Using Boxsim of Visaton I have created a parallel RL and notch filter (attached) to flatten a speaker curve of a Fane speaker 12-250TC.
Since this is a full frequency speaker, it will be run without crossover direct from a power amp. So it makes sense to create an active version of the RL and notch filter and put it upstream of the amp.
However... this means that I should be able to actually calculate (i.e. understand) the passive filter parameters and also, how to design an active filter with the same response. For both I need help.
There may be another challenge: I assume that the filters calculated with Boxsim take the driver behavior on the filter into account (e.g. impedance). Yet an active filter would not be affected by the driver, so a straight copy may not be 100% correct. I assume some additional analysis or actual test measurements may be required to fine tune the electronic version.
I have seen some very sophisticated active filter designs with PCB's on this forum, so I am sure there are members that know exactly how to create a simple active version of my filters.
Could you point me in the right direction? Thanks, Jan
PS I also want to add a bass boost to the filter, but I don't want to over-ask 🙂
Since this is a full frequency speaker, it will be run without crossover direct from a power amp. So it makes sense to create an active version of the RL and notch filter and put it upstream of the amp.
However... this means that I should be able to actually calculate (i.e. understand) the passive filter parameters and also, how to design an active filter with the same response. For both I need help.
There may be another challenge: I assume that the filters calculated with Boxsim take the driver behavior on the filter into account (e.g. impedance). Yet an active filter would not be affected by the driver, so a straight copy may not be 100% correct. I assume some additional analysis or actual test measurements may be required to fine tune the electronic version.
I have seen some very sophisticated active filter designs with PCB's on this forum, so I am sure there are members that know exactly how to create a simple active version of my filters.
Could you point me in the right direction? Thanks, Jan
PS I also want to add a bass boost to the filter, but I don't want to over-ask 🙂
Attachments
I assume this FR is your own measurement. The bass peak is related to either room acoustics, or to your sealed box is too small. It is not the true driver response, which is shown in the factory published FR graph. You can still EQ it if necessary, but changing the box size, or reposition the speaker in the room might work better.
Putting it in a reflex box should boost the bass sufficiently, but if you do EQ the bass boost then a sealed box usually more appropriate. ymmv Bass boost is also needed to compensate for the baffle step cancellation.
http://trueaudio.com/st_diff1.htm
Usually 3dB BSC is sufficient.
You may be able to reduce some of the rising HF response by aiming the high frequency beam away from your listening position. If the beam is aimed at your ears and you apply EQ for flat on axis FR then the speaker will sound rolled off when your head moves outside the beam. FR outside the beam is less sensitive to head movement. But you'll never get flat FR to 20k from whizzer cone. Just try to keep high FR kinda flattish and tune it to your taste by ear.
Line level active filters:
Active Filters
#5, low pass shelving filter is what you need for both cutting the highs and also for boosting the bass. Use his formulas to determine your parts values. He also shows passive line level version worth a try to avoid the complexity of active power supply. You'll need separate filters for each job. If you use both, then you'll need an active buffer to power the filters. If you only use one for treble, then your preamp or signal source may have enough current to power the passive filter without a buffer.
You might also consider doing a passive, speaker level contour filter for the treble adjustment.
Contour Network Designer / Calculator
This is typically how it is handled, and being high frequency, the parts are not too large and expensive like with bass EQ adjustments. Use large gage air core inductor and quality film cap and it will sound very good indeed, but adjusting to taste is not as easy as with the line level filter, since the speaker level parts are far more expensive. You will want to be able to tweak this by ear. Both ways can sound very good.
Putting it in a reflex box should boost the bass sufficiently, but if you do EQ the bass boost then a sealed box usually more appropriate. ymmv Bass boost is also needed to compensate for the baffle step cancellation.
http://trueaudio.com/st_diff1.htm
Usually 3dB BSC is sufficient.
You may be able to reduce some of the rising HF response by aiming the high frequency beam away from your listening position. If the beam is aimed at your ears and you apply EQ for flat on axis FR then the speaker will sound rolled off when your head moves outside the beam. FR outside the beam is less sensitive to head movement. But you'll never get flat FR to 20k from whizzer cone. Just try to keep high FR kinda flattish and tune it to your taste by ear.
