How to design a high end crossover…

My joke as the sun rises…

Requirements for the casual designer:

$4k worth of reference premium inductors, capacitors and resistors just laying about…

Zero out speakers on manufacturers specs before 5 pm.

Add 1/5 of excellent bourbon, branch water, natch.

Test each driver on a, in the old days scope, Ha! 
Computer program or four…

Fiddle with 1st-4th level crossovers for each driver, in this case, in a three way system.

Play your favorite test tracks, Opera thru Rock, change X-over components, pushing and pulling, repeat till the sun rises, or the victim slays the opponent, (manufactures x-over), on the audio analyzer, then refine with the ear, (having been to every concert on that albums release), knowing what the artist intended…

Thank Mom or Dad for the leisure afforded to you to do this ad infinitem.

Love the newfound resolution…

Plan B: Make money, know when to quit, play with this stuff as you personal inside joke.

Wait for post to be retracted… Go to hammock…

     If you're thinking about taking up comedy*; DON'T quit your day job!
                                            *or: making sense
A: Yes it is, and no, it isn’t.

To "design" a crossover you need many more types of inductors and caps in a specific range to zero in on proving the math to be true by ear. It’s an art, after the science.

An example would be, a 1.2 mh inductor, air core and iron core in 18, 16, 14 and 12 gauge round wire, depending on frequency, and that equivalent in air core foil at their equal to determine which is correct. All have different sound.

Surprise, the more expensive ones sound better with better drivers… Iron core 15 gauge inductor, $10. Same value in foil 14 gauge, $38. It becomes obvious that material costs and quality sound reproduction are at odds with one and other from a product price point to consumer cost.

Same is true with caps, because they all have different attributes that has to be taken into consideration.

Then you have to understand that whe combining drivers, one XOver may be fine with a first order, but you may need a fourth to the next driver.

I thought it was all about math, but I was wrong. Even the wire interconnects color the crossover differently.
So, don’t go there as a critical endeavor unless you want to spend countless hours after the calculations to listening to music you know well over and over and over and over again, till it sounds natural. And a couple more overs after that.

I’d venture to say that someone that does this for a living on the high end easily has $20kmin stock parts that they know well to do this. Many good to better caps cost in excess of $300, and the quality of those is a total crap shoot when it comes to a final design.

You could easily wind up pairing a particular driver with a Bennic as a Mundorph EVO Silver Gold Oil. Look up the prices on those, it’s a revelation.

Designing a good passive crossover requires making good driver choices in the first place and it’s all about the ’the slide area’, as Ry Cooder might have called it. How much can you slide the crossover ranges around?

Or, the area of adjustment overlap. How linear and stable the drivers are in the crossover range, and how much overlap there is in the given frequency ranges. It’s an ugly task that takes time to master and it is all compromise, compromise, compromise.

A digital crossover is all compromise as well, with the major fault of being severely quality limited (brick walled) before you even begin. Best not go there.

So we’re left with active analog and passive crossovers, for those who seek quality in their gear.

As for costs, yes it gets quite expensive.

First, one can get to a linear space or place or minimal agreement spot, with inexpensive components....and then venture into the pile of expensive parts to test via. Then slowly zero in on the right spot, the best compromise. Zobels can be important, in my experience. lightly applied zobels, not over-damped zobels, can make a difference. I have a pair of those passive adjustable crossover boxes, which can really help in the initial stages. then add in expensive parts, then add in the premium parts and slowly dial it in. It can be months of time, if one does it right.

High end audio can be and generally is a ridiculously time consuming job with a requirement of extreme capacity in skills and whatnot.. with little room for reward at the end (re sales and the like). 

basically start with the vidsonix virtual crossover boxes, then move to the same but done as a 3/8ths mdf board with strategically located quality binding posts and ground runs, for a test system/board.  

then move to the premium parts on the test board, then move to the more final design and arrangement/layout of the premium parts, and then finalize it all in a single finished package, and then it's the final test run with the gear to be issued as product. Fixing any mistakes found, along the way.

Or something like that. It's a long process - to do it right.

My dream job, if I were still working, would be in high end audio. I truly understand what those people must go through, day after day now.

