Heresy Crossover Tweek

[DrWho]

Instead of thread hijacking, I thought I'd start a thread detailing the crossover tweeks I'm doing to my Heresy speakers. I know there's a lot of material and pictures here, but I would be very interested in any thoughts or criticisms. I'm a complete noob at this so if I'm doing anything stupid by all means pull out the 2x4... []

Here is my component list:

• E-2 Crossover
• K-22-K woofer
• K-53-K squawker, plastic horn
• K-77 tweeter
• everything mounted to the rear of the baffle

My project started after reading about the woofer being 6dB down
from the top end in compensation for the speaker being mounted in a
corner (where there would be 6dB of boundary gain). I've been dragging
my speakers around campus for all sorts of reasons and the only
complaint has always been "man they're bright". Granted, the source
material and amps have been crap and we all know what horns do with
that kind of signal....but I figured it could be reduced somewhat if
the components were brought back into balance.

Anyways, I've
since acquired a cool computer based measurement system that only put
me out $110 and I've been looking for excuses to measure stuff so I
figured I'd share some before and after plots of the progress.

These
first plots are gated measurements in my roughly 14x24x8 apartment
living room. The mic was about 1m away and the speaker was put in the
middle of the room (to minimize as much as possible the reflections in
the room). It's pretty cool to see all of the comb-filtering effects
from room reflections come into play as you extend the length of the
gating. The attached plot is actually a 1/1 octave smoothing of a 200ms
gate to show better detail at the woofer frequencies. You guys will
have to trust me that the smoothing is representative of the behavior
in the higher frequencies when the gating is much shorter (on the order
of 5ms). At 200ms it's pretty much a wash of jagged lines that are mostly meaningless. Anyways, it should be pretty intuitive which is the stock Heresy and which is the one with the tweeter knocked down 3dB.

[DrWho]

The slight differences in the low frequency and midband response are due to the fact that I was unable to get the speaker and mic in the exact same location for both measurements (since I moved the speaker to my table to work on). When unsmoothed you can see how the modal distribution has shifted a few Hz and then the smoothing gets all outta whack. Well...since I've already posted it, here is the response when gated faster and no smoothing applied:

I'm pretty sure the 1800Hz dip is the result of the mic's height from the floor (about 1.5 feet). You can see signs of the floor reflection at 900, 450 and 225Hz too (they're just not as dramatic due to the gating reducing resolution). The pattern control on the K600 horn is probably going to be wide enough in that region for the reflection to have significant energy. You don't see as exciting dips as you go up because its directivity is narrowing (so no reflections). 3.2kHz, and 6.4kHz would be the next harmonics (notice the slight dips).

However, the dips at 7kHz and 9kHz I believe to be comb-filtering due to interactions between the mis-aligned squawker and tweeter (not reflections in the room). In fact, this can be verified by looking at the energy time curves (reflections in the rooms show up as spikes).

The behavior above 10kHz looks pretty normal for a driver's raw response when viewed at high resolution. Most of the peaks and dips are too narrow to be percieved.

While I'm at it, I wanted to post the Spice predictions for the crossover response. This first picture is that of my model. I've basically predicted the response of the drivers as a nominal impedance with an inductor in series. Not ideal I know, but I haven't yet figured out how to measure the complex impedance of my speakers yet (I do have the software though...).

[DrWho]

And then here we have the predicted crossover frequency response of the stock crossover from SPICE:

And then here is the predicted change when the tweeter is moved down a tap:

[DrWho]

As is apparent from the first chart, it seems that dropping the tweeter output wasn't very beneficial. I didn't do any listening tests because I'm fairly certain the 4dB rise in the 4kHz-6kHz region is responsible for most of the brightness issues. Also, the comb-filtering in the transition region is unnacceptable - and I think is the reason "the tweeter" sounds a bit hashy with cymbals and the like. If I had signal-alignment tools, I would love to show how the comb-filtering goes away when they get lined up, but alas that is far beyond my means.

