active filter

[k1jroth]

Hello this is active crossover which iv designed for my
ventd movie subwoofer.

Could someone better in elektronics theck it out and
tell wich mistakes iv made.

[richie00boy]

Pretty good You just need to look at some of the component values.

Some resistors are 4k or less. Try scaling your values to get these up a little as the TL07x starts struggling around 2k.

The 1k output resistors are much too large and will cause noise and slight output attenuation/load intolerance, 47-150 ohms will be fine.

[k1jroth]

added picture of filter

[carlosfm]

You need to add a resistor from +IN to ground, on the input of the first stage.
Those pots to set the lowpass frequency should be linear, not log.

[richie00boy]

I would argue that the resistor at IN is optional, but if you use something like 100k it will not hurt. A

Agree with Carlos on the pot.

A major flaw I missed earlier is the takeoff from the NFB point on the subsonic filter. This should be buffered else the feedback will be screwed up by the (variable as well depending on switch position, bad) loading and the filter fed from it will have skewy response as it's being driven from a relatively high source impedance.

In my own designs I always buffer filter inputs. You might like to do the same (i.e. put buffer at input before subsonic) unless you can be confident that the source will always be <500 ohms or so.

[carlosfm]

Originally posted by richie00boy
...if you use something like 100k it will not hurt.

Yes, that's just a protection, otherwise you can have DC without the source connected.

[k1jroth]

Picture was old verion sorry.

I will use opa2134 as amplifier and polystyrene 1% capasitors.
Is this okay.

[richie00boy]

Yes that will cope with the lowish impedances.

Have you addressed the feedback point issue I mentioned above?

[LukasLouw]

"You need to add a resistor from +IN to ground, on the input of the first stage.
Those pots to set the lowpass frequency should be linear, not log."

The first stage is already biased via the pot and endstop resistor, adding a 100K in parallel with those will be effectively parallel with the pot/endstop and will affect the frequency and Q of teh filter towards the lower end of the frequency settings of the pot, so don't add it.....

Did you mean adding the 100K from teh NC pole of S1, i.e. from the left of the 4K to ground? THAT is where you need bias for the LP filter in bypass mode.

The pots should be reverse log if anything to get a decent variation range.

Lukas

[carlosfm]

Originally posted by LukasLouw
The first stage is already biased via the pot and endstop resistor

You are right, I didn't see that.