proper mic stand and it's going to work with more PC's without relying on an integrated headset jack in a reasonable position on a laptop or Windows tablet.

You should see the crazy "hold still while this sweeps" things I have to do to get a close mic measurement on a cone or port