Hello hello!
I hope I'm not asking too many questions on this site! I have tried to reciprocate and answer questions whenever I can, but since I am a newbie it's not often that I feel I can be useful! Invariably I get kind and helpful answers, and I do not take that for granted! What a wonderful community 🙂
Anyways, to my question:
I have spent some time fiddling around with various simulation softwares. I think I've gotten a hang of the basics of WinISD, I understand at least how to basically navigate the late Mr. Bagby's (RIP) PCD, and this week I have started trying to learn the interface and tools available on VituixCAD, which I am expecting to become my one stop shop once I get more comfortable with it. Someday, I hope to learn how to use AKABAK, but one thing at a time.. haha
If I went into detail about everything I DON'T know about crossover design, I (and possibly the more patient among you) would be here all day, so for now my query regards phase and time alignment.
It seems to me that variable time arrival throughout the frequency spectrum is inevitable? So, presupposing that, we have to decide between coherence at the crossover point or a short group delay across the whole spectrum?
My (correct me if I'm wrong) understanding is that, to slightly misuse an economics term here, it is basically a pareto efficient system where you cannot improve coherence at crossover point without worsening average group delay?
I read an article that went into how different passive electrical filters shift phase above or below your crossover point depending on LPF or HPF, and how the common approach is simply to use a high order filter to minimize the range of crosstalk between drivers, then to focus on a good average group delay. But then, I have also come across folks that seem to pay paramount attention to coherence at crossover point, who say that phase shifting is undetectable at high frequencies so it's ok if it's a full 180 DEGREES OUT OF PHASE with the lower region??
As a rather gullible newbie, I am begging to take someone's word on this! But my research has given me a lot of contradictory information. Is there a general consensus these days on how to approach phase and time alignment when it comes to crossover design?
Or am I completely misunderstanding things? Was my initial presupposition false, and indeed perfect time alignment across the entire frequency range of a passively filtered multi way speaker realistically achievable?
Tentatively, my thinking has been to focus on average group delay, then try for high order slopes with the lowest order electrical filters possible at crossover points at frequencies we're not as sensitive to (sub 500, which i am guessing is also helped by longer wavelengths? and above 5k?)
Thank you in advance. It is a testament to the kind heart of this community that I am progressing at all! (Not the sharpest tool in the tool thing 😋)
I hope I'm not asking too many questions on this site! I have tried to reciprocate and answer questions whenever I can, but since I am a newbie it's not often that I feel I can be useful! Invariably I get kind and helpful answers, and I do not take that for granted! What a wonderful community 🙂
Anyways, to my question:
I have spent some time fiddling around with various simulation softwares. I think I've gotten a hang of the basics of WinISD, I understand at least how to basically navigate the late Mr. Bagby's (RIP) PCD, and this week I have started trying to learn the interface and tools available on VituixCAD, which I am expecting to become my one stop shop once I get more comfortable with it. Someday, I hope to learn how to use AKABAK, but one thing at a time.. haha
If I went into detail about everything I DON'T know about crossover design, I (and possibly the more patient among you) would be here all day, so for now my query regards phase and time alignment.
It seems to me that variable time arrival throughout the frequency spectrum is inevitable? So, presupposing that, we have to decide between coherence at the crossover point or a short group delay across the whole spectrum?
My (correct me if I'm wrong) understanding is that, to slightly misuse an economics term here, it is basically a pareto efficient system where you cannot improve coherence at crossover point without worsening average group delay?
I read an article that went into how different passive electrical filters shift phase above or below your crossover point depending on LPF or HPF, and how the common approach is simply to use a high order filter to minimize the range of crosstalk between drivers, then to focus on a good average group delay. But then, I have also come across folks that seem to pay paramount attention to coherence at crossover point, who say that phase shifting is undetectable at high frequencies so it's ok if it's a full 180 DEGREES OUT OF PHASE with the lower region??
As a rather gullible newbie, I am begging to take someone's word on this! But my research has given me a lot of contradictory information. Is there a general consensus these days on how to approach phase and time alignment when it comes to crossover design?
Or am I completely misunderstanding things? Was my initial presupposition false, and indeed perfect time alignment across the entire frequency range of a passively filtered multi way speaker realistically achievable?
Tentatively, my thinking has been to focus on average group delay, then try for high order slopes with the lowest order electrical filters possible at crossover points at frequencies we're not as sensitive to (sub 500, which i am guessing is also helped by longer wavelengths? and above 5k?)
Thank you in advance. It is a testament to the kind heart of this community that I am progressing at all! (Not the sharpest tool in the tool thing 😋)
Like any compromise it is necessary to learn how much group delay can be tolerated. Ordinarily you can begin by not worrying about it but eventually it will find the time to interest you. Crossing over steep to reduce the band is not necessarily a good compromise.
I have other projects to give me more instant gratification going on, so I am ok with being patient and doing my due diligence right now.
Ok. May I ask what your approach is on this matter? Or does it depend on the project?
Ok. May I ask what your approach is on this matter? Or does it depend on the project?
I was actually just reading an article titled "Audibility of group delay at low frequencies" when I got your notification. Thank you.
After that, I will do some research on the downsides of using steep slopes, because intuitively it seemed like a good idea to me 😕
After that, I will do some research on the downsides of using steep slopes, because intuitively it seemed like a good idea to me 😕
To me it seems the same as running all around a library and then saying to everyone sorry, I thought if I was quick it would bother you less.
So you should look at what it does to group delay.
So you should look at what it does to group delay.
That is a fun analogy haha
It strikes me that it's your reckoning that your signature "A compromise should consider everything that matters, even at the expense of what doesn't" applies to this conversation. I will keep learning about this. Thank you again!
