Correction of large format compression drivers

[vego99]

I recently acquired a new pair of compression drivers for my Klipsch Jubilee horns.

They are Radian 950bepb drivers which house a large 4” beryllium diaphragm.

There are others using similar compression drivers with traditional DSP processors. Those processors use shelving to get them flat to 20khz.

With Audiolense I am noticing a quite a bit of correction is needed to get them to match my modest target curve. 15db of boost correction which results in quite a bit of insertion loss.

It seems as if I have enough voltage in my dac to play at my normal listening levels, Im just not quite used to this much correction.

Should I not be using this much correction, am I risking any damage? Can anyone provide any tips with making a better filter that might not require that much correction?

@Mitchco I know you had worked with Rob(Charbuggy) in the past with his Jubilees / Radian drivers. Did you also run into the same scenario?

Thanks,

-Erik

 

[Paul W]

I also use the 4" beryllium Radians. You can (and considering the cost of the diaphragms, maybe should) use a quality single "protection" cap sized appropriately to help roll off the bottom end.

 

[sledwards]

Here is a result with much less insertion loss. I used partial correction to 13,000 Hz to relieve the tweeters. and a few other tweaks in CPD.

Steve

 

[Iansr]

Unless you are a teenager there’s no point in correcting up to 20khz. You’re lucky if you are even hearing 15Khz.

 

[vego99]

Here is a result with much less insertion loss. I used partial correction to 13,000 Hz to relieve the tweeters. and a few other tweaks in CPD.

Steve

Hey Steve,

Thanks for giving a hand with this. With beryllium arent the breakup modes higher allowing the ability to correct up to their rated FR of 22khz?

ALso, what is the purpose of the mid-band correction window? That is something I have yet to wrap my head around.

-e

 

[vego99]

Unless you are a teenager there’s no point in correcting up to 20khz. You’re lucky if you are even hearing 15Khz.

This may be true… Buy why buy a Ferrari if you’re limited to the posted speedlimit? lol

 

[sledwards]

I have overlaid the measured driver frequency response from Radian on the AL frequency response of your tweeter. You can clearly see where the in-room response of your tweeter is falling off relative to the near-field (maybe 1m) response measured by Radian. I would not have expected that big of difference. How is the tweeter mounted?

 

[vego99]

I have overlaid the measured driver frequency response from Radian on the AL frequency response of your tweeter. You can clearly see where the in-room response of your tweeter is falling off relative to the near-field (maybe 1m) response measured by Radian. I would not have expected that big of difference. How is the tweeter mounted?

Hey Steve,

Tweeter is mounted in a big horn! lol.

Here are a couple base line measurements I did when I first got them.

Measured on axis at the mouth and the other is 1m on axis then at LP normalized at 10khz

 

[vego99]

Hey Steve,

Tweeter is mounted in a big horn! lol.

Here are a couple base line measurements I did when I first got them.

Measured on axis at the mouth and the other is 1m on axis then at LP normalized at 10khz

@sledwards after doing more research.. It seems to be well known that Constant Directivity horns loose about 6db of frequency response per octive at and above 10khz. It also seems as if this large K402 horn does a significant amount of loading in the band pass range.. Both of these are seen when compared to the FR from radian, to which I believe is not a fair assessment because they do their testing with a wave tube rather than a horn.

It seems as if, this is the reason why people use shelving to lower the overall response to bring back the overall high FR of CD horns / Compression drivers.

This now begs the question again, when using AL.. Is it appropriate to do what I have been doing with putting in a correction of 15db of boost? ( is this actually boosting or overall correction)
I dont know the answer. I’ve also started thinking about the need for a passive network to the shelving or put some sort of additional EQ processing to perform the shelving before AL does its magic. I guess with this method, you would be retaining and reducing all of the insertion loss...

@juicehifi any thoughts?

 

[Mitchco]

@Mitchco I know you had worked with Rob(Charbuggy) in the past with his Jubilees / Radian drivers. Did you also run into the same scenario?

Yes, these are constant directivity waveguides and as such requires constant directivity eq: https://sound-au.com/project173.htm

My JBL waveguides (with the large format 4" compression driver) requires the same type of eq along with JBL M2's and other constant directivity waveguides. A good paper on why constant directivity. I let Audiolense take care of the constant directivity eq.

Pro tip: here is the "trick" for good sound using these devices. Use a partial correction. Extend the frequency correction as much as you can before you notice too much insertion loss (get it up there to 15 kHz if you can). But only use up to 10 kHz excess phase correction. Or somewhere around there as you may have to play with where the excess phase correction ends by listening.

Why do this? The frequency correction adds the necessary constant directivity eq and the partial excess phase correction below 10 kHz (or there abouts) lets the Radian drivers "phase plug" do what it was intended to do. So you don't want to mess that up with the excess phase correction.

Cheers!

 

[vego99]

Yes, these are constant directivity waveguides and as such requires constant directivity eq: https://sound-au.com/project173.htm

My JBL waveguides (with the large format 4" compression driver) requires the same type of eq along with JBL M2's and other constant directivity waveguides. A good paper on why constant directivity. I let Audiolense take care of the constant directivity eq.

Pro tip: here is the "trick" for good sound using these devices. Use a partial correction. Extend the frequency correction as much as you can before you notice too much insertion loss (get it up there to 15 kHz if you can). But only use up to 10 kHz excess phase correction. Or somewhere around there as you may have to play with where the excess phase correction ends by listening.

Why do this? The frequency correction adds the necessary constant directivity eq and the partial excess phase correction below 10 kHz (or there abouts) lets the Radian drivers "phase plug" do what it was intended to do. So you don't want to mess that up with the excess phase correction.

Cheers!

