B1 Active Crossover

[Jacques Merde]

Originally Posted by Nikolas Ojala

The B1 buffer is essentially a voltage follower.
A voltage follower is also an essential part of a Linkwitz-Riley filter.
See this: Active Filters

Has anyone tried the B1 voltage follower as a component in a Linkwitz-Riley filter?

Here is my initial cut at a 2 pole LR filter using B1. This one crosses over at roughly 700 Hz (for my own personal project). Other frequencies are, of course, easily implemented. The filter seemed to lend itself to the symmetric supply topology, so that's what I have drawn here. I'm planning to use a beefed up Salas shunt regulator with this, as in the building-symmetrical-psu-b1-buffer thread.

[maxlorenz]

Hi Jacques Merde,

Thank you for posting your filter schematics.
That is the same approach I will take in the distant future for my 1KHz filter, safe that:
1) I can use balanced input signal from TVC so I will put a balanced to unbalanced input buffer instead of a B1.

2) Pots are usually noisier than good quality R so instead of using that 22K pot I will let fixed the HF output and only use a resistor based attenuator for adjusting the bass output. This brings more transparency to the sound, IMHO.

Good luck,
M

[AndrewT]

Originally Posted by Jacques Merde

Here is my initial cut at a 2 pole LR filter using B1. This one crosses over at roughly 700 Hz (for my own personal project). Other frequencies are, of course, easily implemented. The filter seemed to lend itself to the symmetric supply topology, so that's what I have drawn here. I'm planning to use a beefed up Salas shunt regulator with this, as in the building-symmetrical-psu-b1-buffer thread.

brilliant!
It looks so simple. Unity gain and equal value R & C automatically gives Q=0.5 for the L-R function and makes it so simple to match components.
Be careful to check for DC at the outputs. Both at switch on and as the devices warm up. The low pass is particularly susceptible since it is completely DC coupled all the way through.

[Jacques Merde]

maxlorenz and AndrewT,

Thanks for the good suggestions. At the moment, I am more concerned with dealing with the issue of the Sallen-Key topology where, for the low pass filter, the finite output impedance of the buffer in combination with the feedback capacitor provides a for a forward signal shunt path which manifests itself as poor stop band response. In this case and with these component values, the difficulties are well within the audio band. This can be addressed in a number of ways, so I am back to the drawing board to do some thinking....

Cheers,

[njepitt]

Originally Posted by Jacques Merde

Here is my initial cut at a 2 pole LR filter using B1. This one crosses over at roughly 700 Hz (for my own personal project). Other frequencies are, of course, easily implemented. The filter seemed to lend itself to the symmetric supply topology, so that's what I have drawn here. I'm planning to use a beefed up Salas shunt regulator with this, as in the building-symmetrical-psu-b1-buffer thread.

Hi everyone,

I think the circuit Jacques Merde posted here is fascinating, but can I ask a couple of possibly stupid questions?

1. Am I right in thinking the 22k pots shown in the circuit are trimpots, to set the relative levels for LP and HP? I presume the normal volume pots would come before this circuit altogether, right?

2. I presume "B+" and "B-" denote symmetric DC voltages from the power supply. Normal B1 values?

3. If you want to change the crossover point then you mess with the 22nF and 10.2k values throughout and leave everything else unchanged, correct?

4. At first sight this seems to be another place where one could use BF862 in place of 2SK170. The preamp juma designed (and which I may well build) uses the BF862 in a B1-type circuit with gain, to good reports. I've used the BF862 in a mini-aleph style amp (following juma's orientation), and like it a good deal. Is there any reason this circuit couldn't be done all SMD, with BF862 throughout? If you wanted to do this, what caps would you use?

Cheers

Nigel

[njepitt]

Another one: AndrewT mentioned matching components, but what specifically ought to be matched here?

Cheers

Nigel

[AndrewT]

each filter has a pair of equal R and a pair of equal C.
A crossover comprises a pair of filters.
That takes us up to 4off R and 4off C.
Stereo requires two channels. That brings our total to 8off R and 8off C.
For exceptional quality in rolloff and crossover characteristics and in channel matching all these R & C can be matched to <1% tolerance.

[njepitt]

Originally Posted by AndrewT

each filter has a pair of equal R and a pair of equal C.
A crossover comprises a pair of filters.
That takes us up to 4off R and 4off C.
Stereo requires two channels. That brings our total to 8off R and 8off C.
For exceptional quality in rolloff and crossover characteristics and in channel matching all these R & C can be matched to <1% tolerance.

Thanks. So, no matching of jfets (other than recommended for normal B1, I mean). Since close tolerance SMD components (resistors, anyway) are pretty cheap, wouldn't this be another reason for doing a surface-mount version? (Or did I miss something?)

Cheers

Nigel

[Jacques Merde]

Originally Posted by njepitt

Hi everyone,

I think the circuit Jacques Merde posted here is fascinating, but can I ask a couple of possibly stupid questions?

1. Am I right in thinking the 22k pots shown in the circuit are trimpots, to set the relative levels for LP and HP? I presume the normal volume pots would come before this circuit altogether, right?

2. I presume "B+" and "B-" denote symmetric DC voltages from the power supply. Normal B1 values?

3. If you want to change the crossover point then you mess with the 22nF and 10.2k values throughout and leave everything else unchanged, correct?

4. At first sight this seems to be another place where one could use BF862 in place of 2SK170. The preamp juma designed (and which I may well build) uses the BF862 in a B1-type circuit with gain, to good reports. I've used the BF862 in a mini-aleph style amp (following juma's orientation), and like it a good deal. Is there any reason this circuit couldn't be done all SMD, with BF862 throughout? If you wanted to do this, what caps would you use?

Cheers

Nigel

Hello Nigel,

1. Correct on both counts. My thinking is that the trimpots exist for the general case where one doesn't have advance knowledge of the system the crossover will be used in. Once the correct levels have been determined, the trimpots could be removed and replaced by jumpers and/or fixed resistor dividers for a more permanent installation.

2. Correct. I was thinking of +/- 9 or 10 volts, like the DCB1.

3. Correct.

4. I don't see any reason not to build this with a BF862. I haven't really put any thought into SMD cap selection at this point.

Thanks. So, no matching of jfets (other than recommended for normal B1, I mean). Since close tolerance SMD components (resistors, anyway) are pretty cheap, wouldn't this be another reason for doing a surface-mount version? (Or did I miss something?)

Correct, no special matching of JFETs, other than for Idss. As far as component matching goes, Panasonic 1% film caps are available from Digi-Key in those values, and 1% resistors are commonplace. AndrewT's thinking here is essentially identical to mine.

It's easier, for me anyway, to swap out leaded parts, but I don't see any reason not to build a SMD version.

These are all reasonable questions -- not stupid questions

I do have a solution for the low pass bandstop response issue, which I will post later.

Thanks,

[Tea-Bag]

Jaq.
I think this is a cool design, worthy of it's own thread.