I've been building speakers for some years but never attempted to design the filter myself. I want a small bookshelf speaker about 5 liters. I have done some simulation with boxsim, and this is the result.
Does it look OK?
Used frequency and impedance curves from zaphaudio.com
I am also considering buying some measurement tools, and have been looking at the Dayton Audio OmniMic V2 system.
Any advice appreciated !
Does it look OK?
Used frequency and impedance curves from zaphaudio.com
I am also considering buying some measurement tools, and have been looking at the Dayton Audio OmniMic V2 system.
Any advice appreciated !
Last edited:
Can't place the woofer at all. Where is a link? 😕
This is what you typically do with 5" bass: SEAS 5INCH
This is what you typically do with 5" bass: SEAS 5INCH
It some sort of oem driver i think. Here's a link from zaphaudio
http://www.zaphaudio.com/tidbits/Vifa-XT15-FR.gif
Zaph|Audio
Tweeter
https://www.parts-express.com/pedocs/specs/264-1028-vifa-bc25sc06-04-specifications-46535.pdf
http://www.zaphaudio.com/tidbits/Vifa-XT15-FR.gif
Zaph|Audio
Tweeter
https://www.parts-express.com/pedocs/specs/264-1028-vifa-bc25sc06-04-specifications-46535.pdf
Hmm... a WOODEN driver? 😕
Guess you'd treat it like the paper SEAS CA15RLY version in that Troels Gravesen link I gave you. Add some resistance to the shunt arm of the bass to reduce that bump at crossover. I'd try something like 1mH, 10uF and 3.3 ohms for the bass. Third order filter to suit. Crossover around 3kHz. Much bigger box though. Try 12L. Anything less would be silly.
You can strip out Troels' LCR correction and the RC Zobel on the tweeter probably. Those are driver specific. I usually set the bass filter, and let Boxsim optimise the treble. But you can usually use much smaller capacitors than Boxsim comes up with. 4uF, 0.2mH and 10uF would probably be about right.
Guess you'd treat it like the paper SEAS CA15RLY version in that Troels Gravesen link I gave you. Add some resistance to the shunt arm of the bass to reduce that bump at crossover. I'd try something like 1mH, 10uF and 3.3 ohms for the bass. Third order filter to suit. Crossover around 3kHz. Much bigger box though. Try 12L. Anything less would be silly.
You can strip out Troels' LCR correction and the RC Zobel on the tweeter probably. Those are driver specific. I usually set the bass filter, and let Boxsim optimise the treble. But you can usually use much smaller capacitors than Boxsim comes up with. 4uF, 0.2mH and 10uF would probably be about right.
I'd forgotten just how hard these 5" drivers can be if you don't do a low crossover! 🙂
Looks like some sort of bass notch around 1.5kHz is essential to stop the midrange rising badly.
This was a sim I did of a KEF R101 mini with a 1.4kHz LCR trap. It's an alternative to Troels' notch for 2.8kHz crossover. You'll want entirely different values, I expect. Thought it might help anyway. It's very flat and phase and impedance is good.
Sort of standard circuit for BBC type mini speakers. 😎
Looks like some sort of bass notch around 1.5kHz is essential to stop the midrange rising badly.
This was a sim I did of a KEF R101 mini with a 1.4kHz LCR trap. It's an alternative to Troels' notch for 2.8kHz crossover. You'll want entirely different values, I expect. Thought it might help anyway. It's very flat and phase and impedance is good.
Sort of standard circuit for BBC type mini speakers. 😎
Attachments
Thanks! Just recieved my omnimic, so I will build the boxes and measure the drivers in the box. I also have some vifa bc25tg15-04 i might use instead of the bc25sc06-04. The boxes will be around 5 liters because they will need to "fit" in the kitchen
Update: This is the first speaker that i will design the crossover by myself. I have read a lot about the topic, but the thing I have have the hardest time understanding is the phase, and phase tracking? of the drivers.
If someone could explain this to my in simple terms, please do!
Anyway I have spent some time measuring the drivers in the box they will sit in with omnimic, and some more simulations. These are not from my own frd files but simulating with the graphs from my own measurements comes very close. (I still need to learn how to make good/correct measurements)
So here is the new crossover:
Comments appreciated.
If someone could explain this to my in simple terms, please do!
Anyway I have spent some time measuring the drivers in the box they will sit in with omnimic, and some more simulations. These are not from my own frd files but simulating with the graphs from my own measurements comes very close. (I still need to learn how to make good/correct measurements)
So here is the new crossover:
Comments appreciated.
