I'm not ready to claim it is a diffraction effect. I'll find out more from people who know more on the topic and report back. As a diffraction effect it would change as you move off-axis. That isn't the case for my measured samples. It might have more to do with the transition from piston type behavior over to the break-up mode of the driver. It may be more pronounced by drivers with larger surrounds but tjat may be due to the boundry effect as you move to different materials (surround has different properties than the cone) rather than the size/diffraction caused by the surround.
I think what we need is good modeling software for softpart design. We have good software for other system modeling. Seems like a good opportunity for a budding graduate student.
I think what we need is good modeling software for softpart design. We have good software for other system modeling. Seems like a good opportunity for a budding graduate student.
Turns out that this effect is usually due to an edge resonance from the surround. If you have a large surround its more likely to be a problem then on a small surround, hence you see it more in the high excursion drivers vs. the smaller displacement ones that may use a smaller roll surround. Also... stiffer surrounds suffer less, with their resonance being pushed up relative to the softer ones.
It can also be due to the outer edge of a cone being out of phase with the inner. That effect will change as you move off-axis though so you need to check the measurements to see which effect your dealing with.
It can also be due to the outer edge of a cone being out of phase with the inner. That effect will change as you move off-axis though so you need to check the measurements to see which effect your dealing with.
I've used both the RS28A and the H1212 crossed at 1440 Hz 4th order electrical. The RS28A has a very smooth, liquid sound but lacks a little "air" at the top. H1212 got lots of "air" though but it has imho a lot of the metal-tizzle wich I can't stand. Both of the tweeters are fine with the low crossover so that is not a problem.
I would go for the Peerless. I haven't heard it but it's supposed to be a sturdy tweeter with a nice top end.
I would go for the Peerless. I haven't heard it but it's supposed to be a sturdy tweeter with a nice top end.
Thanks, Jonasz. Hmm... Actually I was leaning towards the RS28A but definitely reconsider the Peerless. Yeah, I also heard other people say that ths Seas is metallic sounding. But I wonder if this is really what people hear. According to Mark K's test, the Seas has near non-existent higher order (including beyond 5th order) harmonics. Perhaps, what people hear is TOO low distortion vs some touch of distortions... Just a guess.
I've wondered the same thing. Particularily when a tweeter has higher F2 and low higher order versus the opposite, which is a tradeoff Scanspeak seems to make. Or some like the Peerless which is extremely low in the first <6 orders but just little noise all through the higher spectrum. I wonder how these different combinations of distortion really sound. There is a great little program I downloaded from stereophile that adds harmonic distortion to any .wav you have and you can listen to the differences.
Here is the link to the article: http://stereophile.com/reference/406howard/
There is link in the article to a program called "adddistortion"
There is link in the article to a program called "adddistortion"
Hi,
It seems a fundamental principle of the design is a c/o point at 1.5khz
that is also used to "fill" a mild dip in the bass/ mid drivers axial response.
Bass mids of this type generally do do something at the transition
point from pistonic cone motion to cone flexure mode of motion.
Now at 1.5khz wavelength is ~ 20cm and you have to ask is the
above dip "real", i.e. is it consistently there for off-axial output.
IMO the on axis design of a speaker should not be "flat as possible".
IMO what you want is the flattest "set" of responses, e.g. if you
have off axis flaring of lower treble at the c/o point the the axial
response should have a mild dip at that point. What matters most
IMO is the room averaged response, if this can be achieved with
a smooth axial response all the better.
What you do not want is high off axis variability, this would
render a "textbook" axial response relatively academic.
So from a simulation point of view you need a set of measured
off axis responses and baffle simulators etc. that are capable
of accurate off axis predictions.
As an extreme example consider :
http://www.partsexpress.com/pe/showdetl.cfm?&Partnumber=264-550
and http://www.partsexpress.com/pdf/264-550.pdf
The "dip" at 8KHz for 30 degrees off axis is a phase issue and the
dip lowers in frequency the more extreme the off axial output.
