Jump to content

Cornscala and the incredible disappearing act of high end - help and advice please


Gints

Recommended Posts

Team, I am seeking some advice here.

 

Some background is avilable here (first struggling with overpowering mids, then bi-amping and active EQ), though I have since moved house and things have moved on: 

 

Cliffs notes:

- Bought cornscala, found them muddy - very overpowering around 300-400hz, not much down very low, not much up top

- Started using Dirac and a MiniDSP 2x4 DDRC to EQ this better, and use a more powerful solid state amp to push the additional boost down low.

- System sounds great, good imaging, balanced, quite happy.

 

However, I would like to minimise the amount of correction that DIRAC is required to do. Ideally 0. It is doing an amazing job and I get alot of enjoyment from it, but I would like to understand root cause given this is decent gear.

 

I am trying to understand what could be causing this poor high-end frequency response, and what steps I could take to ascertain it?

 

Since Cornscala is not a real model, this particular build consists of:

Midrange Driver is a PD-5VH mated to a wooden horn (see pics)
Tweeter is a CT125
Woofer is a CW1526C
Crossover is B&K Sound Type CS 400/4500Hz 

 

Amps are a 15W Transcendant Son of Beast for the Highs, AMPower SA1 200W Solid State for the woofer. Low amp attentuated in the active corssover to compensate.

MiniDSP is providing active crossover at 450Hz.

Low is directly wired to amp (crossover bypassed).

Existing crossover is maintained and components bypassed to still provide the 4500Hz crossover point between Squawker/Tweeter (details in other post).

 

Speakers themselves sound great when corrected, but I want to understand why the response might be so bad. I do not know who originally built them, but the wood, workmanship, heft, quality, attention is amazing - so I would like to get these to their potential.

 

1. Could the crossover be busted and maybe look to ALK or similar for a new one?

2. I still have the Squawker/Tweeter coming off the -6db and -3db pins of the autoformer - any issue there (besides wasting some power...but power is not my issue)?

3. Could my crossover modification have created an impedance issue (though this symptom existed even when they were single-amped)

 

Note these problems were the same before I bi-amped. 

 

Ideas/Tips/Advice from those far smarter than I. Happy to supply any more info as required. There are pics of the units in the Bi-Amping thread linked.

 

This is the pre-read from DIRAC (9 full spectrum sweeps in various positions, averaged)

DIRAC_Freq_response.thumb.png.537cf83f790bca97c88dc72b681bad6c.png

Below is the response (supposedly) after the filter is applied.

image.thumb.png.68587166a7e60c0f657bde5a5041bed3.png

 

Crossover pre-modification/bypass:

image.png.b08e5cda8067e4d57d31cdd493bb500d.png

image.png

Edited by Gints
Descirbed the DIRAC screenshots
Link to comment
Share on other sites

Replacing the stock xover with an ALK Universal would be a good start.

Sell the Crites xover ( someone will buy them) to fund the ALK. Expect to wait a few weeks for them.

 

My Crites Cornscalas were unlistenable until i replaced the xover with the ALK.

 

Bob Crites really needs to redesign his xover. Weakest link in his design IMHO

 

 

 

 

 

 

Link to comment
Share on other sites

Hello, I have custom CornScala type speakers very similar to what you have but in "pro" style cabinets I have the Crites CT-120 tweeter and A55-g mid driver and that combo is really really good, best I've ever owned and are IMO head and shoulders above the stock Klipsch drivers k-77 and single phase plug k-55v.

 

Looking at your cabinets they do not appear to be standard CornScala type which are modeled after the original Cornwall, I'd double check the internal cabinet volume to make sure you have enough air space for that woofer as that could really change the bass response, I believe the  Cornwall cabinet is roughly 6.5 cu ft internally those woofers need room to breathe.

Link to comment
Share on other sites

Ok i will measure. You are correct regarding shape... they are normally shorter and wider. 

