Hi ya'all 
Just an idea... I have quite a few speakers and amps built under my belt with , as usual, a mixed mixture of success from awful to simply astonishing!
Most of what I designed and built was for sound PA systems and studio monitors.
I always used pro electronic crossover with all the compensations and everything needed to get them to sound right. I am talking here of a 10 unit rack full of gizmos in it!
Now for something completely different!
A gain-clone amp is truly a piece of wonder! It is basically a 35Watts Op-Amp, and as such can be set to be an electronic crossover.
I am thinking here, say a gain clone stage with 12db/oct slope low pass for the bass, a 12 or 18db/oct slope bandpass for the mids and so on...
We can use pretty much any filter configuration and order here...
Then, put a passive network at the speakers, for correcting some flaws in the freq response of individual drivers, damp some resonances or peaks etc...
I read somewhere that it is better to use full range amplifiers and feed them with the output of an electronic active crossover, but I fail to see why it would not work using a tightly controlled band amplifier for each speaker.
Any one have an opinion on this?
Luc
Just an idea... I have quite a few speakers and amps built under my belt with , as usual, a mixed mixture of success from awful to simply astonishing!
Most of what I designed and built was for sound PA systems and studio monitors.
I always used pro electronic crossover with all the compensations and everything needed to get them to sound right. I am talking here of a 10 unit rack full of gizmos in it!
Now for something completely different!
A gain-clone amp is truly a piece of wonder! It is basically a 35Watts Op-Amp, and as such can be set to be an electronic crossover.
I am thinking here, say a gain clone stage with 12db/oct slope low pass for the bass, a 12 or 18db/oct slope bandpass for the mids and so on...
We can use pretty much any filter configuration and order here...
Then, put a passive network at the speakers, for correcting some flaws in the freq response of individual drivers, damp some resonances or peaks etc...
I read somewhere that it is better to use full range amplifiers and feed them with the output of an electronic active crossover, but I fail to see why it would not work using a tightly controlled band amplifier for each speaker.
Any one have an opinion on this?
Luc
Maybe I'm just thick, but I think you'll have to explain what you're proposing better. If you're talking about line level filters built into the amps, that's equivalent (potentially better, I guess) to a conventional active crossover. If you're then talking about what is essentially adding some speaker level filtering after a normal active crossover, I've been doing that a LOT lately.
OK here goes,
the normal way of doing things is to take a signal, go trough an active crossover to split the bands, go to the power amps and drive the individual speakers.
The amps will be full range (20hz-20Khz) even if each band is limited. It is relatively easy to build an active crossover, here are many schématics and documentation on the subjects.
BUT, these filters will not take in account some deficiencies in the speakers that are corrected with RLC networks in the passive crossovers.
What I am suggesting here is to corrects those deficiencies with passive networks, then use filters/amp combination.
Then to push this one step further, the LM3886 is a power op-amp in effect, so probably any filter of any type and any orders can be designed with it.
The idea is to limit the bandpass of the amp in a tightly controlled manner and we would not need a filter ahead of the amp stage.
Hope this is a bit more clear.
Luc
the normal way of doing things is to take a signal, go trough an active crossover to split the bands, go to the power amps and drive the individual speakers.
The amps will be full range (20hz-20Khz) even if each band is limited. It is relatively easy to build an active crossover, here are many schématics and documentation on the subjects.
BUT, these filters will not take in account some deficiencies in the speakers that are corrected with RLC networks in the passive crossovers.
What I am suggesting here is to corrects those deficiencies with passive networks, then use filters/amp combination.
Then to push this one step further, the LM3886 is a power op-amp in effect, so probably any filter of any type and any orders can be designed with it.
The idea is to limit the bandpass of the amp in a tightly controlled manner and we would not need a filter ahead of the amp stage.
Hope this is a bit more clear.
Luc
well AllenB, I will not have anytime soon to devote to this project ( building a new house this year) but I am at the "mind picking' stage.
You have hit the nail square when you say "taking care with the output impedance versus frequency and watch your loads" as a speaker is not such a well behaved load compared to the 10-20Kohm input resistor seen at the input of most power amp!
And this is where the passive components introduced R ? L? C ?? can help very much to tame those loads...
Maybe it is not worth the trouble also!
The reason I am leaning toward this direction is that I built a couple of 2 and three ways filters, some of my own design, one from Elektor and others from French magazines and I always been a bit disappointed by the results.
Then lately I 'rediscovered' the passive crossover and the techniques of filtering out some of the bumps in the frequency curves a speaker can have and also the step compensation etc..
