Thanks,
Google and Wikipedia seem permanently open on my desktop!
I am having trouble with the question of wether or not more than 1st order crossover is possible with an RC filter. Some seem to say that even 2nd order isn't possible although clearly Dave is having luck.
What actually accounts for the "droopy" nature of 2 RC filters in a row?
What happens if I cascade two 2nd orders? i.e. 2nd order RC-->buffer-->2nd order RC-->buffer?
How des filter Q come into this?
MrKramer
Google and Wikipedia seem permanently open on my desktop!
I am having trouble with the question of wether or not more than 1st order crossover is possible with an RC filter. Some seem to say that even 2nd order isn't possible although clearly Dave is having luck.
What actually accounts for the "droopy" nature of 2 RC filters in a row?
What happens if I cascade two 2nd orders? i.e. 2nd order RC-->buffer-->2nd order RC-->buffer?
How des filter Q come into this?
MrKramer
But speaker phase response has to be seriously taken into account or the resulting active crossover will be useless (like most generic ones). Speakers have inherent 2nd order lowpass and high-pass built-in, but the frequencies and the Qs are usually far from useful and gentle compensation is required in the crossover.
In passive crossovers this is common practice, there is no other way to do it, since the filter is designed and tuned to reach a target acoustical response, not an electrical response at speaker terminals, which is meaningless.
Active crossovers should be done in the same way, so the electrical response of the filter should be ignored and only the acoustical response of each driver has to be considered (measured with a microphone, both in amplitude and phase, being phase matching more important than flat response), making whatever filter adjustments are required.
Textbook active filters are a waste of time and components, they exist only because there are some people with a far deeper understanding of electronics than of speakers
In passive crossovers this is common practice, there is no other way to do it, since the filter is designed and tuned to reach a target acoustical response, not an electrical response at speaker terminals, which is meaningless.
Active crossovers should be done in the same way, so the electrical response of the filter should be ignored and only the acoustical response of each driver has to be considered (measured with a microphone, both in amplitude and phase, being phase matching more important than flat response), making whatever filter adjustments are required.
Textbook active filters are a waste of time and components, they exist only because there are some people with a far deeper understanding of electronics than of speakers
Textbook active filters are a waste of time and components, they exist only because there are some people with a far deeper understanding of electronics than of speakers
100% agree. I went down that road and can vouch for just how inadequate a generic design is.
A great way to work up a crossover, whether it is active or passive is with an easily adjusted digital crossover (like the DCX2496), a measurement mic and measurement program, like Holmimpulse.
OR just skip the passive or analog active altogether and use a DCX.
Last edited:
Iv'e had excellent results with Rod Elliots P09b crossover. His project articles and Assemble Instructions are very informative and the PCBs he sells are reasonably priced and easy to work with.
Linkwitz-Riley Electronic Crossover
col.
Linkwitz-Riley Electronic Crossover
col.
All my results seemed excellent at first... Even the subtractive crossover from some National Semiconductor application note that leaked low frequency in the high output when clipped (which was easy because it had 12dB local gain in one stage) resulting in several broken dome midranges... (funny speaker adventures from my late teens)
Now I think that finding really optimum crossover settings is a challenge, even with a DCX2496 or similar filter (my quality standards have changed, probably).
btw: The simplest general-purpose approach to true Linkwitz-Riley 24dB/oct acoustical response usually requires 18db/oct electrical filters because the additional phase shift and roll-off is already built in into the speakers... Electrical LR24 is nearly useless.
(funny speaker adventures from my late teens)Now I think that finding really optimum crossover settings is a challenge, even with a DCX2496 or similar filter (my quality standards have changed, probably).
btw: The simplest general-purpose approach to true Linkwitz-Riley 24dB/oct acoustical response usually requires 18db/oct electrical filters because the additional phase shift and roll-off is already built in into the speakers... Electrical LR24 is nearly useless.
Last edited:
Something that should be noticed is that: The higher the filter order, higher the cancellation between the ways and is difficult positioning the driver in baffle to adjust the polar response.
System of higher order are used in professional audio allowing for greater power in the driver and does not damage the diaphragm. System at home is a better first-order filter, but may need a second-order for TWT not to damage, (depends on the power ).
See figure what happens
System of higher order are used in professional audio allowing for greater power in the driver and does not damage the diaphragm. System at home is a better first-order filter, but may need a second-order for TWT not to damage, (depends on the power ).
