What technical issue ruins sibilants?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
diyAudio Moderator
Joined 2008
Paid Member
Hi,

First order on a compression driver ?
My topology partially resembles an Earl Geddes crossover using this driver, at least insofar as the nominal first order electrical.

What is the nominal impedance of the driver and what is the value of the cap you are using now?
8 ohms and 5u6F for 3k5Hz.

Manufacturer's Recommended lowest frequency for Horn and Driver would be useful too.
Horn=DIY, driver=1k6Hz.

I merely cross lower than 1kHz.
 
diyAudio Moderator
Joined 2008
Paid Member
Hi,

Your not doing a great job of writing them either, your link to gedlee
is totally missing the point, c/o point over lower than 1kHz you say ?

1st order EQ is one thing, with a "c/o" frequency of 10kHz.
Useless excursion below horn cutoff entirely another.

Could be simply the driver is not EQed correctly.

rgds, sreten.
 
Last edited:
diyAudio Moderator
Joined 2008
Paid Member
My post (a few back) states that my electrical filter centres at 3.5kHz.

1kHz is the point that the woofer meets the tweeter. I am doing largely the same thing as Geddes. My waveguide is not identical to his although it is broadly similar.

I think the post I linked to does apply with these clarifications, but there is an even more appropriate post which I cannot seem to find right now as I am having trouble using the search system at the moment.

Power consumption is miniscule, power compression is not obvious regardless of level, and as I intended to imply in my earlier posts, some material sounds correct even at very high levels. This system simply plays louder and cleaner than any I have built before.
 
Much if not most of it is probably found in the original recording (just listen with good headphones to hear), but in my experience the two speaker problems that most accentuate or add sibilance are cone breakup on the midrange driver (generally the result of too shallow (and too high) a low-pass) and IM distortion on the tweeter (generally the result of too shallow a high pass). Put simply, it's usually a crossover problem. Using the horn cutoff as part of the acoustic transfer function does not protect the tweeter driver from overexcursion below cutoff, and the resulting IM above.
 
Try using 4u and see what happens. If the sibilance is still objectionable, try 3u. Beware of the hole when you move the c/o higher.

I personally like first orders but they are extremely difficult. Woofers need to have a wide frq response, as flat as possible with no cone breakup. Compression drivers and horns must exhibit the same properties. You also have to content with the acoustic centers of the woofer and the comp driver out of alignment.

It is not simply the crossover point but the overlap that is crucial. When it works, music is very realistic with excellent vocal presence. Otherwise, the sibilance "spits" at you.

Happy New Year
Mike
 
diyAudio Moderator
Joined 2008
Paid Member
in my experience the two speaker problems that most accentuate or add sibilance are cone breakup on the midrange driver (generally the result of too shallow (and too high) a low-pass)

My woofer has breakup at 1k8. My woofer crossover in its current incarnation consists of impedance rise compensation, a low-pass inductor and a parallel notch filter for the cone breakup. The peak is down 20dB from the final midrange level of the system and the slope is 12dB/octave acoustic.

As it measures, I doubt some -20dB signal would be as intrusive. I wonder whether cone breakup can be excited by other things. For example, some other frequency rattling the cone, or cone position causing the normal breakup to behave differently etc. The first one could probably not be fixed electrically.

Try using 4u and see what happens. If the sibilance is still objectionable, try 3u. Beware of the hole when you move the c/o higher.
I went there over the last few weeks and all I found was the hole ;)


@Charles Darwin, I could have been less direct (apologies), I am curious about the difficulties in making recordings.

In fact, I quite expected someone to mention microphones as I often hear what I guess sounds like a microphone resonance, especially on vocals, which might be what you are saying.
 
My woofer has breakup at 1k8. My woofer crossover in its current incarnation consists of impedance rise compensation, a low-pass inductor and a parallel notch filter for the cone breakup.
A notch filter does nothing for cone misbehavior above the notch (of which there will be plenty), and first-order electric isn't sufficient, either, for the horn or the woofer. There are reasons that pro's use higher order crossovers on horn systems . . . they don't spend the extra money for nothing. To find out what your drivers are actually capable of you might look to borrow a real crossover for testing . . . a DCX2496 or something like it (even the new miniDSP box would do) . . . anything that will give you LR4 and delay for acoustic center matching.
 
diyAudio Member
Joined 2007
Perhaps another way to ask the question is to rephrase it and ask
" What technical issues would add sibilance to a speaker when it is NOT part of the original recording?"

My experience has usually been that most first order XOs do this unless the tweeter is both extremely robust and ov very low resonance.
note that this meant a lot of my early speakers had this problem ( due to my inexperience and tendency to frugality )
 
My 2 cents worth is that anything in a system can muck with the sibilance heard.

Amplifiers for sure.
Preamps, yep.
Source (phono gear, digital gear).
Drivers, definitely (speakers that is).

For example the same tweeter (assume a high quality direct radiating dome for example) can sound like death on a given solid state amp, and silky wonderful on a toobed triode amp... and I'm not talking about a toobe amp with a rolled off top or rounded off leading edge (but usually one without global FB though...).

So, what is it doing it?

Yeah, what...

It's not cut and dry or simple...

_-_-bear
 
diyAudio Moderator
Joined 2008
Paid Member
@dewardh, I think cone breakup is worth some attention. I don't have enough answers to have a worthy discussion on the matter so I'm going to read up.

@Moondog, I have tried to run small tweeters on first order crossovers (quite a bit). I know that glaring sound, combined with rising distortion and compression affects. Frugality (or creative admiration of classic equipment :D) is one thing but even expensive modern domes are not immune from an inadequate crossover.

