Compression Driver Compensation

[Robh3606]

Hello Zilch

Happy New Year!!

I believe that's harder than I've ever pushed 2435s, Guido, and if my sims are correct, 6 dB harder at 20 kHz than the Timbers filter:

Nothing is getting pushed. Guidos crossover does not increase the power into the driver. All of these crossovers work by attenuation of the midband response. That is all you are doing.

Also, I don't understand the reference to "Unity," or zero dB at the high end. The amount of boost is relative to where it begins, and, using the notch at 2 kHz as benchmark, Timbers has 14 dB to 20 kHz from there, whereas you've got 20 dB, looks like....

There is no boost. 0db is the reference to a given signal level in this case 2.83 Volts or 1 watt input. What you seem to miss is the 14db is really a negative number. It is 14db down from the input voltage. Simply said the network provides 14db of attenuation at the frequency.

Gregs filter is 7db from unity at 20K. Take a look at the voltage drive.

Rob

[Zilch]

Happy New Year, Rob! It's still 2007 here; I'm always behind you guys....

Whether we call it boost* or attenuation, the result is the same -- the HF driver is being driven harder at 20 kHz than it would be without compensation, by the dB differential. To achieve that, we attenuate the midband, yes, taking advantage of the higher sensitivity of the compression driver relative to that of the woofer. If it were not for the compensation, though, as with an exponential horn, for example, we'd be attenuating the full bandwidth of the HF driver to bring the two into balance.

The net result is that the compression driver is receiving differentially more drive at 20 kHz than it is at 2 kHz, and in the in the instance of 20 dB of differential, it's 100 times more power there. Others have suggested there are consequences to doing that, and in the case of 4430, I have earlier shown that increased distortion at the higher frequencies is among those consequences. Thus, when I am cautioned to apply compensation conservatively i.e., not to force 2435HPL beyond pistonic mode, and instead to let the response roll off above that, I look and find the specifications and actual practice of Timbers and others to be consistent with that approach.

I also find in my own experience that the Be diaphragm does not push all that easily beyond pistonic, or, at least, not as easily as the aluminum 2431H does, and I am seeing further evidence of that here....

*I found a JBL technical paper which used the term "boost" in describing the practice of CD compensation, which I consider as comprising renewed license to use the term in this context without guilt or shame....

I think there is a fair amount of unit to unit variation above 10-12KHz with these drivers. One pair of drivers may require quite a bit more kick than another pair.

Here's 18 of them:

[Robh3606]

Hello Zilch

Thus, when I am cautioned to apply compensation conservatively i.e., not to force 2435HPL beyond pistonic mode,

You have no data at all to back that statement up. That is complete conjecture on your part. There should be no effect at all as far as the diaphragm remaining pistonic or not. There is no change in relative power input. You are running the drivers at a greatly reduced power levels across the board. With the power density of music there is very little power above 10K most of it is right where the attenuation is. You also reduce the actual diaphragm excursions by attenuating the midband. Also a good thing.

I look and find the specifications and actual practice of Timbers and others to be consistent with that approach.

That is what we are all doing. Then we season to taste.

Here are the power and distortion graphs for the 4430 horn. You can see the abrupt change in level in the second harmonic. You can also see how added power really makes this jump. These are at power levels that you would expect to be used in a professional application. I doubt any of us would push them that hard where your average power input would be 25 watts.

Attached are the THD plots for my 2435's from the above measurements. I don't see that same knee as the 4430. I don't have as much power going into them but the measurement is at typical levels I would use them at. If I dumped 25 watts into them I am sure the THD would rise but that is not going to happen. Especially out at 20K.

Rob

[Robh3606]

This started in another thread. Moved the posts over so we didn't hijack the thread.

Rob

[Ian Mackenzie]

Hi Rob,

I've always seen this driver EQ was cutting the mid band with a contour to match the mass rolloff.

You simply cannot have more than you got to start with.

I think what Zilch Von Curvee' is about is he's using active Eq.