See figure what happens
Attachments
Hi,
You need an OPamp to build notch filters!
Eva is totally right when he points out that standard textbook filtering is not the way to go. With passive filtering everybody would instantely agree that this doesn´t work right. With active filtering it seems that people forget about that. A proper crossover fulfills the task of filtering and the task of equalizing at once. Most active crossovers just do the filter-job and omit with the equalization. Thats the reason why many tries of activating a passive loudspeaker had to fail.
Now You can decide to do both tasks with separate building blocks -leading to a lot of circuitry effort, but probabely a higher flexibility- or You can design the filter stages off of the textbook standards, so that they perform both tasks at the same, similar to how it´s done with most passive crossovers. The only difference between passive and active filtering is that the speaker´s impedance influences the design of the passive filter, while it doesn´t with active filtering.
jauu
Calvin
Wonder though why my filters with just a buffer do work as notches
You need an OPamp to build notch filters!
Eva is totally right when he points out that standard textbook filtering is not the way to go. With passive filtering everybody would instantely agree that this doesn´t work right. With active filtering it seems that people forget about that. A proper crossover fulfills the task of filtering and the task of equalizing at once. Most active crossovers just do the filter-job and omit with the equalization. Thats the reason why many tries of activating a passive loudspeaker had to fail.
Now You can decide to do both tasks with separate building blocks -leading to a lot of circuitry effort, but probabely a higher flexibility- or You can design the filter stages off of the textbook standards, so that they perform both tasks at the same, similar to how it´s done with most passive crossovers. The only difference between passive and active filtering is that the speaker´s impedance influences the design of the passive filter, while it doesn´t with active filtering.
jauu
Calvin
"btw: The simplest general-purpose approach to true Linkwitz-Riley 24dB/oct acoustical response usually requires 18db/oct electrical filters because the additional phase shift and roll-off is already built in into the speakers... Electrical LR24 is nearly useless."
Electrical LR24 is very useful for drivers that have flat response for an octave or so beyond the intended crossover points.
For pushing the limits on what a set of drivers can do we might end up with a 3P HP on a tweeter, and a 2P LP on the bass-midrange driver (the acoustical response being LR24).
Electrical LR24 is very useful for drivers that have flat response for an octave or so beyond the intended crossover points.
For pushing the limits on what a set of drivers can do we might end up with a 3P HP on a tweeter, and a 2P LP on the bass-midrange driver (the acoustical response being LR24).
and a discrete buffer is a unity gain opamp.Wonder though why my filters with just a buffer do work as notches
You need an OPamp to build notch filters!
Let me tell you a little about my speakers. Maybe that will help.
I set out to make a 3 way 'sealed' system. Got some really good help on this forum. Along the way I strayed from the path and went open baffle. Love it.
I am using a fostex ff85k on top. Really like everything about this driver when I cross it anywhere between say 300-800hz. For a mid, Ive been using a 123a JBL 12". Sounds great crossed to the fostex. Good combo. Tried a couple different slopes and points. Hard to go wrong. I like it more when its not handling real bass. Down low I use a Goldwood 18"er. This one is trickier. To me it sounds really nice up to a point, higher than you might think, where it starts sounding bad. I had real trouble finding the best cross for it. Big problem is that at those low frequencies and with higher orders, passive parts become big and expensive. I couldn't afford to experiment that much.
So with this setup, tweeter protection isn't really an issue. The JBL sounds fine up past the point where it beams, so shallow slope between the 2 has worked out well. The low end of the JBL can be a shallow slope too, its not critical at all. Just sounds a little nicer with a little less bass duty.
MrKramer
I set out to make a 3 way 'sealed' system. Got some really good help on this forum. Along the way I strayed from the path and went open baffle. Love it.
I am using a fostex ff85k on top. Really like everything about this driver when I cross it anywhere between say 300-800hz. For a mid, Ive been using a 123a JBL 12". Sounds great crossed to the fostex. Good combo. Tried a couple different slopes and points. Hard to go wrong. I like it more when its not handling real bass. Down low I use a Goldwood 18"er. This one is trickier. To me it sounds really nice up to a point, higher than you might think, where it starts sounding bad. I had real trouble finding the best cross for it. Big problem is that at those low frequencies and with higher orders, passive parts become big and expensive. I couldn't afford to experiment that much.