@Bear. I'm a big admirer of vacuum tubes, especially raw triodes. I am not running any with this system and (as a new experience) haven't felt them as necessary with my current system as I normally would. On one hand I suspect (and always have) that valve distortion covers up other issues (I still like them despite this, for other reasons as well). Therefore I think my current tweeter must be quite low on the distortion.

Michael Chua said:
It is not simply the crossover point but the overlap that is crucial.
I suppose you're talking about power response? This is a point I would like to do more work on although I suspect the crossover region matches reasonably. For what it's worth here is a smoothed in room (listening position) plot taken this week.
 

Attachments

  • sib.gif
    sib.gif
    15 KB · Views: 207
Last edited:
Except for one thing... triodes are more linear than almost any other device... :D

And, the issue appears not to be "distortion" but the spectra of distortion and not the absolute level of distortion but the spectra of said distortion...

I'm not advocating tubes, I am saying that a tweeter that sounds sibilant with amp "A" may sound perfectly wonderful with amp "B" without there being any measurable difference in absolute distortion levels, and/or with the distortion levels being (seemingly) way to low to be audible... that's my point.

I was just trying to provide an easy and straightforward example to illustrate.

_-_-bear
 
diyAudio Moderator
Joined 2008
Paid Member
Well, I think the only way I'll put that one to rest in my mind is to go ahead and build a third order version of the crossover. What have I got to lose?

Bear, rest assured that one day I'll have triodes running again unless I get hit by a bus on my way out this morning.
 
I suppose you're talking about power response? This is a point I would like to do more work on although I suspect the crossover region matches reasonably.

A frequency sweep gives a general idea of a speaker's response.

When I work on a crossover, I pay particular attention to the way the drivers are acoustically summed. If it is not correct, the drivers can cancel each other before or after the c/o frq. Or it can peak. Very different from electrical summing.

In the plot below, the woofer and tweeter sum nicely. Woofer is 2nd order. Tweeter is third order.

Mike
 

Attachments

  • SUM.jpg
    SUM.jpg
    106.8 KB · Views: 150
Please do not misinterpret my post - again I am NOT advocating the use of tubes - only using them as a point of reference.

Fwiw, in a recent "shoot out" at a highly regarded listening room (mine :p ) the version of the PASS F5 that we had there was substantially better than several triode toobe amps, including one with vaunted WE300Bs installed. Surprised the heck out of me btw... but I am happy. :D

The acoustic response is what counts in the xover region, not the electrical slope. I prefer 4th order filters over 3rd order based upon my personal experiences... ymmv. (...as Micheal Chua mentioned - the acoustic response...)

None of this has much to do with sibilance, imo.

_-_- bear
 
diyAudio Moderator
Joined 2008
Paid Member
I've been over on the waveguide thread learning to do power response measurements. As a result I've dropped 2dB above 4k, and an extra dB near 4k. A portion of the spectrum has now "keyed in". Presence has increased but sibilance has reduced.

Once used to the new sound, recordings with sibilance problems can bring back a sound that doesn't fit, i.e. making it look like the fault of the recording.

Now the response is falling on the listening axis despite sounding better due to the power response, so I may change my listening axis later.
Hi, for subtle issues your talking frequency and power response, rgds, sreten.


@bear, ever get that impression that something is just so much better but you can't listen to it? That said, I think that it's the combination that counts. Some amps seem to leave a speaker exposed, if you catch the drift, and some speakers leave an amp revealed.

Still, a bad amp may be boring or irritating, but a bad speaker can be nasty.
 
AllenB, been down this road many times - it is in a sense a big part of the "holy grail" of audio reproduction, that being "clean clear highs" without any sense of electronics, mechanics or speakers. Hard to do and it is to a great extent (imo) a function of system integration.

The problem that most seem to run into is what I call "complementary colorations" - that is a situation where one aspect of a system is used to compensate for another, the idea being that the resulting sum is neutral. The problem is that it simply will not work IF any of the compensations (and there are multiple) are of an absolute value greater than some threshold (and there is no metric for this, it's a hypothetical). Once you exceed the threshold in any one "vector" (or more) it becomes impossible to find an inverse "vector" or series of "vectors" to compensate back to a neutral center point. This is where one finds the artifacts appearing that sometimes are there, and sometimes are not, imho - again using this hypothetical construct to explain it.

So, at least for me the goal has always been to create accurate reproduction/systems and keep the deviation from that magic "center point" to a minimum (there is ALWAYS some) so that the overall presentation of the system can clearly differentiate things like sources, source material, amps, etc... at the same time one has to choose wisely amongst the known compromises and tradeoffs picking those that will not (in effect) produce annoyances that are clearly audible.

Now, all that mumbo-jumbo above works out to two things: engineering choices and "artistic" choices based on experience. It is a search for the "holy grail" because there are no "pat answers" and no perfect devices of any sort...

Sibilance seems to be a combination of the HF driver and amp interactions + cabling as well as what is actually coming through the supporting components. IF one has a very low distortion amp and speaker, one can still have nasty sibilance (imho, just use a Sony CD player!). Otoh one might have a much higher distortion HF section and amp and still not hear sibilance, depending on the spectra of distortions produced. But the ultimate quality of a system is going to be not just the lack of sibilance but the lack of sibilance (added in - it might be in the source) AND then all the other aspects being as good as possible. Obviously it is likely better to have a system without sibilance in terms of just listening, but it's hard to get a lack of sibilance without either losing some information or having a system that is very very good. Again, the holy grail...

These are only my experiences and opinions. Ymmv.

_-_-bear
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.