Attachments
Last edited:
Vifa XT15:
Qts 0.53
Vas 9.96 L
Le 0.64
Re 5.5
http://www.zaphaudio.com/tidbits/Vifa-XT15-TS.gif
BC25SC06-04
Re 2.9
The driver seems to have a resonance at around 2kHz which you must notch with a parallel LCR trap if you are to avoid the too low crossover point. I would try 0.5mH and 10uF and 10R. You may be able to crossover around 3kHz then.
Bass Coil around 1.5mH is reasonable. Maybe 2R + 6.8uF shunt, but that is a wild guess.
The tweeter is hugely efficient and low Re. I would put at least 3.3R is series with it, maybe more in the course of simming it.
You have an impedance dip at 4kHz which says the tweeter filter is underdamped. Impedance is also dipping too low at the very top. You therefore need smaller capacitors (say 3.3uF) and perhaps a slightly bigger coil. I would put a resistor in front of the tweeter filter too. You really ought to get the high end impedance amp friendly and above 10-15 ohms that way.
Phase should be aligned at crossover and in the region where the two drivers are both contributing significantly. Say to the -12dB point of the weaker driver. I show a reasonable effort below with 3kHz XO. Troels' tweeter filter looks about right to me. I can't say how you'll get on with the bass, which probably needs something different.
It's quite a good strategy to get a reasonable flat curve on the 5" bass first, with a -6dB rolloff around 3-3.5kHz, then just tie the treble filter in with it. By the nature of the 5" driver, you'll have a cone breakup peak at around 7-8kHz too, they all do that.
Qts 0.53
Vas 9.96 L
Le 0.64
Re 5.5
http://www.zaphaudio.com/tidbits/Vifa-XT15-TS.gif
BC25SC06-04
Re 2.9
The driver seems to have a resonance at around 2kHz which you must notch with a parallel LCR trap if you are to avoid the too low crossover point. I would try 0.5mH and 10uF and 10R. You may be able to crossover around 3kHz then.
Bass Coil around 1.5mH is reasonable. Maybe 2R + 6.8uF shunt, but that is a wild guess.
The tweeter is hugely efficient and low Re. I would put at least 3.3R is series with it, maybe more in the course of simming it.
You have an impedance dip at 4kHz which says the tweeter filter is underdamped. Impedance is also dipping too low at the very top. You therefore need smaller capacitors (say 3.3uF) and perhaps a slightly bigger coil. I would put a resistor in front of the tweeter filter too. You really ought to get the high end impedance amp friendly and above 10-15 ohms that way.
Phase should be aligned at crossover and in the region where the two drivers are both contributing significantly. Say to the -12dB point of the weaker driver. I show a reasonable effort below with 3kHz XO. Troels' tweeter filter looks about right to me. I can't say how you'll get on with the bass, which probably needs something different.
It's quite a good strategy to get a reasonable flat curve on the 5" bass first, with a -6dB rolloff around 3-3.5kHz, then just tie the treble filter in with it. By the nature of the 5" driver, you'll have a cone breakup peak at around 7-8kHz too, they all do that.
Attachments
Last edited:
I should mention that if you are getting nowhere on phase alignment, it is worth flipping the waveguide tweeter polarity and trying different slopes. I think, if I remember right, 24dB per octave symettrical LR4 slopes usually line up on phase. 18dB/octave 3rd order slopes may line up with tweeter negative polarity.
You seem to be getting asymettric slopes, which can be done, but gets complicated sometimes. Frankly, sometimes I give up and conclude these drivers don't play together nicely. 🙂
You seem to be getting asymettric slopes, which can be done, but gets complicated sometimes. Frankly, sometimes I give up and conclude these drivers don't play together nicely. 🙂
I think you did exactly the right modifications to improve your design significantly, congratulation.😎
The additional RC in the woofer filter is urgently needed to customize the rolloff of the low-pass filter. The peak at 1.8 kHz is completely gone now and the resonances at 5 and 8 kHz are better attenuated as well.
The change of tweeter polarity and the third order tweeter filter are necessary to achieve proper phase matching and a flat frequency response. In your first version, woofer and tweeter actually were working against each other below the crossover frequency instead of supporting themselves, that was not good.
I do not agree with System7's idea to notch the woofer resonance at 1.8 kHz and to aim for a higher crossover frequency at 3 kHz. Even if that was doable (I doubt), the effort would be high and the attenuation of the resonances at 5 and 8 kHz would suffer. It's much simpler to accept the natural woofer rolloff at 2 kHz and to choose a tweeter which can go low enough. You did that correctly by choosing the Vifa BC25TG15-04.
- Status
- Not open for further replies.