Using the crossovers in your sub AS WELL AS in your receiver? (1 Viewer)

[GregBe]

I know, I know every post I read says to push the subs crossover all the way up and use the crossover on the receiver, but I read a review by Brian Florian of the Paradigm Cinema Series over on Secrets. Although I don't have these speakers, I have similar smaller bookshelf speaks/larger sats that would be benefited by a 100 Hz crossover.

Here is the part of the review I am interested in feedback on.

"The only "issue" is that the system needs a relatively high crossover between satellite and subwoofer. 100 Hz would be good. I certainly would not want it any lower than the 90 Hz crossover in the Yamaha receiver we used for most of the listening tests.

A high crossover in and of itself is not a "bad" thing, but it does bring two stipulations: A) The subwoofer really must go in the front of the room (which is generally preferred anyway), and B) Extra special care must be taken when adjusting the splice between sub and satellite. It was unusually easy to get subjectively bad performance from the system, resulting in a very fat, bloated sound. Run the sub too lean, and the cinema set will have a hollow, empty character. Run the sub hot and you'll have irritating one-note bass.

With 99.9% of surround sound processors these days providing their own high-pass / low-pass combination, I recommend starting with the PDR-10's own low pass dialed to the max and calibrate its level with AVIA. With the sub in the a-typical room corner, the bass will likely sound fat if the crossover is high (which it should be for these speakers). LEAVE the level there! By ear, bring the subwoofer's low pass down until the sound smoothes out. If you have to go much below the SSPs low pass, recalibrate the level and continue to adjust the low pass. Yes, the two filters (one in your SSP, one on the Sub) will interact, bringing phase issues to the table, but it is better to adjust this way than ride a subwoofer level knob all day."

Similar to what the review says, I get a bit of boom in the sub at 100Hz vs. 80 Hz. I tried what Brian suggested, and sure enough when I set the crossover on the receiver at 100Hz and start to inch down the crossover on the sub past 100Hz things really tighten up. 90 Hz sounds the best.

Could it be possible that this is the best setup for my system? I know that crossovers aren't a brick wall, so the hole I am creating may not be an absolute hole. Do you think I am better off just choosing between 80 and 100 on the receiver and not messing with the crossover on the sub?

Thanks
Greg

 

[John Garcia]

You're not creating a "hole", you are rolling off the sub more steeply with the two crossovers cascading. It's sort of a cheapo EQ that happens to work for you because one of your problem points appears to be at or near the x-over point. If it sounds better, leave it.

 

[ricky fong]

GreBe,

WOuld you mind to post the review that you have referred to ?? Thanks!!

 

[Edward J M]

Bingo.

If you have the capability to accurately analyze the in-room FR and the effects of cascading the two filters, the low pass on the subwoofer can be a useful tool.

With that said, most people leave it disabled.

 

[BruceD]

Mostly this phenomena is due to the fact that most of the HT processors use a "THX type" 12dB/octave high-pass filter and a "THX type" 24dB/octave low-pass filter.

As you can see, before you even get started they are mismatched. They assume you have speakers with a -3dB low frequency rolloff (12dB/octave) at the processor's selected high-pass crossover (for example 100Hz or even the standard 80Hz).

Now the question is how many speakers meet the necessary low-frequency rolloff criteria (12dB/octave @ 100Hz or even 80Hz) for these processor crossover schemes? As a guess, I'd say only 3% or less.

I've been harping on this problem for a long time and just wish the the manufacturers would offer a symetrical 24dB/octave high-pass/low-pass Linkwitz-Riley crossover as an option. Yes, this is more useful for those of us with floor standing mains that go low.

What Brian described may or may not work for others, as John and Edward indicated. I went an entirely different route, implementing a true 24dB/octave high-pass/low-pass Linkwitz-Riley external crossover between my mains and sub(s), re-directing all bass (LFE and all small speaker bass) to my mains.

I happen to use 60Hz as my crossover frequency, because my main towers have a -3dB low frequency rolloff of 31Hz. Oh yes, I leave the sub's crossover disabled.

 

[Edward J M]

So you set...

1) Mains to large.
2) All other speaks to small.
3) Sub to off/no.

Then route the mains out to the LR xo?

 

[GregBe]

Thanks for the input guys. Nothing is ever black and white in this hobby is it? I am constantly amazed how many shades of grey home theater has to it.

Ed,
Short of plotting the frequency response, could I use something like the frequency sweeps on DVE or Avia with a SPL meter and compare how smooth the sweeps appear, and go with the one that appears to have the tightest response. I would prefer to plot out the response, but I don't have the ability to burn test tones to a disc on my current computer setup. I do have both DVE and AVIA and a SPL meter.