Create a crossover, love it until you hear George Harrison’s voice on All Things Must Pass, and get totally torqued at his transition between his normal singing voice and his upper range.

Go to this company’s website and look at their capacitor summation; it's brilliant and back breaking at the same time. They easily have to have $30k in caps to rate all of these at multiple mf's.

I'm stuck on this hobby for myself, gonna build my own speakers after I get my soon to come Omen's, but like my mentor said to me 45 years ago: You can learn this or become a doctor, it will take the same amount of time.


What one ends up finding is that it can come down to the tweeters being in the top spot re importance in good design, and that there are only a small handful of ’best’ tweeters available, period. Like five or six of them. Maybe.

The rest are just compromises with little room for fitting correctly and are just flash in the pan one legged single hit wonders.

Where even the $300/ea tweeters still need to be torn down and re-assessed as to their design and build choices, where they are re-executed to a higher standard.

we see things akin to this in all other areas of endeavor and technology - the requirement to take the best in components and move each to a higher standard. Cars being a huge one..

Loudspeaker design is no exception and this should come as no surprise.
digital for loudspeaker crossovers is a sack of effluent when it comes to absolute quality attainable.

I can’t help if it one can’t hear or understand that. maybe one needs to go back a class for a year or two. or maybe find a way to up their hearing skill set. Hint: Projection does not work, but self assessment does.

Digital might work in the future when it comes to moving to the top spot in quality but it is not there yet and I see nothing in the immediate future that changes that assessment.

If digital for crossovers worked as some might desire them to, then all the best loudspeaker companies in the world would be issuing product that way--right now. Note they are not.

I’m sure the wiser among them are playing with digital crossovers...but that’s been true for well over a decade. Yet, nothing issued. They are all waiting for the technology to improve to be close enough to meet what passive design can currently do.

Simply put: not yet. Timeline: indeterminate.
DSP active left passive behind years ago. Thankfully the professionals know it. The best speaker companies in the world are using active.
I don't doubt that. My studio stereo has one built into the amp, and it's just amazing. 
It's a Yamaha pro love audio unit, and fiddling that part of it from time to time, it just amazes me what digital can do, and I never thought I’d ever use that function, I just bought it to use as a cheap amp with a lot of watts.

Yes, the highs and the mid’s are where you open your wallet and let the supplier take the money out, and thank God for that, since the higher in frequency you go, the smaller the mf needed. 

It's the difference between being familiar with what you’re hearing as opposed to seeing the musician perform in front of you in your minds eye. I simply can't believe that I am listening to the same speaker.

It all started when I stumble on DIY YT videos of people moding speakers for themselves or professionally by accident, and then start binge watching them. Danny at GR Research and a fellow named Joppe Peelen, as well as a bunch of others in the led me down the road to sin… 😂 

This is a random one from Joope:

The only speakers I own, or have owned, that I haven’t taken apart and modded are My KEF R's, just brilliant design and manufacturing quality that was beautiful to behold but I dare not fiddle with, lest I mess them up.

But yeah, I do take them all apart and check out pretty much everything just to see what makes them tick. That goes for just about any mechanical or electronic thing I own.
There can be a whole lot more than picking crossover points, slope, and particular components.  A lot of designs include networks to compensate for phase, to notch out certain frequencies, to fill in certain frequencies, etc. 

I saw an advertisement for the YG Carmel speaker that showed its crossover.  This thing had well more than a dozen caps, about a dozen inductors and a whole lot of resistors, and this is a two driver, two-way system.  Obviously there is a whole lot more going on than just a high-pass and a low-pass network.  

This photo shows one of their designs that incorporates cheap to expensive inductors for their specific attributes, and bottom drawer to top shelf caps.

It's a science that’s an art.
I wind my own inductors. Easy peasy.

I have a capacitance board. Any value +/- 10 uF.  Same for non-inductive resistors.

I always use 2nd order. Not L-R.

Its painless.


Specific to bass reproduction.

I just received 5lbs on 12 ga inductor wire yesterday, 14, 16 etc coming. I have two lathes and am building a setup for the small one to do this on. This will help out with costs when you make 4th order, specific, to eliminate the need for a notch filter.

