DIY Crossover Upgrades

[popbumper]

???? I am guessing you are showing the note of "mutual induction"? Other than that I have no idea of the source, nature, or detail of this information....

Popbumper

[Deang]

I am guessing you are showing the note of "mutual induction"?

Yes

"Other than that I have no idea of the source, nature, or detail of this information..."

That's because I blanked out the information. It's from a schematic for a current production model, so I can't post it. The point is -- the engineers at Klipsch are sharp, and I believe they have always accounted for the effects of mutual induction.

[popbumper]

Note to self - take heed, and design to avoid mutual inductance....

Thanks, Dean. Have been working on this all morning. Using a "two-board" scheme, both boards the SAME, but configurable so that you can wire it for HF or LF.

Good idea, STL, I think this will work, but these things are still going to be quite large.

Popbumper

[STL]

On 12/8/2004 12:06:42 PM popbumper wrote:

Using a "two-board" scheme, both boards the SAME, but configurable so that you can wire it for HF or LF.

Good idea, STL, I think this will work, but these things are still going to be quite large.

So you are saying you have one board design, and that it can be wired one way giving you a LF output or another giving you the MF and HF output? If so, that really is not what I was talking about -- although it might work too. I was talking about two unique circuit designs (one for the LF and one for the HF & MF sections) but done in such a way it could all be laid out on one long PCB -- but with the LF sections totally on one end and the rest at the other end. That way only one PCB has to be made, but it could later be cut to make two unique PCBs (or not cut if you have the room for the one long PCB).

[popbumper]

Ah-hah! Interesting, I see what you were saying. In the scheme I was talking about, one board contains:

One configurable coil position (either HF or LF)

Two configurable cap positions (either HF or LF, and HF or Academy)

One configurable resistor position (to accomodate Forte or Academy)

Other components are FIXED or absent, depending on what is needed (autotransformer, specific resistors for certain crossovers, or specific inductors for certain crossovers).

The positions become configurable by soldered jumpers. The result is TWO of the same boards with different parts on them; one will have HF and MF, and one will have LF components. Advantage is seperation of coils on different panes and away from one another, and component sizes can be substantially larger for those interested in using same.

Comments please as always. As this is still in the "planning" stages, it can be changed. I may even do two versions; one stacked "monster", and one "improved" single.

Popbumper

[STL]

I'd be curious to know how you determined the worse-case scenario for the size of the coils and caps. If you have some parts matrix I'd like to see it. If need be, I could host the file for you so others cas see it (since this site doesn't allow spreadsheet attachments).

[Deang]

That's the layout I prefer.

How will you accomodate the autotransformer? The new autotransformers -- do they have connectors to solder too, or wires protruding out like inductors? I guess I should call Bob, it's time to pay my bill anyways.

[popbumper]

STL:

Parts were procured according to recommendations from Dean, primarily, since he has done much of this. Without going into great detail, the HF and MF caps are Auricaps; the resistors are all 12W Mills non-inductive; the coil area is 3" x 3" and based on a large "copper tape" Jantzen type air core. LF coil has not been received yet; I have both ERSE and Madisound types. LF caps are Dayton and Solen.

I guess that really remains the "magic question". If size was not a consideration, one could provide areas of 6" x 6" for a coil alone; I >think< this is why Al does point-to-point, since he is not "confined" to a PCB like I am, and can use pretty much anythin without fear of eaten real estate/stacking/etc.

I am trying to set something up that is a compromise which allows:

a) For a MEDIUM size board, not small like stock, but not as large as point-to-point

Choice of components and mountings that will accomodate most "average" or "small" sizes - which is what I think the majority of folks would lean towards

This is a BIG learning process and it is my goal to satisfy the majority, though like anything, I cannot fill every possible niche. Hope that helped a bit, at least for now.

Popbumper

[popbumper]

Dean:

The Autotransformer will sit between the two boards, affixed to the "lower level" with the HF components. I have estimated a height between the boards of 2.5", which gives adequate clearance for the transformer.

The taps on the transformer are stamped "loops", which are easy to attach and solder appropriate wires to. By far the transformer will be the priciest item of the assembly, and of course heaviest!

Glad you agree with the layout, this will be unlike anything I have ever seen, and will be able to fit into a lot of applications with relative ease.

Popbumper

[STL]

On 12/8/2004 4:47:35 PM popbumper wrote:

By far the transformer will be the priciest item of the assembly...

Really? Auricaps are rather expensive too; a 6.8uF costs $34.

And again explain to me why we are talking about stacking the PCBs? I thought part of the point of two PCBs to increase the distance between the coils. Is there something I am missing?

[popbumper]

Really. I bought a 7uF Auricap and paid less than that, I guess I got lucky or something....

