I'm in the process of building a fairly simple pair of speakers using Vifa P17WJ-08-00 and D27TG-35-06 drive units. The enclosures are designed and semi-built as vented boxes tuned at 45Hz, but now I'm getting onto the crossover design and could do with the benefit of a little of somebody elses experience.
I'm designing the crossover as a 4th order L-R and have put the circuit into PSPICE to model it. The questions I have go like this :
1. I've modelled the LF unit as its Re and Le values to account for its inductance, but I've only modelled the HF unit as a resistor because I believe the inductance will be pretty much negligible. Is this a fair assumption?
2. As a result of putting the driver inductance in, there'a fair lump in the low pass filter response both above and below the crossover frequency. I've put a parallel resistor/capacitor across the drive unit to compensate for this which has improved matters, but it seems to be slightly attenuated just before the crossover frequency. Have I done something wrong here?
3. Is it also worth putting in the resonant LRC section in to compensate for the drive unit resonance?
4. There's going to be quite a few components in this crossover (which I'm not sure is a good thing). What's the best way to build this, as in how to mount the components? Do I need to do anything to stop the vibrations in the box shaking the solder joints apart?
Thanks in advance for you help.
I'm designing the crossover as a 4th order L-R and have put the circuit into PSPICE to model it. The questions I have go like this :
1. I've modelled the LF unit as its Re and Le values to account for its inductance, but I've only modelled the HF unit as a resistor because I believe the inductance will be pretty much negligible. Is this a fair assumption?
2. As a result of putting the driver inductance in, there'a fair lump in the low pass filter response both above and below the crossover frequency. I've put a parallel resistor/capacitor across the drive unit to compensate for this which has improved matters, but it seems to be slightly attenuated just before the crossover frequency. Have I done something wrong here?
3. Is it also worth putting in the resonant LRC section in to compensate for the drive unit resonance?
4. There's going to be quite a few components in this crossover (which I'm not sure is a good thing). What's the best way to build this, as in how to mount the components? Do I need to do anything to stop the vibrations in the box shaking the solder joints apart?
Thanks in advance for you help.
you wrote...
As a result of putting the driver inductance in, there'a fair lump in the low pass filter response both above and below the crossover frequency. I've put a parallel resistor/capacitor across the drive unit to compensate for this which has improved matters, but it seems to be slightly attenuated just before the crossover frequency. Have I done something wrong here?
reply...
the RC network compensates for impedance but also acts like a filter. Try reducing C or increasing R. I have found that playing around works.
you wrote...
3. Is it also worth putting in the resonant LRC section in to compensate for the drive unit resonance?
reply...
No need unless you have a unusually high peak. I use a SS 9900 tweeter. most people recomend a LRC trap for the 30+ohm peak at 500hz or so but I found it is not needed.
you wrote...
4. There's going to be quite a few components in this crossover (which I'm not sure is a good thing). What's the best way to build this, as in how to mount the components? Do I need to do anything to stop the vibrations in the box shaking the solder joints apart?
reply...
try and mount XO on fiberboard (what they make PCB out of). I use 6 mm MDF becuase that is easily available to me.
keep the XO out of the speaker box. if you can do that (WAF rules) then try and isolate the XO within the speaker box.
after XO is final pour epoxy cement over components to prvent them from vibratig. In case you are not sure you canuse silicone glue too.
As a result of putting the driver inductance in, there'a fair lump in the low pass filter response both above and below the crossover frequency. I've put a parallel resistor/capacitor across the drive unit to compensate for this which has improved matters, but it seems to be slightly attenuated just before the crossover frequency. Have I done something wrong here?
reply...
the RC network compensates for impedance but also acts like a filter. Try reducing C or increasing R. I have found that playing around works.
you wrote...
3. Is it also worth putting in the resonant LRC section in to compensate for the drive unit resonance?
reply...
No need unless you have a unusually high peak. I use a SS 9900 tweeter. most people recomend a LRC trap for the 30+ohm peak at 500hz or so but I found it is not needed.
you wrote...
4. There's going to be quite a few components in this crossover (which I'm not sure is a good thing). What's the best way to build this, as in how to mount the components? Do I need to do anything to stop the vibrations in the box shaking the solder joints apart?
reply...
try and mount XO on fiberboard (what they make PCB out of). I use 6 mm MDF becuase that is easily available to me.
keep the XO out of the speaker box. if you can do that (WAF rules) then try and isolate the XO within the speaker box.
after XO is final pour epoxy cement over components to prvent them from vibratig. In case you are not sure you canuse silicone glue too.
2 comments...
Since both the drivers you are using are fairly wide range why dont you try series XOs?
Try and build the box so you can block the port and reduce box volume (I use bags of sand to do that) so youc an expreiment with sealed as well as ported bass. Also if you ever want to *** a sub a sealed system might integrate better.
just my thoughts.
cheers.
Since both the drivers you are using are fairly wide range why dont you try series XOs?
Try and build the box so you can block the port and reduce box volume (I use bags of sand to do that) so youc an expreiment with sealed as well as ported bass. Also if you ever want to *** a sub a sealed system might integrate better.
just my thoughts.
cheers.
I had thought about series xovers, but decided against them mostly because (I think, although I may be wrong) it's a bit harder to get a steep rolloff than with parallel filters. If anyone can point me in the direction of info on a series filter design with 24dB rolloff I'd be very interested...
