Crossover $$$$ shock for first timer

[mikec]

Pricing out building a pair of Paul Carmody's "Tarkus" speakers using all PE parts.

Drivers he spec'd were about $200 total but X-over parts totaled about twice that!

Is this a normal ratio? (Now I understand why some folks try to keep the XO's parts count low!)

How do you know when the more expensive part is worth the extra $$$?

[hitsware2]

What are the parts and values ?

[a4eaudio]

Some of these are not quite the right values and some are out of stock, but I was just trying to ball-park the crossover components. (I could easily have made a mistake but this bill of materials ("BOM") is $115.)
One big factor is the 65uf and 125uf capacitors. Paul has a drawing of the xo board and it appears these must be the inexpensive NPE capacitors based on size. Otherwise they are huge. But this is a BIG difference in the total BOM.

[Chris Roemer]

2 things:
1st, THIS XO IS pretty expen$ive, AS ARE most 3-ways. The reason being that the mid driver's filter typically doubles the parts count - it needs a high pass section (like a tweeter) AND a low pass (like a woofer). If you count the parts, the mid's filter in the Tarkus does indeed double the parts. A typical 2-way XO would cost half as much (actually less - the expensive parts are from the LOW cross point between the woofer and mid (usually in the 200-600Hz range). Crossing lower means larger values (for caps and/or coils). Bigger parts cost bigger bucks.

2nd, if you look at that 6mH (LARGE) coil in the schemo., it's not just a "coil" drawing, it's a coil w/a "bar" along side it. THAT means it's not an air core, but some kind of iron core coil. IMO you could easily use Dayton's 6mH iron core here, which is more like $8 a pop, saving you about $20 right there.

Also, you CAN use the cheap (10w) resistors that PE (used to) sell(s), instead of the $2-$3 ones. They perform just as well, problem being that they are being phased out, so the range of values is becoming limited. These would be part #016-xx (10w'ers) and #015- (5w'ers - that COULD be used in the tweeter L-pad).
Also, the 1.0mH coil inline w/the mid CAN be a #20, just make the 4.7n(ohm) series resistor a tad smaller (subtract the diff. in "DCR" between a #18 coil and a #20 - MAYbe the resistor could be dropped from 4.7n down to 4 ??

[Geoff Millar]

Pricing out building a pair of Paul Carmody's "Tarkus" speakers using all PE parts.

Drivers he spec'd were about $200 total but X-over parts totaled about twice that!

Is this a normal ratio? (Now I understand why some folks try to keep the XO's parts count low!)

How do you know when the more expensive part is worth the extra $$$?

This isn't a surprise: when I built Paul Carmody's Classix II, the XO parts were more expensive than the drivers, which were relatively cheap ($A) at the time. Driver cost was A$120; XO and ports cost, $135.

For my other builds, the XO parts cost is/was usually 70% as much as the drivers, and they're all two way speakers or MTMs. I've priced three way speakers like the Tarkus , Dayton Classic Revival, Knuckleheads (on PE's Project Gallery page) and the XO parts cost about 80% as much as the drivers, on average. However, where drivers here are about twice the price of the same product in the US, crossover parts are only about 25% dearer.

The parts cost ratio may change as you go to more costly speakers: my driver cost for Curt Campbell's 'Slapshot' MTMs was $A640, XO parts cost was A$150. The RS180Ps were $70 each at the time, whereas the DC160s used in the Classix were $35, and the Slapshots use a much simpler, and cheaper crossover.

I use good quality, though not expensive caps, inductors of the recommended value and gauge, and good, but not Mills type resistors. Without wishing to re-open the debate about component cost, I can choose to pay $20 for a good poly cap, or $400 (!) for a 'super' cap of the same value, so a three-way build using the 'super bits' would require a new mortgage.

From what I can gather, which may well be wrong, you may be able to use NPE caps or iron core inductors in the 'shunt' parts of the crossover circuit, and that will save you many $$. With some designs which use high value inductors or caps, the designer may indicate where you could use the cheaper parts. For example, the designer of the 'Knuckleheads' specifies a 8mH I-core inductor in the shunt part of the mid-range's crossover.

Other times the designer will advise that, for example, the difference between I-core and air core, or using a smaller gauge inductor will adversely affect the sound.

I think you have to look at it as a necessary expense: the crossover is the speaker's 'brain' and will largely determine how it sounds, so it's money well spent..

Geoff

[billfitzmaurice]

From what I can gather, which may well be wrong, you may be able to use NPE caps or iron core inductors in the 'shunt' parts of the crossover circuit, and that will save you many $$

As a general rule use NPE in all shunt applications, and use them as pass throughs so long as they don't pass anything above 3kHz or so, which means they're OK for midranges but not for tweeters. You may opt for a poly up to 40uF or so if the price isn't outrageous. Where polys are concerned I never buy anything other than the least expensive Dayton. Use cored inductors whenever the value is large enough that cored are available. Not only are cored less expensive, they also have lower DCR. Where the value is so small that cored aren't available look at DCR. Use the least expensive/smallest wire gauge air core that has DCR no more than 10% of the associated driver impedance for pass throughs, 20% for shunts.

