To use notch filter, or not?

[Calleberg]

I have seen measurements where a passive filter component (a coil) reduces THD with about 5-8dB. Even after the slope caused by said component is EQ:ued back to flat.
So maybe what Plcamp is saying has some bearing but maybe not related so much to the notch but instead the fact that there are component(s) in series with the drive unit changing the drive impedance of that driver which may have positive effects on brake-up behavior

Edit: Wrote this before Plcamp´s post, we seem to be somewhat in agreement

Found the above mentioned measurements, 2 coils totaling 1,6mH in series in the second picture:

 

[Tangband]

Usually people only show 2nd order distortion to 10kHz, 3rd order to 6.7kHz and so on because any distortion beyond that is uninteresting. Amir typically does this too, but if you look at the extended distortion view he did for his 8351B review at 96dB, you'll clearly see the sharp spikes in distortion at 1/2 and 1/3 of the breakup mode. See included image. I maintain that this distortion is unimportant, but it's there.

You dont see this distortion from 8330, 8341 or 8030c , - they also use a smaller metal tweeter , 0,75 inches .

 

[Tangband]

I have seen measurements where a passive filter component (a coil) reduces THD with about 5-8dB. Even after the slope caused by said component is EQ:ued back to flat.
So maybe what Plcamp is saying has some bearing but maybe not related so much to the notch but instead the fact that there are component(s) in series with the drive unit changing the drive impedance of that driver which may have positive effects on brake-up behavior

Edit: Wrote this before Plcamp´s post, we seem to be somewhat in agreement

Yes- lilltroll at faktiskt.io showed this - but the coil only make sence if its needed. In a low inductance driver like seas er18rnx with a better motor there is no such distortion lowering effect .
Again - one has to test it , measure and listen from case to case .
My experience with seas er18rnx with a series Coil and a dsp crossover combined ,didnt show any improvement in distortion.

 

[SoundGuy]

Let's say I have a driver with a breakup/resonance at 27kHz. The driver will move more easily at this frequency, similar to the main resonance frequency of a loudspeaker driver. If I play a tone with this same driver, it will produce some nth-order distortion, because it isn't perfectly linear. Let's say this tone is 9kHz, exactly one-third of the breakup frequency. The driver will now produce a 2nd order 18kHz tone, a 3rd order 27kHz tone, a 4th order 36kHz tone and so on.

As we noted earlier, the driver wants to move at 27kHz. That's smack dab in the middle of the 3rd harmonic, and so it moves a lot more at this frequency than the frequencies immediately next to it (let's say 24kHz and 30kHz), producing higher distortion. The harmonic distortion in itself is harmless, as it is out of the human hearing range in the first place. However in producing this 3rd harmonic 27kHz tone, the driver is in fact moving at that frequency, which can influence intermodulation distortion.

I think it's worthwhile to consider, however, that this 3rd harmonic distortion may still be very low in amplitude (let's be charitable for the sake of example and say 3%, which is -30dB). In addition any music will have considerably lower amplitude at such high frequencies. The audibility of the intermodulation effects introduced by such a breakup not being notched would seem to be low, intuitively. Personally I wouldn't bother notching tweeters.

The story can be a little different for non-tweeters, where the breakup occurs well within the audible band (e.g. the magnesium SEAS midwoofers).

Yes IMD is the issue - glad you raised it. Any big out of band resonance can easily cause in band IMD. Metal tweeters often have a bit of audible glare which I put down to increased IMD. Intrinsic cone damping is your best insurance against this. It is often sounds better with a transducer design that rolls off the highs than a highly rigid transducer designed to push the on axis higher frequency response specification. This is why doped fabric dome tweeter solutions worked out 40 years ago are still competitive and often sound audibly better than the latest that measure “higher spec“ and appear more advanced (in the lab).

IMD is much more audible than harmonic distortion by an order of magnitude.

 

[Tangband]

Yes IMD is the issue - glad you raised it. Any big out of band resonance can easily cause in band IMD. Metal tweeters often have a bit of audible glare which I put down to increased IMD. Intrinsic cone damping is your best insurance against this. It is often sounds better with a transducer design that rolls off the highs than a highly rigid transducer designed to push the on axis higher frequency response specification. This is why doped fabric dome tweeter solutions worked out 40 years ago are still competitive and often sound audibly better than the latest that measure “higher spec“ and appear more advanced (in the lab).

