Hi, Everyone.
So I have a simulated response in XSIM based on measurements I have taken of my loudspeaker. I won't go into detail on how I arrived at the measurements but they are accurate.
My simulated crossover and the actual real life measurements of the loudspeaker after the crossover was implemented are different. After some manual tweaking I got the crossover where I like it, and then re-did my crossover in XSIM. Obviously, the measurement part of XSIM is not accurate to real life.
Here's my question, is there a way in this case to take my measurement into XSIM and essentially teach it to to say, with these crossover components these are the actual measurements and then tweak it further from there?
Cheers,
Joey
So I have a simulated response in XSIM based on measurements I have taken of my loudspeaker. I won't go into detail on how I arrived at the measurements but they are accurate.
My simulated crossover and the actual real life measurements of the loudspeaker after the crossover was implemented are different. After some manual tweaking I got the crossover where I like it, and then re-did my crossover in XSIM. Obviously, the measurement part of XSIM is not accurate to real life.
Here's my question, is there a way in this case to take my measurement into XSIM and essentially teach it to to say, with these crossover components these are the actual measurements and then tweak it further from there?
Cheers,
Joey
If you think XSIM is broken, then tweaking it's output to match your measurements isn't going to give a valid simulation that you can further tweak. Try another crossover simulator, (or, more likely, check your method of creating and applying the FRD and ZMA files used by XSIM).
Crossover simulation software, such as XSim or VituixCad, are known to be quite good. Unfortunately, when someone is not getting good agreement, measurements are the first place to start looking. Making good measurements seems easy, but there are many ways to do it wrong.
Other things that can go wrong:
(2) The crossover wiring does not match the simulation
(3) The crossover component values do not match the simulation (including DCR insertion loss)
(4) The actual driver to driver phase (i.e. timing) is different than what the simulation is working with. This can be the result of a measurement mistake, but it can also be an incorrect setting in the simulation.
Think about these, and make sure that none of these potential errors apply to your situation.
j.
Xsim applies its filters accurately.
I may be wrong but I think it does 15th octave smoothing, which is relatively benign and possibly helpful.
I may be wrong but I think it does 15th octave smoothing, which is relatively benign and possibly helpful.
Thanks for the good words, AllenB and hifiJim. (I'm the author of XSim).
Something to keep in mind is that XSim and other simulators are just basic mathematics (complex arithmetic and matrix solution). Other than possible programming errors (which should have shown up by now!) there's not a lot to get wrong. They aren't anything really new, any novelty is the user interface that lets circuits be entered and adjusted more easily than by, say, a netlist.
Something to keep in mind is that XSim and other simulators are just basic mathematics (complex arithmetic and matrix solution). Other than possible programming errors (which should have shown up by now!) there's not a lot to get wrong. They aren't anything really new, any novelty is the user interface that lets circuits be entered and adjusted more easily than by, say, a netlist.
Thanks Bwaslo, your XSim software is invaluable! Along with DATs and OmniMic even a rank amateur like me can make superlative loudspeaker designs.
Years ago I when I was still employed in electroacoustics I used MLSSA, Clio and Neutrik gear to measure and develop loudspeakers, but invariably tweaked the crossover manually. It's so much easier with XSim, and not infrequently tweaks are counter-intuitive (or at least my intuition hasn't always been good!)
How do you know your measurements are accurate? If you measured impedance using a PC sound card and a resistor (I have seen 27 ohm and 100 ohm recommended. Also, suggested was using an amplifier and a 10 Ohm resistor) your measurements are wrong. The resistor is much too large. It will change the Qes of the driver under testing and consequently Qts. You need to use a current shunt of 0.1 Ohms or less or a good quality current probe. With the small resistor you will need an amplifier which you will need to calibrate. Texas Instruments makes IC amps for that purpose and you can purchase a evaluation board all ready to use.
Knowing the actual impedance of the drivers at the crossover frequency is essential to designing the crossover filters; which should be computed for them to have the proper Q. If you don't have the proper Q you will probably have a bump or dip at the crossover frequency. If the HF and LF filters don't have the same Q, you might have phase issues because all 4 poles won't coincide.
For L-R 2nd order filters, the Q should be 0.5.
Last edited:
What you are suggesting here will work.
Qes of the driver does not change when you put a resistor in series with the voice coil.
