Yeah, Fusions use IIR. IIR filters acts causally as an analog or passive filter, you can't change the phase without changing the frequency response.
If you want "perfect" impulse response with a multi-way speaker you need to use FIR filtering.
Thanks YSDR.
My driver unfortunately has rising phase in the crossover region.
For a 3 way speaker:
Could I:
1) set a 2nd order LR crossover in Fusion,
2) add basic EQ/correction using the Biquad filters in fusion for each driver
3) then go to my PC and use RePhase to correct the phase to flat for the whole speaker together, or am I screwed trying to fix the phase of the drivers as a whole speaker in the crossover regions?
The alternative being VERY steep filters, so the bands I can't EQ/RePhase are smaller?
My driver unfortunately has rising phase in the crossover region.
For a 3 way speaker:
Could I:
1) set a 2nd order LR crossover in Fusion,
2) add basic EQ/correction using the Biquad filters in fusion for each driver
3) then go to my PC and use RePhase to correct the phase to flat for the whole speaker together, or am I screwed trying to fix the phase of the drivers as a whole speaker in the crossover regions?
The alternative being VERY steep filters, so the bands I can't EQ/RePhase are smaller?
Last edited:
@lbstyling
My advice is like stated in the HFD manual.
First, you need to flatten the frequency responses of the drivers individually at least 1 octave from the xo point(s).
If you are done with the flattening, just apply your desired filter slope.
If you did this with all drivers, then just need to set the timing to get the deepest null at the xo point with reverse-null test, assuming you choosed a Linkwitz-Riley slope. If you have a deep reverse null, then change the polarity of one driver and your phase tracking is as good as your combined frequency response.
Of course there are other methods too. If you want more help, send a message to me.
My advice is like stated in the HFD manual.
First, you need to flatten the frequency responses of the drivers individually at least 1 octave from the xo point(s).
If you are done with the flattening, just apply your desired filter slope.
If you did this with all drivers, then just need to set the timing to get the deepest null at the xo point with reverse-null test, assuming you choosed a Linkwitz-Riley slope. If you have a deep reverse null, then change the polarity of one driver and your phase tracking is as good as your combined frequency response.
Of course there are other methods too. If you want more help, send a message to me.
Having recently gone through this process, I would advise others to start with YSDR's approach above. As you adjust your filters, keep good notes. I used a word document in the form of a lab notebook, and I pasted in graphs, plots. Also, with every significant change you make in HFD, save a copy of the filter file with a new name. I used a date and time stamp for my filename. You always want to be able to go back to a previous filter.
1) use the boost/cut filtering and shelf filtering to adjust each driver to be flat for at least 1 octave beyond the crossover. bandpass drivers (midranges) need to be an octave below the low crossover and an octave above the high crossover.
2) view your measurements with little smoothing (1/48th octave) to look for resonances and narrow band problems, but evaluate the flatness with smoothed data (1/6 octave). In other words, when you have eq'd the drivers to the point where the 1/6 octave data is flat, you are done.
3) make sure your measurement technique is good. There are other threads which talk about this, so I wont go into details... If you are doing a 3 way, it is crucial to get an accurate measure of how much baffle step correction you need, and to accurately blend your near-field (or ground plane) woofer response into your gated measurements. Without this you will struggle getting the woofer to mid transition right.
4) Take measurements on axis, and through a window of off-axis positions. A window of +/- 30 degree horizontal and +/- 15 degree vertical worked well for me. If you discover diffraction problems on-axis that do not exist in the full listening window, don't try to eq those diffraction problems. If they are mild, you wont hear it, if they are severe, you will not be successful at fixing them with eq (you need a different cabinet).
5) as a starting point for applying crossover filters, use LR4. Take your measurements, and adjust your eq parameters by small amounts to get the total system response as flat as you can. Apply delay to the drivers which are physically closer to you to line them up with the driver which is furthest away. Use the polarity reversal feature in HFD to create a null at the crossover, then adjust the delay distance to maximize the null. This becomes your ideal delay value. Once you have gotten the LR4 filters as good as you can get them, take some time to do some critical listening.
6) After your critical listening, you may want to adjust your eq parameters a bit, but get it to the point where you are happy.... This is your baseline.
7) Now is the time to experiment with other filter topologies. If you want to try LR2 or 3rd order BW, please do. For each of these, don't be shocked if you have to make slight adjustments to the eq parameters in order to get them to sound their best. In order to compare an LR2 filter to an LR4 filter with your set of drivers, the only fair way is to compare the best possible LR4 to the best possible LR2...
BTW, in my case, I have not found anything that sounds better than LR4...
Do all of these things first, and then worry about what might happen with an FIR filter and a "perfect" impulse response.
1) use the boost/cut filtering and shelf filtering to adjust each driver to be flat for at least 1 octave beyond the crossover. bandpass drivers (midranges) need to be an octave below the low crossover and an octave above the high crossover.
2) view your measurements with little smoothing (1/48th octave) to look for resonances and narrow band problems, but evaluate the flatness with smoothed data (1/6 octave). In other words, when you have eq'd the drivers to the point where the 1/6 octave data is flat, you are done.
3) make sure your measurement technique is good. There are other threads which talk about this, so I wont go into details... If you are doing a 3 way, it is crucial to get an accurate measure of how much baffle step correction you need, and to accurately blend your near-field (or ground plane) woofer response into your gated measurements. Without this you will struggle getting the woofer to mid transition right.