Agree regarding IMD distortion , but dont agree about metal tweeters in all cases - dont you like the sound from Kefs uniQ tweeter ? ( made of metal ?)

 

[Plcamp]

My simple thought now is that…

If you think of the driver as a voltage source, it can’t compete with the amp under control of a pre-amplifier cross and eq.

But if you think of the driver as a current source, it’s the zero output impedance amp that cannot compete.

Not all distortion mechanisms can be controlled by high quality voltage sources like power amps.

So in the case (like my woofer) where there’s a 10db+ breakup peak, this should not be ignored, because the dsp won’t help with a current source distortion mechanism.

Aside from all above, it will be interesting to see if I think any difference results when I try it!

 

[Tangband]

My simple thought now is that…

If you think of the driver as a voltage source, it can’t compete with the amp under control of a pre-amplifier cross and eq.

But if you think of the driver as a current source, it’s the zero output impedance amp that cannot compete.

Not all distortion mechanisms can be controlled by high quality voltage sources like power amps.

So in the case (like my woofer) where there’s a 10db+ breakup peak, this should not be ignored, because the dsp won’t help with a current source distortion mechanism.

Aside from all above, it will be interesting to see if I think any difference results when I try it!

Yes, trying and measuring is knowing.

 

[Scgorg]

You dont see this distortion from 8330, 8341 or 8030c , - they also use a smaller metal tweeter , 0,75 inches .

View attachment 225991

This doesn't show an extended view. Notice that 2nd order ends at 10kHz, 3rd order at 6.7kHz and so on. This is why I specifically mentioned the 8351 review, because it's one of the rare cases where Amir actually showed distortion quite far out.
Both 8030C, 8330A, and 8341 has their tweeter breakup at ultrasonic frequencies (as does 8351). Because of this you need to look at 2nd order beyond 10kHz, 3rd order beyond 6.7kHz etc.
It is easy to confirm this yourself. Measure various drivers with significant breakup beyond the audible range and look at an extended distortion graph. All of them will show harmonic distortion spikes corresponding to 1/n harmonic.

 

[Tangband]

That’s not what is claimed…the claim is that sub harmonic IM levels will be decreased with series notch…not that it would be unusable without them.

Looking at the driver being a source, one can imagine that the presence of the series notch limiting resistor would cause suppression of the driver’s ability to produce third order distortion vs no limiting resistor at all. The series notch resistor won’t allow the driver to create as much breakup frequency 2nd and 3rd order current as when it is absent?

Edit: I will try it on this woofer at some point. Which raises its own question, given I run two of these in parallel. I think for this to work in the case of two parallel woofers, each woofer might need its own series notch. Gotta think on that more, but you can’t control the notch r value if (thinking of the driver as a source) you have two sources in parallel?View attachment 225985

This will be a gigantic job for you, because the notch filtering needed is most likely to be on slightly different frequencies for the different bassdrivers.

 

[SoundGuy]

Dont agree in all cases - dont you like the sound from Kefs uniQ tweeter ? ( made of metal ?)

Agreed. I find my Genelec 8020 have typical metal tweeter glare. I have heard and evaluated the Kef uniQ for a couple hours with all kinds of music and it sounds very natural (presumably no overtly audible IMD or “ringing”). Kef have made an excellent transducer!

 

[DanielT]

Have fun with resistors that can stand a lot of power. Not to bad in sweden winter time.

Imo there is just no need for.

Probably unnecessary overkill, but as I said, I was still a little curious, with my subwoofer example.

Aha, a resistor that should handle 200 W, was not exactly free. $20. See attached picture.

 

[tomtoo]

Probably unnecessary overkill, but as I said, I was still a little curious, with my subwoofer example.

Aha, a resistor that should handle 200 W, was not exactly free. $20. See attached picture.

The question for a afford should be, has it realy a positive effect? You have already step filters.
At the end its a hobby and everybody can put his afford(financial) where he like to. But imo i would put it in different places. When you do a active system keep the passive shiit out if its not realy needet.

 

[DanielT]

The question for a afford should be, has it realy a positive effect? You have already step filters.
At the end its a hobby and everybody can put his afford(financial) where he like to. But imo i would put it in different places. When you do a active system keep the passive shiit out if its not realy needet.

Nop, most likely unnecessary overkill. Maybe even make it worse? What do I know. Nothing I will do. I was just generally curious.

