What's the difference between a Butterworth and Linkwitz-Reilly?

[cpt_paranoia]

Marshall Leach wrote this and you need to keep it in mind when considering crossover types:

I'd also point out component tolerances as an overlooked issue when designing crossover filters. All very well designing a super-duper crossover filter that will give great results with perfect components, but real components like capacitors tend to have large tolerances. Which means your filter may end up having response lumps in unexpected places, and your L-R filter may look more like a Butterworth. Since, as the OP noted, the circuit configuration is the same, the filter type is only defined by the component values. If those values aren't the ones you designed, your filter isn't the one you designed.

 

[nfafan]

Did Hollywood make a movie of this for us non-EE types?

 

[dc270]

So you are saying is to use the better tolerance components, as opposed to the 10% typical, use the 5~2% tolerance caps, high grade copper coils, non inductive resistors and the like?
How does the given drivers voice coil inductance factor into the crossover filter components you use?.Are there better reasons to use a given filter type than everybody using the L-R filter?
Thanks again for your input, hope the answer contributes to Tox's question...…..
DC

 

[cpt_paranoia]

So you are saying is to use the better tolerance components, as opposed to the 10% typical, use the 5~2% tolerance caps

If you can get them that good, yes. Or tune them with small, parallel caps.

Caps are the most loosely toleranced, in general (although big, air cored inductors aren't great). Certainly electrolytics. Modern devices are better, due to the improved manufacturing consistency.

The other difficulty is that a capacitor isn't just a capacitor; it's actually a distributed RLC network, depending on its construction. So your design should try to take that into account, if you want a really accurate filter.

I remember designing RF filters using the bible 'Zverev'. You had to allow PCB placements for SOT (select on test) caps to tune the filter, especially a narrow response, high order filter. I also remember a spell on the production test line, hand tweaking air-cored inductors to get the right response. And, on the odd occasion, sending them back for rework after having the coil 'break' in frustration... This was a long time ago...

https://www.amazon.co.uk/Handbook-Filter-Synthesis-Anatol-Zverev/dp/0471749427

 

[JimPA]

The OP might enjoy reading this.blog0402.htm

 

[bakes]

Linkwitz-Riley is lousy on pancakes...

Basically, different values for the components yields a different filter response.

 

[Joe6805]

So, to BE a Linkwitz-Riley, does each have to be second order?

Mine uses a first order band pass for the mid, and a second order for the tweet (woofer is wide open). Not sure WHAT type it is. Mostly was curious about phasing, if mine needs any type of polarity reversals.

Thanks,

Joe

 

[JimPA]

So, to BE a Linkwitz-Riley, does each have to be second order?

Mine uses a first order band pass for the mid, and a second order for the tweet (woofer is wide open). Not sure WHAT type it is. Mostly was curious about phasing, if mine needs any type of polarity reversals.

Thanks,

Joe

A second order Linkwitz-Riley crossover has the tweeter positive wire connected to the tweeter's negative terminal..
This means that the tweeter is out of phase.

The previous link I had post shows this.

 

[edwyun]

What you usually want is a 4th order/24dB per octave L-R alignment to address both amplitude and phase response. But depends on the speakers.

 

[darg]

Tox,

Sorry to raise such an old thread, but have been wondering the same thing lately.

My Optimus 1000s use a 2nd order on the tweeter, and a band pass on the mid. Tweeter is using a 2.2uF cap and 4ohm resistor in series, with .4mH coil in parallel. 5” mid is using a 3.9uF cap and .56mH coil in series. 15” woofer is wired direct, no coil or cap.

My questions are; with the ratios you mentioned, it LOOKS like the tweeter network is L-R? If it IS, do I need to wire it out of phase? New to the whole “phase shift” thing with different type and order networks.

Original drivers were all wired in phase, if I recall. Replaced the 3” cone tweet with a dipole though, and still working out the blending.

Thanks!