/sreten.
It seems a fundamental principle of the design is a c/o point at 1.5khz
that is also used to "fill" a mild dip in the bass/ mid drivers axial response.
Bass mids of this type generally do do something at the transition
point from pistonic cone motion to cone flexure mode of motion.
Now at 1.5khz wavelength is ~ 20cm and you have to ask is the
above dip "real", i.e. is it consistently there for off-axial output.
IMO the on axis design of a speaker should not be "flat as possible".
IMO what you want is the flattest "set" of responses, e.g. if you
have off axis flaring of lower treble at the c/o point the the axial
response should have a mild dip at that point. What matters most
IMO is the room averaged response, if this can be achieved with
a smooth axial response all the better.
What you do not want is high off axis variability, this would
render a "textbook" axial response relatively academic.
So from a simulation point of view you need a set of measured
off axis responses and baffle simulators etc. that are capable
of accurate off axis predictions.
As an extreme example consider :
http://www.partsexpress.com/pe/showdetl.cfm?&Partnumber=264-550
and http://www.partsexpress.com/pdf/264-550.pdf
The "dip" at 8KHz for 30 degrees off axis is a phase issue and the
dip lowers in frequency the more extreme the off axial output.
/sreten.This new 2" dome looks good
xo at 1500hz should be perfect
http://www.tb-speaker.com/detail/1208_03/50-1426sa.htm
xo at 1500hz should be perfect
http://www.tb-speaker.com/detail/1208_03/50-1426sa.htm
sreten said:
I agree 100%! Rather than design with a single axis measurement you have to look at the range.
In terms of the subjective classification of a driver, I don't typically do it. The crossover plays such a fundamental role that its virtually impossible to characterize the driver alone when using subjective impressions. You have to also carefully view the distortion plots. For one, they are difficult to consistently measure outside of an anechoic chamber and second, because some of the effects can easily be controlled by the crossover.
sreten said:Hi,
It seems a fundamental principle of the design is a c/o point at 1.5khz
that is also used to "fill" a mild dip in the bass/ mid drivers axial response.
Bass mids of this type generally do do something at the transition
point from pistonic cone motion to cone flexure mode of motion.
Now at 1.5khz wavelength is ~ 20cm and you have to ask is the
above dip "real", i.e. is it consistently there for off-axial output.
IMO the on axis design of a speaker should not be "flat as possible".
IMO what you want is the flattest "set" of responses, e.g. if you
have off axis flaring of lower treble at the c/o point the the axial
response should have a mild dip at that point. What matters most
IMO is the room averaged response, if this can be achieved with
a smooth axial response all the better.
What you do not want is high off axis variability, this would
render a "textbook" axial response relatively academic.
So from a simulation point of view you need a set of measured
off axis responses and baffle simulators etc. that are capable
of accurate off axis predictions.
As an extreme example consider :
http://www.partsexpress.com/pe/showdetl.cfm?&Partnumber=264-550
and http://www.partsexpress.com/pdf/264-550.pdf
The "dip" at 8KHz for 30 degrees off axis is a phase issue and the
dip lowers in frequency the more extreme the off axial output.
Good advice from you as always. I totally agree that we have to consider the off-axis family of responses in designing an XO. But my problem is that I have to rely on on-axis measurement for this design---no measurement setup yet. This is one of the reasons why I chose a low XO point. According to PE measurement the Usher midwoofer's 30 deg off-axis response is virtually identical to on-axis FR up to 2 kHz. And also it seems that there's indeed a phase issue regarding the dip at 1100 - 1200 Hz since it becomes less deep off axis. But this phase interaction does not override the dip's presence; it does not become a "bump" but still a dip. That's why I didn't go too wild with the correction although I could if I wanted.
Does this make sense?