PS, if I knew how to make a signature I think mine would read "I'm confused" 😛
Well, as an economist I had to search to make sure I was correct. 🙂 "Pareto's Law" or the "Pareto Rule" has been cited on diyaudio.com, which has to do with the "80/20" rule. But you appear to be the first person to refer to Pareto Efficiency. (Sorry, I don't think there is any award though.) I think most people on the forum just refer to "trade-offs".
The tip of my tongue is replete with the words I'd sooner use to describe my thoughts if I were just a little better at calling them to mind. It always tickles me that it's my POOR capacity to conjure the contents my word bank which keeps my diction so varied.
"Jesus Bryguy, just say windmill!"
"I swear! I really forgot the word, and "air mower" was the first substitute I could think of!"
Minor aphasia? My father had serious aphasia in the last years of his life, but he usually managed to find a way to express what he meant: 'sport where guys throw each other on the floor' for judo, for example.
There are crossover filters that don't mess up the phase response, at least theoretically. They either have very shallow roll-off (like first-order continuous-time filters; there are ways to go a bit beyond first order, but not much) or pre-ringing (linear phase FIR).
There are crossover filters that don't mess up the phase response, at least theoretically. They either have very shallow roll-off (like first-order continuous-time filters; there are ways to go a bit beyond first order, but not much) or pre-ringing (linear phase FIR).
My father has aphasia as well, after his stroke. It is an unfortunate affliction. Myself, I think I just have a bad memory! (Plus, I am 20 years old, so aphasia I think is unlikely)
I have been looking at some of the different approaches people have taken like the "harsch" filters mr. xrk971 uses. I'm finding myself yet again to be quite out of my depth! Still much learning to be done.. I do not have a background in anything mathematically oriented, so all of this stuff is uncharted territory for me! And resources like the loudspeaker cookbook quickly become overwhelming and filled with words and concepts which it presupposes you are, but that I am not really familiar with. Maybe if I tried reading it again now I'd have better luck
Don't get overwhelmed, take your time and look into most interesting topic at a time at suitable depth. Try to understand the main idea / concept, you can always return to it and dig deeper if there is need to. Year or two into the hobby and all the bits and bobs start to make sense 🙂
For example, you can think crossovers to be perfect and concentrate on other things until its time to do one. And you can actually make them perfect, what ever that is, with free software and low cost measurement gear, its just another x hours of time to make happen. You could also assume drivers ideal until you dig deeper into them, boxes problem free, etc. Let your imagination run free, anything is possible and you can afford it 😀
For example, you can think crossovers to be perfect and concentrate on other things until its time to do one. And you can actually make them perfect, what ever that is, with free software and low cost measurement gear, its just another x hours of time to make happen. You could also assume drivers ideal until you dig deeper into them, boxes problem free, etc. Let your imagination run free, anything is possible and you can afford it 😀
Last edited:
Thank you tmuikku 🙂
I am letting myself commit a number of design sins for the time being, but since I am not committing any of it to a real speaker design at the moment I don't feel too bad about it. I am trading some of my speaker gear soon for an Omnimic and I already have a DATS, so all I'll have holding me back pretty soon is myself!
If you consider taking measurements and simulartng stuff based on them you would be better of by investing sound interface and microphone instead of the USB mic. Two channel sound interface can take advantage of loopback in the second channel to keep timing syncronized between measurements. USB mic cannot do it and lacking timing reference means measurements are unreliable as for example operating system can affect different delay, ruins phase information. End result made based on measurements is precisely as good as the measurements are.
Remember to have fun with the hobby! 🙂
Remember to have fun with the hobby! 🙂
I'm definitely having fun! Hm, I'm not sure I understand. Is the problem inherent to USB, or rather is it that there is only one microphone? Would for example, two Dayton EMM-6 microphones attached to 2 channel microphone preamp connected to my computer (via USB) be suitable?
One mic is fine, other channel is connected with a cable from output to input, while the other output feeds amplifier and input the microphone. See VituixCAD manual or measurement guide, or perhaps REW or ARTA and other software also have guides how to do dual channel or semi dual channel measurements.
The point is to preserve accurate time information in the measurements, between measurements of various drivers and rotation angles. When microphone and DUT location is known the measured data (potentially) accurately represents reality.
What goes to microphone I think they are all good enough for occasional use as long as it comes with calibration file. USB interface is fine, if there is delay due to operating system its in both the reference and mic channel and doesnt bother.
The point is to preserve accurate time information in the measurements, between measurements of various drivers and rotation angles. When microphone and DUT location is known the measured data (potentially) accurately represents reality.
What goes to microphone I think they are all good enough for occasional use as long as it comes with calibration file. USB interface is fine, if there is delay due to operating system its in both the reference and mic channel and doesnt bother.
Just a regular condensor measurement microphone, running through a DIY phantom power supply and balanced to unbalanced conversion.
To put that in simpler terms, it means that you can plug it straight into a line input socket (like a standard 3.5mm connector) and timing has not been changed. This is important so that you can combine it with the wanted loopback signal so that your mic signal and loopback signal become the new Left and Right inputs.
There is a way you can do this without building electronics and that is to buy an appropriate sound interface.
To put that in simpler terms, it means that you can plug it straight into a line input socket (like a standard 3.5mm connector) and timing has not been changed. This is important so that you can combine it with the wanted loopback signal so that your mic signal and loopback signal become the new Left and Right inputs.
There is a way you can do this without building electronics and that is to buy an appropriate sound interface.
Any non-USB mic, a mic that is used with interface is my message. Calibration file makes the data more accurate.