The new DSP software package from Bodzio Software, separated from the SoundEasy package where it originated. It's now stand-alone, improved and includes a full measurement suite. It was used by John to create the response shown above in his test demonstration.
Dave
This discussion reminds me of why the US is so far behind in speaker building than the rest of the world these days.
Well Joachim, you may have Acapella and MBL. But we in America at least have Lee Ritenour!
YouTube - lee ritenour - rio funk
I'm going down to DC to see him in April - can't wait!
Well Joachim, you may have Acapella and MBL. But we in America at least have Lee Ritenour!
YouTube - lee ritenour - rio funk
I'm going down to DC to see him in April - can't wait!
Maybe yes, maybe no. That's case-by-case. In any case, that implies that someone has suggested actually using a driver into the breakup. It's a straw-man argument as no one has. The breakup discussion related to linear distortion and whether or not it can be EQ'd on an axis. It's well understood that the off-axis will change, nothing new in that. For that matter, the same issue arises with a driver that becomes directional below breakup, but none of that was part of the initial discussion.
To make the point, here's what I said in this post early on after posting some measurements of my dipoles using the UE:
All of that is simply part of normal design issues, it's no different than what Linkwitz had to do with the Orion. He uses a driver with severe breakup.
The measurements that John posted should make it clear. The breakup, rather extreme, was essentially complete EQ'ed to the target. Were the breakup in-and-of itself non-linear, EQ to a target, any target, would not be possible.
Dave
Last edited:
Beg away Earl. I heard the Acapellas, sat and talked with the chief engineer from Lancshe while listening to his speakers, heard some beautiful Vivid Giyas and Dynaudios this year at RMAF. While the Revels and Vandersteens were very nice - the models mentioned above were simply in a class of their own. And where were you?
Last edited:
It is posible to build a driver that works in the breakup mode and that is quite linear. They are called bending wave transducers. The OX tweeter for exampe i use breaks up at around 8kHz but the response is still well controled without rising distortion. When it come to the midrange in my design i try to stay away from the fundamental resonance and not use a too high crossover frequency. That has more to do with the radiation pattern i am after. The Scan wideband is a bending wave transducer too so i could afford a higher crossover frequency based only on low distortion.
Each case is unique and requires testing to determine the source and nature of any problems experienced. Electrostatics are another example. The modulation of their surfaces can be quite chaotic and precise measurements are required to determine first if a problem needs to be addressed and subsequently how best to go about it if the need occurs. It appears that Johnk has backed off from what appeared at first to be an assertion that the bulk of all distortions could be fixed with EQ. If he accepts that there are limitations to UE's usefulness and that testing is a good idea no matter what - then there is clearly no disagreement here and we should stop wasting bandwidth on that subject - derailing your thread even further.
I really have no desire to discuss common sense issues of speaker design and what miss-use of any design tool will cause. Deward, you of all people know my position directivity and limitations of the useful bandwidth of any given driver. I don't spend my time making polar plots like this:
An externally hosted image should be here but it was not working when we last tested it.
just because they are pretty.
If you want to preach about using driver into and past breakup please address the full range driver crowd.
I'm going to make a couple of comments. First, and Earl can back me up on this because we have discussed in in may threads over the years, EQ is only useful if the response of a driver(or entire speaker) is uniform over the intended listening window, or at most, if the response decreases monotonically from the on axis response over that window. In other words, no serious off axis peaks or nulls. That being the case, beneficial equalization of the on axis response will be equally beneficial off axis because a dip or peak in the on axis response corresponds to a similar dip of peak off axis.
Second, with regards to the examples I have presented (dlr reposted some of he data here). They were intended to show that the observed driver behavior was linear, nothing more. And in fact they do that very convincingly. In both cases the response at the observation point was shown to be highly irregular. Based on that response a linear correction was derived with the intent correcting the original response to a pre-specified target. In both cases this was successful. In one case as much as 10dB or more equalization was applied to correct the initial response to match the target. Now, understand that it was not known before hand whether the observed response was that of a dominantly linear or nonlinear system. But a correction was developed on the assumption of linearity. That the observed result matched the targeted response demonstrates that the system is indeed linear. If it were highly nonlinear the linear correction would not suffice to bring the observed response into agreement with the specified target.