Line level active filters:
Active Filters
#5, low pass shelving filter is what you need for both cutting the highs and also for boosting the bass. Use his formulas to determine your parts values. He also shows passive line level version worth a try to avoid the complexity of active power supply. You'll need separate filters for each job. If you use both, then you'll need an active buffer to power the filters. If you only use one for treble, then your preamp or signal source may have enough current to power the passive filter without a buffer.
You might also consider doing a passive, speaker level contour filter for the treble adjustment.
Contour Network Designer / Calculator
This is typically how it is handled, and being high frequency, the parts are not too large and expensive like with bass EQ adjustments. Use large gage air core inductor and quality film cap and it will sound very good indeed, but adjusting to taste is not as easy as with the line level filter, since the speaker level parts are far more expensive. You will want to be able to tweak this by ear. Both ways can sound very good.
Thanks Richidoo, that's great.
1) You are very right. What you see is the response of the speaker, the filter correction and a 75 ltr closed box as calculated by Boxsim. The box is too small indeed, it should be more like 150 ltr, but the idea is to have an active movable full range speaker for use outdoors.
2) BR would also need a much bigger enclosure which makes it unpractical for my purpose. Hence the proposed eq boost.
3) True. HF from this 12" speaker will be quite directional which is not too bad in my case as it allows to target the audience rather than the surroundings (although the bass will go its own way).
4) Will try.
5) Thanks. I checked that section, but shelving filters are different from RL and parallel notch filters, so I think there is no direct translation possible. It would be a different solution to the problem as far as I can tell.
6) I tried that particular site and method, but I did not get good results with it.
Last edited:
Hi, i think, that thou have virtually over-EQed by 3dB, Jdelange. Reduce resistor and possibly first choke value, heck thou might even cut the notch altogether and only use shelving. Such a simple EQ is best done passively, because that raises source impedance and so reduces distortion and lifts reliability (granted using good well-fitted choke and resistor, which will not burn). The bass extension must be done actively of course. Just my two cents, Uli
@Uli: interesting thoughts!
I tried both. Attached 1) a version with lowered RL-values to raise the high end part, and 2) the same but in addition the notch section removed.
Drawn lines are the Uli-versions, dashed is what I had before.
I think I want to keep the notch, but change of the RL values makes sense to me. Overall still quite bumpy though, 4-5 dB between peaks and valleys feels quite a lot to me.
The earlier design was an optimization done by the Boxsim program, so it appears there is still room human improvement!
I tried both. Attached 1) a version with lowered RL-values to raise the high end part, and 2) the same but in addition the notch section removed.
Drawn lines are the Uli-versions, dashed is what I had before.
I think I want to keep the notch, but change of the RL values makes sense to me. Overall still quite bumpy though, 4-5 dB between peaks and valleys feels quite a lot to me.
The earlier design was an optimization done by the Boxsim program, so it appears there is still room human improvement!
Attachments
Last edited:
Alternative
You've got me thinking... there are more ways to go about this. I replaced the notch filter in series with the speaker with two parallel correction filters. The overall frequency response is smoother, the impedance curve is flatter and there is less material in the signal path. I assume this fact is an improvement if the correction remains passive as suggested.
For comparison I stitched the previous response on top of the latest alternative with the corresponding hardware, which is rather small stuff even if it is a bit more.
You've got me thinking... there are more ways to go about this. I replaced the notch filter in series with the speaker with two parallel correction filters. The overall frequency response is smoother, the impedance curve is flatter and there is less material in the signal path. I assume this fact is an improvement if the correction remains passive as suggested.
For comparison I stitched the previous response on top of the latest alternative with the corresponding hardware, which is rather small stuff even if it is a bit more.
Attachments
Hopefully! +-2dB is not bad, and before thou decide on too much, better obtain say 30° off-axis plots, so thou see, where is plenty and where is shortage. Most fullrangers start to beam at the theoretical point (cone still radiating as piston, but its size approaching wavelength) but widen up later on, especially ones with whizzer cone. I would pull down level only above 1 KHz, keeping sensitivity at 97,5dB, at higher frequencies probably even higher (just on axis!).
Oh, it will beam for sure. It's a 12" speaker. From what I have read about full ranges like this (Hobby Hifi, you are most likely familiar with), there is a very small sweet spot. I'll have to live with that.
Fane doesn't provide off axis response, but I can guess.
One thing I did notice in the last version is that the impedance of the system drops to about 5 ohm. Does not worry me too much, but with an active version this would not occur.
