Active Filter Network

[Matttcattt]

I currently have my sub crossed at 100Hz, and I have a linkwitz transform. I feel that it is time to create proper preamp with high pass filters. Basically, is my current circuit ok, or should I add some unity gain buffers anywhere? I calculated the values for the filters using the SLOA093 document from Texas Instroments. I have built each filter seperately on breadboards, and they work ok. I just dont know if I can stick them together like I have.

 

[joensd]

You can have a look here and compare that with the TI papers.
The sheets includes everything like subsonic filter, input buffer, linkwitz transfrom and variable gain.

 

[Matttcattt]

Thanks for the link, but ive already had a play with the Linkwitz Tranzform Program, and other spreadsheets.

 

[theChris]

all filters show are 2nd order, likely making them 12dB/oct unless they are signifigantly overdamped which i'm betting against given the closeness of values.

note that the source impedance is not shown. you may acheive better results adding a buffer on the inputs.

have not checked the values but it is likely the rumble filter and linkwitz transform filter will be near the same frequencies.

 

[Evan Shultz]

Yes, all your filters are indeed 2nd order, 12dB/octave.

You need a buffer at the very end to prevent the pot from getting loaded down.

 

[Matttcattt]

So adding buffers at each end is the only change I need to make?

Are the values of the components ok?

[edit]: Sorry if ive confused anyone with the orders of my filters. I thought 2nd order was 6db/oct.

 

[richie00boy]

Yeah, a buffer at the front end would be a good idea because there are a few stages to drive and you can never be sure what the source impedance is. A buffer on the low out mono is also needed as mentioned for the pot.

I would also add a resistor of 22-100 ohms at the output of each op-amp that will drive a cable to swamp against capable capacitance possibly sending the op-amp into instability.

You might also like to simply parallel two 0.1uF for your sub low-pass as well instead of using a 0.22uF, but this is being a little picky and I guess you are short on space. I'd also look into cascading two of those filters to cut off more sharply.

 

[carlosfm]

You need a buffer at the very end to prevent the pot from getting loaded down.

Why not place the pot before the rumble filter instead?
No additional buffer needed.

 

[richie00boy]

I may disagree. Depending on the value of the pot, the filter will be affected.

 

[Matttcattt]

I would also add a resistor of 22-100 ohms at the output of each op-amp that will drive a cable to swamp against capable capacitance possibly sending the op-amp into instability.

The buffers, or all the opamps?

You might also like to simply parallel two 0.1uF for your sub low-pass as well instead of using a 0.22uF, but this is being a little picky and I guess you are short on space. I'd also look into cascading two of those filters to cut off more sharply.

Which filters? The lowpass?

 

[richie00boy]

The buffers, or all the opamps?

output of each op-amp that will drive a cable

So that would be the buffers then.

Which filters? The lowpass?

for your sub low-pass as well instead of using a 0.22uF

It would seem so.

 

[Matttcattt]

I dont apear to have read you reply very well.

Thank you for your advice.

I'd better get on with building them...

[edit]: I've updated the diagram to check i've got the right idea.

 

[carlosfm]

I've updated the diagram to check i've got the right idea.

For the summing op-amp I would use lower value resistors (like 22k).
You have an input buffer now.
Also, you will optimize dc-offset if instead of connecting NI directly to ground you use a resistor.
That resistor value should be the parallel combination of the other three resistors.

 

[richie00boy]

Agree with that. Also I'd put the 100 ohm resistors on the outputs of the left and right lines as well. 22k would also be a good value of input impedance for your new input buffers.

 

[Matttcattt]

Ok, i've made some changes, and ringed them as the diagram is quite large... Is everything i've changed ok?

[edit]:forgot the image...

 

[richie00boy]

Looks good but those 10k resistors on the output to ground are redundant.

 

[carlosfm]

Looks good but those 10k resistors on the output to ground are redundant.

No, they are not redundant.
The chips need a path to ground when there's nothing connected on their outputs.

 

[richie00boy]

Why is that Carlos? In the dozens of such circuits I have built it has yet to appear a problem. Sure the input if floating needs them to provide bias current, but I don't see why the output needs them.

 

[carlosfm]

Why is that Carlos? In the dozens of such circuits I have built it has yet to appear a problem. Sure the input if floating needs them to provide bias current, but I don't see why the output needs them.

The output should not be floating, as it will if there's no device connected.
It needs a return path.
Just try what happens when an op-amp has it's output(s) floating. Touch the op-amp with your finger and feel the heat.

 

[pinkmouse]

What heat? I've never noticed any.