 

Took a quick measurement:

540 depth

510 width

1050 height

 

Adjusting for around 35mm of panel thickness (these boys are thick and heavy) leaves about .202cubic m or 7.0777777777 cubic feet. Less some volume occupied by horns.

 

So an unconventional shape, but the size seems fine. Note there is no separation in compartment (I have seen some implementations where the bass bin is separate.

Edited by Gints
updated with detailed measurements
Link to comment
Share on other sites

12 hours ago, Gints said:

[1] System sounds great, good imaging, balanced, quite happy.

 

[2] However, I would like to minimise the amount of correction that DIRAC is required to do. Ideally 0. It is doing an amazing job and I get alot of enjoyment from it, but I would like to understand root cause given this is decent gear.

 

[3] I am trying to understand what could be causing this poor high-end frequency response, and what steps I could take to ascertain it?

So, in short:

 

A. Everything sounds great with Dirac providing correction...

 

B. ...but you think that you have to do something more because Dirac is correcting the system so that it provides flat frequency response. 

 

C. ...furthermore, you've already decided what the solution is, and you want to know why your assumption--that all the drivers, when put into a box and crossed using an electronic, digital, or passive electrical crossover filter, don't automatically provide linear, flat frequency response without the Dirac in the loop.

___________________________________________________________________________________________________________________________

 

So to answer your apparent underlying question: there is no such thing as a loudspeaker system that produces flat frequency response (i.e., at the level you achieved using Dirac) without correction somewhere-even in the passive "balancing networks".  Statements B and C are the problems themselves. The assumption that flat inherent response of individual drivers placed in a box together and simply crossed over using a passive crossover without EQ or other notch filters (passive or active) is actually incorrect. 

 

Your woofers are a bit too sensitive while using the bits of the passive crossovers that you've got for the midranges and tweeters in that the woofers currently produce about 5 dB too much SPL on-axis relative to the midrange horn/driver and tweeter.  This is apparently due to some attenuation of the passive crossover network (that you're still using of the midrange horns/drivers) relative to the woofer because your direct radiating woofers will always be less efficient than the midrange horns/drivers. Your tweeters aren't as sensitive as your midrange horns/drivers, so their frequency response appears rolled-off, but the real cause of the tweeter frequency response roll-off is that the tweeters just aren't quite as sensitive as the midranges, and that the tweeters appear to have drooping response above 10 kHz.

_________________________________________________________________________________________

The root issue above currently is that you've got three-way loudspeakers and are only bi-amping them, instead of tri-amping them so that you can control time alignments and relative channel gains separately, as well as providing needed EQ of the horns/drivers and woofers to correct their non-flat frequency response--which all drivers/horns (or direct radiating drivers alone) exhibit.

 

If you instead tri-amp your Cornscalas using a 2-in, 6-out DSP crossover and six channels of amplification instead of your present four, you can also dial in the tweeter delays and SPL relative to the midrange, and the same for the woofer relative to the midrange, then you'll have time-aligned loudspeakers (and the difference in clarity and transparency is very audible). 

 

...Or you could release the tweeters from inside the box and bring them outside on top, then move them back to physically time align the tweeters to the midranges (placing the tweeters in suitable small baffles to hold them in place, and covering up the tweeter holes in the loudspeaker boxes.  This would still leave the midrange-woofer time misalignment, but that time misalignment is much less than the tweeter/midrange misalignment when viewed in terms of wavelengths of misalignment at the crossover frequency interference bands.

 

Chris

  • Like 1
Link to comment
Share on other sites

Thank you Chris. Plenty to think about.

 

For me, it is not necessarily about doing something more... just trying to see if there are any clues as to why it is occurring, which you have helped.

 

I agree the 3 levels are out. The woofer I can correct via the active crossover. The squawker/tweeter I currently only have the active crossover and relative autoformer attentuation. My concern however was that while I could increase the tweeter power (remove autoformer/different tap etc.) it would still not amend the 'droop' you describe - presumably it will lift relatively evenly across the spectrum. Why would the tweeter have such droop? Is that normal for this device?

