Help me understand X-Over point selection

[coolhandjjl]

I've read of crossing a mid to tweet anywhere from 5800 to 8500Hz. Is the spot typically chosen on driver limitations and horn dispersion patterns? If you had a mid+horn that had linear performance well past 8500, or a tweet+horn that had linear performance well down past 5800, how do you decide on a point? If both driver+horns had similar off-axis measurements throughout the customary crossover range, is there a sweet spot determined by music content? Any rules like "avoiding crossing at xxxxHz because that's where essing can be aggravated"?

[Don Richard]

Crossover points are selected to protect the HF from damaging mids frequencies and to match the polar beamwidth between the two as you have said. Natural driver rolloffs need to be considered because the acoustic crossover slopes are what you are really going for, not the electrical crossover. This can lead to wierd electrical settings like gapped or overlapping crossover frequencies. Overall phase response may also be important.

[coolhandjjl]

Crossover points are selected to protect the HF from damaging mids frequencies and to match the polar beamwidth between the two as you have said. Natural driver rolloffs need to be considered because the acoustic crossover slopes are what you are really going for, not the electrical crossover. This can lead to wierd electrical settings like gapped or overlapping crossover frequencies. Overall phase response may also be important.

ALK Engineering has their Extreme Slope Network to address some of those issues. If the polars matched across a broad range of points, what makes someone choose 5800 over 8500 if the drivers could handle either?

[Guest " "]

ALK Engineering has their Extreme Slope Network to address some of those issues. If the polars matched across a broad range of points, what makes someone choose 5800 over 8500 if the drivers could handle either?

The human hearing band is broken up into segments called octaves. There are about 10 octaves within the range of human hearing. Typical 3 way systems breakup the octaves so that 3 to 3.4 octaves are reproduced by each driver. This results in a balanced sounding speaker system. limitations are the capabilities of each driver and how they handle overlaps. If you don't balanced the assignment of octaves to a driver, the driver could have the perception of being too loud, just becuase for example it's reproducing 5 octaves instead of 3.

[Don Richard]

If the polars matched across a broad range of points, what makes someone choose 5800 over 8500 if the drivers could handle either?

Mesasurements interpreted by someone who understands their meaning. Listening tests. System output.

[Deang]

Polars typically don't match across a wide range. You present a hypothetical that doesn't really exist. You pretty much answered your own question, with some great followup from Don. Phase is also an important parameter. I don't like the sound that's generated from the Extreme Slope Networks. You need some overlap between the drivers to get a seemless presentation.

[WMcD]

It is interesting to read the tech blog on the Palladium, in three parts.

http://community.klipsch.com/blogs/default.aspx?GroupID=36

There is other very interesting things there too.

"The Art" probably does not answer all you questions. Further, it is mainly about how very complicated x-over circuits can be used to equalize out irregular response in the driver-horn. But you need something like LinearX to do this. That program accounts for phase of the acoustic output and delay because the voice coils of the drivers are not usually in alignment.

Of course I would say that any design of X-over does have to take into account the natural roll off of the driver-plus-horn.

The most recent approach to crossing over seems to have a goal of doing so where the polar response of the two respective drivers are about equal. This might be similar to matching actual acoustic output. So, if the mid horn has a 90 degree by 40 degree polar response, we'd like to have the woofer polar be somewhat close to that. I'd guess 70 degrees. Or maybe the goal is only the horizontal.

It is true that woofers, as we go up in freq, do narrow in output polar response.

The "Art of" blogs don't discuss this that I can see. But there is a hint that the overall crossover design is 3.5. (Not referring to three woofers but rather that there is something extra in the "0.5") My guess is that this means the three woofers work at low freqs but only the top one works at higher freqs. That way the transition to the mid horn only has to take into account the top woofer which is plysically close to the mid horn and does not have a complicated interference pattern caused by three woofers working. The RF7 might use something similar with two woofers.

As far as freqs to be avoided. PWK in the Heritage liked to keep the x-over freq between mid and tweeter away from 3500 Hz, and placed it up at 6000 Hz.. because that 3500 is the freq where our ears are most sensitive. I reason differently. If the problem is destructive interference (a dip) because both mid and tweeter are working, this freq is where our ears can hear despite the dip from interference (if any).

He also liked to use a high pass filter on the mid and let it go as high as it can, up to 6000 Hz. But I think this allowed for the glitch in the midrange driver at 8000 Hz to get through.

So, there is a lot going on and people may differ as to the best way of doing things.

WMcD

[Daddy Dee]

In my very lay understanding, think I understand why phase issues are a problem which much be accounted for in selection of drivers, crossover points and slopes.

One thing I do not understand but often read about are issues with polars, which also seem to be affected by horns of different design. If anyone would be available to say more about polars for dummies, I'd appreciate that.

be sure to type your post slow because you know i can't read fast.

[Deang]

How about we do that with a picture. Tell me if it helps.

[coolhandjjl]

Nice visual. Thanks.

[Axz Hout]

r

[Axz Hout]

y

[Daddy Dee]

thanks Dean. that does indeed help

[WMcD]

I don't know the answer to C.F.'s questions.

If you want to keep lower freqs out of the tweeter you can use a 6 dB per octave at 6000 Hz or a 12 dB per octave at 3500 or so. Of course PWK went to an 18 dB at 6000.

Regarding the EV's, it was Al Klappenberg who pointed out that these are a constant k design. See http://en.wikipedia.org/wiki/Constant_k_filter. It is interesting reading if only for the mention of Dr. Zobel. The tip off is probably that in the EV the same values of L and C are used in the both high pass and low pass.

It is my read that the design works if the load on both filters is the same. So if you say the revised values are because the tweeter load changed, then the woofer load would have to change also.

As I recall, the EV crossovers shown were part of the EV "building block" series where you could start with a woofer, add a tweeter with the X36 x-over , and then add a midrange horn and X8 (have to look that up). But there were several different woofers. Could they all have the same impedance?

My guess is that these designs were the best engineering of the 1960's but we can't read too much into them.

WMcD

.

Does this indicate a shift in the cross over freq? or the impedance of the load? I'm not sure

[Axz Hout]

u

[JohnA]

To the OP's question, assuming you actually have drivers that cam perform as described, ......

Sometimes an engineer just has to take a decision and go. Like your situation, there are several possible answers and none are wrong. Assuming the dispersion patterns were similar, I would choose a tweeter high pass that was at least one octave above F3 (-3 dB) or one octave above the fiaphragm resonance frequency and use at least 12dB/octave. then, I'd look for response anomalies and tweak the element values to try to minimize them.

[Deang]

Hi John.