Ricky,
Here is the review from Brian that I saw the info in.
Paradigm review on Secrets

I want to make it really clear that this method is not ideal in most situations. I hesitated even to post it as to not confuse people just starting out. For most people, you don't want to use two crossovers. I, on the other hand, can never leave well enough alone and must try every tweak that comes across my desk. I am sick but I love it.

Greg

 

[Brian L]

Great thread, Greg.

What I would also wonder about is the actual performance of the sub crossover vs. the labeling on the knob. Having read Tom Nusaine's reviews for forever, it is not uncommon that the actual crossovers performance vary a great deal from the numbers on the dial.

I guess thats why Brian says to tweak by ear, and then just record the number.

BGL

 

[Edward J M]

Great point, and I have also found a similar issue with the filter slope; it doesn't always act exactly the way you might expect.

Regards,

Ed

 

[Kevin C Brown]

I continue to disagree with the statement that the common 24 dB/octave LPF and the 12 dB/octave HPF are mismatched. For small ranges of freq, you can consider it linear and not logarhythmic. Hence for an 80 Hz crossover:

12 dB/octave HPF = 12 dB/40 Hz
24 dB/octave LPF = 24 dB/80 Hz

12/40 = 24/80

That's why they do it that way. I've plotted the freq response in my room, and I get essentially flat response through the crossover. I submit that those that are not having flat freq response through the crossover are experiencing room effects, phase effects, etc.

 

[BruceD]

Kevin,

Simple plotting of crossover parameters will show that the HPF and LPF are mismatched by as much as 3dB through the xover frequency (this is unavoidable).

While this may not be all that noticable to some ears, it is very noticable on an plot (even a mathematical model).

Yes, as Ed says sometimes you get lucky with a speaker's F3 (and more importantly it's slope) and the processor's xo frequency.


1) Yes
2) Yes
3) Sub=No

Yes, processor L&R out ---> LR active xover
then xover HPF --> L&R amps
and xover LPF --> Parametric EQ --> powered sub

 

[Kevin C Brown]

Exactly at the crossover freq, each is -3 dB, which adds up to 0. ??

 

[Edward J M]

Yes.

 

[BruceD]

Ed,

No I don't use an ICBM for that and other reasons, but use instead crossovers from Marchand

I use their XM9-L stereo 4th order L-R xover.

This is quoted from the Marchand site, "Note that both the high-pass response and the low-pass response are down exactly 6 dB at the crossover point of 100 Hz. This means that at this frequency the amplitude is exactly half. Adding the high-pass and low-pass together sum to unity. As a matter of fact the sum of the high-pass and the low-pass response is unity for all frequencies. This is why the filter is called a “constant voltage” network. It is also called a Linkwitz-Riley network."

This is not true for 2nd order xovers, and even less true for a 4th LPF and 2nd HPF combination.

 

[Wes Nance]

OK, I want to jump in here.

I think I mostly understand what is being discussed. My question is whether or not this problem area in the crossover can simply be dealt with a parametric eq on the sub such as a Behringer 1124p.

If I plotted my FR with the mains and sub on, couldn't I dial in the overall FR from there by EQ'ing the sub? I currently run my mains small (they are bookshelves with an f3 around 65hz or so) and use my receiver's stock 80hz crossover. It sounds fine with movies, but I notice I greatly prefer running the bookshelves set to large with the sub's crossover dialed in for music. I think there's too much musical information around 80 hz that gets cut over to the sub, and I can tell.

Does that all make sense? I'm wondering if I should always run my center and mains large, then use the sub's crossover and my 1124p to dial it in.

Wes

 

[JohnDG]

Also note that this may be sub specific.

The PDR-10, with the cross-over "maxed" on the sub, is strong from 45-100Hz to the detriment of the lower ranges. My experience with this particular sub is that by "taming" the upper end by using the sub's cross-over along with the receiver's cross-over (cascading) I was able to get a pretty flat FRC over the cross-over threshold and, additionally, down to 25Hz.

After I later replaced the PDR-10 with an SVS, I now only use the cross-over settings on the receiver with sub's setting "off." This also gives a flat FRC over the cross-over threshold.

In other words, YMMV.

jdg

 

[Richard Greene]

A 24dB/octave filter is nothing more than two 12dB/octave filters cascaded. A 12dB/octave low pass filter in the receiver cascaded with a 12dB/octave low pass filter in the subwoofer is the cheapskate's 24dB/octave low pass filter.


