Simulations are very sensitive to how measurements are done and how parameters are set. VCAD2 is quite a challenge to master...
Pictures of measurement process indicate room reflections come within about 2-3 milliseconds. This makes fine measurement data for tweeter /mid xo >>1kHz but below that is flying blind. 3-way speaker needs much longer reflection free window to reduce error for midrange to be able to do the low xo, like 6ms or preferably much more. This means either ground plane measurements or lifting the thing high in air in a very big room or outside, and with spinorama it's real effort.
Perhaps experienced can manage with less but for beginner, and somewhat experienced like me, there is ton of opportunity to make humanly errors in the process and the data ends up with error so does the end result. I'm measuring, processign data, implementing xo, measure to confirm it's what supposed to, rinse and repeat until it is. And it has always been multiple iterations. Luckily changes in DSP filter are free, i can only imagine pain adjusting passive xo several times, although thats really something one must absolutely do to get best out of a speaker.
So, heads up, nice build, take your tine to make the sound as good as it can be. Have fun!🙂
Perhaps experienced can manage with less but for beginner, and somewhat experienced like me, there is ton of opportunity to make humanly errors in the process and the data ends up with error so does the end result. I'm measuring, processign data, implementing xo, measure to confirm it's what supposed to, rinse and repeat until it is. And it has always been multiple iterations. Luckily changes in DSP filter are free, i can only imagine pain adjusting passive xo several times, although thats really something one must absolutely do to get best out of a speaker.
So, heads up, nice build, take your tine to make the sound as good as it can be. Have fun!🙂
Hi tmuikku, indeed passive filters are hard and the interaction with the impedance and enclosure is tricky,
But VituixCAD helped me in the right direction and I am already 90% there.
With my previous speaker I did not have VituixCAD an took 3 months to get to the final version (that I did not change afterwards)
The fun and now confusing part is that my Friend and I build the same speaker and measure and tweak individually.
He seems to have a simulation that is closer to the end measured response.
But for a flat first response he needs different components than I………
That I cannot understand and I plan to investigate that.
After this weekend it will pause for 2 weeks as I need to go on a business trip.
So stay tuned…
But VituixCAD helped me in the right direction and I am already 90% there.
With my previous speaker I did not have VituixCAD an took 3 months to get to the final version (that I did not change afterwards)
The fun and now confusing part is that my Friend and I build the same speaker and measure and tweak individually.
He seems to have a simulation that is closer to the end measured response.
But for a flat first response he needs different components than I………
That I cannot understand and I plan to investigate that.
After this weekend it will pause for 2 weeks as I need to go on a business trip.
So stay tuned…
I think ground plane measurements would be more than helpful. A time window of 8ms is enough for most 3-ways.
Turn some inductors orthogonally for less interference?The passive filter with rest material simulated in VituixCad.
View attachment 1441807
//
I am only focusing on the LPF for the mid and how it connects to HF.
HF is already fine and acts like the filter I used in my DSP
So I am less worried about what happens below 300Hz and that makes it a lot easier and less influenced by the room.
HF is already fine and acts like the filter I used in my DSP
So I am less worried about what happens below 300Hz and that makes it a lot easier and less influenced by the room.
That is what I will do in the end filter as the parts will be closer together, but good point….. unfortunately it didn’t solve the unexpected differenceTurn some inductors orthogonally for less interference?
//
Post some filtered measurements, and the filters, and maybe it will be apparent to someone who does a lot of x-overs.
My experience is in such a case that the measurements have an issue, not vituixcad itself. Can you share some screencopies or dataThe VituixCad simulation did not bring the required result. Although it helps where to tweak but something is off. I did manage to get a frequency response after some modifications of some values. But I like to have this more in line with each other.
Again it is starting to sound really good also with the passive filter.
Back from a business trip (and still a -9 hour jet-lag) I changed the setup to use the Moon Audio W8 amp to power mid and high and relying only on the MiniDSP HP filter at 162 Hz and the temporarily passive MF/HF section. The Hypex DSP is set to 162Hz LR 4th order LP as well.
Again an improvement as the Moon has more control and is even more silent, and I can use balanced connections.
I modified the filter a bit to remove a small wide dip at between 3k and 8kHz and levelled the LF to MF/HF again as the Moon has a lower sensitivity than the 4 channel NAD.
the difference in gain between the Hypex and Moon is still 7dB. So I my new analog discrete HPF will have a 12dB higher gain in its LF (pass-through buffer section) output.
A 4 stage attenuator will allow for attenuating the LF output in 1 dB steps to -15dB so it can be adjusted to level for future amp gains.
I have not yet re-measured the units to do a more reliable VituixCAD simulation but in the end the real measurements are what counts.
Using AudioNET CARMA (really easy and fast), Arta and REW, and all give the same results, so that counts.
More on the later later.
Tomorrow I want to start playing with room correction and the work done by "Obsessive Compulsive Audiophile" YouTube instructions that sounds really promising (check him out)
Stay tuned
Again an improvement as the Moon has more control and is even more silent, and I can use balanced connections.
I modified the filter a bit to remove a small wide dip at between 3k and 8kHz and levelled the LF to MF/HF again as the Moon has a lower sensitivity than the 4 channel NAD.
the difference in gain between the Hypex and Moon is still 7dB. So I my new analog discrete HPF will have a 12dB higher gain in its LF (pass-through buffer section) output.
A 4 stage attenuator will allow for attenuating the LF output in 1 dB steps to -15dB so it can be adjusted to level for future amp gains.
I have not yet re-measured the units to do a more reliable VituixCAD simulation but in the end the real measurements are what counts.
Using AudioNET CARMA (really easy and fast), Arta and REW, and all give the same results, so that counts.
More on the later later.
Tomorrow I want to start playing with room correction and the work done by "Obsessive Compulsive Audiophile" YouTube instructions that sounds really promising (check him out)
Stay tuned
Still working on the filter.
My friend Joost and I still have a different inductor for the LPF of the midrange. But my temp filter gets the same results on his speaker as on my so something is different in his spiderweb version compared to mine.
But I still owed you a picture of the back.
Here it is
Have a nice Easter Day
My friend Joost and I still have a different inductor for the LPF of the midrange. But my temp filter gets the same results on his speaker as on my so something is different in his spiderweb version compared to mine.
But I still owed you a picture of the back.
Here it is
Have a nice Easter Day
Oh and I bought a mic level calibration device that gives a reference tone of 94dB when you mount the microphone.
The system sits at about 91dB @ 2,83V
The system sits at about 91dB @ 2,83V
This the curve with the filter I made of left-over passive components.
The Hypex is an LPF of 24d/Oct at 162Hz, the Mini DSP is just a flat HPF 24dB/0ct at 162Hz both Linkwitz-Riley.
Measured at 1m at tweeter height. Below 200Hz this measurement is influenced by the room modes and do not say a lot.
I will try to combine it with a nearfield measurement,
Still need to play with room correction FIR filter as I mentioned before that added a fantastic improvement at my friend's Joost setup.
Stay tuned.......
Nice work! I wouldn’t average 1/3d on measurements though. While at mid frequencies the resolution probably is determined by the gate, at high frequencies you miss quite some details like interference or possible resonances.
Hi Markbakk,
see below
Psycho acoustic smoothing (how we perceive the sound)
1/48th smoothing
see below
Psycho acoustic smoothing (how we perceive the sound)
1/48th smoothing
- Home
- Loudspeakers
- Multi-Way
- Building the best 3-way (NOT) full range speaker in the world