I’m also interested in trying absolutely everything at a particular number, I’ve chosen 1.2 mh, that I can use to determine pluses and minuses, the only ones I don’t have yet are a 10ga foil wound and a 12 ga Litz wire, but am ordering the Litz tomorrow, Solen, with my regular order from that supplier.

In a closed range measurement, 20-200 hz for extra precision, the foils are superior in my results. Not so much by manufacturer though, Dayton’s are not bad at less than ½ the price of some others.

All of this is not going to go to waste, after I scratch this itch, I’m using some existing good quality orphaned drivers for testing, as well as existing speakers I have, and will build some speakers out of the parts and sell them if I don’t replace my existing speakers with them.

This is where the DSP comes in handy, as great inductors for subs are ridiculously expensive, and I find them perfect for subs. I just wish more sub makers would consider that many people are buying dual subs now, and so make the upper range XO on the smaller ones at 200, so they can make an excellent combo with small bookshelf speakers, where the sub is used as part of the speaker stand.
I’d venture to say that someone that does this for a living on the high end easily has $20kmin stock parts that they know well to do this. Many good to better caps cost in excess of $300, and the quality of those is a total crap shoot when it comes to a final design.
It doesn't need to be that complicated. Every designer has a "house sound" as I've heard it labeled. The actual electrical values often play an even more important part in the overall voicing of a speaker, than the quality.

Matching drivers that play well together (pardon the pun) is the very first step, before you even slice up the signal, to get a well resolving cohesive and dependable voicing.

Anti-phase measurements between crossover points (slopes obviously) where the drivers are at their best sounding are a good key way to get them to seamlessly blend across the frequency range. The argument that one driver alone that can do it all, that argument has merit.

Even DSP, requires knowledge of where to cross the drivers, if a fully active approach is to be undertaken. To me, it's the driver interaction, before the crossover is even implemented to shape the signal the drivers are fed, is the key to high end.

You can use relatively cheap components throughout, even the drivers to some extent, but if they won't play well together, if one simply won't vanish into the overall sound and reveals itself with colouration - your crossover cannot possibly win. DSP, fully active, with a crossover or not, driver matching matters.

Is this guy joking?
This is why a speaker is good or not. IMO, the hardest thing to accomplish is High End Audio. It is v far from just selecting the best components. It takes an artist to integrate drivers, crossovers and enclosures. To make the final speaker sound like music is rare.
For an amateur to accomplish this and, even, professionals with years of development.
Add 1/5 of excellent bourbon, branch water, natch.

Is this guy joking? 

Anyone with a fifth (or even 1/5) of excellent bourbon under his belt, yeah, I think he is joking.
Well, I didn't drink the whole bottle!

But yes, joking a bit.

First time in years I worked on a project and was surprised to look out the window and see the sun was rising.
I new one XO designer that would tune each XO to the cabinet and drivers he was using. The left and right crossovers very seldom looked alike. When the drivers were pulled and matched top to bottom and left to right as close as his staff could get, they would load wire and seal the drivers.

The Passive XOs were point to point mounted to 3/4 pre formed MDF or HDF boards. He used two LPads and parallel quasi 2nd order for a 3/4 way.

Here is the real trick, he would put an untuned xo in, position the speakers in the same place every time and take tone burst readings.

He would add a small cap in the band pass to one side only, and normally unwind and remove a small amount on a hand wound inductor (made from a real dry 1/2" dowel and hand wound copper wire 20-18 ga) It had 16-20 raps. He would remove 3-5 wraps on a dowel, and retest the DB levels. That is how he balanced the DB level in the BP and lower or raise the pass side with a small cap of the same quality.
It’s not a bypass cap (super expensive), it’s a trim cap normally a long life poly or something.. It act differently, the same with the inductor.. It’s for trim ONLY doesn’t effect the SQ just the tonal balance..

THAT is how I learned to TUNE an XO. Making one it’s not rocket science but, you really need to be able to TUNE and know HOW to tune your work for a SOTA audiophile quality XO.. Production crap who cares..

The guy could build and tune 3 pairs of XO in a night, after spending all day selling, ordering and talking to customers..

The staff would ready the call in orders on the spot.. or paint the inside with sound coat or black hole or the different wire or wool vs fiberglass in the mids. 3 guys I think Brian C, was 1. RIP.