4) Take measurements on axis, and through a window of off-axis positions. A window of +/- 30 degree horizontal and +/- 15 degree vertical worked well for me. If you discover diffraction problems on-axis that do not exist in the full listening window, don't try to eq those diffraction problems. If they are mild, you wont hear it, if they are severe, you will not be successful at fixing them with eq (you need a different cabinet).
5) as a starting point for applying crossover filters, use LR4. Take your measurements, and adjust your eq parameters by small amounts to get the total system response as flat as you can. Apply delay to the drivers which are physically closer to you to line them up with the driver which is furthest away. Use the polarity reversal feature in HFD to create a null at the crossover, then adjust the delay distance to maximize the null. This becomes your ideal delay value. Once you have gotten the LR4 filters as good as you can get them, take some time to do some critical listening.
6) After your critical listening, you may want to adjust your eq parameters a bit, but get it to the point where you are happy.... This is your baseline.
7) Now is the time to experiment with other filter topologies. If you want to try LR2 or 3rd order BW, please do. For each of these, don't be shocked if you have to make slight adjustments to the eq parameters in order to get them to sound their best. In order to compare an LR2 filter to an LR4 filter with your set of drivers, the only fair way is to compare the best possible LR4 to the best possible LR2...
BTW, in my case, I have not found anything that sounds better than LR4...
Do all of these things first, and then worry about what might happen with an FIR filter and a "perfect" impulse response.
Last edited:
Thanks!
This is basically where I got to 2 weeks ago.
I ended up with a LR 12db filter.
I have been researching filter phase and delay, and would like to try to correct the impulse response to see what it does to the sound.
It appears that IIR filters may actually be better than FIR filters in some regards as they do not introduce 'pre-ringing. Pre ringing (if I have got it right) cannot be compensated for/corrected.
So I am now looking into phase correction with IIR filters. This basically means delaying the signal and using a 'all pass' filter. I have not figured out how to do it yet though.
Bruno is apparently on board with this idea somewhat if I have read this correctly.
My initial trials with FIR filtering on the PC gives me the impression that they loose the initial crack of sound on cymbals etc, at least with the amount of filtering I'm using on a compression driver.
Here is a file playing 4 drum hits with no pre-ringing vs 4 hits with pre-ringing for you to hear yourself.
ShortTest.flac - Google Drive
Interesting. It should sound like that? Sounds really weird with the pre-ringing.
Hi, as someone that has been long using both the Ncore modules and the Fusion amps, how woukd you rate them relatively to each other in perceived sound quality..?.. I have a pair of KEFs 205 reference in my living room I want to activate got a pair of new 205.2 coaxials so this will be a 3 way soon) as well as a pair of Genelec 1037B I would like to substitute theinternal amps and compare( also have a pair of old Dynaudio C2 passive studio monitors lying around in my office..).
I already have a Minidsp openDRC di I use before my preamp/amp integrated ( just EQ with REW for now, will try rephase later) wich I am loving (3d illusion clearly drops when I insert it in the system just doing nothing, but EQ is two steps forward, so all in all, it stays..). I want to get to near state of the art sound, and was thinking of gettin an OKTO Research unit and a couple of NCores, as I also have an old balanced class A Power amp lying around I could use for highs. But the Fusion 253 looks like an unbeatable, self contained, easy to configure solution, especcialy also having the possibilityto supplement its internal processing, all in one solution, that specwise could take me there, with minimal hassle and much, much cheaper.... I am really thorn, prefer to spend once but get what I want instead of trying to go cheap and ending spending more.. Also, reports I keep reading of this amps blowing and taking drivers with them is something I´d rather not deal with especially with my coaxs( I got lucky, wont be able to replace them if they fail...). So, are they really worthy at the quality level I want, or must I really dig deeper in my pocket..? Thanks a lot for your time,
falvespinto
All I can say is that they are both good enough that the sound quality is determined by other components in the chain.
My advice is "don't"
. The people at Genelec know what they are doing.3D illusion is just that - an illusion. Usually created by phase differences and/or uneven frequency response/dispersion.
I have not seen that happen, and would love to analyze the case(s) where that seems to have happened.
Hi,
Thanks a lot Julf, got that itch scratched, think I'll keep it simple then,..get a pair of Fusions, streamline the system and call it a day. Love those Genelecs too, and yes they clearly know what they are doing, I was just curious to check if modern amps with digital filtering and REW could give them a run for their money, 20 years down the line...but hey, the brains/knowledge departmeng would probably need 80 years head start, anyway...☺️��
Thanks a lot Julf, got that itch scratched, think I'll keep it simple then,..get a pair of Fusions, streamline the system and call it a day. Love those Genelecs too, and yes they clearly know what they are doing, I was just curious to check if modern amps with digital filtering and REW could give them a run for their money, 20 years down the line...but hey, the brains/knowledge departmeng would probably need 80 years head start, anyway...☺️��
It's still a mystery (or not?) how that headroom is used in the DSP. If there is free 8 bit to increase the level, then there would be no difference other than the sound level is higher. But I found that if I go anywhere over 0 dB, the base noise is increase.
Or is this a normal behaviour?
Or is this a normal behaviour?
Last edited:
Hi All,
i am curious if someone of you knows if the fusionamps DSP can be bypassed, but still being able to have the same voltage gain for each unit?
I would like to have the option to work with the DSP but want to use an ASP at some point.
I know the advantages of the dsp, but would like to go full analog when playing vinyl.
Cheers
i am curious if someone of you knows if the fusionamps DSP can be bypassed, but still being able to have the same voltage gain for each unit?
I would like to have the option to work with the DSP but want to use an ASP at some point.
I know the advantages of the dsp, but would like to go full analog when playing vinyl.
Cheers