You can use various tone controls, graphic and parametric equalizers to do the same thing, and do it a lot faster and easier than by changing components in a crossover.
Maybe you'll argue that's "not the same thing" and I agree. For tweaking the sound, line-level eq's are a lot more versatile.
ETA: You might also argue that changing the crossover changes the dispersion by shifting more frequencies to the woofer or tweeter, and I'll agree with that, but if you're doing this I'd be surprised if you much care or know about dispersion. But if you REALLY want to do that, you can get line-level crossovers and do biamping, with a power amplifier for each driver. There's lots of ways to shave a cat, or however that saying goes.
Last edited:
I did design my actual speakers crossover without measuring and did build it on datasheets of the drivers and predicted response of the boxes. And i was actually very close to what i wanted. The top was a bit too loud and needed attenuation.
I did also measure it, and my response is not flat at all. The bass response is rising from about 300hz as the woofer in my system (CR point is 180hz) is playing louder than the top (a FR driver). At 100Hz it's already 3dB louder and it stays there till 35Hz and then slightly rolls of until an F3 of +/- 25Hz. I wish i was home so i could show you the graphs. But when i would calculated it flat, it would sound bad (to much mid and treble) in my system. Flat response is not always the best response. In music studio's it's needed for the engineer to hear what they do, but studio monitors often doesn't sound very good at all (that is not their goal).
If you know the formula's and the laws of physics arround it, you can tune by ear and get good results. And measurements and computers can help you with that. But that it's impossible to make a good crossover without measurements is ********.
And i learned those formula's and laws of physics by trying to build a crossover from an online calculator. It sounded bad, so i tried to fix it by studying and experimenting the case. When i fixed it (without measurement, because i had no equipment then), i knew a lot more about it so it was a very good lesson. So when noobs try the same way, they will probally do the same if they are serious with DIY'ing.
I did also measure it, and my response is not flat at all. The bass response is rising from about 300hz as the woofer in my system (CR point is 180hz) is playing louder than the top (a FR driver). At 100Hz it's already 3dB louder and it stays there till 35Hz and then slightly rolls of until an F3 of +/- 25Hz. I wish i was home so i could show you the graphs. But when i would calculated it flat, it would sound bad (to much mid and treble) in my system. Flat response is not always the best response. In music studio's it's needed for the engineer to hear what they do, but studio monitors often doesn't sound very good at all (that is not their goal).
If you know the formula's and the laws of physics arround it, you can tune by ear and get good results. And measurements and computers can help you with that. But that it's impossible to make a good crossover without measurements is ********.
And i learned those formula's and laws of physics by trying to build a crossover from an online calculator. It sounded bad, so i tried to fix it by studying and experimenting the case. When i fixed it (without measurement, because i had no equipment then), i knew a lot more about it so it was a very good lesson. So when noobs try the same way, they will probally do the same if they are serious with DIY'ing.
Last edited:
Hi waxx, there is absolutely no doubt that using traced curves and simulations will give you much better results that a simple textbook crossover. You may get quite close to what you would get with dedicated measurements.
I've long wanted to design and build a speaker using traced curves and sims and then do actual measurements on it to see how close it really ends up.
I think the main thrust of the first post (and the title of this thread I think is taking away from that) is that if you just use a text book crossover or pre-made crossover, and then try to tweak that, that you probably won't get where you really want to be.
What you have related above is far more advanced and a very good approach to take if you don't have the tools necessary to do actual measurements.
Tony.
I've long wanted to design and build a speaker using traced curves and sims and then do actual measurements on it to see how close it really ends up.
I think the main thrust of the first post (and the title of this thread I think is taking away from that) is that if you just use a text book crossover or pre-made crossover, and then try to tweak that, that you probably won't get where you really want to be.
What you have related above is far more advanced and a very good approach to take if you don't have the tools necessary to do actual measurements.
Tony.
Yes I agree.
The only part I don't understand is how you would find your relative acoustic centres without measuring? Also, the measurements of all the drivers in the datasheets would be taken on-axis but for crossover design you would measure all drivers at the listening height (usually on the tweeter axis), so they sum correctly at that position.
