Originally Posted by go94022
Wondering about DirectServo for vented boxes... With a sealed box, cone movement translates directly into audio output, so it is easy to see how controlling cone movement will linearize response and improve distortion.
With a vented box, cone movement at tuning resonance is greatly reduced. Virtually all the output comes from the port. In this case, it seems DirectServo may not have a big effect on linearity. I can see that servo can still affect T/S parameters, but how well does it reduce THD at resonance? Or maybe distortion from the driver at resonance is not so important, and the linearity of the cabinet/port dominates?
First off, I had a post a few pages back about how I started Directservo as a patent. Before that is certainly years of experiments (just like everyone else before we working on servo, ad hoc or trial and error and only on one type of speakers -- sealed speakers). One day I thought there got to be a way to formulate the problem (after all, that is how we are trained in graduate school), and then I was able to formally formulate the problem of "servo feedback" and best of all, it can all be explained in the context of T/S parameters frame work. I have told some customers that the effect of servo feedback from accelerometer is it creates a large equivalent moving mass. Customers of Velodyne servo subs (I had ULD15 II in 1990s) can push their cone and do the experiment. The problem with creating a large moving mass is once you push it hard enough to start the mass-spring oscillation; it is hard to make it stop unless you have sufficient damping.
Anyway, so I started the process of filing the patent with the whole idea of protecting the intellectual property by the patent book by Pressman. It clearly describes you can file patent to protect your IP, for others that you cannot file patents for, just keep your mouth shut (which is called trade secrets) is the conclusion. So you will see that I don't openly discuss our patents beyond what is disclosed in the patent.
So back to your question, the problem you have described for vented sub is only there when you apply infinite servo loop gain. If we reduce the loop gain just enough (but still significant), it will work for ported subs too. It turns out today’s ported subs, without any boost, all have very weak output at the port tuning frequency point (unless you go with 300L box). And that is the opportunity for the servo to come in play.
It appears that the 15" CI models can support both sealed and vented. What is the difference in servo control between these versions? Based on the above paragraph, is there a difference in level of servo control at or above cutoff frequency for sealed vs. vented? I realize there is not much you can do with vented below cutoff.Does the vented version have a limiter to avoid driver damage below cutoff?
The servo boards are different for vented and sealed subs (and for OB/IB). So the cutomers need to specify which mode of operation he wants. There are simple conversion steps you can do to convert from one version to another.
Zaph Audio (www.zaphaudio.com
) has lot of measurements of hobbyist drivers (but not much for subwoofers). I agree that distortion characteristics explain a lot about how a driver sounds.
I am wondering about the effectiveness of DirectServo on higher order distortion components that are more objectionable. I can imagine servo helps 2/3rd order a lot, but higher order problems may be due to mechanical cone/suspension/motor issues that are not as easily corrected. As an aside, I saw a note that the LV12R has a lower quality driver than the F12G. Perfectly reasonable, of course. Are the audible differences due to these kind of higher order issues that are harder to correct?
Now that is another area that we keep as a trade secret that I will not go in details to explain how servo help reducing higher order distrortion. It is a simple math observation. The method/insight is mentioned in one of AES papers, but it was used in a different area (in distortion component simulation). But let us just say, indeed the distortion reduction is also effective on higher order distortion. If you check the comparison of distortion comparison on our websiste, you can see the more consistent improvement is on higher order harmonics, exactly where we need.
Or maybe DirectServo helps across the board (e.g. lowers distortion of all orders by X db), but if the driver starts off with lower numbers it remains lower after correction. What level of improvement is typical (what is "X")?
I guess the question is whether competitive products with a high quality driver without servo (but perhaps with limited EQ) can be equivalent to a driver with servo.
A high quality driver is always a necesary part in my formula. If we started with a less quality driver (like those off the shelf from my suppliers), the poor quality is sufficient to cast a masking effect on the final sound quality. So my view is very different from anyone else who wants to use servo sub as a gimmick. Chad once talked to me that he wants to license our patent. During the conversion, all he talked about is the "lowest distortion" number technology (completely ignore the part of "how to make good sound" discussion, which to me is more exciting). His intent is very obvious. A lot of people wants superficially low distortion number. Is the low distortion the key to good sound? Absolutely not. That is just one layer of onion. After that we come to the next layer which is time coherence. So what servo improves goes beyond the simple distortion numbers. What we improve furthers is in spider/surround memory effect, and thermal memory effect, and how to maintain very good cone control in the presense of standing waves and reflections.
Engineers are in general very poor statisticians. They view the world as black and white. If they can just for a moment believe that the characteristics they observe at time t can be slightly different from time t' and start thinking how to improve that, or .. if they ever view each individual sub having a slightly different characteristic and try to come up with a way to narrow that difference (to create a consistently good product too), or start to recognize it is difficult to tell the difference between correlation (like hearing the result of good quality sound) and causation (engineers want to find the exact reason leading to good quality sound) in statistical observation. It is just like the great Albert can never embrace quantum physics because he does not believe god will throw a dice on us. But inject a bit of statistical view help engineers.
Bill has posted a video link a few pages back. The presenter talked about the experiments that he has when he has a control to flip a switch and how people are fooled. But if some one look at the scatter plots with x axis be the real physical difference and y axis be the observed sound quality difference, this presenter only talks about the spread of y axis at a single x point. In the real life, A-B comparison is like look from the y axis, when we observe a difference (which mapped to a y value) ,then what is the chance of having a real world difference (the spread on the x-axis)? They are two totally different questions.
Now, everytime members bring up the issue of sound difference compnents, interconnects, and blah blah, it always leads to heated discussion. In the back, I would hav said "cool down guys, the observation you guys have is statistical insignificant to reach any meaningule statistical conclusion".
In the end, all the above is why these days I spent more time on customer service than to do any technical debates.
-Edited by Rythmik - 2/17/13 at 1:17pm