IF it had added something positive, manufacturers of subwoofers would have added such notch filters a long time ago.

 

[Calleberg]

Exactly a STOP filter .That's what's interesting.

Hmm, I wonder. There must be something I'm missing. Another example. Let's say you have a subwoofer, where you want to set the crossover at 80 Hz. You use some kind of low-pass filter but in addition to it a notch filter with a bandwidth of 80-20 kHz, all frequencies above 80 Hz are attenuated down by 60 dB using two resistors and two capacitors. Where is it that makes it bad? Or hm, do the capacitors cost a lot of money?

With an LP with a slope of at least 24dB, there is probably no need for a notch filter in a subwoofer, but I'm still a little curious.

Does the power somehow reduce to frequencies below 80 Hz in this case?



View attachment 225987

Notch Filter Calculator

This notch filter calculates computes the values of the resistors and capacitors for the notch filter based on the input frequencies to be entered.

www.learningaboutelectronics.com

Just look at the numbers, you are suggesting putting a 1kOhm resistor in series with your subwoofer , and besides your notch filter is just one Lowpass and one Highpass filter combined, what good is the high pass going to do for your woofer? But I´m still happy you posted your first post because it all got de-railed into a much more interesting topic

 

[DanielT]

Just look at the numbers, you are suggesting putting a 1kOhm resistor in series with your subwoofer , and besides your notch filter is just one Lowpass and one Highpass filter combined, what good is the high pass going to do for your woofer? But I´m still happy you posted your first post because it all got de-railed into a much more interesting topic

Forget about the subwoofer, see my answers to tomtoo above.

Other than that, I agree with you.

It boils down to whether there is something passive in a crossover that can add something that cannot be added in any other way, or? If I interpret the conversation between those who write in this thread.

Tangband has a sensible view of the whole thing, I think. Test and measure that is.

 

[Plcamp]

This will be a gigantic job for you, because the notch filtering needed is most likely to be on slightly different frequencies for the different bassdrivers.

Heh, I already know it is (four jobs). Right now it’s an experiment on one of them, which moves to 4 if it is significant when I try it.

Having used these drivers with a Minidsp 2x4 hd and just about every crossover topology you might imagine…those 15” drivers (when crossed into the fullrange I’ve got) sound best when 12db low passed acoustically at about 650 hz. So the breakup at about +11db at 1.2 kHz isn’t very well suppressed at all. Using a dsp notch did diminish it, but didn’t eliminate it.

Anyway…as someone said, it is a hobby and hobbyists do weird things sometimes. For example, I have saved magnet wire from a few old motors. My wife laughs at me when I do stuff like that (as she should) but every now and then it suddenly becomes a wise choice.

I do intend to buy a flex eight for the 4-way ability it grants, and expect passive crossover components only for tweeter protection, and maybe this. I am convinced dsp 4 way is a must for Open Baffle.

 

[DanielT]

Has anyone tried this:

Op Amp Active Notch Filter Circuit
Op amp circuit for an active notch filter can be used to remove single frequencies or small bands of frequencies and the electronic circuit design is easy.

Op Amp Notch Filter: Active Filter Circuit » Electronics Notes

Op amp circuit for a notch filter circuit can be used to remove single frequencies or small bands of frequencies & the electronic circuit design is easy . . read more.

www.electronics-notes.com

 

[Thomas_A]

Has anyone tried this:

Op Amp Active Notch Filter Circuit
Op amp circuit for an active notch filter can be used to remove single frequencies or small bands of frequencies and the electronic circuit design is easy.

Op Amp Notch Filter: Active Filter Circuit » Electronics Notes

Op amp circuit for a notch filter circuit can be used to remove single frequencies or small bands of frequencies & the electronic circuit design is easy . . read more.

www.electronics-notes.com

Yes tried some using the Linkwitz site. But it was many years ago.

 

[DanielT]

Yes tried some using the Linkwitz site. But it was many years ago.

Aha, did it work well? These days there are probably digital solutions that work just as well, or better. Which is discussed in this thread, EQ in the digital world vs notch filters. Maybe,as Purifi shows, some advantages of notch filter, what do I know.

 

[Thomas_A]

Aha, did it work well? These days there are probably digital solutions that work just as well, or better. Which is discussed in this thread, EQ in the digital world vs notch filters. Maybe,as Purifi shows, some advantages of notch filter, what do I know.

Yes it worked as it should.