Fane doesn't provide off axis response, but I can guess.
One thing I did notice in the last version is that the impedance of the system drops to about 5 ohm. Does not worry me too much, but with an active version this would not occur.
And i think, that laying loads too big in parallel to the driver is gonna impair the sound. Bernd Timmermanns of HobbyHiFi is rumoured for exact equalizations, caring very neatly for on-axis response but somehow draining life from it. Truth needs a messenger.
We'll see. So far it is all theory. Next up is the boost and the building. Then practice will tell.
You can
You assume correctly.
You can simulate your filter in Box sim, and it will simulate baffle effects etc.
If you have resigned the notch in the 'passive' section, then you will need to redesign using the 'active filter section.
I have done similar, but low pass or high pass filters.
I used the Q values from a filter table for LR filter of the intended order, and then slid the individual Fc for each filter section around until I evened out the summed response. In the end it was only minor tweaks away from what would be called Textbook alignments.
Then I cheated and saved myself some long winded math iterations and used this calculator
https://www.google.co.uk/url?sa=t&s...ggdMAA&usg=AFQjCNHjiXpHv8S1jJeF67DnHnR0T0ftxg
Then some more fiddling of values to get the Fc and Q to as close to what I desired.
In testing the active filter worked just as planned, and the electrical graph in Box sim, matched perfectly (or as close as possible) to my measured electrical output from each filter.
Job done.
That being said, it was still alot of work.
Good luck
You assume correctly.
You can simulate your filter in Box sim, and it will simulate baffle effects etc.
If you have resigned the notch in the 'passive' section, then you will need to redesign using the 'active filter section.
I have done similar, but low pass or high pass filters.
I used the Q values from a filter table for LR filter of the intended order, and then slid the individual Fc for each filter section around until I evened out the summed response. In the end it was only minor tweaks away from what would be called Textbook alignments.
Then I cheated and saved myself some long winded math iterations and used this calculator
https://www.google.co.uk/url?sa=t&s...ggdMAA&usg=AFQjCNHjiXpHv8S1jJeF67DnHnR0T0ftxg
Then some more fiddling of values to get the Fc and Q to as close to what I desired.
In testing the active filter worked just as planned, and the electrical graph in Box sim, matched perfectly (or as close as possible) to my measured electrical output from each filter.
Job done.
That being said, it was still alot of work.
Good luck
Thanks!! But...
1) As far as I can tell, the active section of Boxsim only allows for standard low pass and high pass filters (see input screen below).
2) So that's where it may work for you but not for me?
3) That's a great link. Bookmarked. But I do not recognize the type of filters I am using?
I assume that if I did, I could enter the values from Boxsim in the appropriate passive template (lot of other inputs required that look puzzling at first sight...), and then try to get an equivalent response from an active equivalent in the lower section?
It's never easy, is it?
Attachments
That was an excellent start that initially brought me here: Likwitz with a lot of basic theory, and then here: ESP Subwoofer equalizer They even provide a PCB!
(they have more fantastic stuff like standard PCB's for other type of filters)
You can....
Box's I'm has only LP and HP filters. That is true.
For a band pass you combine a LP and HP and set the 'distance in Fc by you desired bandwidth (more or less).
Q and Fc set the damping and turnover points. A table of alignment tables for Q and Fc factor will allow you to do this.
In much the same way you can build a notch filter, except that your bandwidth is negative. That is you set your LP Fc frequency higher than your HP Fc frequency (the opposite is true for a bandpass).
Box's I'm has only LP and HP filters. That is true.
For a band pass you combine a LP and HP and set the 'distance in Fc by you desired bandwidth (more or less).
Q and Fc set the damping and turnover points. A table of alignment tables for Q and Fc factor will allow you to do this.
In much the same way you can build a notch filter, except that your bandwidth is negative. That is you set your LP Fc frequency higher than your HP Fc frequency (the opposite is true for a bandpass).
As a test set the active filter LP and HP to second order, using the tick box.
Set Fc for both filters to the value you require, E.g. 2kHz. Set Q on both filters to 0.71 (just an arbitrary value I plucked out of the air....)
Try it 🙂
When I get home to my desktop I'll mock up an example if your interested - I'm fairly certain it's all possible in Box sim.
Set Fc for both filters to the value you require, E.g. 2kHz. Set Q on both filters to 0.71 (just an arbitrary value I plucked out of the air....)
