Hi all,
Well, I thought I had the right tool. A Behringer mic / mixer, external sound card,the Woofer tester, X-over Pro, LSPCad.
A bunch of caps, resistors and inductors.
A very solid enclosure for a tower MTM with Focal 5N421 mid/woofers and a Neo3 PDR tweeter.
When I built the crossover, either I had too much lows, or too little bass. When I finally managed to find an equilibrium, and tried to integrate the tweeter. I could never do it. Out of nowhere, there was a major (+10 dB) resonance from 3900 to 6000 Hertz.
The strange thing is that, when measuring the units separately, the FR plots look very smooth and nice... once I try to integrate them, all hell breaks loose.
I tried everything... removed the impedance eq, added it, added a notch filter to the tweeter.. soldered and resoldered everything...
Then, I checked out impedance plots. It almost drops to 1 ohm at the point where I have the tweeter resonance.
I simply don't understand what is going wrong. I'm fed up with my design.
I've decided to keep the cabinet and drivers, and start the crossover from scratch.
Considering it's an MTM design, I was wondering whether a 1st order crossover would be simpler to build.
And I was also wondering if someone could help me by pointing out a book (any book!) web page or whatever that can help me understand what went wrong this time. I know there's lots to learn about building a speaker, and I guess I need to study a lot this time in order to get things right.
Thanks for your help!!!!
Well, I thought I had the right tool. A Behringer mic / mixer, external sound card,the Woofer tester, X-over Pro, LSPCad.
A bunch of caps, resistors and inductors.
A very solid enclosure for a tower MTM with Focal 5N421 mid/woofers and a Neo3 PDR tweeter.
When I built the crossover, either I had too much lows, or too little bass. When I finally managed to find an equilibrium, and tried to integrate the tweeter. I could never do it. Out of nowhere, there was a major (+10 dB) resonance from 3900 to 6000 Hertz.
The strange thing is that, when measuring the units separately, the FR plots look very smooth and nice... once I try to integrate them, all hell breaks loose.
I tried everything... removed the impedance eq, added it, added a notch filter to the tweeter.. soldered and resoldered everything...
Then, I checked out impedance plots. It almost drops to 1 ohm at the point where I have the tweeter resonance.
I simply don't understand what is going wrong. I'm fed up with my design.
I've decided to keep the cabinet and drivers, and start the crossover from scratch.
Considering it's an MTM design, I was wondering whether a 1st order crossover would be simpler to build.
And I was also wondering if someone could help me by pointing out a book (any book!) web page or whatever that can help me understand what went wrong this time. I know there's lots to learn about building a speaker, and I guess I need to study a lot this time in order to get things right.
Thanks for your help!!!!
Don't call this project a failure too soon. Just because there is some work to be done.
Cover all the bases, if the drivers are ok and the cabinet is sturdy and well damped (as well as tuned to the right frequency, if it's a bassreflex). It is indeed time to look at the crossover.
I would try to build one gradually, and see what each step does. First a simple 1st order with perhaps only one of the two woofers. Then the other woofer, add some impedance correction.
I would start with a higher order filter for the tweeter if the crossover freq. is near the resonance f of the tweeter.
Do you have the schematic of the filter you have built?
Building loudspeakers (as any other equipment) is an iterative process, and especially if you can change a lot of things rather cheaply (i.e. the crossover), it will take quite a while before the "failure" status is appropriate.
You could also use an active crossover, then tuning is even easier (although you'll need an extra amp).
Regards,
Jarno.
Cover all the bases, if the drivers are ok and the cabinet is sturdy and well damped (as well as tuned to the right frequency, if it's a bassreflex). It is indeed time to look at the crossover.
I would try to build one gradually, and see what each step does. First a simple 1st order with perhaps only one of the two woofers. Then the other woofer, add some impedance correction.
I would start with a higher order filter for the tweeter if the crossover freq. is near the resonance f of the tweeter.
Do you have the schematic of the filter you have built?
Building loudspeakers (as any other equipment) is an iterative process, and especially if you can change a lot of things rather cheaply (i.e. the crossover), it will take quite a while before the "failure" status is appropriate.
You could also use an active crossover, then tuning is even easier (although you'll need an extra amp).
Regards,
Jarno.