Two common subwoofer problems:
(1) Too much output over 80Hz.
(you can hear male voices when all other speakers are off)

The crossover problems disappear when you buy an external active crossover like my 70Hz. 24dB/octave Marchand XM9
(any turnover frequency and slope that reduces output by at least 24dB at 160Hz. will satisfy most listeners)


(2) Loud frequency response peaks (or troughs) under 80Hz.
caused by standing waves

The frequency response peaks heard at your listening position can be tamed with a parametric EQ like my
Behringer #1124. The nulls you are stuck with -- you can move your seat to get away from the deepest portion of nulls, which can be over -20dB, but will still have at least one or two weak bass frequencies (down at least -6dB) under 80Hz.

If you want to use your main speakers full range,
then sealed enclosures have a 12dB natural bass roll-off and ported enclosures have a 24dB natural bass roll-off
... so a 12dB/octave (or 24dB/octave) low pass filter on the sub could be set to provide the mirror image of the main speaker's bass roll-off. But note that crossover frequency controls on subwoofers are not particularly accurate (SPL measurements would help).

Two way speakers with 6.5" or smaller bass drivers usually sound best when NOT used full-range (use a high-pass filter that's a mirror-image of the subwoofer's low-pass filter.

 

[Brian L]

Ed, BruceD, and others, a couple comments/questions on test tones, and low frequency measurements in general.

To start with, I do use a basic 1/6th octave EQ (non parametric) for the bass (22.5 to 80 Hz 1/6 octave frequency centers), and I prefer the poor-man's method of measurements with discrete test tones and a decent SPL meter.

I have found that, depending on the tones one uses, getting decent stable readings with stuff below 120Hz or so is tough. Even with the meter set to C weighting/Slow response, the meter still bounces up and down by a significant amount (+/- 3db or so).

I would guess that this would be a function of the test tones used, but have observed similar results with the tones on the Sheffield "My Disc" Test Disc, Surround Spectacular, and the tones that come with the Infinity RABOS kit CD.

However, I did find some tones that do give my rock solid readings with the RS Meter, as well as with the Infinity RABOS meter (which is what I normally use for bass measurements).

Those tones are on the Bass Mekanix IV CD. They are discrete tones from 20hz to 99hz, and are at 1/2 hertz increments (the whole frequencies are on the left channel, the 1/2 frequencies are on the right....you need to use one channel or the other).

I can't say if these are pure sin waves, or if the other tones mentioned are warbled, but the Bass Mekanix tones produce the most stable, solid readings I have been able to find.

So, the question I would have is, what type of tones are really prefeferable for rudimentary low frequency measurements? I am guessing that there may be reasons why warbles are preferred over sin waves, even though in my experience, I have a tough time (make it IMPOSSIBLE) getting what I consider accurate, repeatable readings.

Any thoughts?

BGL

 

[BruceD]

Primarily this is due to the room/speaker interaction or what Richard called standing waves. These are simply bass wave reflections from the room's wall surfaces (ceiling and floor also).

Test tones come in all kinds of flavors as I indicated earlier; pink noise, white noise, sine waves, warble tones, etc. Warble tones will probably produce the most stable reading on an SPL meter, because they never stay centered on one specific frequency long enough to allow standing waves to build up. If I'm trying to optimize a xover frequency between 2 speaker drivers, this is a useful test tone.

The AVIA calibration DVD uses two different pink noise tones, the developer stated it like this,
"The lower portion of the tests is bandpass from 35-70 Hz (one octave) and the upper is 500 to 2kHz (two octaves). However, the rms levels of the two portions of the tests are identical so they will read the same level on your meter ."

A 1/6th octave graphic EQ may not provide the best solution for bass EQ, but your ears will tell you if it's good enough for your purposes.

I use the same setup as Richard (Marchand active xover and BFD Parametric EQ) but do my 24dB L-R xover @60Hz instead of 70Hz.

IMHO, to truly measure the room/speaker frequency response, I think you need more than just an amplitude or RTA measurement(SPL) i.e. what you get with sine wave test tones and an SPL meter.

IMHO, I think having the "time" variable helps you identify these room modes. This involves PC measurement programs like ETF

 

[Brian L]

Thanks BruceD.

No arguments on the short comings of 1/6 Octave vs. multiple bands of Parametric, but I do find 1/6 octave centers superior to 1/3 Octave, which I in turn find superior to not EQ'ing the sub at all

I will have to review the various tones that are at my disposal and see if I can ascertain which may be warbled. I can say with certainty though that the Bass Mekanix tones produce absolutely stable readings...none of the others, in my room do so.

While I am pretty sure I do not have a 100% brain wrap on some of the concepts here, what is the down side of using a tone that does allow for a standing wave to occur? Is it not part of the rationale for using EQ to identify these frequencies and attempt to correct them?

Or am I confusing the term "standing wave" with room peaks and nulls that occur when certain frequencies combine in an additive fashion, vs. a null when the same frequencies occur in a fashion that produces a cancellation?

BGL