However, I should be able to swing the $20 or so needed to put a .4mH choke in series with each squawker:

Which should have this predicted change on the frequency response:

[DrWho]

Just to point it out - I intend to put the tweeter back onto its original tap. I figure if I need any attenuation from the tweeter, that I'll just change the cap doing its low pass to shift its crossover point higher. Maybe I'll move to a 1.5uF or 1.0uF cap and see what that does to the measured response. Couldn't I also just put a resistor in parallel with the tweeter to bring down the impedance of the spike towards the bottom of its response? Which will then effectively shift the high pass up in frequency?

[Guest " "]

What are you using as L5, L4, L3, L2, thats not the symbol for an autoformer

Here is a sample.

If you are using a transformer, then you are fine.

[DrWho]

The numbers under the L5-L2 are the values of inductance on a transformer in Henry's. I haven't figured out how to model an autoformer in SPICE so I'm just using transformers to net the same effect on the voltage. The K1 K2 crap underneath is referring to the amount of mutual inductance between the coils - I went ahead and just assumed 100%.

[Guest " "]

T2A inductance, here you go.

[DrWho]

Ya I saw that on Warren's site, but I'm not sure how it helps...The amplifier connects to tap 0 and there is no inductance offered for that leg...I get the impression that that's the series inductance?

[Guest " "]

Well I have more questions about that chart than I can offer answers.

I don't see how you could have 21mh of inductance on the -3db tap and 2.8hm on the -12db tap. The number of turns is higher at the -12db tap than the -3db tap. The whole thing only makes sense to me if you flip all the mh values over .

[WMcD]

Mike. Since you asked for comments.

Granted that a hot high end is often another way of saying there is weak bass.

But I harbor a personal belief that the shrillness is a matter of the mid range horn having on-axis gain above 3 kHz or so and thus "amplifying" driver anomalies or even horn anomalies above that freq.

This may explain why there are favorable comments about Al. K's extreme slope filters and the P-trap,and even the tractrix horns which don't narrow down.

Therefore I'd suggest you shoot for a passband filter for the mid with a 3rd order or greater roll off starting (minus 3 dB) at 3.5 kHz or even lower freq.

I'm quite convinced this is not a tweeter problem. When I listen to a K-77 (with filter) it seems to me it has so little contribution that it can not cause annoyance. But listening to a K-55 type on a horn (I've only done this with a high pass) it does sound harsh.

- - -

A little off topic. PWK seemed very concerned that a crossover point should not be at 3500 Hz because that is the freq where the ear is most sensitive. The Fletcher M curve shows the general thought that the ear is indeed most sensitive there. We do hear the most quiet sounds there very well.

It was, reportedly, very difficult to argue with PWK when he was with us. And harder now. Smile.

My argument is that the dip in the Fletcher curve shows that at 3500 Hz our ear is least sensitive to CHANGES in level. So if there is an amplitude change because of crossing over, that is the best place to put it, rather than the worst.

On the other hand, if you look at the FM curves there is a fair amount of ear sensitivity there and an octave above which could mean this band is the best place to filter down power to the driver if it is getting distorted by the driver.

Of course, this is not a matter of SPICE simulations.

Gil

[DrWho]

Granted that a hot high end is often another way of saying there is weak bass.

Except in this case, the squawker to woofer transition is very smooth - so I don't think we can say there is too little bass. If anything, I would argue that the woofer is a bit too hot in the 100Hz region, but room influences might be coming into play at those frequencies. Regardless, I think I debunked the theory that the woofer is 6dB down - at least with my particular speaker.

But I harbor a personal belief that the shrillness is a matter of the mid range horn having on-axis gain above 3 kHz or so and thus "amplifying" driver anomalies or even horn anomalies above that freq.

I think you're on to something there, but it's a bit beyond my means right now to change out the squawker. They also wouldn't be Klipsch after I modded them either - and in some ways I kinda like the Klipsch sound, but that's a matter of subjectivity so no need to go there [] I think I would just start from scratch if I were going to start using different components.