 

Tri-amping sounds interesting. Presumably this allows great levels of correction (frequency/power/timing). Whilst DIRAC may be frowned upon, what is has done to this system under bi-amping is absolutelu incredible.

Link to comment
Share on other sites

I thought that was his point, that the results of DIRAC were so admirable.

 

If the tweeter crossover parts are between the autoformer and tweeter it should matter less that you swap taps, but if the parts are before the autoformer you'll also be changing the crossover frequency and perhaps functionality.

  • Like 1
Link to comment
Share on other sites

10 hours ago, Chris A said:

 

 

C. ...furthermore, you've already decided what the solution is, and you want to know why your assumption--

 

Sorry mate, what do you mean I've decided what the solution is? Do you mean 'I have assumed the solution is to make my speakers perfectly flat out of the box' as the solution.

 

Obviously they will never be flat, but I am interested in exploring how they could be improved (aside from DIRAC)

 

"and that the tweeters appear to have drooping response above 10 kHz."

 

Is this fixable or is this just a hard fact of the tweeter? While I understand tri-amping will allow individual level control and timing control - would that fix the roll-off problem I have with the highs? Or would it just life the droopy response upward? I suppose you could correct that with EQ etc.

Link to comment
Share on other sites

12 hours ago, Gints said:

My concern however was that while I could increase the tweeter power (remove autoformer/different tap etc.) it would still not amend the 'droop' you describe - presumably it will lift relatively evenly across the spectrum. Why would the tweeter have such droop? Is that normal for this device?

I have the same tweeters in a pair of 1979 Cornwalls, and they exhibit the same response and require the same levels and type of correction:

 

1869178313_RightCornwallUncorrectedandCorrected(REWEQ).thumb.JPG.2ffa78eb4bb915e9dc7fe452e33b3ac9.JPG

 

2 hours ago, Gints said:

Sorry mate, what do you mean I've decided what the solution is? Do you mean 'I have assumed the solution is to make my speakers perfectly flat out of the box' as the solution.

Obviously they will never be flat, but I am interested in exploring how they could be improved (aside from DIRAC)

New tweeters that have flatter inherent response using their supplied horns...is the answer to your question. It is desirable to have flattish response, but not required if you are using something to correct the response, which is the point that I was making.  I see a lot of "audiophiles" (a misnomer in this case) that have only one measure of merit (MoM) for loudspeaker performance...you guessed it: flat on-axis frequency response.  In fact, this is not really as important an MoM as others because it is easily corrected, unlike the most important factors for loudspeaker performance.

 

Paul Klipsch wrote a paper in 1970 called "The Mud Factor" in which he said the following:

 

Quote

The attributes of a loudspeaker, in order of importance, are:


(1) Total Distortion, at a given acoustic power output level [a.k.a., modulation (FM and AM sidebands), and compression distortion, etc.]
(2) Polar Response [a.k.a., consistent directivity vs. frequency]
(3) Amplitude vs. Frequency response [a.k.a., flat frequency response, but at more than just on-axis, and this is correctable on-axis, but not off-axis]
(4) Harmonic Distortion [actually an indirect measure of modulation distortion]

There are other measures that are (in my experience) even more important than the last two factors, but PWK had limitations of what he could easily measure (and explain to his customers) in 1970, limitations that we don't have today, namely:

  • phase distortion,
  • group delay distortion,
  • impulse and step response distortion,
  • compression distortion,
  • resonance/decay response

 

Most of the above factors are not independent of each other--as we will see in further discussions.

__________________________________________________________________________________________________

One of the good things about the CT125 tweeters is that they actually sound better to my ears, before or after correction, than the stock K-77s that come with the Heritage Klipsch loudspeakers of that time period.  But the problem is that this tweeter must have correcting EQ to flatten its response.    Note the 5 dB PEQ needed at 16 kHz to correct its high frequency droop.

