Klipschorn bass response

[John Warren]

I read the Delgado article describing the Jubilee bass unit. In that article the anechoic response of the Klipschorn bass unit without a filter is shown. I've cut and pasted it in this thread. On the log-log plot, the response shows a large "hump" with a peak between 200-250Hz. Given the measured response, I was interested in seeing what the AK bass filter sections looked like. I was also interested in determining the magnitude of the components needed to make a (large) notch filter that could nullify the response. That said, I have built nothing yet, this is all model predictions. First plot is the Klipschorn response from the AES article (R. Delgado and P. W. Klipsch, J. Audio Eng. Soc. Vol. 48, No. 10, 2000 Oct pp. 922-929).

[John Warren]

Now posted below are the low-pass filter section predictions of the AA, AK-2 and AK-3. I assumed that the filter sections are loaded with an 8 Ohm pure resistor which, looking at the impedance response of the Klipschorn between 100-500Hz, isn't that bad. What I don't quite understand is that the AK-2 and AK-3 appear to provide additional "boost" where it's really not needed(?). Also, I've ingored phase response which becomes a concern within the driver overlap areas. For the moment however, let's consider the bass unit alone (and not the midrange unit) contribution to the range we are looking at (100~300Hz).

[John Warren]

Looking at the response plot from the article, I modeled a simple parallel notch filter that will provide about -6dB at a notch filter centered on 206Hz. The filter has the following response with L=2mH, C=300uF and R=8Ohm.

[John Warren]

Increasing the resistor value increases the amount of attenuation. The plot below shows how where R is 6 Ohms, then 8, 10 and 12.

[John Warren]

The plot below is the predicted response of the notch filter in series with the 2.5mH choke that is used in the A and AA networks.

[John Warren]

The value of the notch filter must obviously be assessed against a frequency response plot. That's takes some work.

[Deang]

Interesting, especially your question about boost at the worst possible spot (with AK-2 and AK-3).

What happens if you run the first plot using a 4 ohm resistor instead of 8 ohm? I mean, what should we expect to see happen?

[John Warren]

The bunps in the responses go away when the 8 Ohm resistors are changed to 4 Ohms. Note that the use of a resistor (pure resistance) as the load over this region is not accurate. As everyone in the world knows, the impedance is not purely resisitive so the predictions have error. The correct model would be the passive network loaded by both the resistive and reactive parts of Z. The notch filter response was also modeled using an 8 Ohm resistor. I have the equiv. circuit model for the K33e but not loaded in the Klipschorn. I will rerun the notch filter into the K33e in free air and see how the response varies.

[bodcaw boy]

I read the Delgado article describing the Jubilee bass unit. In that article the anechoic response of the Klipschorn bass unit without a filter is shown. I've cut and pasted it in this thread. On the log-log plot, the response shows a large "hump" with a peak between 200-250Hz. Given the measured response, I was interested in seeing what the AK bass filter sections looked like. I was also interested in determining the magnitude of the components needed to make a (large) notch filter that could nullify the response. That said, I have built nothing yet, this is all model predictions. First plot is the Klipschorn response from the AES article (R. Delgado and P. W. Klipsch, J. Audio Eng. Soc. Vol. 48, No. 10, 2000 Oct pp. 922-929).

hello john,

you might want to try the really really stupid and unfathomable crossover i designed for the khorn lf; the ak 4 lf section. but if you are going to try it for acoustics and not microwaves, then it might just work. there is a program called leap that uses imported real impedance curves and real imported freq curves to develop networks with. if you need any info, let me know.

roy delgado

[DrWho]

John, how does the crossover transport function change when you throw in the inductance of the K33 in addition to the wildly changing impedance from the acoustical loading? It looks to me like the old crossovers were trying to extend the HF bandwidth of the bass bin - and if you reduce the impedance seen by the network, don't the bumps at the top increase in frequency?

And won't reducing the huge hump in the response bring down the effective sensitivity? I would think it would be easier to just go with an active EQ []

[John Warren]

Dean-Here's the AA, AK-2 and AK-3 looking into a resonance circuit that simulates the K3E in the folded unit. It's not perfect but it's better than the resistor load (read the disclaimer below). The top graph is the impedance approximation with phase angle. The bottom graph is the predicted voltage across the driver (in relative dB) loaded with the simulated impedance.