[Dukk]

I too recently priced up a Tarkus build. With the components I chose - all air core inductors, all Dayton Poly caps, using multiple components where needed to hit values exactly, I arrived at a total of $488.20, with $287.80 being the crossover components. It was $50 cheaper than the crossover components for a Sunflower build, for comparison.

Personally, I love the look of passive crossover components and generally try to highlight them in a build by displaying them so, for me, it is worth the bucks

[PWR RYD]

I have listened to Paul's Tarkus speakers a few times in different rooms with lots of different types of music. They are really good for classic and modern rock music and nice for everything else. That said, based on the drivers' level of detail and distortion, I would not get hung up on spending tons of money on large gauge air core inductors and big poly caps. It's a great diy design and those Peerless drivers are hard to beat for their price, but I doubt the better crossover components would make any audible improvement. Go with Bill's advice and use Dayton 18 ga. iron core inductors for the larger coils and Dayton non-polarized electrolytic caps for the larger value capacitors.

[badman]

Having such a broad mix of required values makes things expensive. Hybrid filters (active where possible, passive for some elements) can save a lot of components- for instance, if you need your mid high-pass to be very steep but the mid/tweet is shallow, an active filter on the midtweet/woofer XO can save a lot of the bigger parts. Looking at a relatively wide-overlap possibility for midtweet/tweet setup, my parts for a 5khz Xo are all below .5mH or 6uF for a few variants. Do that at 500... 10x larger values.

Active is where it's at for the future. Bigger investment on the first speaker.... but by the third or fourth, if you're using decent parts and not able to re-use most values, you're even, and have a higher-performing solution along with not having a huge inventory of caps and coils.

Anyone want some 10 awg aircores in higher values (up to 5.4mH)? Wish I could get back what I paid!

[devnull]

Don't forget that you can always substitute parts and roll your own. If that boutique priced inductor is outrageously priced why not go up a size to a generic and remove a few turns to bring it down to the correct value. This also works for down to around .5mH starting out with a 1mH iron core for a really low DCR. Also you can use 2 smaller generics instead of one large boutique. As an example, had a crossover with a 18mH to the tune of 80 bucks, but 2 generics were 20 bucks, really simple decision to make.

Just remember to measure your substitute inductors and use those values in your simulation, kinda a chicken and the egg principle as you're chasing the inductance and DCR values. Start out with removing a small number of turns and watch how the values change. Then you can guesstimate the final values to plug back into your simulation.

[PWR RYD]

Don't forget that you can always substitute parts and roll your own. If that boutique priced inductor is outrageously priced why not go up a size to a generic and remove a few turns to bring it down to the correct value. This also works for down to around .5mH starting out with a 1mH iron core for a really low DCR. Also you can use 2 smaller generics instead of one large boutique. As an example, had a crossover with a 18mH to the tune of 80 bucks, but 2 generics were 20 bucks, really simple decision to make.

Just remember to measure your substitute inductors and use those values in your simulation, kinda a chicken and the egg principle as you're chasing the inductance and DCR values. Start out with removing a small number of turns and watch how the values change. Then you can guesstimate the final values to plug back into your simulation.

Pretty good advise IMHO although I've never seriesed inductors . I don't build other peoples' designs but this logic makes good sense.

[badman]

Pretty good advise IMHO although I've never seriesed inductors . I don't build other peoples' designs but this logic makes good sense.

No prob with series inductors- just follow the same protocols you would (right angles and space) for multiple inductors in your layout. You might be able to squeeze more inductance out of some coils by "Stacking" them (allowing field interactions), or other manipulations, but generally speaking, I wouldn't recommend using them any different than other components as it could be unpredictable and misbehave.

I will say, having big poly caps is really handy for active systems. A little thump protection on my horn tweeter is welcomed since there's no other passive XO, so there's 250ish watts behind a 107+dB driver. If my amps weren't well-behaved, the cap (200uF) would need to be substantially smaller for protection, but then you get into the protection cap acting more like a filter element.

[djg]

Wolf winds his own coils. I'm waiting for him to start making his own capacitors.

[Chris Roemer]

$ave $33-$45 by:

Coils: for the 6mH (which Paul shows as iron core on his schemo. anyway), 1.0, 0.35, and 0.20, use: 257-572, -048, -030, -024: saving $11 +1 +1.5 +3,
which is $16.50, $33 on the pair.

Your caps are almost as low as you could go. You COULD go npe on the 12uF shunt on the mid, saving $4, $8 on the pair.

For the resistors: you can go w/5w resistors on the tweeter L-pad: #015-5 & -4, saving about 3 bucks,
and you COULD use #016-4 (OR -5) for the series resistor on the mid...
If you go w/a #20 1.0mH coil, the higher DCR (.2ohms ?) might cut the mid by -1/4dB. Making the "4.7n" resistor a "5" might cut the mid a total of -1/2dB.
OR, you could make it a "4" ohmer, in which case the mid might end up a total of about +1/2dB louder. Neither one should be audibly diff. from Paul's.

[djg]

Cease and desist.