I'd not mentioned the crossover frequency because it always seems to be such a controversial subject and everybody has their own preferences / rules. For the record, I'm crossing over at around 1800Hz. Yes, it's quite low, hence why I want a fairly steep rolloff rate to avoid rattling the tweeter about too much. This may or may not work well, but at the moment, I'm much more concerned with trying to make the LF filter response look a bit more reasonable - it's improved with the RC in parallel with the LF driver though.
I'd not mentioned the crossover frequency because it always seems to be such a controversial subject and everybody has their own preferences / rules. For the record, I'm crossing over at around 1800Hz. Yes, it's quite low, hence why I want a fairly steep rolloff rate to avoid rattling the tweeter about too much. This may or may not work well, but at the moment, I'm much more concerned with trying to make the LF filter response look a bit more reasonable - it's improved with the RC in parallel with the LF driver though.
Redeye, IME impedence compensating the woofer extends and cleans the driver response in highs and lows.
Compensating the tweeter allows smooth response past hearing range and for both drivers much better depth imaging.
The seriers LRC network should enable cleaner LF response from the tweeter, and enable you to cross lower down.
Eric.
Compensating the tweeter allows smooth response past hearing range and for both drivers much better depth imaging.
The seriers LRC network should enable cleaner LF response from the tweeter, and enable you to cross lower down.
Eric.
you wrote....but decided against them mostly because (I think, although I may be wrong) it's a bit harder to get a steep rolloff than with parallel filters
reply...
any specific need for a steep roll off. yes series XOs (what I know of them) have shallow rolls ofs unless you want a very forward sounding speaker. but since your drivers are wide range a series XO at 3k would be fine.
at 1.8k you would need a 18db+ HP for the tweeter to prevent serious damage to the tweeter.
cheers.
reply...
any specific need for a steep roll off. yes series XOs (what I know of them) have shallow rolls ofs unless you want a very forward sounding speaker. but since your drivers are wide range a series XO at 3k would be fine.
at 1.8k you would need a 18db+ HP for the tweeter to prevent serious damage to the tweeter.
cheers.
I've added the driver resonance of the woofer into the crossover model, and it doesn't do anything too scary. But, I can't find the data I need to model the resonance for the tweeter (a Vifa D27TG-35-06). Does anyone know where I can get this data from or am I going to have to measure it? Is it actually worth doing this modelling for the tweeter too or are the values so small as to be negligible?
I've built a 4-th order l/r myself and i think you've made a good choice! i actually have a 4-th l/r active between woofers and mids at 500Hz and the same in passive between mids and tweeters at about 3KHz.
1.8KHz isn't necessarily too low, but that will depend on whatever the tweeter's resonance is. ideally with the x-over freq that low, you'd want a pretty low resonance tweeter. i've read in a few places that you really should cross them so that the freq is as much higher than the tweeter resonance as you can stand, at the very least an octave above. that would mean your tweeters would have to be Fs at 900Hz or less. unlikely but possible! the fact that you have a 24dB/octave slope will help things a great deal though!
the attenuation may just be the zobel doing it's job, and with it in place you no longer have the typical response rise as the driver's impedance goes wacky. do you mean that you think you have a dip around the x-over freq that starts below it, or is it just that the driver is producing less at that point now (this is probably what it should be doing btw).
another possibilty is that the value for your resistor in the zobel network is too high. a general estimation for the value is about 3/4 the nominal impedance of the driver, e.g. for an 8ohm driver you'd want to use a 6-7ohm resistor with the cap (which should be a relatively high value, somewhere in the 30-50uF range i think).
one important thing apart from what was already covered is that you should try as much as possible to keep the inductors as far away from each other as you possibly can, and also align the coils so that they are offset by 90 degrees to each other if they are close. i have used those sheets of prototyping circuit board that have grids of holes, and i secured the coils by tieing them with heavy gauge fishing line, through the holes and then through the middle of the coil. it works pretty well, and then if you later decide (and you will... believe me !) to tweak or change anything it isn't impossible to remove them like it can be if you use epoxy or varnish to cover them.
best of luck - i reckon you'll be happy with it when you get it sussed - i was!
1.8KHz isn't necessarily too low, but that will depend on whatever the tweeter's resonance is. ideally with the x-over freq that low, you'd want a pretty low resonance tweeter. i've read in a few places that you really should cross them so that the freq is as much higher than the tweeter resonance as you can stand, at the very least an octave above. that would mean your tweeters would have to be Fs at 900Hz or less. unlikely but possible! the fact that you have a 24dB/octave slope will help things a great deal though!
the attenuation may just be the zobel doing it's job, and with it in place you no longer have the typical response rise as the driver's impedance goes wacky. do you mean that you think you have a dip around the x-over freq that starts below it, or is it just that the driver is producing less at that point now (this is probably what it should be doing btw).
another possibilty is that the value for your resistor in the zobel network is too high. a general estimation for the value is about 3/4 the nominal impedance of the driver, e.g. for an 8ohm driver you'd want to use a 6-7ohm resistor with the cap (which should be a relatively high value, somewhere in the 30-50uF range i think).
one important thing apart from what was already covered is that you should try as much as possible to keep the inductors as far away from each other as you possibly can, and also align the coils so that they are offset by 90 degrees to each other if they are close. i have used those sheets of prototyping circuit board that have grids of holes, and i secured the coils by tieing them with heavy gauge fishing line, through the holes and then through the middle of the coil. it works pretty well, and then if you later decide (and you will... believe me !) to tweak or change anything it isn't impossible to remove them like it can be if you use epoxy or varnish to cover them.
best of luck - i reckon you'll be happy with it when you get it sussed - i was!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.