Popbumper

[STL]

So where is a good place to buy Auricaps?

[popbumper]

Parts Connexion has the 6.8uF Auricap for $20.50; the 7uF is a bit more. Check them out, www.partsconnexion.com.

Popbumper

[popbumper]

STL:

A bit more of a scientific approach to your question about "how I chose components" - last evening I went to my crossover parts matrix, and determined which value for which components were the largest. For example, across many speker platforms, the largest cap in the LF section was 110uF, while the largest cap in the MF/HF sections was 7uf. For inductors, the largest iron core was 4mH; the largest air core was 0.39mH.

These values were cross referenced to cap and inductor lists from Parts Express, Parts Connexion, and Madisound. From each of these lists, I determined which value from which manufacturer represented the largest footprint. By building the largets footprint into the real estate, the end user can choose from a number of manufacturers without fear of exceeding the "size restriction" imposed by the PCB.

And there you have it....hope that helped!

Popbumper

[STL]

That's what I expected you would have done. I just thought you might have made some matrix in a spreadsheet to share -- not that I need to see it or anything.

I would be curious to know if you came across an instances where for one value of a part there was one brand that was consideribly larger. For instance, say you looked a 5 different 7uF caps and 4 of them were about the same size but one of them was 1.5 times larger. In that case I can't help but wonder if it might be better to just design the PCB for the 4 "regular" sized caps. That is what I meant back when I was talking about having this not designed for the extremes. In my fictional case here, taking my approach would work for 80% of the parts while allowing the board to be 50% smaller! Those using extremely large caps could still make them work by being be creative, and this new universal PCB would still be a plus for them because those larger parts would not have fit a stock PCB at all (even with creative mounting techniques).

[popbumper]

Funny you should ask - as a matter of fact I did run into ONE instance of that with the caps - there was a Hovland cap that was 3.25" in length, where the nearest "competitor" was only 1.8" long! As you suggested, I will NOT design the board for the largest size, at least in this case.

Popbumper

[STL]

Yes, that's what I was talking about. I thought it might have been a little more common than just one part, but without seeing your data I couldn't know what you did.

[Erik Mandaville]

Dean:

I just found this! Great work, and excellent pictures! Yes, I think you do need to give me a lesson in detail digital photography.

Instead of using a soldering iron, though, you might try just heating up a steel rod in your BBQ. get it good and hot, and then weld together the leads of any caps and resistors together with that. Once the leads are glowing red, give them a good 'smack' with a small hammer to complete the weld. This is a very old technique for joining metal, which works great for modern audio.

I call it "solderless soldering." The air and high frequency detail that results is astounding.

[Number 9]

Pardon my ignorance Dean, but why so much hot melt glue?

Seems to create a lot of unecessary extra work IMO to clean off if you want to change something later on.

[Deang]

You aren't "ignorant" -- just stating what appears to be the obvious.

There isn't really anything that can be changed without substantial modification. Anything else that could be changed would add very little to the overall performance of the speaker.

I'm handicapped by being an audiophile first. This means I'm prone to adopting strange and interesting ideas. Some time ago, I redid the networks in my Dad's JBL Aquarius towers. When I finally got to the first network -- I found it completely covered in adhesive. It took me a half a day to get to the parts. I used and exacto knife and just started slicing pieces off. It was terrible. At any rate, from time to time, I caught myself wondering about all that glue I found. Sometime later, I did a pair of Advents with very low serial numbers (34,xxx), and found the same thing. After buying my last set of amps, I started reading about potted transformers -- and the next thing I know I'm thinking about the glue again. So -- I ended up doing a lot of reading on the potting of electronic parts and sub-assemblies.

Ceramic capacitors in particular are known to be microphonic -- just like tubes. Mechanical vibration causes them to resonate at certain frequencies, and it is both measurable and audible. Now, there is no evidence that I can find suggesting that modern day, quality capacitors are subject to this problem. I would say it's doubtful -- but I really don't know for sure. However, I do know that I've handled many networks that started out well glued, to now only being held to the board by their leads. I have one of these right now -- a near brand new set of RF-7 networks sitting in my workroom that were delivered to me with a tweeter inductor on one of the boards no longer being held on by the silicone.

All of these networks are in physical contact with the cabinet, and usually less than 6 inches from the back of the woofer. Something tells me this isn't the greatest place to be. Potting protects the parts from mechanical/acoustical vibration, moisture, and changes in temperature. In short, it doesn't hurt anything -- so why not?

As far as removing the hot melt goes -- it takes about 15 minutes with a hot knife. It actually peels off rather easily. What's hard is getting it off the surface of the board and wiring. Oh well.