Joe

In your case you need to let us know where the resistor is located. Depending on its location relative to the inductor it can be 4 ohm tweeter together with a 4 ohm resistor to make the package 8 ohm impedance compatible for the amp or a 4 ohm resistor to reduce output level of the tweeter. In general, is the inductor parallel to just the tweeter or parallel to the resistor and tweeter together. This in general gives you an idea about the real cutoff frequency.

 

[Grenadeslio]

What's the difference between a Butterworth and Linkwitz-Reilly?


One you put on pancakes, the other is related to a character on MASH.

 

[Joe6805]

In your case you need to let us know where the resistor is located. Depending on its location relative to the inductor it can be 4 ohm tweeter together with a 4 ohm resistor to make the package 8 ohm impedance compatible for the amp or a 4 ohm resistor to reduce output level of the tweeter. In general, is the inductor parallel to just the tweeter or parallel to the resistor and tweeter together. This in general gives you an idea about the real cutoff frequency.

The inductor is parallel to both the resistor and tweeter. I’d draw it out, but am unsure how to post pics here being fairly new.

So goes through the cap, then the resistor and coil in parallel. The 4 ohm resistor I believe is only for tweeter attenuation, as the original tweet is 6 ohms.

Thanks!

Joe

 

[darg]

So it's a level and impedance adjustment. Do you know the cutoff for the high pass? Based on just the empirical 8 ohm assumption you. Ight would get different values for the cap and the inductor than what is in the circuit. This way recalculating or changing anything becomes a little bit more experimental. On the other hand who knows what values for the cap the mfg had on had or bought on the market so dint handle it as a generic crossover. There was thinking and measurements going into building these that are hard to backtrack. Good luck

 

[Joe6805]

So it's a level and impedance adjustment. Do you know the cutoff for the high pass? Based on just the empirical 8 ohm assumption you. Ight would get different values for the cap and the inductor than what is in the circuit. This way recalculating or changing anything becomes a little bit more experimental. On the other hand who knows what values for the cap the mfg had on had or bought on the market so dint handle it as a generic crossover. There was thinking and measurements going into building these that are hard to backtrack. Good luck

I agree completely, who knows what they did, but assume they had it figured out.

Having measured the impedance of the original tweeter (labeled 6, measures 6.6) and combined with the measurement of the original resistor (labeled 4, measures 4.4), I’m getting ~11 ohms combined. So my plan is to match that total load after the cap with the new tweeters. So measuring ~7.5 ohms (labeled 8) on the new tweeters, I’ll use a 3.5 ohm resistor. Should theoretically retain the original high pass point then?

Honestly don’t know what crossover points the mid or tweet used originally. Have tried using calculators to figure it out, and the values never seem to match what’s in there.

Thanks for all of the help,

Joe

 

[RTally]

I agree completely, who knows what they did, but assume they had it figured out.

Having measured the impedance of the original tweeter (labeled 6, measures 6.6) and combined with the measurement of the original resistor (labeled 4, measures 4.4), I’m getting ~11 ohms combined. So my plan is to match that total load after the cap with the new tweeters. So measuring ~7.5 ohms (labeled 8) on the new tweeters, I’ll use a 3.5 ohm resistor. Should theoretically retain the original high pass point then?

Did you measure the impedance or dc resistance? If you used a ohmmeter to measure resistance, then you measured dc resistance. Although it provides a guideline, dc resistance can vary a lot from impedance. The driver's impedance is best measured at the frequency of interest, which in this case, is the crossover frequency.

 

[Joe6805]

Did you measure the impedance or dc resistance? If you used a ohmmeter to measure resistance, then you measured dc resistance. Although it provides a guideline, dc resistance can vary a lot from impedance. The driver's impedance is best measured at the frequency of interest, which in this case, is the crossover frequency.

Thanks. Yes, measured DC resistance with a meter.

Is my logic sound, retaining the original crossover frequency by trying to match the resistance after the cap between old and new?