@Mitchco thanks for chiming in. And thanks for the Protip on the partial correciton. I learned about this a while back with great success…. Here is a screen shot of the partial correction I am using. I’ve got to push the frequentcy stop correction to 20 - 22khz to get things to lift up past 15k otherwise the FR drops like a rock..

Please help me under stand this portion of youre response “ only use up to 10khz excess phase correction"

As you can see I’m stopping TTD correciton at 900hz which is slighly above my crossover point ( 475hz). If I am correcting the way I am now, am I doing any harm with the 15db of correction boost or is it just the overall correction deviation? As mentioned above, even with the insertion loss, I can still achieve about 95db spl + at my listening position and I still have more volume to go and I can also increase the voltage of my dac by another +6dBu

 

[Mitchco]

@vego99 from the Radian website: "The 950 driver delivers clean high frequency sound with ultra-low distortion. This reduction in 2nd and 3rd harmonic distortion is only possible with our large format 4" diaphragm, 5 slit phase plug and Mylar suspension."

Key phrase: 5 slit phase plug. These drivers have been designed using the latest computer aided design tools (I forget the name at the moment) that optimize frequency and phase response - especially near where the wavelengths approach the diameter of the voice coil and beyond. Here is a good reference: https://fohonline.com/articles/speaking-of-speakers/understanding-compression-drivers-phase-plugs/

"Modern drivers are now less concerned with how much the phase plug can increase efficiency and more with improving compression driver extension and response smoothness."

So we don't want to mess that up too much with excess phase correction in the high frequencies as it is already "designed into" the system. But we do want to compensate for the constant directivity horn eq. It is OK to correct for the excess phase up to where the phase plug becomes active. Typically with large format compression drivers that is around 10 kHz with a 2" inch exit.

The JBL 2453H-SL aquaplas coated 1.5" exit I use makes it out to about 11 kHz. If you are using the latest version of AL, you should be able to took at the phase of the driver and see if there are rotations around 10 kHz or wherever that frequency is. Then don't use excess phase correction beyond that point. You might have to "bracket" some filters with varying ranges of excess phase cut off frequency and give a listen. If you can level match and have instant switching, you should be able to tune into the soundstage differences. One will be the "sweet spot" where everything sounds totally focused.

Good luck!

 

[vego99]

@vego99 from the Radian website: "The 950 driver delivers clean high frequency sound with ultra-low distortion. This reduction in 2nd and 3rd harmonic distortion is only possible with our large format 4" diaphragm, 5 slit phase plug and Mylar suspension."

Key phrase: 5 slit phase plug. These drivers have been designed using the latest computer aided design tools (I forget the name at the moment) that optimize frequency and phase response - especially near where the wavelengths approach the diameter of the voice coil and beyond. Here is a good reference: https://fohonline.com/articles/speaking-of-speakers/understanding-compression-drivers-phase-plugs/

"Modern drivers are now less concerned with how much the phase plug can increase efficiency and more with improving compression driver extension and response smoothness."

So we don't want to mess that up too much with excess phase correction in the high frequencies as it is already "designed into" the system. But we do want to compensate for the constant directivity horn eq. It is OK to correct for the excess phase up to where the phase plug becomes active. Typically with large format compression drivers that is around 10 kHz with a 2" inch exit.

The JBL 2453H-SL aquaplas coated 1.5" exit I use makes it out to about 11 kHz. If you are using the latest version of AL, you should be able to took at the phase of the driver and see if there are rotations around 10 kHz or wherever that frequency is. Then don't use excess phase correction beyond that point. You might have to "bracket" some filters with varying ranges of excess phase cut off frequency and give a listen. If you can level match and have instant switching, you should be able to tune into the soundstage differences. One will be the "sweet spot" where everything sounds totally focused.

Good luck!

Thank again Mitch.
I have an older version of AL with no phase correction.

also, just to confirm… this differs from the TTD correction stop point in the partial correction field?

 

[juicehifi]

No, it's the same.
You should zoom in on the tweeter correction filters. If you're overcorrecting you will see a lot of high frequency oscillation around the peak. Lots of shark teeth before and after.
The best solution here is most likely to stop the TTD correction at 10 kHz or lower.

Sometimes a frequency correction all the way to the top doesn't work well either. But maybe it does here, since the resopnse in the top looks quite smooth. You should try to find that out. In any case you should stop the frequency correction before the tweeter drops seriously.

If you find that semi full range frequency correction creates an over-buzy tweeter correction you can use the target to do a coarse eq and by setting the no-correction boost to a level that levels things out as good as it gets.

But remember to check the impulses of the tweeter correction.

 

[vego99]

No, it's the same.
You should zoom in on the tweeter correction filters. If you're overcorrecting you will see a lot of high frequency oscillation around the peak. Lots of shark teeth before and after.
The best solution here is most likely to stop the TTD correction at 10 kHz or lower.

Sometimes a frequency correction all the way to the top doesn't work well either. But maybe it does here, since the resopnse in the top looks quite smooth. You should try to find that out. In any case you should stop the frequency correction before the tweeter drops seriously.

If you find that semi full range frequency correction creates an over-buzy tweeter correction you can use the target to do a coarse eq and by setting the no-correction boost to a level that levels things out as good as it gets.

But remember to check the impulses of the tweeter correction.

Hey @juicehifi Bernt,

Thanks for chiming in.. Ok, thanks for the info.. I was just getting concerned that I was over corrected causing potential harm to the driver.

Correction filter…. Yes! This is something Ive been keen to keep an eye on… With my correction out to 22k and a TTD stop at 900hz the correction filter on the HF is pretty good…. Not sure what your opinion is…. The low frequency i’m struggling with.. not sure if this is just residual noise from what the mic picked up. I can adjust the time domain window really low to (.5/50MS) which results in a good correction impuse but the group delay and step response suffers somewhat.