| Re: Compression Drivers [message #74432 is a reply to message #74426] |
Sat, 10 November 2012 22:53   |
 |
Wayne Parham
Messages: 18678
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
| zheka wrote on Sat, 10 November 2012 14:43 | Have you noticed any changes in authentic DE250s recently?
|
No, I have not. And as you are aware, I gave them a pretty thorough run of measurements when I tested the H290C waveguide. They act exactly like they always have, at least since I started using them.
In fact, look at the current spec sheet from B&C:
Same as it ever was.
Some characteristics of compression drivers are pretty stable and consistent, but others aren't. Part of a good design is making the crossover tolerate shifts that are most likely. Because they will shift, actually quite a bit, in some respects. That's normal.
The front chamber size and phase plug slots are usually pretty consistent from unit-to-unit. So you can count on the acoustic reactance from those features to stay constant. Alnico magnets may (permanently) lose strength when run hard, but ceramic and neodymium magnets are resistant to demagnetization. So (BL) motor strength is pretty stable in ceramic and neo motors, less so in alnico. Diaphragms can deform when pressed hard, and sometimes their environment can affect them too. Things like sunlight and humidity sometimes affect them. But in general, you can expect the diaphragms to be pretty consistent unless they are damaged. But the one thing that is all over the map is the voice coil. It will change drastically at different drive levels - even seemingly small drive level changes - and this changes motor strength and electrical damping. Lots of non-linear properties shift as well.
I think the real problem is Geddes crossovers are too sensitive to driver parameter shifts and unit-to-unit variations. This is a common problem, one that I've seen in both DIY and commercial speakers. Some system designs are just more sensitive than others, and in my opinion, one of the most important features of a good design is its tolerance of parameter shifts.
Geddes may use tank circuits in his crossover for impedance/response shaping. He's not alone, I've seen it done by other manufacturers and DIYers as well. It is a fairly common approach, but ill-advised, in my opinion. That kind of crossover design is way too sensitive to changes in drivers, both from unit-to-unit variation and even from shifts due to temperature change at various power levels.
It is described in my Speaker Crossover document as a "resonating damper for the tweeter circuit" and is used as a way to mitigate peaks in impedance, which often show up as aberrations in response. As I said above, the problem with this approach is it is fairly sensitive to driver parameter shifts, and that's why I don't do it in my crossover. Variations between driver units make them hard to match with the crossover. Even temperature changes in the voice coil (which happen as drive voltage changes) will move the peaks enough that the tank circuits do not match.
Probably the greatest strength of my crossover is having that R1/R2/C1 network do all the work. It provides specific damping, and that approach is more tolerant of driver shifts than using tank circuits. Of course, it helps when the horn/waveguide is built properly, and doesn't have resonant modes in the passband.
|
|
|
|
|
|
|
|
| Re: Compression Drivers [message #74444 is a reply to message #74440] |
Mon, 12 November 2012 01:17 |
|
Wayne Parham
Messages: 18678
Registered: January 2001
|
Illuminati (33rd Degree) |
|
|
I see tank circuits in a variety of loudspeakers, but I see them most often used with CD horns and waveguides. Conical horns are usually peaky if truncated at all, and all CD horns and waveguides share a common heritage with conical horns. So it is natural to conjugate these peaks with (notch filter) tank circuits. The problem is as driver characteristics change, either from unit-to-unit variation or even from thermal drift, the tank circuit required would need to change too. That's why I prefer resistive dampers to reactive dampers (notch filters) in this case.
There are a lot of horn/waveguide products on the market that are highly reactive in their passbands. An example of a horn like this is the SEOS12. It is more a curved baffle than a horn, pretty much just a dimple on the baffle. At 3.5" long, it's barely long enough to even hit the primary resonant mode at its ~1kHz crossover point. So it provides very little acoustic loading for the first octave and is reactive high into the passband.
A conical horn and most of the waveguide derivatives load the driver poorly, and the ratio of mouth area to depth is all it has to set the load. As with all things, there are competing priorities here. A 90° wall angle sets the depth accordingly, but then to add a secondary flare to combat waistbanding decreases depth or increases mouth size, whichever way you want to look at it. That's why the SEOS horns are so reactive. Lots of other waveguides suffer this problem too, to tell the truth.
Conventional wisdom says larger mouths smooth ripple, but this is only true when the mouth is undersized. One can go too far with this, and actually make the mouth too large. That will increase ripple too. A horn is reactive not only when the mouth is too small, but also if it is too short. The former problem is usually the case with basshorns, because mouth area requirements are prohibitively large. But the latter problem is sometimes seen in tweeters, especially in the newer waveguides that may not have been analyzed for acoustic loading.
In any case, I personally find it better to not try and chase down a horn with peaks using tank circuits. It's like trying to push an air balloon under water with a pole. I much prefer the R1/R2/C1 method, which I developed after years of working with these kinds of speakers. The speaker document I referenced above was sort of a treatise on the various strategies I had used over the years, and it sort of builds up to the solution I use today.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Posting Rules
Members
Search
Help
Register
Login
Get a T-Shirt!
