B1 Active Crossover

[Jacques Merde]

go to Linkwitz site.
If you use gain stages then the gain controls the Q. It becomes a very different design if you adopt other than unity gain.
You can use ECV Sallen Key or MFB as your gain filter stages.

Another good site is Rod Elliot's. See esp. Project 9 and Project 81.

If you need gain, you could splice a JFET BOZ to the input (or output). Or maybe something like Juma's BF862 Preamp.

 

[njepitt]

...
If you need gain, you could splice a JFET BOZ to the input (or output). Or maybe something like Juma's BF862 Preamp.

I'm thinking about putting juma's BF862 preamp at the input, as you suggest. Since it has a gain of 2, would anything in the crossover circuit need altering to handle a higher input?

I quite like the idea of doing a preamp with switchable crossover, so it can be used with or without breaking into high and low freqs, with BF862 throughout. Still thinking it all through, though, and am interested in hearing any and all ideas....

Cheers

Nigel

 

[Jacques Merde]

I'm thinking about putting juma's BF862 preamp at the input, as you suggest. Since it has a gain of 2, would anything in the crossover circuit need altering to handle a higher input?

I quite like the idea of doing a preamp with switchable crossover, so it can be used with or without breaking into high and low freqs, with BF862 throughout. Still thinking it all through, though, and am interested in hearing any and all ideas....

Cheers

Nigel

Nigel,

I think that the BF862 will make a good choice. Its high gm makes it attractive, IMO. The circuit should handle larger input signals fine, though with some increased distortion . You might consider limiting the supply voltages to +/- 9 Volts with the BF862, just to avoid any possibility of exceeding the max Vds of the BF82.

 

[njepitt]

Nigel,

I think that the BF862 will make a good choice. Its high gm makes it attractive, IMO. The circuit should handle larger input signals fine, though with some increased distortion . You might consider limiting the supply voltages to +/- 9 Volts with the BF862, just to avoid any possibility of exceeding the max Vds of the BF82.

Hi Jacques,

I certainly don't have enough understanding of these things to judge how much distortion would be increased, but it's definitely to be avoided if possible... The obvious alternative would be to use separate gain stages after splitting the frequencies; I mean put one in each of the LP and HP outputs. Is this likely to be better from the point of view of distortion? (Against doing it this way is the obvious increased complexity,which sort of goes against the elegant simplicity of the original idea... Still trying to think this through...)

Cheers

Nigel

 

[schro20]

If everyone thinks just WIMA box caps are fine here then I think I have all the info I need, but I wonder if people are going to want to use big caps like Axon, Mundorf, etc..
Uriah

I think having multiply spaced holes for the caps is generally a good idea. I have to agree with Jaques that the Panasonic 1% are likely the baseline (see what Mr. Linkwitz himself has to say about this). On that note his site is fantastic reading when it comes to linelevel cross overs and the considerations going into their design. As regards more exotic caps, keep in mind that we are not trying to deal with 250V or higher parts as required in passive cross overs... Tolerance being tight on the caps is one of the main requirements here (again, see Linkwitz on the subject).

My 2cents.

Peter

 

[maxlorenz]

About caps:
I have read from various sources that polystyrene caps are to be preferred on this kind of filter. RS components has good ones but not many options on values...I will go for polyester then reevaluate for an upgrade.

 

[udailey]

Peter,
Thanks for your input. I have them spaced at 5mm, 7.5mm, 10mm and 15mm and width is 9mm.
Uriah

 

[udailey]

PCB a few options

Guys what do you think?
The only difference between these two is the obvious absence of filled plane in the center. Ground on both encircles the board. In the one that is filled in the center - the fill is not connected to ground.
So let me know what you think about the layout. The board is right now 58mmX115mm.
Uriah

 

[spencer]

Yamaha F1030 (believe F1040) is using a buffer to do active crossover. Thus B1 can be used to replaced the buffer in F1030. Refer to HERE for F1030 circuit diagrams.

 

[wintermute]

Hi Jaques, I note that Rod Elliot included an inverting buffer in his 12db/octave crossover for the LP section (which seems a bit odd as I would have thought the tweeter would be the one you invert). Anyway I guess the question is, what were you planning on doing with respect to the 180 degree phase shift of low compared to high? reverse the polarity on the speakers for one (as would be done in a passive crossover)?

Tony.

 

[R-K Rønningstad]

Maybe Rod thought it better to include additional amplifier stages in the LF path due to the added distortion being less audible here?

Anyway, be that as it may, when making this dedicated filter one could as well reverse polarity at one driver.
But, why not opt for a LR4 filter? Have a look at Linkwitz' pages. And as he states, if you do not do correct for other things when running active, it is really no use. Well, have a look at the man's page!

All the best!

RK

 

[Jacques Merde]

Hi Jaques, I note that Rod Elliot included an inverting buffer in his 12db/octave crossover for the LP section (which seems a bit odd as I would have thought the tweeter would be the one you invert). Anyway I guess the question is, what were you planning on doing with respect to the 180 degree phase shift of low compared to high? reverse the polarity on the speakers for one (as would be done in a passive crossover)?
Tony.