Therefore I'd suggest you shoot for a passband filter for the mid with a 3rd order or greater roll off starting (minus 3 dB) at 3.5 kHz or even lower freq.

Yikes, that's a bit drastic of a change....I was thinking first order starting around there so that I didn't have to move the tweeter crossover point lower. If I've still got the peak, I was thinking of moving to second order. Higher order = more expensive and I'm trying to get by with as little changes as possible. Y'all know what it was like to be in college.

I will definetly keep it in the back of my mind though. I think I should take a measurement of the squawker alone and see what kind of odd behavior it's exhibiting and then try to match a filter to compensate. While I'm at it, I might as well do some off-axis measurements too - I'm kinda curious how beamy it gets up high. Maybe the solution could be as easy as sitting a little bit off-axis from the speaker?

I have another free program here that I need to learn how to use that should theoretically provide me with some distortion measurements as well as the ability to sum multiple response plots together and predict crossover performance with the true response of the drivers. If the rise in the squawker response is distortion related, then I don't think a filter is going to do much good - and that's when it's time to upgrade to the K69 + K402 combo []

I'm quite convinced this is not a tweeter problem. When I listen to a K-77 (with filter) it seems to me it has so little contribution that it can not cause annoyance. But listening to a K-55 type on a horn (I've only done this with a high pass) it does sound harsh.

Yea, I don't think it's a tweeter problem either. If you look at the before and after of the tap change, it's quite apparent that most of the tweeter energy is above 6kHz. I always try to tie frequency response aberations back to my live mixing experience, and I'm quite confident that 6kHz+ doesn't contribute at all to any percieved harshness of the mids. It's more of a sibilance / hashy part of the sound. Of course, it's rather difficult to describe sounds and everyone does it differently []

A little off topic. PWK seemed very concerned that a crossover point should not be at 3500 Hz because that is the freq where the ear is most sensitive. The Fletcher M curve shows the general thought that the ear is indeed most sensitive there. We do hear the most quiet sounds there very well.

It was, reportedly, very difficult to argue with PWK when he was with us. And harder now. Smile.

My argument is that the dip in the Fletcher curve shows that at 3500 Hz our ear is least sensitive to CHANGES in level. So if there is an amplitude change because of crossing over, that is the best place to put it, rather than the worst.

On the other hand, if you look at the FM curves there is a fair amount of ear sensitivity there and an octave above which could mean this band is the best place to filter down power to the driver if it is getting distorted by the driver.

I wonder how you're interpreting the F-M curves? Here is what I believe to be the most recent version:
http://sengpielaudio.com/Acoustics226-2003.pdf
I read the plots to say that the ear is the most sensitive in the 3.5kHz region???

Of course, this is not a matter of SPICE simulations.

[]

Thanks for the comments.

[xenon]

Can you assist please I am putting together some speakers to make a center system for my home theatre

I have K53H

K75K

K24K

with a heresy E2 crossover

That changes are required on this xover to drive these smoothly

will the heresy box be Ok for the woofer ? The box will be made from ground up so I can alter everthing as required

You may wonder why I am going to this trouble but here in Australia we only have over priced units or crap

I guess its to do with the high cost of importing due to the weight

Regards bob

[Daddy Dee]

Gil,

You are talking about rolling off the K-55 at 3500? That's interesting. I also have found that the K-55 is not on it's best behavior when it poops out at 6000. Crossing at 4500 has been a pleasant sounding change in combination with a CT-125 tweeter.

I was imagining that you were thinking about an alternative to the K-77 in such a configuration. Any particular driver as an alternate?

I'm pretty sure that the CT-125 could perform at that lower crossover point, too. Would be interesting to hear.

[mike stehr]

Gil,

You are talking about rolling off the K-55 at 3500? That's interesting. I also have found that the K-55 is not on it's best behavior when it poops out at 6000. Crossing at 4500 has been a pleasant sounding change in combination with a CT-125 tweeter.