I stopped in after work 100 times, he would be there trimming OXs or making them or writing an article.. or getting ready for CES. the soldering iron he used, you could have soldered gutters it was so big.... no kidding.. ;-)

He liked TRT Caps too,, Teflons caps sent straight from Satan,, really tough to break in. But when they do. 3-500 hours.

I saw a few slow to learn customers. He'd tell them, cover the speaker with moving blankets, use the cables your going to use, plug them in put them side by side and let them play at 10-30% (vary the volume) for a week or two.. NO KIDDIND. Then go back, position the speakers close, let the cables settle for an hour or so and listen.. MAGIC, every time..

First time in years I worked on a project and was surprised to look out the window and see the sun was rising.

In that case then I tip my hat to you, sir. Even in my prime at WSU I never was quite able to drink straight through to dawn.
Allow me to over explain.

The process, then you can judge me to your hearts content, based on the facts and without all of the speculation. And of course, I never retract anything, so I’ll slowly try a nudge this post back to discussing what is pertinent to the original intention of it, designing XO’s.

I’m retired, so I can spend 40 hours a week doing this, and still have time for other hobbies and working on my house. And if I stay up all night working on something? Just working while I’m in the groove, as opposed to in a rut, which you can do when you’re the boss.

Start at 4:30 in the afternoon, Check previous records and notes; work on the breadboard to build the XO prototype, hook up, run test sweeps, make changes and repeat till it looks better than the last best version from previous sessions, based on thinking about how to improve the design from the last session.

Make drink and move to listening chair, start playlist of selected speaker instrument tones and demo songs and listen for about 62 minutes, taking notes on good points and bad. Back to computer, log notes.

Repeat till 7am, clock out.
So yeah, I thought it was funny and posted it as half joke, half truth. Four listening sessions over that period of time? My God, could he walk back to the house?!
If you want to find fault, send me a note through Audiogon and I’ll be happy to provide you with all of my faults so you can really tee off on me with maximum effect in the future.

But if you can’t relax with a drink and listen to your stereo because you are alcohol intolerant, your religion forbids you from drinking you’re in AA, or your psych meds won’t let you have a drink, stop telling the world that on my post.

We should really be talking about trying to make the best XO that one can here, costs be damned, and time is not a major concern. Designing an XO like this, as Danny at GR R says:

People think I just sit down and punch in a few numbers to a computer program and assemble the parts, etc. That is nothing like what happens…

It >is< nothing like that; Danny knows what numbers to punch in after testing the drivers individually because the published specs are merely a guideline for selection, and are not dead on to the product that shows up at your doorstep. But he is also wise not to divulge too much because what goes on in his head is Intelectual Property, and important for his livelihood. 

I have one of his XO’s, and no matter how hard you try and get the specs out of me, that will never happen, because I was an artist, then a Senior Designer in ID, and I understand how important trade secrets, copyrights and patents are to making a living/profit. Someone uses one of my photographs or part of something I designed for profit? I’m going to sue them.

That is why I am amazed the Humble Home Made HiFi publishes photos of their XO builds, because of the IP they are disclosing.

Look at these photos, if you know what you are looking at, it’s all there. Of course you may have to put off that trip to Hawaii, for 4, to buy one, let alone with the accompanying speaker it’s in, but you can read the caps and inductors makers name, and what model from their selected line. Of course these are specific to a particular set of speakers they are designing for, and they don’t show those, but still…


You really do need a a soundproof supply room to break a lot of this stuff in, speakers to caps.

I'm speculating early on in my research on cables so far, without a complete understanding, so FWIW: I do think there is something to your ears becoming accustomed to a new component, so breaking them in; but I also think there is something to breaking something in like a speaker cable, from a physics stand point, and that could be something as simple as the cabling's jacket shrinking from mild heat and off gassing, changing the relationship of the conductors distance from one and other. 

I do not think, at this point, that the physical nature of the conductor changes after manufacturing changes, unless something like cryotreatment is applied, or the wires are annealed after the drawing of them, tempered or surface smoothing, or a combination of those, to make a superior cable. 
Caps? Yep, physical and chemical relationships are being ingrained in the parts memory, bringing out it's full character.

Inductors? See wire finishing.