My plan is to reverse the leads to one of the drivers. I don't see an advantage to doing anything more elaborate than that. For the proper operation of the crossover it is necessary to reverse the polarity of one of the outputs -- it doesn't matter which one.

 

[udailey]

Tone Control

Jacques,
I use full range drivers. People are always telling me to run a BSC circuit to tone down the top frequencies and then the frequency response will flatten out a little more. What I was thinking was to put this between source and amp and use it to in effect replace the BSC by dividing the frequencies, adding some gain to the bass and putting the signals back together before they hit the amp. Tone control. I could do that with this circuit right? The way I read the circuit I could increase/decrease gain on either HF or LF. Is it this easy or do all the other values change as well?
Uriah

 

[Jacques Merde]

But, why not opt for a LR4 filter? Have a look at Linkwitz' pages.

There is no reason not to. It would be straightforward to implement as it essentially consists of two 2 pole filters in series. In that case it may be possible to omit the additional buffers for the feedback caps in the LP section. The stopband attenuation without them will be in the neighborhood of -100 dB.

And as he states, if you do not do correct for other things when running active, it is really no use. Well, have a look at the man's page!

Yes, there is a lot of good information on the Linkwitz site. I believe that there is significant advantage to be achieved by active crossovers by virtue of bandwidth limiting the power amplifiers (less distortion) and by better coupling of the amplifier outputs to the individual drivers, even if you do not follow through as thoroughly as SL indicates. Bear in mind that this crossover is just one building block in a system.

 

[Jacques Merde]

Jacques,
I use full range drivers. People are always telling me to run a BSC circuit to tone down the top frequencies and then the frequency response will flatten out a little more. What I was thinking was to put this between source and amp and use it to in effect replace the BSC by dividing the frequencies, adding some gain to the bass and putting the signals back together before they hit the amp. Tone control. I could do that with this circuit right? The way I read the circuit I could increase/decrease gain on either HF or LF. Is it this easy or do all the other values change as well?
Uriah

Uriah,

I would not implement a BSC that way at all. Linkwitz shows implementaions of active and passive shelving networks on his site for just this purpose. A passive shelving filter nestled between two B1s would probably better suit your purpose.

 

[R-K Rønningstad]

There is no reason not to. It would be straightforward to implement as it essentially consists of two 2 pole filters in series. In that case it may be possible to omit the additional buffers for the feedback caps in the LP section. The stopband attenuation without them will be in the neighborhood of -100 dB.



Yes, there is a lot of good information on the Linkwitz site. I believe that there is significant advantage to be achieved by active crossovers by virtue of bandwidth limiting the power amplifiers (less distortion) and by better coupling of the amplifier outputs to the individual drivers, even if you do not follow through as thoroughly as SL indicates. Bear in mind that this crossover is just one building block in a system.

Yes. I did a LR4 with dipole compensation with SL's circuitry. It works very well. It uses opamps. But it is possible to implement with JFETs as well, and the B1 or similar is well suited. If you build a x-over with "normal" discrete transistors, you normally have lots of stages (input diff, output follower etc) and lots of transistors of course, and multiplied by four for LR4, by two for stereo and then dipole compensation etc.
But with B1 circuits it becomes less complicated.

 

[Jacques Merde]

Yes. I did a LR4 with dipole compensation with SL's circuitry. It works very well. It uses opamps. But it is possible to implement with JFETs as well, and the B1 or similar is well suited. If you build a x-over with "normal" discrete transistors, you normally have lots of stages (input diff, output follower etc) and lots of transistors of course, and multiplied by four for LR4, by two for stereo and then dipole compensation etc.
But with B1 circuits it becomes less complicated.

Yes, those JFETs start to add up quickly . FWIW, the JC/EB complementary JFET buffer looks like an attractive option as well, IMO.

 

[wintermute]

Thanks RK and Jaques, I see SL simply says to reverse one driver too.

LR4 would be an option, but I do like the idea of using bessel filters compared to butterworth, after having seen the difference in group delay (though SL says it isn't detectable)... it looks nasty on the graphs I have seen. and the LR implementation seems to not suffer from the usual bessel tradeoff of a non flat response.

Tony.

 

[d to the g]

Have a look at Linkwitz' pages.

SL uses the Sallen-Key topography...

In a very clever paper, Peter Billam (JAES v26 n6, p426, 1978) showed that the positive feedback “aggravated” the distortion through the crossover region, and by a surprisingly high amount. His measurements suggest an 80-fold increase in distortion at the corner frequency.

Quote taken directly from SY's archeron crossover page. The archeron offers a pretty nifty way of doing active 2nd (and 4th) order crossovers without resorting to the sallen key or state variable methods.

syclotron.com

 

[Jacques Merde]

SL uses the Sallen-Key topography...



Quote taken directly from SY's archeron crossover page. The archeron offers a pretty nifty way of doing active 2nd (and 4th) order crossovers without resorting to the sallen key or state variable methods.

syclotron.com