Driver Measurements

[TimRawson]

Trying out LspCAD (will also try out SoundEasy in a couple weeks). Going to try some measurements, but it is not clear as how drivers should be measured, or rather the correct process. LspCAD seems to imply measuring in actual box, but haven't built box yet. Can they be measured on an "infinite baffle" (mounted in a large sheet of wood with back to open air? Should I use a dummy box (about .5 cubic feet)? Near field or 1M? A quick run down of basic approach to obtain freq and impedence would be much appreciated.

On a similar note, if I have a diy 2 way speaker that already is constructed and want to redesign it's crossover, imagine I could take inbox measurements easily enough - but how does one consider the baffle step? Is it subtracted from the response?

[TacoD]

I measure in a testbox, an infinite baffle doesn't show the bafflestep. Also it is important to measure the port output. There is a really nice guide from LspCAD. http://www.ijdata.com/l_justmls.htm See also http://www.hifiklubi.ee/fat_margaret_project/cross.html

[ergo]

Hi,

First off - you have to realize that even if you do the measurements on big baffle it can be helpful for designing the box etc, BUT for the final design of the crossover you can not use these measurements.

My process is usually as follows (after the drivers are selected and bought):

* to measure the impedances of mid woofer and extract T/S parameters
* design the box with whatever software tool you like (LspCad in my case)
* simulate the best tweeter location (with KHF Tool, Edge, BDS etc)
* build a box
* measure the mid woofer impedance in box to verify correct tuning
* measure the driver responses in final box
** for the mid woofer I first measure near field and if there is one then the port near field
** for tweeter I usually do a 70cm on axis measurement
** then mid woofer without touching the microphone
** then the mid woofer and tweeter in parallel
** then to be sure also the mid woofer and tweeter in parallel but tweeter reversed (helps to double check the relative offset later on)

These last two measurements are important to determine the correct relative offset between the tweeter and woofer. LspCad tutorials show this in more detail.

Now if you have measured all this only then you are ready to design a crossover. This is because the correct crossover simulation needs the driver x,y,z coordinates relative to each other. Without this and the phase data the result of crossover simulation is not the same as in reality.

****

Perhaps Jon and other experienced members can also give some pointers.

One thing that I have usually done to make the process more simple is to use the same mid woofer response for simulation that is measured 70cm on tweeter axis. The more precise way would be to measure the mid woofer twice - first on tweeter axis (for relative offset) and then lower the mic to mid woofer on axis and measure again. This way you are exactly on woofer axis and one might say it is more precise.

I do not know which way is better... as we listen with ear on tweeter axis level anyway I'm not so sure the simpler method is worse in precision.

Ergo

[ergo]

With my current ongoing project I made a different approach. I traced the SPL graph from Seas datasheet with SPL Trace.

Imported this to LspCad, made some smoothing and exported.

Imported the output file to KHF Tool (could be BDS spreadsheet also) and found an optimum tweeter location on baffle with real tweeter data.

Usually the baffle diffraction simulation shows only the variance from 0dB level. This way you can take into account any tweeter specific response variances and try to optimize the location on baffle so that the edge diffraction might even smooth the native response of the tweeter.

I cut the baffles on Sunday but due to busy schedule it will take a little time before I can actually measure the result and verify if this approach was as good as planned.

Ergo

[JonMarsh]

Well, my own process is like a mix of what Ergo describes above. It starts with relatively near field measurements on large dipole baffles, to get a feel for the inherent driver response and distortion.

This data is then imported into BDS, or now, the KH tool. Then by simulation I refine my baffle/panel/enclosure design until I think the results meet the system plan as well as possible. (this is why you see those asymmetrical layouts even for dipole baffles from me).

Then, the test baffles are built and measurements done pretty much as Ergo describes above. When you can correlate the driver positioning and offset to the extent that your simulation matches the real world test response reversing a driver phase, then it's possible to design the crossover with some accuracy. For flat baffle designs, I find I can usually save this data and get good results for future designs as regards acoustic offset.

It's all pretty straightforward, just a bit time consuming. As most things worth doing in life are. :B

Some other notes: LspCAD JustMLS uses MLS for impedance measurements also, and the same warnings regarding gating time and LF response apply- you must have a long window for validity of measurements. For Impedance, I sugggest downloading the demo of Praxis and trying that for impedance measurements.

Myself, I love LspCAD for crossover design (version 6), but I've only dallied with Just MLS briefly, and mostly use CLIO and Praxis for measurements. LspCAD does provide an all in one package, but I like some of the other features and capabilities of those other packages. If you get involved in this hobby, you may come to feel the same way.

LspCAD and Soundeasy have very different "workflows"; in Soundeasy, you have to take some specific steps and save data to make it ready for the next step in the flow; LspCAD, to me, has a more natural workflow. i don't like the CAD interface much in Soundeasy, but that's just a personal issue (character flaw? I've owned two versions of Soundeasy (older ones) but switched to LspCAD in version 5, not without some concerns, but version 6 has addressed my concerns and allowed my creativity in crossover design to have free reign (go look at NatalieP crossover- many programs cannot simulate this, including Version 5 LspCAD).

~Jon