Hi fjhuerta,
You have three times the test gear I have but you lack something. I don't say this to p**s you off.
Lets start with the MTM design, are the two midwoofers 4 or 8 ohms? If they are 4 ohms you will end up with +-2 ohms load on amp, not a good thing for the amp.
Startup LSPcad and setup midwoofer and tweeter into driver database. Measure tweeter center to midwoofer center to find the rough xo point. Midwoofer Xover=13,500/ctc"= ????hz maximum. Tweeter Xover is about 2 octaves above Fs.
Start with those numbers in LSPcad. Get the summed frequency response flat. Now you are nearly there! Reverse the tweeter polarity while watching the summed frequency response. You should see a very deep (and narrow) 20db dip at xo frequency. You are done with LSPcad when you get this and flat response when polarity is back.
You have three times the test gear I have but you lack something. I don't say this to p**s you off.
Lets start with the MTM design, are the two midwoofers 4 or 8 ohms? If they are 4 ohms you will end up with +-2 ohms load on amp, not a good thing for the amp.
Startup LSPcad and setup midwoofer and tweeter into driver database. Measure tweeter center to midwoofer center to find the rough xo point. Midwoofer Xover=13,500/ctc"= ????hz maximum. Tweeter Xover is about 2 octaves above Fs.
Start with those numbers in LSPcad. Get the summed frequency response flat. Now you are nearly there! Reverse the tweeter polarity while watching the summed frequency response. You should see a very deep (and narrow) 20db dip at xo frequency. You are done with LSPcad when you get this and flat response when polarity is back.
Maybe you have something that's connected different from what you designed. Getting 1 ohm impedance is very rare, I'm surprised that your card/amp can drive it.
Another possibility is that your test setup is not properly calibrated.
Another possibility is that your test setup is not properly calibrated.
Are you measuring complex impedance (magnitude and phase) and accounting for that in the crossover calculations?
Are you measuring driver acoustic center offset and accounting for that in the crossover calculations?
Have you measured the electrical signal at each of the drivers and compared that to the prediction from the crossover calculations?
Are you measuring driver acoustic center offset and accounting for that in the crossover calculations?
Have you measured the electrical signal at each of the drivers and compared that to the prediction from the crossover calculations?
Don't Panic!
First off, it's not a failure - you will probably (by neccessity) learn more about your crossover design by having to debug it that your would have if everything had gone together perfectly the first time. Look at it as a learning experience 🙂
I've been away a while, but Dickinson's Loudspeaker Design Cookbook was always considered an excellent place to start understanding.
Also it would be helpful if you have a schematic of your crossover that you could make available on-line. Input impedence droping to 1 ohm doesn't sound good (both figuratively as well as literally).
Make sure you double checked your wiring 🙂
-bill
First off, it's not a failure - you will probably (by neccessity) learn more about your crossover design by having to debug it that your would have if everything had gone together perfectly the first time. Look at it as a learning experience 🙂
I've been away a while, but Dickinson's Loudspeaker Design Cookbook was always considered an excellent place to start understanding.
Also it would be helpful if you have a schematic of your crossover that you could make available on-line. Input impedence droping to 1 ohm doesn't sound good (both figuratively as well as literally).
Make sure you double checked your wiring 🙂
-bill
Fjuerta, dont give up, you ll fix it before you know it.
It seems that you have a phase problem. First you need to make sure that you measure correctly. Re-calibrate your system and measure the drivers again. You should measure both drivers mounted on
the final box without moving either the box or the mic between the measurements. Then you should make sure that the measured data that will be used at
the crossover simulator have not been manipulated differently (for the phase to be accurate). For example the widnowing of your FFT should be exaclty the same for all drivers.
If your phase data is not accurate, you cant simulate the crossover succesfully. Then you can simulate and do what mrlots2do said about reversing the tweeters polarity. If you get
a large dip, then your drivers are in phase (when you restore the polarity back to normal).
About the 1ohm measurement you get, try this: measure a 10 ohm / 2 watt , 1% tolerance resistor as a driver's impedance. Do you get a flat impedance graph at 10 ohms? If you dont, you need to
calibrate again, before trusting your impedance measurements.
If those focals are 4ohm, you get a 2ohm in parallel, which is not good for your amplifier. Either you should use just one of them, or connect them in series, which is not that good for various reasons.
