• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Lattice Phase Equalizer

BKr0n

Senior Member
Joined
Sep 6, 2023
Messages
336
Likes
126
Before I get into the weeds of it, I would just like to say that this place is much more invested in the scientific aspect of audio as opposed to the "flavor" of it so thank you for existing because all of these DIY audio forums are deteriorating what little sanity I have left by giving too many opinions as opposed to raw data.

Moving on! (oh and warning: INCOMING WALL OF TEXT)

So I'm building a set of speakers. I was lucky enough to find a good deal on KEF SP1632s so I decided to pull the trigger on a new project. I'm pairing them with the Dayton E150HE-44 so I can have a compact 3-way speaker with good overall frequency response. While trying to design the crossover I came across something called zobel networks, and the lattice phase equalizer. I thought it would be a neat idea since these filters are good for not just splitting frequencies, but maintaining the phase relationship of a balanced line. So I did a rough draft of the design, did some math, downloaded VituixCAD, and tried putting together a prototype.

It was shit. Complete and utter shit.

When I finished prototyping the circuit in CAD, the frequency response curves looked NOTHING like the LR4 network I had in mind. That being said, if someone would be so kind as to give my circuit a once over I would be most appreciative. I'll post links to all the material I used in order to get the numbers I have in the schematic, as well as said schematic and the info the CAD is spitting back at me.

Crossover calcs (https://www.diyaudioandvideo.com/Calculator/ApcSpeakerCrossover/)
Zobel/Lattice info (https://en.wikipedia.org/wiki/Zobel_network) (https://en.wikipedia.org/wiki/Lattice_phase_equaliser) (https://physics.stackexchange.com/questions/301342/transfer-function-of-a-lattice-phase-equalizer) (https://audiojudgement.com/impedance-equalization-circuit-zobel-network/)
Sub specs and CAD info (https://daytonaudio.com/product/191...-mmag-extended-range-subwoofer-4-ohm-per-coil)
Mid/Tweeted specs and CAD info (https://www.audioexcite.com/?page_id=3614) (https://www.diyaudio.com/community/threads/small-3-way-with-kef-sp1632-sb-acoustics.233572/page-2) (https://us.kef.com/products/r3)

While I'm here, I figure I would also ask about the part selection as well. Since I'm only going to be going for about 50-100 watts handling, could I use high power RF on-board components as opposed to the large chunky passives ordinarily used in audio circuits? I figured A) since I'm not going overboard on power and RF chips usually can handle a few amps why not? B) It would give me an opportunity to use precision components as opposed to the +/- 3-5% on typical audio components, and C) Despite them having much higher specs, they're still cheaper than Mundorfs lol

Looking forward to the responses and thanks for the help!
 

Attachments

  • VituixCAD XO-schema-1.png
    VituixCAD XO-schema-1.png
    36.2 KB · Views: 98
  • CAD curves.png
    CAD curves.png
    879.5 KB · Views: 95
First off, you missed the obvious place for a topic like this,

My first thoughts:
Good Lord(e), you'll go broke on parts! :eek:

What kind of masochistic tendencies drove you to attempt a passive crossover of this level of complexity with no apparent prior experience? And why do you insist on a balanced circuit (doubling the number of parts) in what is effectively a floating transducer that only cares about the voltage between its input terminals?

So do I have that right - you first designed the whole thing using calculators that assumed ideal, purely resistive drivers of identical and constant sensitivity, then entered the whole shebang into VituixCAD with real driver / baffle data only to find the result a real mess?
Well, duh. :facepalm:

That approach was, I'm afraid, totally and entirely backwards. The good news is that, now that you've got everything in VituixCAD, you stand a decent chance of actually coming up with a crossover that works, after starting from scratch. (I think there's also some problems with how things were entered, looks like you effectively have a super high source impedance. The whole baffle aspect seems to need some work as well, all your drivers appear to be in the same place for one.) A tweeter in a coax is going to behave much like in a waveguide, so expect its raw response to rise a fair bit towards the low end (could be 10+ dB for around 100ish per 2.83 V). Things were never going to work out like that.

Hmm, your midrange comes out to 93 dB/2.83 V (at 4 ohms), while the sub driver is rated 83.8 dB/2.83 V (it's an 8 ohm 5.25" sub driver, so no surprises there). Not a very good match. You'd end up with a passive speaker of pretty lousy sensitivity, with high crossover complexity in no way helping matters. You may ultimately be looking at 81ish dB / 2.83 V, with a fair bit of power wasted in dropper resistors (about 2 W for each watt going to the coax). That would be a prime case for going active, only the mid-tweeter XO could remain passive just fine. Or use 4 of these sub drivers, or two 6.5" woofers, or a 10" woofer.
 
Last edited:
First off, you missed the obvious place for a topic like this,
...oopsie
What kind of masochistic tendencies drove you to attempt a passive crossover of this level of complexity with no apparent prior experience? And why do you insist on a balanced circuit (doubling the number of parts) in what is effectively a floating transducer that only cares about the voltage between its input terminals?
One who is very bored and not afraid to go back to the drawing board :D
you first designed the whole thing using calculators that assumed ideal, purely resistive drivers of identical and constant sensitivity, then entered the whole shebang into VituixCAD with real driver / baffle data only to find the result a real mess?
...I think i missed that part xD then again it's been almost a decade since I've touched any kind of spice or cad software
The good news is that, now that you've got everything in VituixCAD, you stand a decent chance of actually coming up with a crossover that works, after starting from scratch.
Already there! Decided to scrap the idea due to parts and complexity. Just made the basic crossover so I can see what the curve would look like "on paper" in VituixCAD. I'll go from there.
Hmm, your midrange comes out to 93 dB/2.83 V (at 4 ohms), while the sub driver is rated 83.8 dB/2.83 V (it's an 8 ohm 5.25" sub driver, so no surprises there). Not a very good match. You'd end up with a passive speaker of pretty lousy sensitivity, with high crossover complexity in no way helping matters. You may ultimately be looking at 81ish dB / 2.83 V, with a fair bit of power wasted in dropper resistors (about 2 W for each watt going to the coax). That would be a prime case for going active, only the mid-tweeter XO could remain passive just fine. Or use 4 of these sub drivers, or two 6.5" woofers, or a 10" woofer.
Didn't even think of that. It's going to be for a small space so I figured even at 80dB it would be fine. I've considered going active, but if I did that then design complexity would be a whole different ballgame. I went passive to save money and to be able to upgrade down the road (can see how great that's going :rolleyes:) Additionally, the sub itself is dual voice coil so i could wire them in parallel to reduce overall impedance.
 
A Zobel is not going to help with that. It's purely a matter of load current required and output loading (with resulting crossover distortion), not to mention being (un)forgiving of poor contact. Most HiFi amplifiers just aren't built with 2 ohm loads in mind, and some AVRs you don't even want to bother with <6 ohms if you can help it.

You can either drive a dual voice coil sub with two amplifiers in BTL (if in series) or with two amplifiers combined in parallel (if in parallel). The result is fundamentally equivalent (each amplifier sees a 4 ohm load), with the BTL solution keeping you from having to pick resistors to combine the outputs (which must be small enough to not degrade damping factor too much but big enough to keep unwanted current flowing between the outputs at bay).
 
*sigh* I guess I'm going active... my poor wallet
 
Back
Top Bottom