Is that measuring before OR after the series resistor ?👽?
Any series L C or R added to a driver MUST by definition, change TS parameters [ depending where measured ].
Obviously measuring impedance requires series a series resistance.
https://www.diyaudio.com/community/threads/speaker-rebuild.416263/post-7972571
Any series L C or R added to a driver MUST by definition, change TS parameters [ depending where measured ].
Obviously measuring impedance requires series a series resistance.
https://www.diyaudio.com/community/threads/speaker-rebuild.416263/post-7972571
To clarify >
The results of xSIM is actually affected by the impedance of the 'source drive signal' > usually assumed [ default ] as 1 milli ohm.
PS.
I think I'm talking about very subtle things.
The results of xSIM is actually affected by the impedance of the 'source drive signal' > usually assumed [ default ] as 1 milli ohm.
PS.
I think I'm talking about very subtle things.
What I meant was if you conducted TS parameter test/analysis in a known sealed box volume directly at the voice coil terminals
[ which is the way it was done years ago before manufactures quoted TS parameters ]
and then repeated the test with series components, you would measure different results.
[ which is the way it was done years ago before manufactures quoted TS parameters ]
and then repeated the test with series components, you would measure different results.
Pre Windows 95 there was program code that had to be laboriously entered into the 'computer of the day' that then used
a 1 cubic foot sealed enclosure to perform tests as described.
From memory, the details including program code was published in a book called "How to design loudspeakers"
[ I don't know what became of my copy of the book ]
It was probably the first comprehensive exposé & application of the work of Theil & Small.
My friend - with computer - entered the code & measured accordingly a cheap Magnavox woofer > and came up with a HUGE optimum enclosure -
that being for bass reflex tuning with the 'classic' twin impedance peaks of equal amplitude. [ it sure showed the limitation of what we now call
a 'small magnet/low power motor' = high QTS.
PS.
The program used the comparison of measurements made in 'free air' and box enclosed.
a 1 cubic foot sealed enclosure to perform tests as described.
From memory, the details including program code was published in a book called "How to design loudspeakers"
[ I don't know what became of my copy of the book ]
It was probably the first comprehensive exposé & application of the work of Theil & Small.
My friend - with computer - entered the code & measured accordingly a cheap Magnavox woofer > and came up with a HUGE optimum enclosure -
that being for bass reflex tuning with the 'classic' twin impedance peaks of equal amplitude. [ it sure showed the limitation of what we now call
a 'small magnet/low power motor' = high QTS.
PS.
The program used the comparison of measurements made in 'free air' and box enclosed.
Last edited:
It sounds like you're describing the "known volume" variety of the test to measure Vas. When you measure the impedance of a driver you can derive the resonance, ie Q and Fs. When you combine a speaker with a box, their mechanical behaviour combines and this changes the impedance, including the resonance peak. The new resonance details along with knowledge of the box is used to derive Vas.
It is usual to measure the impedance by supplying constant current (ie via a large source resistance) and reading the voltage across the coil which varies with the impedance variations. If you were to instead connect the amp straight to the coil, you'd measure constant voltage across the spectrum since that's what the amp is meant to do.
If you used an intermediate value of resistor, the voltage you read would be different again.. but when you scale it to consider the resistance value you used, you discover that the speaker/box combination impedance is still the same. It is the same for all three of these tests.
It is usual to measure the impedance by supplying constant current (ie via a large source resistance) and reading the voltage across the coil which varies with the impedance variations. If you were to instead connect the amp straight to the coil, you'd measure constant voltage across the spectrum since that's what the amp is meant to do.
If you used an intermediate value of resistor, the voltage you read would be different again.. but when you scale it to consider the resistance value you used, you discover that the speaker/box combination impedance is still the same. It is the same for all three of these tests.
Yes, but the impedance will be the same so the calculated Qt will be the same regardless of the source impedance since it will measure the same.
I guess I should post my opinion of xSIM >
xSim doesn't make mistakes, it purely depends on data entered/presented to it >
Erroneous data entered into it produces an erroneous output. It is purely mathematical analysis.
One thing it can't do is GUESS frequency response anomalies of a driver.
xSim doesn't make mistakes, it purely depends on data entered/presented to it >
Erroneous data entered into it produces an erroneous output. It is purely mathematical analysis.
One thing it can't do is GUESS frequency response anomalies of a driver.