If your focals are 8ohms, you have a 4 ohm combination, which means you need a low pass with about 1.7 - 1.8mH inductor for enough baffle step correction, (assuming you do it 2nd order) and a capacitor appropriate to cut at the desired xover freq. Then when you ll shape your woofer's response as desired, you ll build your hi pass to match your tweeter's response with the woofer's (making sure you always reverse the tweeter to check the phase matching too). Your bass "quantity" depends on the inductor's value. When designing your low pass filter, the cutting frequency is not enough, you also need to take into consideration the bafflestep correction you need (= a specific inductor's value, higher than the standard filters) and you ll adjust the capacitor of the low pass keeping the inductor constant, to get the desired x -frequency for your low pass. Then you go for your hi pass. Anyway start with calibrating and checking your phase data for accuracy and then its up to you to do whatever you like with the simulator.
It seems that you have a phase problem. First you need to make sure that you measure correctly. Re-calibrate your system and measure the drivers again. You should measure both drivers mounted on
the final box without moving either the box or the mic between the measurements. Then you should make sure that the measured data that will be used at
the crossover simulator have not been manipulated differently (for the phase to be accurate). For example the widnowing of your FFT should be exaclty the same for all drivers.
If your phase data is not accurate, you cant simulate the crossover succesfully. Then you can simulate and do what mrlots2do said about reversing the tweeters polarity. If you get
a large dip, then your drivers are in phase (when you restore the polarity back to normal).
About the 1ohm measurement you get, try this: measure a 10 ohm / 2 watt , 1% tolerance resistor as a driver's impedance. Do you get a flat impedance graph at 10 ohms? If you dont, you need to
calibrate again, before trusting your impedance measurements.
If those focals are 4ohm, you get a 2ohm in parallel, which is not good for your amplifier. Either you should use just one of them, or connect them in series, which is not that good for various reasons.
If your focals are 8ohms, you have a 4 ohm combination, which means you need a low pass with about 1.7 - 1.8mH inductor for enough baffle step correction, (assuming you do it 2nd order) and a capacitor appropriate to cut at the desired xover freq. Then when you ll shape your woofer's response as desired, you ll build your hi pass to match your tweeter's response with the woofer's (making sure you always reverse the tweeter to check the phase matching too). Your bass "quantity" depends on the inductor's value. When designing your low pass filter, the cutting frequency is not enough, you also need to take into consideration the bafflestep correction you need (= a specific inductor's value, higher than the standard filters) and you ll adjust the capacitor of the low pass keeping the inductor constant, to get the desired x -frequency for your low pass. Then you go for your hi pass. Anyway start with calibrating and checking your phase data for accuracy and then its up to you to do whatever you like with the simulator.
Just noticed that you have EXTERNAL sound card... This a major issue since most external sound cards (as many internal) have a latency problem.
This means that the measurement signal is not given to the drivers at exactly the same time for every measurement. So if your tweeter's measurement has different latency than your woofer's ( the MLS signal is not sent to the driver at exactly the same time ) your phase data is useless. You need to find out a way to compare the reference signal timing between your measurements, so you can get two valid measurements for both drivers (phase comparable). If you wont do that, you ll never simulate correctly.
This means that the measurement signal is not given to the drivers at exactly the same time for every measurement. So if your tweeter's measurement has different latency than your woofer's ( the MLS signal is not sent to the driver at exactly the same time ) your phase data is useless. You need to find out a way to compare the reference signal timing between your measurements, so you can get two valid measurements for both drivers (phase comparable). If you wont do that, you ll never simulate correctly.
First of all, I'm sorry about not answering to this post sooner. I had a hectic week at work, and I really wanted to thank you all for your support and your great posts as it should be done...
Jarno,
I suppose I tried to do too many things the first time around 🙂 Considering I had no previous experience, I guess going to an MTM with BSC, impedance EQ and notch filters was a mistake. But I do agree - I'll begin by taking things simple this time. I'll attach a diagram of my crossover design.
Frank, I tried reversing the phase. It made things even worse! There was now a -4 dB irregularity before the +5 dB hump. I'm pretty sure my problems lie in the crossover design.
Mrlotstodo,
No offense taken - at all! It just goes to show you that equipment does not make a good speaker. That's why I'm here - to learn!