The entire filed of classical acoustics is based on the assumption of linearity. Our measurement system (FFT based) are all based on the assumption of linearity. Speaker design is based on the assumption of linearity. Be build speaker and amplifiers with the goal of linearity. And then we go back and run distortion tests and typically find that in all but extreme conditions we find that 1 part is 1oo or 1 part in 1000 or less is attributable to nonlinearity in the system.
Second, with regards to the examples I have presented (dlr reposted some of he data here). They were intended to show that the observed driver behavior was linear, nothing more. And in fact they do that very convincingly. In both cases the response at the observation point was shown to be highly irregular. Based on that response a linear correction was derived with the intent correcting the original response to a pre-specified target. In both cases this was successful. In one case as much as 10dB or more equalization was applied to correct the initial response to match the target. Now, understand that it was not known before hand whether the observed response was that of a dominantly linear or nonlinear system. But a correction was developed on the assumption of linearity. That the observed result matched the targeted response demonstrates that the system is indeed linear. If it were highly nonlinear the linear correction would not suffice to bring the observed response into agreement with the specified target.
The entire filed of classical acoustics is based on the assumption of linearity. Our measurement system (FFT based) are all based on the assumption of linearity. Speaker design is based on the assumption of linearity. Be build speaker and amplifiers with the goal of linearity. And then we go back and run distortion tests and typically find that in all but extreme conditions we find that 1 part is 1oo or 1 part in 1000 or less is attributable to nonlinearity in the system.
I don't agree with what you have written here. I never said the the UE could do anything but correct linear distortion and at that, only relative to the obseration point. But let's just accept that we agree in the final analysis that the end product needs to be tested to assure it operates as intended. I do agree that enough bandwidth has been wasted arguing over what is or is not linear.
Yes, John, but you then post, and continue to post charts and statements that imply or outright assert that UE can "fix" cone breakup (or the obvious response anomalies that it causes), whether in-band or out. As here:
No, what you have "convincingly demonstrated" is that you can fool one test at one specific location, not that it's "linear" or that you can "fix" the problem. Maybe it's fixed (in that case), maybe it's not. It depends on the physical mechanism that caused the response irregularity to begin with. It's like "equalizing" diffraction . . . flatten one place, make it worse in another. And if you flatten on-axis, while what we actually hear is primarily off-axis (room reflections) . . . well . . . there we go again. Those "whee . . . look what it can do" flat response curves are . . . misleading . . .
Misleading only if one does not understand what they represent. That is no different than anything else one can do with any other method of driver control. The UE is a tool, just as is any crossover, whether based on passive components or DSP. If the user doesn't understand the multitude of issues with a speaker and isn't competent enough to use a tool, whatever it is, to its full potential and within whatever limitations it may have, it's not an issue of the tool, it's an issue of the user.
Dave
This is the third time I posted this obvious contradiction. Any excuses DIY as to why you keep deleting it?
Nothing changing, there's not even anything contradictory. You do precisely as described. Now if other factors were ignored or the user is unaware of those issues, the tool will produce results that are not what that person expected. But that's still no different than measuring drivers with a standard package, importing the measurements into software such as the PCD, optimizing on a single axis in ignorance of what occurs off-axis, then assuming that the results will be as desired. No difference whatsoever, yet for some reason you seem unwilling to recognize or acknowledge this. A tool is a tool, you use how it's intended and take into account all that you should. If you don't, you're at fault, not the tool.
I stand by every word in those quotes. If it's necessary to explain in detail for you to understand what is meant, just ask. I would expect better from you at this point, but maybe my expectations are just too high.
Dave
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.