[John Warren]

Hi Roy-

I'm glad you chimed in! Are you trying to tell me that the primary function of a loudspeaker filter is NOT to match the impedance at all frequencies for maximum power transfer? Are you trying to tell me that the criteria for a flat frequency response is different? Are you trying to tell me that voltage sources don't really care what the impedance is? Are you trying to tell me that the filter design process requires knowledge of both impedance AND frequency response information? and that BOTH must be considered in the design? (so where the you two years ago when I asked "what's his name" these very questions?). I could go on and on but I don't want to get my BP going skyward, it's the weekend.

LEAP is a powerful tool and I am aware of it. Based on your response, the AK-4 has some frequency compensation circuits built into it. I'd be interested in seeing the topology (and yes I know I could "buy" a pair but what fun would that be!) and I am also aware that you can't reveal the information. Is there any possibility that you could show us the measured frequency response of the bass horn with the AK-4 filter branch?

Also, the Jubilee looks like a home run. I'm surprised a consumer version wasn't marketed.

jw

[Deang]

The frequency scaling at the bottom of the graph isn't there, can you remove and reinsert so it shows up?

I believe Dennis once posted that the AK-2/AK-3 networks lift response and plug that hole at 200Hz. I will have to wait to get to work on Monday to pull the text from my notes.

It looks to me like the old crossovers were trying to extend the HF bandwidth of the bass bin...

Yes, but best way to do that is to remove the inductor and just play it full out.

...and if you reduce the impedance seen by the network, don't the bumps at the top increase in frequency?

That's what I was thinking, and why I asked what he saw if he changed the dummy resistor to four ohms. Also, if impedance drops at that spot where the hole is, the amplifier would double its output at that frequency and possibly lift the response some. That's pure conjecture, I have no idea if it works that way.

[Deang]

I would love to revisit those issues and get Roy's take. I still have a lot of trouble with a couple of them.

[John Warren]

John, how does the crossover transport function change when you throw in the inductance of the K33 in addition to the wildly changing impedance from the acoustical loading? It looks to me like the old crossovers were trying to extend the HF bandwidth of the bass bin - and if you reduce the impedance seen by the network, don't the bumps at the top increase in frequency?

Mike-The graph just above is an attempt to quantify the effect of a real load on the LF filter sections. The crossover transfer function is determined by the topology of the network. The radiation resistance due to horn loading does not wildly change, in fact, it's quite benign (it's a pure resistance!). The inductive portion of the loudspeaker vc is now contributing to the impedance.

And won't reducing the huge hump in the response bring down the effective sensitivity?

Yes it will but consider a siren, it's exceedingly efficient at one frequency.

I would think it would be easier to just go with an active EQ []

You cannot fix frequency domain problems in the time domain. Tweaking the filter is a frequency domain fix to a frequency domain problem.

[bodcaw boy]

Hi Roy-

I'm glad you chimed in! Are you trying to tell me that the primary function of a loudspeaker filter is NOT to match the impedance at all frequencies for maximum power transfer? yes.... Are you trying to tell me that the criteria for a flat frequency response is different? yes.... Are you trying to tell me that voltage sources don't really care what the impedance is? to an extent; amps have trouble with very low impedance loads...the wimps..... Are you trying to tell me that the filter design process requires knowledge of both impedance AND frequency response information? and acoustic phase response and power response of the different sections of the loudspeaker would be nice.... and that BOTH must be considered in the design? yes..... (so where the you two years ago when I asked "what's his name" these very questions?). i was getting a haircut..... I could go on and on but I don't want to get my BP going skyward, it's the weekend.

LEAP is a powerful tool and I am aware of it. Based on your response, the AK-4 has some frequency compensation circuits built into it. yes it does.. I'd be interested in seeing the topology (and yes I know I could "buy" a pair but what fun would that be!) and I am also aware that you can't reveal the information. the schematic is known and been published here and in fact, mr crites, if he is on tonight, can supply it. if not, i can get it somewhere...... Is there any possibility that you could show us the measured frequency response of the bass horn with the AK-4 filter branch? i am sure we got that somewhere....

