Howdy folks, gonna try to sum this up quick as I've got some other threads elsewhere on it.
Built some C-notes, measured at the end to confirm performance. I was getting a HF shelf boost. PE offered to measure the speaker to confirm it's working as intended and they did so. It appeared to measure pretty much like the advertised spec. I ended up considering my EMM6 might be off so I had them replace it. I went ahead and measured them with the new mic and the shelf is still there. This is strange to me because I was informed that the omnimic PE is using in house is calibrated the same as the EMM6. I've got some speaker projects that have been on old because I don't know how if I can trust my EMM6. I was speculating that it's from pointing the mic at the speaker, and that rotating it so it's facing up might show more accuracy. It kind of starts lining up with PE's data but the shelf is still there, just lessened.
Here's my own outdoor data with EMM6 pointed at the speaker.
Here is PE's data. I was told it was an omni mic pointed at the speaker. Black is original, red is mine.
Built some C-notes, measured at the end to confirm performance. I was getting a HF shelf boost. PE offered to measure the speaker to confirm it's working as intended and they did so. It appeared to measure pretty much like the advertised spec. I ended up considering my EMM6 might be off so I had them replace it. I went ahead and measured them with the new mic and the shelf is still there. This is strange to me because I was informed that the omnimic PE is using in house is calibrated the same as the EMM6. I've got some speaker projects that have been on old because I don't know how if I can trust my EMM6. I was speculating that it's from pointing the mic at the speaker, and that rotating it so it's facing up might show more accuracy. It kind of starts lining up with PE's data but the shelf is still there, just lessened.
Here's my own outdoor data with EMM6 pointed at the speaker.
Here is PE's data. I was told it was an omni mic pointed at the speaker. Black is original, red is mine.
The Panasonic capsules do tend to have a peak at the top end. It isn't a problem if you can navigate around it, you can still build a good speaker with it.
How does one work around this? Seems tough to design a xover when all your measurements have a HF shelf topping out at nearly 10db.
Any errors will be common to all measurements, so in a sense you can ignore it. Think about a regular crossover, you need to voice it with global EQ after you cross (and flat isn't always what you end up with).
Since all discrepancies here are global, there will be no errors in your crossing because all that matters is the relative contributions of the drivers. You can compensate them simply after crossing.
With all that said, get to know your mic curve. It will be fine up to some frequency then will rise in a certain way. If it concerns you, you might be able to find a generic calibration file, or make your own. You might compare the result to a known measurement, like a common tweeter on a large baffle, like the manufacturer curve.
Here is an example of the ECM8000 (many mics response plotted together), as it uses the same Panasonic IIRC.
Since all discrepancies here are global, there will be no errors in your crossing because all that matters is the relative contributions of the drivers. You can compensate them simply after crossing.
With all that said, get to know your mic curve. It will be fine up to some frequency then will rise in a certain way. If it concerns you, you might be able to find a generic calibration file, or make your own. You might compare the result to a known measurement, like a common tweeter on a large baffle, like the manufacturer curve.
Here is an example of the ECM8000 (many mics response plotted together), as it uses the same Panasonic IIRC.
If you haven't already, have your microphone calibrated and load the calibration file into your measurement software.
Tom
Tom
If your measurement software doesn't support "calibration files" and you can program, you can post-process it yourself. The math is called deconvolution, or complex division. The mic file has to be normalized to 0dB in the flat pat of its spectrum, and phase has to be unwrapped. Doing it this way will also result in an accurate time domain waveform after an IFT. I can post some example code if it would help.
a quote from another member:
"The Dayton EMM is calibrated, but the accuracy of the calibration is questionable (thus the recommendation to get the calibrated version from Cross Spectrum)."
see post 71
"The Dayton EMM is calibrated, but the accuracy of the calibration is questionable (thus the recommendation to get the calibrated version from Cross Spectrum)."
see post 71
Honestly the calibration is quite small in it's corrections and measurements look nearly identical without it. I'm not convinced ti's a simple calibration issue. Even measures basically identical to the previous EMM6.
This whole thing has gone on forever and kind of spoiled my desire to DIY anything. Can't get a solid answer anywhere, not even from PE.
Don't sweat it. When you voice a speaker you go out on a limb. You're not even trying to make it perfectly flat as that isn't always the way it should be.
What if the goal is flat anechoic response? The speakers I'm working on are intended to be used as studio monitors.
You will find that flac anechoic response goal to be challenging. I did some comparative tests of different microphones. They are in this PowerPoint on measuring speakers: https://www.linearsystems.com/lsdata/files/Demian Martin Burning Amp Slides 2019.pdf In it I should the difference between on axis and grazing with a similar microphone, amnd the B&K correction charts for different angles. I hope this helps a little. Tuning to taste is really problematic for response because your ear adapts really well to different "errors" making them inaudible except as differences.
Even if you measure flat, your plot isn't going to show the tendency of diffraction to change the tone, or the reflections off the desk.
I wouldn't account for a desk in the final tuning, not that I use one really but that doesn't quite make sense to me, just fix desk issues with digital eq, I don't lump that into integrating drivers via xover network.
I usually take a lot of measurements on and off axis that should provide me with insight into diffraction behavior. It should be fairly mitigated as the design is basically the tweeter elements from the cinetor Evo with the same cabinet width and 3/4" radius on the edge.
Very insightful paper. Im aware a fairly flat on axis response can be challenging but not out of the realm of my ability. I'm not shooting for Neumann flat, just neutrality in general.
Whether diffractions or reflections, their audible effects change with level. You can't EQ that. You can't EQ their angle or time delay.
I have two interfaces I've tried, a motu m4 and rme babyface. They both give me the same results, I really don't think it's my gear.
I ended up contacting PE to ask what my cause the difference between their measurements and mine.