The only part I don't understand is how you would find your relative acoustic centres without measuring? Also, the measurements of all the drivers in the datasheets would be taken on-axis but for crossover design you would measure all drivers at the listening height (usually on the tweeter axis), so they sum correctly at that position.
Last edited:
Crossovers are the great puzzle of audio, no question of that.
I segued from by ear to using REW. I cannot imagine why anyone would not like to have some kind of baseline to correlate with what you are hearing. I agree what REW hears and what you hear can seem to be two different things and they are.
Then there is the problem of the speakers themselves. In my case, horns, and as much as I like them I am limited to where I can place them in the room and they have plenty of problems themselves, especially the edgar 75 hz horn using the GP 515ghp. Each of them loads the room very differently.
I have acquiesced to using DSP for each driver with the Xilica XP. I was very concerned that I would lose more than I gained but I think I am well ahead. I do not hear anything really funny about the sound of the XP the one thing that surprises me is that, with CD, every recording sounds the same. There is not the variety as I heard without it. Nonetheless I get this spacious presentation with great clarity. One can tune any harshness out by using a gentle filter on the top octaves.
At least with horns as measured by REW a flat response is much too bright. It doesn't sound bad but it is not right. This is part of the room curve discussion which seems in my experience to be something to consider.
I tune each speaker to be as flat as possible without high Q filters at the listening chair so the room's affects are minimized. (I hope)
Everyone is always impressed with a new toy but I have been working with this for about a month. One thing for sure when you have so much opportunity to try things you could find yourself fiddling every bit as much as you did with the passive. The subwoofers previously used the Yamaha YDP2006. I thought that was a good device and twenty years ago I am sure it was. I bought them used about seven years ago. The Xilica is vastly superior. No contest.
if you want a good sounding system you have to measure and listen. And don't be afraid of good DSP. The Xilica is absolutely quiet and easy to use with the computer based control panel.
Yes, I still feel a bit like a heretic but I feel less and less of that each time I listen.
I segued from by ear to using REW. I cannot imagine why anyone would not like to have some kind of baseline to correlate with what you are hearing. I agree what REW hears and what you hear can seem to be two different things and they are.
Then there is the problem of the speakers themselves. In my case, horns, and as much as I like them I am limited to where I can place them in the room and they have plenty of problems themselves, especially the edgar 75 hz horn using the GP 515ghp. Each of them loads the room very differently.
I have acquiesced to using DSP for each driver with the Xilica XP. I was very concerned that I would lose more than I gained but I think I am well ahead. I do not hear anything really funny about the sound of the XP the one thing that surprises me is that, with CD, every recording sounds the same. There is not the variety as I heard without it. Nonetheless I get this spacious presentation with great clarity. One can tune any harshness out by using a gentle filter on the top octaves.
At least with horns as measured by REW a flat response is much too bright. It doesn't sound bad but it is not right. This is part of the room curve discussion which seems in my experience to be something to consider.
I tune each speaker to be as flat as possible without high Q filters at the listening chair so the room's affects are minimized. (I hope)
Everyone is always impressed with a new toy but I have been working with this for about a month. One thing for sure when you have so much opportunity to try things you could find yourself fiddling every bit as much as you did with the passive. The subwoofers previously used the Yamaha YDP2006. I thought that was a good device and twenty years ago I am sure it was. I bought them used about seven years ago. The Xilica is vastly superior. No contest.
if you want a good sounding system you have to measure and listen. And don't be afraid of good DSP. The Xilica is absolutely quiet and easy to use with the computer based control panel.
Yes, I still feel a bit like a heretic but I feel less and less of that each time I listen.
Here is a challenge idea. Start with a very highly regarded 2 way. Take the crossover out. Pass the striped box around to a dozen or so guy with a mini DSP 2x4 and let each guy have a shot at designing a crossover for the system. If you like, also specify the source material to be used for the design evaluation. Each guy will keep his configuration file confidential. In the end, all the configurations would be submitted for listening and measurement. Think there is any chance any would be close to each other or the original?
My question is, how close are actual drivers to there data sheets. Within 1%, 5%, 10% ? If your using data sheets to design your design will be out by the corresponding %. Designing by ear means your source and room will affect your design. A room that absorbs a lot of HF will yield speakers that are too bright etc.