Try it 🙂
When I get home to my desktop I'll mock up an example if your interested - I'm fairly certain it's all possible in Box sim.
Last edited:
I will try tomorrow. I can follow your reasoning.
But it's still challenging: I have 3 filters to model of which I only have values found by visually smoothing the speaker curve. Nothing else. So I need to understand what these values accomplish (frequency, bandwidth, Q...). Then I need to translate this into an active equivalent and then I need to feed this into an (upside-down) bandpass arrangement in Boxsim.
Yes, an example would be very welcome!
But it's still challenging: I have 3 filters to model of which I only have values found by visually smoothing the speaker curve. Nothing else. So I need to understand what these values accomplish (frequency, bandwidth, Q...). Then I need to translate this into an active equivalent and then I need to feed this into an (upside-down) bandpass arrangement in Boxsim.
Yes, an example would be very welcome!
Here is my way to small-signal ("active") HiFi:
* I use only passive low- and highpasses, and since i use no chokes, these filters are of first order only. If i need steeper slopes, i cascade filters. Voltage followers provide fresh low impedance if needed.
* I build negative resonances (notch filters) using a serial resistor and a serial resonance circuit to ground using a gyrator. A gyrator provides serial inductance and resistance to ground and needs two resistors, one of which is the serial resistance, one capacitor and one voltage follower.
* I use only passive low- and highpasses, and since i use no chokes, these filters are of first order only. If i need steeper slopes, i cascade filters. Voltage followers provide fresh low impedance if needed.
* I build negative resonances (notch filters) using a serial resistor and a serial resonance circuit to ground using a gyrator. A gyrator provides serial inductance and resistance to ground and needs two resistors, one of which is the serial resistance, one capacitor and one voltage follower.
Grasso has a good point
In my own experience of this type of work, on my only "decent" design I eliminated notches when making the transition from passive to active filtration.
The passive filter was a second order LP with notch, and second order HP. The usual phase inversion of tweeter was used to align phase.
In approximating the original filter in active components, upped the order to fourth, and eliminated the notch. Phase alignment sort of fell into place really. I ended up with fourth order LP plus a stage for BSC, and a fourth order HP. It saved me the need to try and make a active botch, which was a relief (and is also the reason I am certain it CAN be done in Box sim, but haven't actually done it)
Knowing what the passive filter sounded like, I set the target frequency response to be close as possible, with similar or better phase response.
Sounds easy perhaps, but wasn't that easy, but better than having to fiddle with making notches.
In my own experience of this type of work, on my only "decent" design I eliminated notches when making the transition from passive to active filtration.
The passive filter was a second order LP with notch, and second order HP. The usual phase inversion of tweeter was used to align phase.
In approximating the original filter in active components, upped the order to fourth, and eliminated the notch. Phase alignment sort of fell into place really. I ended up with fourth order LP plus a stage for BSC, and a fourth order HP. It saved me the need to try and make a active botch, which was a relief (and is also the reason I am certain it CAN be done in Box sim, but haven't actually done it)
Knowing what the passive filter sounded like, I set the target frequency response to be close as possible, with similar or better phase response.
Sounds easy perhaps, but wasn't that easy, but better than having to fiddle with making notches.
Hello Jan,
This 2007 DIYA thread might give you an impression of how to approach your issue. It will require a buffer and at least two gyrator circuits.
http://www.diyaudio.com/forums/multi-way/100525-active-notch-filter-design-needed.html
Let me know in case you run into problems in re-calculating the circuit to suit your particular requirements.
Do you have a .frd file of the Fane driver? That is absolutely necessary to feed any decent x/o simulator. The on-line component calculators are pretty useless, unfortunately. To my best of knowledge, none of them can calculate a decent gyrator.
Regards,
Eelco
This 2007 DIYA thread might give you an impression of how to approach your issue. It will require a buffer and at least two gyrator circuits.
http://www.diyaudio.com/forums/multi-way/100525-active-notch-filter-design-needed.html
Let me know in case you run into problems in re-calculating the circuit to suit your particular requirements.
Do you have a .frd file of the Fane driver? That is absolutely necessary to feed any decent x/o simulator. The on-line component calculators are pretty useless, unfortunately. To my best of knowledge, none of them can calculate a decent gyrator.
Regards,
Eelco
- Status
- Not open for further replies.