WinPCD - New tab for finding Relative Acoustic Offset

[Jeff B.]

Re: WinPCD - New tab for finding Relative Acoustic Offset

Hi Dave......nice tool. Been playing around with it a little. A bit different than my Lsp cad 5.25. As for minimum phase extraction........Freq. Response Modeler isn't the only avenue, I hope. If you don't have Office products that can be pricey.....is there another program that can perform this function?

best regards, John

I believe that Neil Davis' PSD Lite has a phase extraction tool.

Jeff

[isaeagle4031]

Re: WinPCD - New tab for finding Relative Acoustic Offset

You're welcome. I hope to get time to add more functionality. Lots of ideas, not nearly enough time to do them all. The acoustic offset function was a spur-of-the-moment idea. Took more time than I expected, though, I thought I could rip it out quickly.

dlr

Series crossovers I hope?????

[bohdan]

Re: WinPCD - New tab for finding Relative Acoustic Offset

As for minimum phase extraction........Freq. Response Modeler isn't the only avenue, I hope. If you don't have Office products that can be pricey.....is there another program that can perform this function?

best regards, John

Hi John,

Minimum phase extraction has to be done correctly, otherwise you are just getting into trouble believing, that the result is actually
the true minimum-phase phase response. More information is in http://www.bodziosoftware.com.au/Min...riveration.pdf

SoundEasy and Ultimate Equalizer both have this function and more.

You will find some more information in other files on http://www.bodziosoftware.com.au/


Best Regards,
Bohdan

[hongrn]

Re: WinPCD - New tab for finding Relative Acoustic Offset

The problem is that SoundEasy is not exactly easy to understand. It's geared towards professionals, not DIYers. It has literally dozens of buttons one must set before running the software, but lacks a simple "Default" button. The manual is so complicated that one must buy a third party quick start set of instructions to even get going. I understand your target market, but since you jumped in and commented on this thread, I felt compelled to tell you that if your goal is to reach out the the DIYer community, your approach has to be more simple, practical and concise without of course sacrificing accuracy. In terms of minimum phase extraction, ARTA is simple to master, and does the job beautifully.

Hi John,

Minimum phase extraction has to be done correctly, otherwise you are just getting into trouble believing, that the result is actually
the true minimum-phase phase response. More information is in http://www.bodziosoftware.com.au/Min...riveration.pdf

SoundEasy and Ultimate Equalizer both have this function and more.

You will find some more information in other files on http://www.bodziosoftware.com.au/


Best Regards,
Bohdan

[dcibel]

Re: WinPCD - New tab for finding Relative Acoustic Offset

Easy is what SoundEasy isn't, however it is hugely powerful software and represents great "bang for the buck". Regarding this topic, SE will, among other things, extract minimum phase. I understand it's not for everyone, its awkward user interface simply provides the tools without any Wizards etc to walk you through design processes. There is a great learning curve, and I must say that reading through the SE manual I am learning more than just how to use the software. It's almost like a textbook in acoustic design, aided by SE. ;)

ARTA is also hugely powerful software, however only provides the features provided by EasyLab in SE. You still need something else for box modelling, diffraction sims, crossover design. I can't speak for how much easier ARTA is to use, but it does appear to have a very detailed 180pg manual. At first glance it looks equally as complicated.

[charlielaub]

Re: WinPCD - New tab for finding Relative Acoustic Offset

You find the offset between a pair of drivers at a time, not three. For three drivers you have to find offset twice. For example, once between tweeter and midrange and again between tweeter and woofer.

Although this may be true for your software, the offsets for all drivers in a loudspeaker can be simultaneously determined provided there is sufficient data present and an algorithm that can pull off the task.

Case in point - the new version of ACD-LADSPA that I hope to be releasing in the next few weeks. ACD can model a loudspeaker up to 4-way. This is just how many drivers I happened to include since 5- or more way is pretty rare, but there is no hard limit to the approach per se. As is customary, you take one frequency response measurement on each driver. I do this with the mic in the exact same spot for all. Then, without moving the mic, you record a single measurement with all drivers operating "in parallel" meaning that they are all given the same input, without any filters or crossover network connected. All measurement data is saved with minimum phase. The data sets are then loaded into ACD and an optimization engine is used to fit all the offsets AND the driver polarities simultaneously and automatically, with the click of a button. Compared to the old manual method this is a great step forward. The only pre-requisite(s) to my approach is that one driver must be selected as the "reference" for the offsets (they are differential) and one driver's phase must be chosen as the reference phase. These two references are typically provided by the same driver - the tweeter, but it's totally arbitrary.

As long as the frequency range for each driver overlaps (and it should if they will be used together in a loudspeaker!) there will be some interaction recorded in the multi-driver data file and the optimization engine should be able to determine the offsets. It's certainly a cute trick, but it does indeed save a little time and a couple of measurements.