Thanks,

Joe

 

[RTally]

Thanks. Yes, measured DC resistance with a meter.

Is my logic sound, retaining the original crossover frequency by trying to match the resistance after the cap between old and new?

Thanks,

Joe

Nope. Two issues: impedance and sensitivity.

If the original are a nominal 6 ohm driver and the new one is a nominal 8 ohm driver, then you have a 2 ohm difference. The measured dc resistance is too imprecise to use as a replacement for impedance. If the original crossover for the original driver had a 4 ohm resistor, then you should replace it with a 2 ohm resistor for the new driver. But all that will do is ensure that the crossover point remains the same without considering the sensitivity of the drivers.

You need to determine the sensitivity of the drivers and adjust the resistor value to ensure the new driver (in circuit) has the same sensitivity as the original driver (in circuit). Then you should change the other crossover component values to ensure you have the desired crossover frequency.

It sounds like you replaced an original tweeter with one that is not matched at all. In that case, you need to design a new crossover circuit for the new tweeter parameters. To avoid doing that, it is best to replace an original driver with one that has closely matched parameters.

 

[Joe6805]

Nope. Two issues: impedance and sensitivity.

If the original are a nominal 6 ohm driver and the new one is a nominal 8 ohm driver, then you have a 2 ohm difference. The measured dc resistance is too imprecise to use as a replacement for impedance. If the original crossover for the original driver had a 4 ohm resistor, then you should replace it with a 2 ohm resistor for the new driver. But all that will do is ensure that the crossover point remains the same without considering the sensitivity of the drivers.

You need to determine the sensitivity of the drivers and adjust the resistor value to ensure the new driver (in circuit) has the same sensitivity as the original driver (in circuit).

So if it were to blend well, ideally a 2 ohm resistor would be the trick? Understand you’re saying to go by the nominal values and not the measured ones.

Only other thing to mention, the new unit is an add-on dipole that originally had its own crossover (simply a 4.7uF cap). Is the rated impedance assuming the original cap, or wouldn’t it matter? That’s the only reason I thought to measure them instead, because the original cap is now bypassed to use the internal crossover.

They’re a hair bright with a 2 ohm resistor, currently running them with a 3 ohm. I understand this will lower crossover frequency?

Thanks, sorry for being slow at this

Joe

 

[RTally]

So if it were to blend well, ideally a 2 ohm resistor would be the trick? Understand you’re saying to go by the nominal values and not the measured ones.

Only other thing to mention, the new unit is an add-on dipole that originally had its own crossover (simply a 4.7uF cap). Is the rated impedance assuming the original cap, or wouldn’t it matter? That’s the only reason I thought to measure them instead, because the original cap is now bypassed to use the internal crossover.

They’re a hair bright with a 2 ohm resistor, currently running them with a 3 ohm. I understand this will lower crossover frequency?

Thanks, sorry for being slow at this

Joe

Blending two drivers is a function of crossover frequency and sensitivity matching. There are ways to reduce sensitivity without changing impedance. But that will take a lot more explaining that I care to do on a forum. Crossover design is not trivial or straightforward.

Experimenting is often fun, but I do not trust my ears. If I were to do what you are attempting I would use a real time analyzer (RTA) and a calibrated mic to ensure that the crossover between the drivers is where it needs to be and the levels are too. For me, trying to calculate and determine the precise values of the components without also testing the results would be an exercise in frustration.

If you trust your ears, try it and see if you like how it sounds.

 

[Joe6805]

Experimenting is often fun, but I do not trust my ears. If I were to do what you are attempting I would use a real time analyzer (RTA) and a calibrated mic to ensure that the crossover between the drivers is where it needs to be...

That makes sense, thanks for all of your help.

I have RoomEq Wizard and an SPL meter, may be best to go that route. I trust my ears for the sensitivity, and how I like my high end, but not sure I trust them 100% for crossover frequency.

Appreciate your time,

Joe