So I guess by now you can see where I am heading
Luc
You have hit the nail square when you say "taking care with the output impedance versus frequency and watch your loads" as a speaker is not such a well behaved load compared to the 10-20Kohm input resistor seen at the input of most power amp!
And this is where the passive components introduced R ? L? C ?? can help very much to tame those loads...
Maybe it is not worth the trouble also!
The reason I am leaning toward this direction is that I built a couple of 2 and three ways filters, some of my own design, one from Elektor and others from French magazines and I always been a bit disappointed by the results.
Then lately I 'rediscovered' the passive crossover and the techniques of filtering out some of the bumps in the frequency curves a speaker can have and also the step compensation etc..
So I guess by now you can see where I am heading
Luc
Well, active and hybrid crossovers can be simulated with real measurements in order to achieve predictable results, just like conventional passive crossovers. Step compensation is a much better job for the line level side, but no need to get into specifics without a specific system, I guess. Incorporating a (any) filter into an LM3886 circuit is sort of a separate issue, and not the kind of thing I know how to do, but it seems feasible.
Last edited:
I'm a bit of a fan of passive line level filters myself at least for secondary filtering that would prefer some isolation, but constraining the bandwidth of an amp that can afford this could have a few things going for it.
I don't take any direct action over the baffle step when I design a crossover. I measure the speakers from all angles and the information automatically falls in to place.
I don't take any direct action over the baffle step when I design a crossover. I measure the speakers from all angles and the information automatically falls in to place.
Here is my reasoning behind all this.
As a pro I use 2 heavily modified Behringer DCX2496 digital crossover. These are terrible straight out of the box and have reliability problems. I will not get into the details of the mods I do to them, but the one very big thing they have for them is the algorythms in the DSP section, probably one of the best in business. The other processors I use are dbX "driverack PA+" ( yeah they are all super, pro, hyper, and plus something, don't you find it annoying when the marketing dept takes over the engineering? )
What I am trying to achieve here is a reasonably high quality systems (3) for my coming new house, garage and a room n the basement.
I am definitely hooked on bi and tri amplification, and I am spoiled with all the parameters I can adjust on the digital crossovers.
On the other hand, I can build very inexpensively a bunch of gain clones, the speakers to go with them, but getting and modding three DCX2496 is out of my reach!
Hince the idea of making the gainclone frequency 'restricted' looks like a good idea, and compensating passively for some rough bumps in the individual speaker response, and taking some precautions to 'time-align' if it's important, should yield some very listenable systems.
This is where I am trying the community brain here
Luc
As a pro I use 2 heavily modified Behringer DCX2496 digital crossover. These are terrible straight out of the box and have reliability problems. I will not get into the details of the mods I do to them, but the one very big thing they have for them is the algorythms in the DSP section, probably one of the best in business. The other processors I use are dbX "driverack PA+" ( yeah they are all super, pro, hyper, and plus something, don't you find it annoying when the marketing dept takes over the engineering? )
What I am trying to achieve here is a reasonably high quality systems (3) for my coming new house, garage and a room n the basement.
I am definitely hooked on bi and tri amplification, and I am spoiled with all the parameters I can adjust on the digital crossovers.
On the other hand, I can build very inexpensively a bunch of gain clones, the speakers to go with them, but getting and modding three DCX2496 is out of my reach!
Hince the idea of making the gainclone frequency 'restricted' looks like a good idea, and compensating passively for some rough bumps in the individual speaker response, and taking some precautions to 'time-align' if it's important, should yield some very listenable systems.
This is where I am trying the community brain here
Luc
Hi Luc.
I have had similar thoughts for a while. Should work very well. Sound like you have it all thought it through pretty much. Just a few "reflections".
Btw in the "nothing new under the sun department". In the '70's Plessey/Ducon UK marketed a small IC power amp chip SL40XX? something or other (3 watt into 8 Ohms) and they published a diy three-way active system in a UK magazine (Popular/Practical Electronics?) using a similar idea......although I recall their "Power amp" chip may have had some small signal silicon on the front of it t as well to build the cross over components around.
Now back to your system.
I am not an expert on this area, just read a lot. A couple of thoughts. I like the idea of the power amp as an frequency modifying op-amp.
(a) Tomlinson Holman in the '70's built a RIAA pre-amp with an active 2nd order "rumble filter" (JAES) built into it as you propose so at least one commercial venture has had a similar idea of filtering plus gain.
(b) As I understand it INVERTING gain clones may be worth looking at. Many active filter use non-inverting (Sallen-Keys) set ups but there are a set of filters (3rd and 2nd order) that use multiple feedback around inverting amps. These give measurably lower distortion and for various reason several people prefer the sound of inverting (shunt feedback) amps' (And this is not pure "hog wash". They argue their case with issues of grounding and feedback etc...I like my subjective prejudices to have a theoretical basis! Ha ha...)
