advanatge of 1s order over 2nd or 3rd order?

[phase_accurate]

A little off-topic regarding the original thread subject but here is a link to a new digital crossover that has significantly improved off-axis behaviour. To make full use of it they still need symmetrical driver arrangements but the results look very promising:

http://www.acourate.com/HorbachKeele...eXOFilters.pdf

http://www.acourate.com/HorbachKeele...XOFilters2.pdf

Regards

Charles

[thoriated]

A quasi first order xovered speaker such as my Iron Lawbreakers will exhibit little difference in amplitude response through the crossover region as you move vertically through any reasonable angle, since it is designed so that the theoretical response nulls are nearly directly above and below the cabinet.

It should also be pointed out that in a case such as this, any differential narrowing of the individual driver vertical dispersions near the xover frequency significantly reduce the depth of any such null, very different than the case with a high order FIR xover where a foot or less of vertical movement at listening distance essentially neutralizes the necessary pre-ringing cancellation.

[speaker dave]

Originally Posted by phase_accurate

A little off-topic regarding the original thread subject but here is a link to a new digital crossover that has significantly improved off-axis behaviour. To make full use of it they still need symmetrical driver arrangements but the results look very promising:


Regards

Charles

Hi Charles,

I like those arrays. They give a very well behaved vertical performance and inherent symmetry. Not phase coherent though. They are an enhancement of what I was doing at Snell with the XA series. Don and I had discussed these extensively at the time.

D'Appolito Arrays with Waveguides

The post that Dennis H gave a few pages back should be looked at as well. It is a similarly laid out array with a crossover optimized for best phase response rather than vertical directivity.

David S.

[phase_accurate]

Hi Dave

I was searching for Denis' post but couldn't find it. Was it within this thread here ?

Regards

Charles

[catapult]

Originally Posted by thoriated

Here's a link to an AES paper from 2008 titled: "Perceptual Study and Auditory Analysis on Digital Crossover Filters" at:

http://www.acoustics.hut.fi/~mak/PUB/AES124-000056.pdf

Among the researchers' conclusions:

"FIR crossovers seem to be *highly susceptible* to
off-axis errors with higher filter orders. The
flight time difference of only 0.02-0.03 ms be-
tween low- and highpass bands at 3 kHz was
found to produce audible ringing with high FIR
orders of the scale 1000-2000. Rough safety
limits would be to keep the order of a linear-
phase FIR filter at/under 600 at 3 kHz accord-
ing to both the headphone simulation and the
real loudspeaker experiment."

We discussed that paper on another forum and I learned a lot about FIR filters from "JJ" the head of R&D at DTS. Most important to this discussion, the "order" of a FIR filter is the number of taps, not its slope. The max slope varies with XO frequency. I did some calcs based on the measured latency of the DEQX and its 300 dB slopes are well within the paper's safe range at any reasonable XO frequency.

[catapult]

Originally Posted by phase_accurate

Hi Dave

I was searching for Denis' post but couldn't find it. Was it within this thread here ?

Regards

Charles

Hi Charles,

Here's the thread at HTG. It's an electrical sim of a 5-way. To build a practical speaker like that, you'd probably want to use 9 drivers in a symmetrical array to avoid the lobing problems Speaker Dave mentioned. Dunlavy always did that and it worked pretty well.

HTGuide Forum - Duelund meets Dunlavy (aka Duelund meets transient perfect)

[phase_accurate]

Ah - this one !

I stumbled over it some time ago. It is yet another way to implement a constant-voltage crossover.
The main disadvantage here is also the lobing issue. If space permits a symmetrical topology as proposed is indeed the most elegant solution for that (or maybe a coaxial).

What I use are approximations of low-order (currently 2nd/2nd) constant-voltage active crossovers. An old incarnation (2nd/"1st") can be found on this forum. Just google "transient perfect crossover". Because the drivers are spaced quite closely regarding the low crossover frequency it doesn't behave too badly in terms of lobing.

Regards

Charles

[thoriated]

Because the drivers are spaced quite closely regarding the low crossover frequency it doesn't behave too badly in terms of lobing.

Sorta reminiscent of the Iron Lawbreakers. With 'classic' drivers like the Altec 288-16G and the JBL 2220A (all alnico), it can get kinda loud on flea power, too, with the overall efficiency ~100db/w/m. Not quite in the Klipschorn league, but then it is in only a 100 liter cabinet. I pulled enough tricks out of my bag, though, to get full response down to 60hz, and useable response to the 30hz port tuning.

[speaker dave]

I was looking at the array that Dennis H. had put up and it finally made sense why it works. I know that it works from a transfer function point of view (all the S domain sections add up to unity transfer function) but wanted to understand it intuitively.

I've coppied Dennis' crossover diagram and circled a few of the sections. Note that each driver has 4 first order crossover section and as you go from the high frequency unit at the top you start with 4 highpasses. Second unit has 3 highpasses and one lowpass, third 2 hi and 2 low, etc. Also the high and lowpasses are all with shared frequencies so the tweeter has the crossover point it needs plus the crossover point of the upper mid, the "middle mid", the lower mid and the woofer. These extra roll offs make the crossover definitely higher than first order.

Note that only one driver is shown but the assumption is that each of the 5 crossover pathes is terminated by a driver.

But if you look at the encircled section note that for every pair of drivers considered, the effective crossover between the units is first order. In every case, outside the encircled bits, the extra crossovers would be indentical for that pair of units so although they would cause phase shift they cause no phase shift between the pair. Also, the phase shift they cause is not in the passband of either driver but "downstream". At that downstream point, another pair of drivers will dominate response and again they are crossing with a first order crossover (between them) and sharing all other poles.

So in the end you get a transient perfect crossover, but with enough other poles added to both greatly improve power handling and also to allow for some compensation of the inherent driver rolloffs. By that I mean that you have a much better chance of optimizing to the 2nd 3rd or 4th order acoustical target rolloff than you would of making a driver hit a 1st order target over any appreciable bandwidth.

Note that this only works well with a multiway system with lots of sections. Using it as a 5-way means the extra poles downstream aren't too far removed and the bandwidth requirement of each driver is lessened. As a 5 way the power handling is considerably better, primarily because the multiple extra poles give effective electrical filtering. This works best with more sections because the extra rolloff points are less removed from the fundamental crossover point.

Neat.

David S.

[catapult]

Good explanation, David. Thanks.