I was imagining that you were thinking about an alternative to the K-77 in such a configuration. Any particular driver as an alternate?

I'm pretty sure that the CT-125 could perform at that lower crossover point, too. Would be interesting to hear.

Dee,

Do you have Crites networks with the 4500 Hz mod for the CT-125?

I need to source some 500 uH inductors, so I can do this mod.

The Eminence APT200 or whatever driver the CT-125 uses, has a specced xover point of 3.5kHz.

I'm using a pair in 1989 raw birch LaScalas for now, but when I first got them I hooked them up in my Cornwalls. Just a swap out from the K77, using the stock B2 network.

Right off the bat I could tell that the CT-125 could benefit from a lower crossover point. Too laid back sounding at that 6kHz crossover point to my ears. While it did seem more extended. The K77 may be a little zingy....but IMO, it's voiced with the midhorn pretty well. Or it comes on strong at 6kHz, making the transition more balanced...I guess....

I suppose the 4500Hz mod for the CT-125, could be lashed up for a Cornwall network as well, or any of the networks possibly.

In the case of the K401/K55V combination, I just get the feeling converting the midhorn to bandpass at 4.5kHz will net a better balanced sound with the transition from mid to tweeter. With respect to the CT-125 tweeter.

[Islander]

I thought I'd start a thread detailing the crossover tweeks I'm doing to my Heresy speakers.

I don't know what your budget is and it looks like you're having fun, but have you thought of trading your Heresies for Heresy IIs and then doing the Heresy III upgrade?

[Erik Mandaville]

Dee:

It's been awhile since I checked this, but I believe the suggested crossover point for the CT-125 (non Klipsch Heritage application) is 3500HZ with an 18dB/octave filter. My first experiments with a narrow pass-band for the midrange involved the use of a 4.5kHz crossover point, and then to 4kHz after that. This is the network I built without the autoformer which I personally preferred. It has great clarity and sense of immediacy, and was found to be mildly less 'lossy' than the type 'A' with an autoformer. This is ground already covered thoroughly, and I bring it up only to say that I am very comfortable with the 4kHz crossover point. I was reluctant to go lower than the 4kHz point for reasons mentioned above.

I used resistive series/parallel L-pads to balance the drivers. Nothing new there; they have been and continue to be used by many well-known brands of speakers, including high-efficiency horns. PK liked the autoformer, and I simply preferred the other approach, with the 3rd order tweeter connected directly to the amp input of the crossover rather than after the input cap/autoformer connection.

I agree with you on the lower crossover point for the squawker, which I first heard, thanks to you, using a pair of of Bob C's networks. Since there is already published data on the behavior of the drivers I am using, it was not difficult to come up with something that performed at least as well, and to my ears better than, the type 'A' I was using before and liked, particularly with low-power triode tube amps.

Erik

[Daddy Dee]

Erik,

Good to see you on the forum, bro.

Yes, you're right on the 3500hz on the CT-125. I haven't heard it below 4500 hz, but am curious about that. Tony Reed and I are working diy PA project for church. He's doing the real work. Anyway, it will be a two way using an Eminence Beta wooofer with a frequency response from 41hz to 4khz. We've been considering the CT-125 as one HF option as well as the K-65 (Eminence PSD 2002) which is 1.2khz to 20000. We're going to use the K-65 since that is what Klipsch uses in the K-3002 PA speakers.

I'd like to hear the crossover you built. Anything that you'd prefer over the Type A has to be good stuff.

[DrWho]

I thought I'd start a thread detailing the crossover tweeks I'm doing to my Heresy speakers.

I don't know what your budget is and it looks like you're having fun, but have you thought of trading your Heresies for Heresy IIs and then doing the Heresy III upgrade?

I would sooner get rid of the Heresy speakers and put a K510 in my Chorus II's to replace the squawker and tweeter [] Sadly I don't have enough money for either.

[Islander]

Sadly I don't have enough money for either.

Not to worry. Rome (or your dream system) wasn't built in a day. You'll get there.