I'm sure you could implement something similar in WinPCD.

[hongrn]

Re: WinPCD - New tab for finding Relative Acoustic Offset

Yep, it takes real skills to design a good user interface, and even better skills to write an instruction manual.

[charlielaub]

Re: WinPCD - New tab for finding Relative Acoustic Offset

the Frequency Response Blender can extract minimum phase. There is no requirement to blend anything. Only the slope of the tails will influence the phase at the frequency extremes.

[Jeff B.]

Re: WinPCD - New tab for finding Relative Acoustic Offset

the Frequency Response Blender can extract minimum phase. There is no requirement to blend anything. Only the slope of the tails will influence the phase at the frequency extremes.

Will the current version, the one I added stuff to, still work outside the Windows Office environment? If not, you should make a "lite" version available so others can still make use of its features.

[charlielaub]

Re: WinPCD - New tab for finding Relative Acoustic Offset

Will the current version, the one I added stuff to, still work outside the Windows Office environment? If not, you should make a "lite" version available so others can still make use of its features.

From the Blender tutorial, which is available HERE:

Using the Blender when VBA macros cannot be executed:

So, yes, you can still use the Blender in OpenOffice or LibreOffice Calc, but the spinners and buttons will not function. You will not be able to make use of the diffraction modeler in OO Calc, for instance.

Everything should still work that is not macro driver, however, meaning you can still blend and extract minimum phase.

[dlr]

Re: WinPCD - New tab for finding Relative Acoustic Offset

Although this may be true for your software, the offsets for all drivers in a loudspeaker can be simultaneously determined provided there is sufficient data present and an algorithm that can pull off the task.

It's true for any software I've used in the past, not just mine. I haven't checked out your software yet, it sounds like a useful addition.

I'm sure you could implement something similar in WinPCD.

I briefly considered something like this, but am not in position to put that time into it. I'd rather put the effort into additional options since the user has the ability to do it now. One problem I thought might occur is the case with drivers with significant offset. It seemed to me that diverging results could be a problem the way it can be with an optimizer. If it can converge it will save time, no doubt. My concern was that if there was divergence with more than two drivers, some users might not recognize the problem. I also think that it's not a bad idea for DIYers to learn the issues such as this, it helps their overall understanding of the spatial aspect of a speaker system.

dlr

[RBG]

WinPCD - New tab for finding Relative Acoustic Offset

This probably was discussed somewhere, but can you use any of these softs with MS office for Mac?

[dlr]

Re: WinPCD - New tab for finding Relative Acoustic Offset

Series crossovers I hope?????

I had originally planned on replicating most of what is in Jeff's PCD, but put some on the back burner, including series. There have been very few requests for series with 2.5-way higher. Due to the program structure (not the most optimal use of objects I'm sure) it will take a fair amount of work to add. I've been trying to stay on the track of more widely used or desired options, though admittedly some of it is what is in my own mind on that one such as the offset tab section. That one I thought would be helpful, especially to newbies.

The current structure does not reuse any crossover component component objects the way I think Jeff did (if I have that right). Maybe if I can find a way to refactor (organize the object aspect) to make that part easier, I might add it. I've had a desire to go back to that part of the code to improve it. My primary focus early on was responsiveness, not optimal code or memory usage. This was my first real object-oriented program, so I had (have) a lot to learn. I do some refactoring with each change.

dlr

[isaeagle4031]

Re: WinPCD - New tab for finding Relative Acoustic Offset

Understood. You are far and beyond my knowledge of programming. And (thanks to Ben) I'm a glutton for series crossovers it seems. 3 of my last 4 builds have been series lol

[charlielaub]

Re: WinPCD - New tab for finding Relative Acoustic Offset

One problem I thought might occur is the case with drivers with significant offset. It seemed to me that diverging results could be a problem the way it can be with an optimizer. If it can converge it will save time, no doubt. My concern was that if there was divergence with more than two drivers, some users might not recognize the problem. I also think that it's not a bad idea for DIYers to learn the issues such as this, it helps their overall understanding of the spatial aspect of a speaker system.
dlr

I completely understand your concerns. With most any non-linear optimization there is no guarantee of convergence to the "true" solution. Often the result is that the algorithm converges to some local minimum that is hopefully near to, or is, the global minimum.

In this case we have on our side the fact that the interference between drivers creates a "fingerprint" that is unique.

The challenge is that, as the dimension of the problem increases (e.g. you add more drivers to the system), the number of local non-global minima increases and the likelihood that the algorithm ends up in one of these instead of the global minimum goes up as well. But in the limited amount of testing that I have done on systems of up to 4 drivers, I did not experience a problem. One can always fall back to pair-wise offset determination and ACD-L can be used in that manner but with the driver offset and polarity still automatically determined. Doing a sanity check of the fit by looking at the fit residuals and a plot of the fit curve vs the target curve is on the user.