Here's PE measurement of the Usher:
http://www.partsexpress.com/pdf/296-602g.pdf
Before you do anything else... spend a dime on any component, go buy a measurement set-up. All the published data is suspect at best. You need to be able to measure what you have on your own baffle. The differences as you move a driver baffle to baffle are substantial. You won't do good design until you can measure the drivers in their intended application/acoustic environment. Studying and understanding the graphs of bad data is almost a waste of time. Its good for learning I guess but that data isn't going to necessarily translate to your finished loudspeaker.
Generally you're right. But I think my simulation is not that sloppy. I did some homework about this. First, I found that Zaph's infinite baffle measurement I used is very accurate, and consistent from a driver to another, compared to other hobbists' or manufacturers' data. As people agree, he is indeed a perfectionist! So, these data are very appropriate for modeling purpose. Second, I simulated numerous existing designs using his data along with diffraction simulator and was able to predict their measured responses, including phase tracking, with great accuracy. But this level of accuracy was not gained with ease. By trials and errors, I had to learn many skills and tricks, which work consistently, to predict measured responses.
Generally I agree that in-box measurement should be used to design an XO and don't want to recommend a complete newb to use my approach because there're many possibilities of making mistakes somewhere in the process.
BTW, here's another good success story about using this kind of approach:
www.eldamar.net/audio/RS150MTM
First, I found that Zaph's infinite baffle measurement I used is very accurate, and consistent from a driver to another, compared to other hobbists' or manufacturers' data. As people agree, he is indeed a perfectionist! So, these data are very appropriate for modeling purpose. Second, I simulated numerous existing designs using his data along with diffraction simulator and was able to predict their measured responses, including phase tracking, with great accuracy. But this level of accuracy was not gained with ease. By trials and errors, I had to learn many skills and tricks, which work consistently, to predict measured responses.Generally I agree that in-box measurement should be used to design an XO and don't want to recommend a complete newb to use my approach because there're many possibilities of making mistakes somewhere in the process.
BTW, here's another good success story about using this kind of approach:
www.eldamar.net/audio/RS150MTM
It just seems easier to me to just measure it. The back calculations are highly dependent upon one axis of measurement. For example, the 6.5K dip in the Peerless tweeter is only there on-axis. Its an on-axis cancellation that only appears in that one measurement, hence has little or no effect on the subjective sound quality. If you don't have a measurement set-up you will never know which glitches in the response are worth going after and which are not. Your also not able to check your work, the quality and consistency of the transducers and back-check your simulations.
Imagine being a carpenter without any measurement tools, rather you depended upon the measurements given by your vendors. Think you could craft anything other than basic projects? You couldn't, you need to measure during every step of construction of a project and the same applies to designing loudspeakers.
Imagine being a carpenter without any measurement tools, rather you depended upon the measurements given by your vendors. Think you could craft anything other than basic projects? You couldn't, you need to measure during every step of construction of a project and the same applies to designing loudspeakers.
Then you understand the limitations and strengths of models. Let me tell you.... the math wiz kids still have not managed to be able to model everything you see at the output of the loudspeaker. Its a complex system and there are a lot of generalizations used.
The ole yardstick is still a pretty useful tool.
Where do you live in Indiana? I'm from the region..... for better or worse.
The ole yardstick is still a pretty useful tool.
Where do you live in Indiana? I'm from the region..... for better or worse.
I know that. Modeling is always based on certain regularity assumptions that may not be satisfied in reality. But predictions in physical science do a far better job, in general, than in my area (cognitive science and mathematical psychology). I'm convinced that my simulations of various existing designs show this. Also, using sims for a simple 2-way with a low XO point is relatively robust to idiosyncratic off-axis behavior and don't go far from measured responses. And John Krutke's infinite baffle measurent is really nice! Using FRDC modeling tools with his data convinced me. Of course, a well calibrated measurement system will be a sure bet.
I live at Bloomington, BTW.
I live at Bloomington, BTW.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.