Yes the z offset is the biggest issue when doing a design using manufacturers curves. You basically have to take a guess by trying to work out the offset between the voice coils. 15mm is an often used guestimate.
The on-axis is not an issue if you convert all of the drivers to minimum phase and use a sim that allows you to put in all of the driver offsets (and the listening axis).
Tony.
What did the poor cat do wrong
?
I agree that some measuring gear would make the job faster combined with tuning by ear.
I like the process. It takes time and patience. A lot of reading and different online calcs.
I think I have learned a lot by doing it this way. Just putting numbers in a software may not have given me the practical experience of what will happen when components are swapped.
Im not trying to offend anyone - I have never used software and measuring gear. Surly its more complicated than what I just wrote.
Noobs should do a lot of reading up first. Then try, fail and experiment a lot before even thinking of putting a lot of money in gear. Remember school anyone? I have a ton of salvaged crossovers to that purpose.
?I agree that some measuring gear would make the job faster combined with tuning by ear.
I like the process. It takes time and patience. A lot of reading and different online calcs.
I think I have learned a lot by doing it this way. Just putting numbers in a software may not have given me the practical experience of what will happen when components are swapped.
Im not trying to offend anyone - I have never used software and measuring gear. Surly its more complicated than what I just wrote.
Noobs should do a lot of reading up first. Then try, fail and experiment a lot before even thinking of putting a lot of money in gear. Remember school anyone? I have a ton of salvaged crossovers to that purpose.
Manufacturers curves alone are not enough. You need to factor in your box alignment (for the woofers) and also baffle diffraction effects for all of the drivers.
Designing a crossover to suit a particular room may seem like a good idea, until you need to change to a different room
The engineering route is to design something that is neutral and the expectation is that the end user will eq to suit their room/tastes. Earl Geddes challenged me when I was wanting to pad my tweeters down to have a reduced treble (I'd designed the crossover using sims and it measured well but I didn't like it). He stated that I may just be used to the treble being subdued from years of listening to inaccurate speakers, and that I should listen to the (more accurate) speakers for a while without the additional attenuation. I did as he suggested, and after a while I got used to it. When I did try attenuating the treble it now no longer sounded as good. We are creatures of habit. If we always try to change something to how we like it, rather that trying something new for a while we may in fact be missing out on a lot.
Tony.
Even more interesting I think, would be to have a number of guys design a passive crossover by ear, and another lot design with measurements.
I've tried to design by ear in the past but would never go back. My experience of this is probably different to most.
A loudspeaker manufacterer wanted to design a speaker using a very simple crossover. He chose the drivers and asked me to measure the individual drivers and send him the results. He would then send me a crossover diagram and ask me to simulate it, build it and have a listen. This proccess went on for weeks and it drove me insane because i'd built a proper crossover that was far better (but obviously with a lot more parts) than anything he came up with but he wasn't interested because it was too complicated and over budget apparently. We parted ways in the end and the company disappeared a while later.
Of course crossovers can be done by ear only, but you get a very basic functionality crossover with this.
Just one example, even in an untreated room, the difference between a shallow reverse null and a deep reverse null can be easily heard (of course in normal mode, not in reverse mode) and if you don't have proper data whats wrong with the speaker (shallow reverse null) you may get a very chaotic speaker with sub-optimal performance with tune by ear.
But some likes tweaking by ear forever, that's sure.
Just my 0.5 cent.
Just one example, even in an untreated room, the difference between a shallow reverse null and a deep reverse null can be easily heard (of course in normal mode, not in reverse mode) and if you don't have proper data whats wrong with the speaker (shallow reverse null) you may get a very chaotic speaker with sub-optimal performance with tune by ear.
But some likes tweaking by ear forever, that's sure
.Just my 0.5 cent.
Last edited:
OK so I started working on something for this pretty much straight away and it appears that when I copy a post it keeps the original date. So we now have a new post as post #3 Why simple crossovers, tuned by ear, don’t work I was hoping it would appear here in the next available position... As it would take a quite a bit of editing to create a new post. I'm hoping people will just go back and have a look at post 3
Tony.
Tony.