Chris
pfm Member
For some reason Comeau didn’t see fit to use the Seas mid-woofer without its filter in his WD25 rethink of the A26. I’m sure he explained why in his articles or on his forum but I wouldn’t have understood it anyway.
A look at the Harbeth P3 crossover
- Thread starter tuga
- Start date
For some reason Comeau didn’t see fit to use the Seas mid-woofer without its filter in his WD25 rethink of the A26. I’m sure he explained why in his articles or on his forum but I wouldn’t have understood it anyway.
I can appreciate real life measurements and tweaks.
John Phillips
pfm Member
I am interested so please forgive me being curious. If I am asking too much please ignore my questions.
The differences between parts, AFAICS and ignoring tolerances, are in their parasitic elements. My recollection from a while ago is that Mylar (polyester) capacitors ("cheesy" in the video) have a high-ish dissipation factor usually modelled as having a higher effective series resistance than polypropylene. And I think Mills resistors are non-inductively wound while the sand cast varieties are not - so the "cheesy" varieties have more parasitic series inductance.
These extra parasitic elements can, I am sure, make a difference; but can they be dealt with when designing a perfectly effective crossover? Are there some limits on parasitic elements that dictate using non-cheesy components, perhaps in specific parts of a crossover?
Tony L
Administrator
The point I was making was the big companies who do every aspect of a speaker including designing and manufacturing their own drivers should surely be able to design for an exceptionally simple crossover. Off-the-shelf drivers are by concept generic/multi-use, I understand some complexity would be needed there. I’d have expected the likes of Kef, IAG etc to build much of the required performance into the drive unit itself. If Tannoy could do this in the 1950s, Epos in the 1980s I don’t understand why there are so many crossover components stuffed into a typical modern speaker box.
tuga
Legal Alien
A crossover should be as simple as possible, Epos’ is overly simple and that shows in the measurements. Technically it’s poor design.
Tony L
Administrator
Sadly measurements only tend to show very crude frequency response plots and maybe dispersion. There is seldom if ever anything in the dynamic domain. I respect the ES14 even if I’d not want to own a pair. They were a nice lively sounding pair of speakers, though I have very low tolerance of metal dome tweeters, especially ones of that early generation, and that’s what breaks them to my ears. I’d have loved to hear them with a good tweeter as the rest of the design is very interesting IMHO.
irb
pfm Member
Exactly. In other words, a crossover should be as simple as possible, but no simpler. As Einstein said (sort of).
irb
pfm Member
I'm not sure I agree. Building a driver with a natural, smooth, tightly defined roll-off at a specific frequency is not trivial, and not necessarily the best approach in practice. (It will undoubtedly force other compromises on the drive-unit design.) It's arguably a better approach to design a driver that behaves pistonically within its intended frequency band, and preferably a bit beyond it, allowing the crossover, active or passive, to get rid of the out-of-band stuff.
The assumption underlying much of this thread, and indeed the video that prompted it, is that passive crossover components are in principle a bad thing. Complex crossovers are therefore perceived as very bad, as they use lots of components. I'm just not sure that the underlying assumption is true. I suspect that more speakers have been let down, over the years, by having too few components in the crossover than too many.
Tony L
Administrator
Yes, I’d largely agree with that, though I’d argue the biggest issue is low impedances and reactive loads (no matter how they end up in a loudspeaker design). I’d probably go as far as arguing that if you can’t drive a loudspeaker perfectly well with a nice 20-30 Watt valve amp it is a pretty crap loudspeaker!
irb
pfm Member
Yes, low impedances and reactive loads are a much bigger issue than is often realised. It's actually quite difficult to build a speaker with commercially available 'hi-fi' drivers that doesn't present a challenging load.
Robert
Tapehead
Expensive ones he carries in stock.
James
Lord of the Erg\o/s
A single capacitor on a tweeter to act as a high-pass filter rarely works in practice. The reason for this is simple. The impedance peak (usually occurs between 500Hz and 2kHz) will turn an otherwise first-order roll-off into something less far less like one. Also, a single capacitor offering 6dB/oct roll-off doesn't offer much protection against LF.
As for wide-open mid-woofers, anything is possible. You could easily build a driver with a voice coil that has sufficiently high self-inductance to act as a low pass. Then add enough damping mass to the cone so it doesn't behave pistonically beyond n kHz. No guarantee it'll sound good, though.
If the holy grail is seamless acoustic crossover, then you'd be spending a lot of time, money and effort on building a whole bunch of prototype drivers to measure and hear if they play nicely together with minimal / zero filters. I think it is much simpler to select off-the-shelf drivers and design a crossover for them.
Tony L
Administrator
Certainly much more simple, though if one looks back through audio history the vast majority of the real greats that have now attained classic status used in-house driver designs. That is what pushes the art forward. Tannoy, Quad, Klipsch, Altec, JBL, Yamaha, AR, Kef, Spendor, Mission, Magnepan, Martin Logan etc etc. The LS3/5A are one of comparatively few speakers to use off the shelf drivers and belong in this company. Obviously some mix and match, e.g. the BC1 used an in-house bass-mid and off the shelf tweeters, as do say Harbeth today. I’d cite the Kef LS50 as an interesting modern design that will end up in this list given time.
James
Lord of the Erg\o/s
Driver choices can make or break a loudspeaker design. My golden rule is to pick drivers that are flat(tish) at least one octave beyond its intended pass-band, and probably more importantly, its resonance frequency is at least two octaves beyond. So for a 2-way crossing at 2kHz*, that means a mid-woofer that is flat up to 4kHz, and a tweeter that has fs of around 500Hz. If, like me, you much prefer sealed boxes, then the driver choices narrows considerably.
In some respects, 3-ways offer much more design latitude. But designing a pass-band filter for the mid is much trickier. It's all part of the fun.
*for those vexing that this is smack-bang in the middle of aural sensitivity, fear not. You won't hear the cross with a properly designed pair of filters.
In some respects, 3-ways offer much more design latitude. But designing a pass-band filter for the mid is much trickier. It's all part of the fun.
*for those vexing that this is smack-bang in the middle of aural sensitivity, fear not. You won't hear the cross with a properly designed pair of filters.
Fatmarley
"It appears my intelligence circuits have melted"
In LspCAD (crossover design software) there's no option to add parasitic elements to resistors in the model. I'm sure I read somewhere that the small amount of inductance in sandcast resistors wasn't worth worrying about (could be wrong though). You can add ESR to capacitors in the pro version of LspCAD, but it's a fixed value.
I would be curious to know if there is any crossover design software out there that allows you to go into that much detail.
The laws of physics plus large frequency response variations becoming less acceptable to the market than they used to be in the days you are referencing. Commercial speaker companies will seek to minimise costs and that will include using the least number of crossover components that is viable. If you put a driver with a perfectly flat response in a typical speaker cabinet you will measure something like an 8 dB variation at the listening position with plenty of unwanted bumps and dips requiring correction in better quality designs. Modern speakers are using drivers with a better technical performance but it doesn't reduce the component count in the crossover. Minimising costs also leads to the over use of 2 ways which makes low order crossovers more problematic pushing up the number of components. Etc...