But the point is once you normalise the compression driver output level to match the woofer if you looked at the overall voltage drives the compression driver voltage reduces with frequency.

We are assuming the compression driver and horn is at least 10 db more sensitive in the midband.

[Zilch]

You have no data at all to back that statement up. That is complete conjecture on your part. There should be no effect at all as far as the diaphragm remaining pistonic or not.

It's the old "Intended purpose" thing I get smacked with all the time here; use the drivers within their designed performance envelope:

http://www.jblpro.com/vertec1/doug%2...%20rev%20c.pdf

Breakup mode is virtually by definition a bad thing.

We wished for the first break up mode of the diaphragm to be at the very top of the band of audibility.

Guido's "Pushing" the drivers into breakup mode at three times the rate of the Timbers filter above 10 kHz.

Me, I'm the premiere "Booster" here, you may recall, slavishly pumping up the 2435 VHF response up there, clear back to early work with the AM6212 filter nearly three years ago....

[Robh3606]

Guido's "Pushing" the drivers into breakup mode at three times the rate of the Timbers filter above 10 kHz.

Like I said you have nothing to substantiate your allegations that these filters are driving the compression drivers into break-up mode any faster than they would normally get there.

It's the old "Intended purpose" thing I get smacked with all the time here; use the drivers within their designed performance envelope:

http://www.jblpro.com/vertec1/vertec...0Chap%205A.pdf

That is the correct document to cover the intended use. It tells you how to set up the Loudspeaker Controllers for the Vertec system. When used in the line arrays they actually do use active EQ to bring up the last octave. They obviously don't want to sacrifice SPL in this application. They add at least +12db @ 16K of active EQ depending on the controller. Then they are low pass filter it at 22K. You know like a CD player does.

You really think that attenuating the midband response is an issue when in the actual application they hit them with +12db of active EQ at 16K??

I would say that what we are doing definitely fits into intended use and designed performance envelope.

Rob

[Mr. Widget]

I would say that what we are doing definitely fits into intended use and designed performance envelope.

I fully agree with that statement.

As far as I know, JBL started in this direction with the 4430/35s and has continued in that direction with many of their consumer and professional designs. Since they are not building professional monitors with compression drivers these days, I feel that only their work on the consumer side really relates to what we are doing when we are talking about high end music playback for the home. The Vertec and other JBL offerings are PA systems and as such have different requirements and sonic quality benchmarks.

In the consumer stuff the designs using these 3" drivers typically have additional tweeters also. The Everest II does carry on in the tradition of the 4430/35 especially when you consider the super tweeter being active only above 20KHz anyway... actually when you consider the 4435 with helper woofer, the Everest II is really an updated and upgraded direct descendant... but I digress.

As for this compression driver compensation thing, I am not alone, but I do think I am of a minority opinion in this group when I complain of the sonic "quality" of these compensated compression drivers... my criticism is not limited to the JBL 3" drivers either. I have tried to do a similar design with both the TAD TD-4001 and with the TD-4003. Both of these drivers are rated by their manufacturer as to having response to 20KHz and beyond... that may be, but like the smaller JBLs, this HF content is unacceptable to my ears. The Everest II's 476Be on the other hand seems to be free of the hard, edgy, metallic, unpleasant extreme top end that I have experienced with all of these other drivers. I asked Doug Button about this at last year's CES and he attributed it to the 476Be's higher mass break point. I have to assume he knows what he is talking about.

For me the solution is to "tack on" a super tweeter above the midrange and therefore only use the midrange in it's optimal performance area. This is a compromise too however as it introduces time alignment and comb filter related issues. I personally am less sensitive to these than I am to the "edgy" sound of a pushed driver.


Widget

[Zilch]

Like I said you have nothing to substantiate your allegations that these filters are driving the compression drivers into break-up mode any faster than they would normally get there.