I have different goals from you.
I was thinking of a small domestic tri-amp set up with single rail class A amps (JLH). Two thoughts struck me re; "hybrid" schemes.
(1) For various reasons I have all the components for several class-A single rail amps (JLH 10 Watters). I was going to have a second order active (or passive) cross-over prior to the amp and then use an appropriate sized high quality cap' after the power amp to achieve an additional order to the cross-over and also act as an DC blocking cap allowing me to get rid of the usual 2,500uF output electro'
(2) Re: equalization prior to the Power Amp. One of my drivers will be a KEF B110 and I have half a dozen high level passive cross-over schematics all seeking to reduce the 8-10db peak in the 2-3kHz range. So I thought I would achieve that passively prior to the amp i.e. a shelving network from around 600c/s. Then it dawned on me that while the SIGNAL feeding the amp will be fine and I'll get a flat over all response from the amp/speaker output any cross over distortion products from the power amp that were generated by signals in the range around 300c/s would be amplified and come out 8-10 db higher at the output of the power amp RELATIVE TO THE COMPENSATED SIGNAL. I.e. I wasn't "shelving" any distortion products generated by the amp' by a similar amount. (JLH wrote about the subjective problems of components in the 2-3k Hz range and the 7th, 9th and 11th harmonics of a 300c/s note would be a potential problem. He argued that 300 c/s was a more appropriate test frequency for audible distortion than 1,000c/a for that reason.) Now, theoretically my class A system should not have that issue (higher order harmonics) but it may be relevant to consider this for a badly biased class B or AB Power Amp.
Hope that makes sense.
Good luck. Cheers, Jonathan
I have had similar thoughts for a while. Should work very well. Sound like you have it all thought it through pretty much. Just a few "reflections".
Btw in the "nothing new under the sun department". In the '70's Plessey/Ducon UK marketed a small IC power amp chip SL40XX? something or other (3 watt into 8 Ohms) and they published a diy three-way active system in a UK magazine (Popular/Practical Electronics?) using a similar idea......although I recall their "Power amp" chip may have had some small signal silicon on the front of it t as well to build the cross over components around.
Now back to your system.
I am not an expert on this area, just read a lot. A couple of thoughts. I like the idea of the power amp as an frequency modifying op-amp.
(a) Tomlinson Holman in the '70's built a RIAA pre-amp with an active 2nd order "rumble filter" (JAES) built into it as you propose so at least one commercial venture has had a similar idea of filtering plus gain.
(b) As I understand it INVERTING gain clones may be worth looking at. Many active filter use non-inverting (Sallen-Keys) set ups but there are a set of filters (3rd and 2nd order) that use multiple feedback around inverting amps. These give measurably lower distortion and for various reason several people prefer the sound of inverting (shunt feedback) amps' (And this is not pure "hog wash". They argue their case with issues of grounding and feedback etc...I like my subjective prejudices to have a theoretical basis! Ha ha...)
I have different goals from you.
I was thinking of a small domestic tri-amp set up with single rail class A amps (JLH). Two thoughts struck me re; "hybrid" schemes.
(1) For various reasons I have all the components for several class-A single rail amps (JLH 10 Watters). I was going to have a second order active (or passive) cross-over prior to the amp and then use an appropriate sized high quality cap' after the power amp to achieve an additional order to the cross-over and also act as an DC blocking cap allowing me to get rid of the usual 2,500uF output electro'
(2) Re: equalization prior to the Power Amp. One of my drivers will be a KEF B110 and I have half a dozen high level passive cross-over schematics all seeking to reduce the 8-10db peak in the 2-3kHz range. So I thought I would achieve that passively prior to the amp i.e. a shelving network from around 600c/s. Then it dawned on me that while the SIGNAL feeding the amp will be fine and I'll get a flat over all response from the amp/speaker output any cross over distortion products from the power amp that were generated by signals in the range around 300c/s would be amplified and come out 8-10 db higher at the output of the power amp RELATIVE TO THE COMPENSATED SIGNAL. I.e. I wasn't "shelving" any distortion products generated by the amp' by a similar amount. (JLH wrote about the subjective problems of components in the 2-3k Hz range and the 7th, 9th and 11th harmonics of a 300c/s note would be a potential problem. He argued that 300 c/s was a more appropriate test frequency for audible distortion than 1,000c/a for that reason.) Now, theoretically my class A system should not have that issue (higher order harmonics) but it may be relevant to consider this for a badly biased class B or AB Power Amp.
Hope that makes sense.
Good luck. Cheers, Jonathan
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.