So with this setup, tweeter protection isn't really an issue. The JBL sounds fine up past the point where it beams, so shallow slope between the 2 has worked out well. The low end of the JBL can be a shallow slope too, its not critical at all. Just sounds a little nicer with a little less bass duty.
MrKramer
Sounds nice. I've a similar setup with a fostex 108 crossed over at 300Hz to a 12" Eminence. Crossover parts certainly can become large and expensive at those lower frequencies. With a three way crossover you will certainly need some kind of buffering. But if everything works well with a shallow crossover slope then PLLXO seems like a pretty good option. On the other hand have you tried a higher order crossover on between the JBL and Goldwood? I would have to imagine the JBL would be a little more resolving of a driver. By using a higher order crossover between the JBL and Goldwood you could push the crossover frequency a bit lower with out putting too much strain on the JBL. Being that I'm in the same boat and feeling indecisive. I'm voting that you try both options and get back to us on which one you prefer.
preamp -> active 4th order crossover between JBL/Goldwood -> PLLXO between JBL/fostex
and/or
preamp -> PLLXO between JBL/Goldwood -> buffer -> PLLXO between JBL/fostex
I'm looking foward to hearing your listening impressions
or opinions from others with more experience on the subject
Joe
Last edited:
When it stays down low in OB, I think the Goldwood is better than the JBL for big acoustic sounds. Drums, Pianos, big instruments actually sound big.
The JBL is way better for low voice.
I think you are right about a steeper cross between them.
I will go with this.
For the shallow slopes:
RC-->source follower
For the steeper slopes:
Salen Key using source follower.
Any suggestions as for the source follower layout?
The Q for 1st order is fixed right? How do I define the Q for the higher order?
Is my thought of using my existing +/-18v supply just wishful thinking?
Any JFET part number recommendations? I see 2SK170 bandied about. Good choice? Easy to find?
Thank you all for your help. Learning alot here.
As I get further with this Ill post my progress.
MrKramer
The JBL is way better for low voice.
I think you are right about a steeper cross between them.
I will go with this.
For the shallow slopes:
RC-->source follower
For the steeper slopes:
Salen Key using source follower.
Any suggestions as for the source follower layout?
The Q for 1st order is fixed right? How do I define the Q for the higher order?
Is my thought of using my existing +/-18v supply just wishful thinking?
Any JFET part number recommendations? I see 2SK170 bandied about. Good choice? Easy to find?
Thank you all for your help. Learning alot here.
As I get further with this Ill post my progress.
MrKramer
a unity gain stage uses different values of impedance in the various legs of the filter to set both frequency and Q.
A Salen & Key equal component value uses the filter components to set the frequency and uses gain to set the Q.
An MFB uses all the components to set all three Q, F and A independantly but it is inverting.
Find papers explaining how each of these manage their adjustments.
Thanks Andrew,
Each time someone mentions something on this thread I immediately start searching on google. I find lots of papers out there and I read as many as I can. Seeing as Im such a newbie, I have difficulty sorting out the wheat from the chaff. Many of these papers are meant for engineers. I'm not an engineer. Many are simply arguments supporting somebody's favorite layout, often not for audio use.
If you have any specific sites in mind that would be helpful, please let me know.
MrKramer
Each time someone mentions something on this thread I immediately start searching on google. I find lots of papers out there and I read as many as I can. Seeing as Im such a newbie, I have difficulty sorting out the wheat from the chaff. Many of these papers are meant for engineers. I'm not an engineer. Many are simply arguments supporting somebody's favorite layout, often not for audio use.
If you have any specific sites in mind that would be helpful, please let me know.
MrKramer
useful calculators and other filter info:
Design and Dimensioning of Active Filters
National Semiconductor
OA21, oa26, oa27, oa28, oa29, none of these are for a complete beginner, but they do include most that you will ever need for S&K filters.
Ti has a few papers on filtering and design of filters and they used to have an excellent calculator, but the current replacement is too automatic and does not allow adequate flexibility for analogue design.
Design and Dimensioning of Active Filters
National Semiconductor
OA21, oa26, oa27, oa28, oa29, none of these are for a complete beginner, but they do include most that you will ever need for S&K filters.
Ti has a few papers on filtering and design of filters and they used to have an excellent calculator, but the current replacement is too automatic and does not allow adequate flexibility for analogue design.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.