Also, the Jubilee looks like a home run. I'm surprised a consumer version wasn't marketed. be careful, you are getting close to this becoming a jubilee worship thread.....

jw

wow! don't know what to say.....i am not used to very many intelligent, data based, knowledge exchanging conversations at this place. usually all i see is lots of opinions and no data or facts. i reread and reread your post to see if i missed something. we'll see.......you're not going to pull off a mask or something?

roy delgado

[seti]

Jubilee worship! Is that what we need to do to help get the Jubilee consumer version to market? How many goats need to be sacrificed? Not sacrificing the virgins though.

Fun thread : )

[mikebse2a3]

John, how does the crossover transport function change when you throw in the inductance of the K33 in addition to the wildly changing impedance from the acoustical loading? It looks to me like the old crossovers were trying to extend the HF bandwidth of the bass bin - and if you reduce the impedance seen by the network, don't the bumps at the top increase in frequency?

Mike-The graph just above is an attempt to quantify the effect of a real load on the LF filter sections. The crossover transfer function is determined by the topology of the network. The radiation resistance due to horn loading does not wildly change, in fact, it's quite benign (it's a pure resistance!). The inductive portion of the loudspeaker vc is now contributing to the impedance.

And won't reducing the huge hump in the response bring down the effective sensitivity?

Yes it will but consider a siren, it's exceedingly efficient at one frequency.

I would think it would be easier to just go with an active EQ []

You cannot fix frequency domain problems in the time domain. Tweaking the filter is a frequency domain fix to a frequency domain problem.

Actually the frequency domain and time domain together is what represents more acurately the frequency response of a loudspeaker. Must reading for anyone on this subject is in the written works of Richard C. Heyser many of which can be found in "Time Delay Spectrometry An Anthology" from the Audio Engineering Society. In particular please refer to Heyser's Paper's on (Loudspeaker Phase Characteristics and Time Delay Distortion: Part 1 and also Part 2) for when correction for a loudspeaker frequency/phase response is possible. The use of an EQ is possible when the area of concern is of minimum phase behavior and will show an improvement in both frequency and phase measurements in the area of concern. Trying to apply an EQ correction in an area of non-minimum phase behavior will most likely make things worse. In Part 1 of the paper I'm refering to Heyser demonstrates how to measure and identify minimum phase behavior and non-minimum phase behavior in several types of loudspeakers and in Fig. 10 he uses a corner horn system with a dip in response that is in a minimum phase area of it's frequency response and as he applies a corrective EQ to the system the Frequency and Phase Response are both much improved.

It still amazes me how little Heyser's Works seems to be known. Heyser has to be one of the greatest minds ever to have worked in Audio and his Papers should be read by anyone studying in Acoustics and other related fields.

mike tn

[DrWho]

Mike-The graph just above is an attempt to quantify the effect of a real load on the LF filter sections. The crossover transfer function is determined by the topology of the network. The radiation resistance due to horn loading does not wildly change, in fact, it's quite benign (it's a pure resistance!). The inductive portion of the loudspeaker vc is now contributing to the impedance.

I thought the acoustical impedance of a horn was all over the place? Thus also affecting the effective electrical impedance? I have attached a picture with the predicted electrical impedance for a bass horn I'm working on. I would imagine the khorn being even more complicated due to all of its folds???

You cannot fix frequency domain problems in the time domain. Tweaking the filter is a frequency domain fix to a frequency domain problem.

Usually I hear these arguments being presented in the other direction (can't fix time domain in the frequency domain), but I still agree. I was just thinking it would be easier and cheaper to implement active EQ at the line level before the amp instead of trying to build it into the network for the speaker. I would think the only advantage to having it after the amplifier is if it makes the bass bin an easier load to drive. But if you're gonna go with active EQ, then why not go with a unit that is also an active crossover? And then you can get some time-alignment and remove yet one more variable []

[DrWho]

Are you trying to tell me that the primary function of a loudspeaker filter is NOT to match the impedance at all frequencies for maximum power transfer? Are you trying to tell me that the criteria for a flat frequency response is different? Are you trying to tell me that voltage sources don't really care what the impedance is? Are you trying to tell me that the filter design process requires knowledge of both impedance AND frequency response information? and that BOTH must be considered in the design? (so where the you two years ago when I asked "what's his name" these very questions?). I could go on and on but I don't want to get my BP going skyward, it's the weekend.

ROFL!

...where's the crazy icon everyone has been using?