The woofers are actually 2.9 Ohm units. I wired them in series.
I think my error was not using LSPCad (I merely got it for JustMLS) and using X-Over pro. I think that program is way too simplified to be of any use. But your post was a lot more helpful than the help files for LSPCad! Now I think I'll devote a lot more time to trying to use that program instead of X-Over pro.
Soongsc, I made sure everything was OK... I double checked the circuit. I think using low impedance woofers and the fact that something was wrong with the impedance EQ I used had a lot to do with my results. Yeah, I have a 1 Ohm impedance result at around 7,000 Hz (my crossover is centered at 3,000 Hz). I was using a big SS amp to test the speakers, so maybe that's the reason it could handle the load.
Sy,
I guess it's "Three strikes and you are out!". 🙂 No, I didn't do any of the above. Thanks a lot for the hints, I really appreciate them. As I said before, all I did was use X-Over pro and think that would be enough. Now I see I was barely scratching the surface of what can be done.
Bill,
I'll upload the schematic ITT. I was very frustrated when I made my post, but now I'm ready to take things easy and start from scratch. I'll definitely be getting Vance Dickinson's book - I know I need it.
Jomor,
THANKS! It's all becoming clear now... I always wondered wat on earth should I do with phase measurements... and I slowly became aware that I could be having phase issues.
One other thing I noticed - the tweeter had a bump in its response. I wonder if its because of the baffle its mounted on, or if the tweeter has issues on that range. I need to check that out.
Jomor, I calibrated my sound card using JustMLS, and apparently, it doesn't have any latency problems. I wonder... is this enough? I also have a PC with a Santa Cruz soundcard. If necessary, I can use it to measure the speakers, too.
Thanks all. I really, really, really appreciate your comments. Now I'm beginning to really understand where I failed and what can I do about it! 🙂
Jarno,
I suppose I tried to do too many things the first time around 🙂 Considering I had no previous experience, I guess going to an MTM with BSC, impedance EQ and notch filters was a mistake. But I do agree - I'll begin by taking things simple this time. I'll attach a diagram of my crossover design.
Frank, I tried reversing the phase. It made things even worse! There was now a -4 dB irregularity before the +5 dB hump. I'm pretty sure my problems lie in the crossover design.
Mrlotstodo,
No offense taken - at all! It just goes to show you that equipment does not make a good speaker. That's why I'm here - to learn!
The woofers are actually 2.9 Ohm units. I wired them in series.
I think my error was not using LSPCad (I merely got it for JustMLS) and using X-Over pro. I think that program is way too simplified to be of any use. But your post was a lot more helpful than the help files for LSPCad! Now I think I'll devote a lot more time to trying to use that program instead of X-Over pro.
Soongsc, I made sure everything was OK... I double checked the circuit. I think using low impedance woofers and the fact that something was wrong with the impedance EQ I used had a lot to do with my results. Yeah, I have a 1 Ohm impedance result at around 7,000 Hz (my crossover is centered at 3,000 Hz). I was using a big SS amp to test the speakers, so maybe that's the reason it could handle the load.
Sy,
I guess it's "Three strikes and you are out!". 🙂 No, I didn't do any of the above. Thanks a lot for the hints, I really appreciate them. As I said before, all I did was use X-Over pro and think that would be enough. Now I see I was barely scratching the surface of what can be done.
Bill,
I'll upload the schematic ITT. I was very frustrated when I made my post, but now I'm ready to take things easy and start from scratch. I'll definitely be getting Vance Dickinson's book - I know I need it.
Jomor,
THANKS! It's all becoming clear now... I always wondered wat on earth should I do with phase measurements... and I slowly became aware that I could be having phase issues.
One other thing I noticed - the tweeter had a bump in its response. I wonder if its because of the baffle its mounted on, or if the tweeter has issues on that range. I need to check that out.
Jomor, I calibrated my sound card using JustMLS, and apparently, it doesn't have any latency problems. I wonder... is this enough? I also have a PC with a Santa Cruz soundcard. If necessary, I can use it to measure the speakers, too.
Thanks all. I really, really, really appreciate your comments. Now I'm beginning to really understand where I failed and what can I do about it! 🙂
- Status
- Not open for further replies.