I'm self taught, so don't have anything to relate from others who I've learned from except what I learn from my testing, and other people affecting those judgments by my studying their designs.

A cheep tip. Not on XO's but…

Want better results from standard SxS speaker cable? Mesure out length you need, adding a couple of feet. Place one end in a vice, one the other attach to an VS electric drill. Slowly twist the wire under tension till you get 6 twists for every foot, then anchor the drill with a weight. Set the twist with a high powered hair dryer, careful not to melt the wire, let cool. 
If it didn't keep the twists, it wasn’t heated enough, or you have a high temp jacket, and so have to add twists in the beginning knowing that the cable will relax.

Now you have much better speaker wire.
Oh, one other thing, all of the Humble Home Made HiFi's that are round wire coils use Litz wire, out of all of their photos, only a couple are not.
william53b, congratulations on your discoveries. It's a humbling thing to realize that one's preconceptions are faulty, and your admission of it publicly may be of help to participants here. An a priori commitment to measurements prevents many from advancing their system, one reason for my adage; the greatest impediment to advancing an audio system is the audiophile.  

You may find my review of the Aspen Acoustics Lagrange L5 MkII at, my most recent work, especially interesting among the speaker articles I have written. I would guess that much of what Scott Kindt of Aspen Acoustics is doing in creating his speakers would be of value to you.  



Some people only post to prove how much they know. I'm conversational by nature, so like to post and reply that way.

Besides, I made headway the other night, and that's what truly matters, is t it?
OP thankfully I haven’t had to break in to many pieces of equipment, cabling or otherwise over the last 6 months or so. A phono/tape pre amp and my V12r had new parts that took over 200 on the Cary and I’m at 300 on the Decware.

Cabling, literally 1000 hours to plug and play within 24 hours of hooking them back up..

I’ve seen people plug drop and call it good.

I’ve seen others carefully remove their nurtured, conditioned and static bagged prizes. String them out NEVER touching the floor (on risers). THEN discharge themselves remover the static terminal covers add contact enhancer and torque the securement nut to 50 inch pounds, back off 3 times and a 4th torque to 60-75 inch pounds.. The amp terminals first then to the speaker end, discharge again through the ground side of the cable and hook that end up.. Picky.. I’m picky and it pays great dividends

Maybe some folks like the "BRAND NEW" sound, me I’m running the other direction a full 2 weeks on a phono stage, cable, cart change and tonearm rewire.. Preamps, amps and cables 24/7 will do. A week of playing, no listening. I only turn them off to do a cable check (tight). I monitor with a thermal gun. I can usually tell if something isn’t quite right. Left to right temps and ambient vs gear difference..

Paying attention more to thermal changes than anything else..

The shops, wood and metal, THEN. Now, Wood and storage, I HATE working metal any more, just hate it.. I hung up my mechanic tool, thank GOD. I never thought I’d say that.. 14 year old working in service stations until I was 62, and came up with a broken neck. Two disk and a lot of bone chips..:-) I HATE wrenching now, at least the HD side.. Electronic side easy peasy. OBD2 and CAN buss is on everything now..

This stuff stereo, is a walk in the park to fix or make for that matter, really it’s NOT rocket science..

More like a door on a 64 VW with a whole lot of lipstick on it.. That people charge a whole lot of money for.. I got no real problem with that. They better have great customer service though, I will chew them up and SPIT them out.. No problem with that and or a plane ticket..

Almost went to Perth, behind a SA that sold me a pile of $hit.. Money was no object for me.. I should have sicked george on them.. ;-)

LOL we’re talking about how to make a turntable quiet. Try to get a power plant quiet for a offshore drill rigs with 5 generator plants on board, and 25 pissed off workers complaining of FEELING the stupid thing..

Caisson, dam face, deep mine, or pressure work.. SUCKS too.. Hard on the ol body..

I’ll take the turntable, the crossover, or the speaker build for that matter OP Yup Yup.... It's all easier than what I use to do..

Have fun.. Puff Puff pass..:-)

Post removed 
Simple, don't even bother with a XO at the speakers. Instead do the crossover at pre amp level and have separate amps for every driver.
^ that's called a fully active system. It was mentioned that even DSP could be used in that manner.