I don't recall alleging that they are driving them there faster, rather, that they are driving them harder once operating in the breakup region. That's certainly obvious in Guido's voltage drive curve where he kicks the rate of "Boost" up to ~13.2 dB/octave above 10 kHz to achieve 6 dB additional drive at 20 kHz, whereas Timbers maintains it linear at ~4.2 dB/octave, by my crude measurements of the slopes. That may not get it there any sooner, but it clearly increases the intensity of breakup mode output at the high end by that 6 dB, as is apparent in the frequency response curves illustrating the result.

You really think that attenuating the midband response is an issue when in the actual application they hit them with +12db of active EQ at 16K??

Uhmm, the Timbers filter hits them with +12 dB of PASSIVE EQ at 16 kHz, as well, relative to the midband. At 20 kHz, he's at +14 dB; you guys are kicking it an ADDITIONAL +6 dB to +20 dB total there.

To the best of our knowledge, I believe, pistonic operation of 2435HPL becomes break-up mode somewhere in the region of 18 kHz, perhaps lower, consistent with what Doug Button states in that paper, with Greg Timbers' implementation of the passive equalization, and with 4313B's observations with respect to the preferred use of the device, according to design, in the UHF region.

http://audioheritage.org/vbulletin/a...1&d=1170461963

See also the Vertec curves. Even in SR, they do not push 2435HPL past 16 kHz; in fact, NO JBL product, either Pro or Consumer, employs it in the manner you contend to be "within the design performance envelope."

http://www.jblpro.com/vertec1/new_vertec/pdf/VT4889.pdf

I fully agree with that statement.

You agree that "pushing" 2435 at 20 kHz by 20 dB over it's midband response falls within the design performance envelope of that driver?

On what basis, that we're not going to exceed the maximum power capacity of the driver doing that at home SPL levels?

Without regard for any compromise of sonic quality which may occur as a consequence of doing that?

[Robh3606]

Place Holder

Do you know what a negative number is??

Rob

[Mr. Widget]

You agree that "pushing" 2435 at 20 kHz by 20 dB over it's midband response falls within the design performance envelope of that driver?

I wasn't being so specific, I meant that in general this is how JBL uses these drivers... as for how much is too much, in my opinion for critical listening applications it is all too much if a tweeter isn't there to help them out. This is not a scientifically derived opinion, simply one that I have concluded after listening to JBL's own systems as well as several of your experiments and also my own with both JBL and TAD drivers.


Widget

[tomee]

This started in another thread. Moved the posts over so we didn't hijack the thread.

Rob

Anyway to link to the original thread? I'd like to see where this all started.

[Zilch]

Anyway to link to the original thread? I'd like to see where this all started.

http://audioheritage.org/vbulletin/s...ad.php?t=19399

Especially:



Versus:

[Robh3606]

Good

Glad you posted both drives.

The difference between the 2 drives at 20K is 3db.

Uhmm, the Timbers filter hits them with +12 dB of PASSIVE EQ at 16 kHz, as well, relative to the midband. At 20 kHz, he's at +14 dB; you guys are kicking it an ADDITIONAL +6 dB to +20 dB total there.

How are you getting +12 db for Gregs Voltage Drive?? It's about -8db down. That means with 1 watt you have under 250mw at the driver.

With the active drive used in the Vertec system you have +12 @ 16KHZ with a fairly wide Q so it effects frequencies on both sides and actually does a good bit of the compensation for the mass roll off.

With 1 watt in with +12 db we have about 15 watts in so each driver see's about 5 watts.

Big difference between the two.

Rob

[edgewound]

I have a question.

What does all this have to do with a passive, full-range network?

You can't have "boost" of a particular frequency band within a passive network....excluding the use of an equalizer in a full-range, non-biamped system.

Is the term "boost" here relating to the relative amplitude of the high frequencies after attenuating the mid-frequencies, therefore giving the illusion of "boosted" high frequencies.

Biamping is a whole 'nother discussion of various voltage drives...but this discussion is a cacophony of semantics.


You can't get something from nothing.