 

So the reason why I wrote what I did was to highlight that what you fasten on as the root problem--tends to become the only problem that you pay attention to. 

 

If instead you use your ears and a measurement tool such as REW, you get much better results, in my experience.

 

Chris

Link to comment
Share on other sites

Here is a natural response of a TAD TD-4002 driver on a K-402 horn (on axis) before correction:

 

TAD TD-4002 on K-402 (No EQ, no Xover FIlters), 1 m on-axis SPL Phase.jpg

 

Here are the required PEQ corrections:

 

TAD K-402 PEQs (minimal).JPG

 

and the resulting response (on and off axis):

TAD K-402 uncorrected and corrected response.JPG

 

The point that I'm trying to make is that frequency response, uncorrected, is not the measure that it's cracked up to be.  The loudspeaker that this driver and horn is used on is the Klipsch Jubilee, and in my experience I've not heard a better loudspeaker (that is, the Jubilee dialed-in properly).

 

Chris

Link to comment
Share on other sites

Oh, yes...  Jubilees, properly set up, are spectacular:

 

Right Jubilee SPL and phase response.jpg

TAD TD-4002 Jubilee (Red Trace) vs. Danley SH-50 (Green Trace) Phase.jpg

TAD TD-4002 Jubilee vs. Danley SH-50 group delay response.jpg

TAD TD-4002 Jubilee Impulse Spectrogram (Fractional Order Crossover Filters).jpg

TAD TD-4002 Jubilee vs. Danley SH-50 step response.jpg

The Danley SH-50 step response is in cyan, the Jubilee in yellow trace color in this last step response plot.

 

That's just the measurements.  If you listen to them after careful dialing in (shortened-down HF horns and zero phase shift crossover filters, etc.) it's difficult to put the SH-50 in the same category as the Jubilee (w/TAD TD-4002 drivers).  I'm really impressed with what I hear currently (and I've owned the Jubs for almost 12 years now). 

 

They're spectacular performers.  But I digress from the OP's subject...

 

Chris

Link to comment
Share on other sites

The Fc or cut-off, is the lowest frequency the horn is capable of playing. That looks like Dave Harris' horn for the Cornwall, it looks too small to be the Fastrac LaScala or Al's Trachorn. I believe the Fc of that horn is 384Hz. With that driver, and a 6dB/octave filter, your crossover should be at least a half octave above cut-off, or, 600Hz. Crossing over too low will increase distortion, and puts the diaphragm in a constant state of distress.

 

Link to comment
Share on other sites

12 minutes ago, Deang said:

The Fc or cut-off, is the lowest frequency the horn is capable of playing. That looks like Dave Harris' horn for the Cornwall, it looks too small to be the Fastrc LaScala or Al's Trachorn. I believe the Fc of that horn is 384Hz. With that driver, and a 6dB/octave filter, your crossover should be at least a half octave above cut-off, or, 600Hz. Crossing over too low will increase distortion, and puts the diaphragm in a constant state of distress.

As you approach Fc, the response and control from the horn itself, not including the driver issues mentioned above will also get quite ugly.  The closer to Fc, the more jagged the response is.

  • Like 1
Link to comment
Share on other sites

On 9/1/2019 at 6:14 AM, EmilC said:

 

Bob Crites really needs to redesign his xover. Weakest link in his design IMHO

Hi EmilC -

 

Just a bit curious about your observations here -- though let me say first I don't own or use any of Bob's crossovers.  I make my own, as well as wind my own chokes.  Describing something - anything - as the "weakest link" is fine!  Just as I've told my Art History students for decades that "not liking" a work of art is all well and good.  I just prefer specific supporting information.  And of course we need to always acknowledge the fact that there are many absolutely satisfied customers of Crites' products.  

 

In light of the above, can you be more specific about what it is (specifically) that makes it the weakest link?

  • Like 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...