An unusual question for a speaker, but yes what you propose is an appropriate way to handle this, assuming no interturn capacitance 😉
I measured some times ago one wirewound resistor (10Ω/10W if I remember well) and the inductance was completely insignificant for audio use in speakers. The measurement was made with a VNA (BODE100).
Updating the download link for Xsim by using the https version..
https://libinst.com/Xsim/XSimSetup.exe
Also attaching the file here in case of outages (if that is allowed?)..
https://libinst.com/Xsim/XSimSetup.exe
Also attaching the file here in case of outages (if that is allowed?)..
My favorite XO simulator program. I have now used Xsim on the development of 7 commercial speakers. Easy enough for DIY’ers, but definitely powerful enough and capable of professional results.
Thanks Bwaslo for a superb product!
Thanks Bwaslo for a superb product!
Here is an example of how well Xsim can do in predicting the response of one of my speakers vs the actual measurement using a production crossover PCB (orange is actual measurement and blue is prediction by Xsim):
The predicted Step Response matches perfectly also - but here is just the measured Step Response - a very nice behavior close to the ideal right triangle shape:
The predicted Step Response matches perfectly also - but here is just the measured Step Response - a very nice behavior close to the ideal right triangle shape:
THANK you. I've thought this for a long time, as even 10 uH inductors have way more turns than wirewound resistors.
Hi guys 🙂 I recently discovered Xsim from reading here, many thanks to the creator of it! so am now playing with it.
I have this 3rd order crossover project just as an experiment. I constructed the FRD and ZMA as text files from manufacturer graphs, so I realise they are a bit theoretical at this point too. Component values started out at Butterworth then drifted with adjustments in capacitors. I was just wondering if you cast your eye over it, and feed back your observations, so I can learn my way around 'steering' it, and what needs steering to get what effect.
I am still learning around what makes a good graph vs a bad one, what statistics are the vitals, and which are the less dramatic.
Cheers!
I have this 3rd order crossover project just as an experiment. I constructed the FRD and ZMA as text files from manufacturer graphs, so I realise they are a bit theoretical at this point too. Component values started out at Butterworth then drifted with adjustments in capacitors. I was just wondering if you cast your eye over it, and feed back your observations, so I can learn my way around 'steering' it, and what needs steering to get what effect.
I am still learning around what makes a good graph vs a bad one, what statistics are the vitals, and which are the less dramatic.
Cheers!
Look for a base sensitivity (level). Your response rises towards 1kHz but it would ordinarily want to stay where the lower midrange is. Make it fall slightly towards the top.
However the outside of your box will probably raise the upper midrange and treble even beyond this. The inside of your box will also raise some of the lower frequencies.
However the outside of your box will probably raise the upper midrange and treble even beyond this. The inside of your box will also raise some of the lower frequencies.
Thanks, that's interesting - so if I were to target more like bringing the overall level more around the 80 to 85bB line?
Is that dip under 4ohm on the system impedance any cause for caution? small impedances make something inside me say 'don't short circuit stuff!'
Is that dip under 4ohm on the system impedance any cause for caution? small impedances make something inside me say 'don't short circuit stuff!'
The process of reducing the tweeter level will probably tend to ease the impedance dip
Did you include the phase responses of the SPL and impedance responses of the drivers? That would likely have been done using a minimum-phase transformation, but some manufacturers include the phase data with their impedance curves.
It would be handy to know which drivers you are using, so please provide some further details.
General announcement -- I uploaded a new version of XSim ("XSim4").
More info at https://www.diyaudio.com/community/threads/xsim4-program-available.417482/latest
More info at https://www.diyaudio.com/community/threads/xsim4-program-available.417482/latest
I did several measurements in REW from several points around the center and grouped them using "overlay". However, in order to import them all into XSim, is it possible to average them down into one curve? The bucket of measurements looks like:
Well, first problem is, you also need phase measurements. Second is that phase info doesn't meaningfully average, except when all curves are very close, at which point, best to just pick the one closest to the middle.
What I've done is to model each actual driver with multiple drivers measured at all the different places, and just drag them in and out of the circuit to look at the results from different places. XSim will let you have driver or othe parts just sitting unconnected, they get ignored in the sim.
What I've done is to model each actual driver with multiple drivers measured at all the different places, and just drag them in and out of the circuit to look at the results from different places. XSim will let you have driver or othe parts just sitting unconnected, they get ignored in the sim.
Maybe he’s just asking if you can mathematically average any curve? There is math functionality to do that. Like subtract, add, scale, it seems adding 5 together then dividing by 5 will give you a
Mathematical average without consideration to phase.
Mathematical average without consideration to phase.
You could of course average magnitudes, but angles? What's the average for an extreme case like 80dBSPL@0deg and 80dBSPL@180deg?
Combine them as vectors, you get -infinity @ undefined degrees. Average dBSPL.and deg separately you get 80dB@90deg. But 180deg is the same as -180deg, so is the average phase also 80dB@-90deg? Which?, can't be both.
Or try
80dBSPL@181deg and 80dBSPL@-179deg: Average the dBSPL and degrees separately, you get 80dBSPL@1deg, which is nonsense, it should stay around 181deg. The periodicity of phase makes averaging strange.
Combine them as vectors, you get -infinity @ undefined degrees. Average dBSPL.and deg separately you get 80dB@90deg. But 180deg is the same as -180deg, so is the average phase also 80dB@-90deg? Which?, can't be both.
Or try
80dBSPL@181deg and 80dBSPL@-179deg: Average the dBSPL and degrees separately, you get 80dBSPL@1deg, which is nonsense, it should stay around 181deg. The periodicity of phase makes averaging strange.
The issue with averaging complex quantities is only a problem if you're trying to arrive at a typical type curve using a bunch of good (but made under different conditions) curves. If you have a bunch of curves that SHOULD be identical but arent, because of noise, though, then averaging the complex curves (average the real parts and average the imaginary parts) is an advantage because averaged noise should go toward zero.