OPA627 Active HPF

[D_GR8_1]

Hi guys,

hehe, this is my first post.

I am trying to design a HPF using the OPA627 op-amp.
I have simulated the design in Protel and ran simulation, however I am getting some weird results. (or so I think).

My HPF is supposed to be a 4kHz HPF.

The schematic & simulation plot is attached!

From my understanding of filters, the crossover point should be the -3dB point, right?

From the plots the -3db point is CROSSOVER/2 or 2kHz.

Also at the 4kHz point, the attenuation is -6dB.

Why is that?

[sss]

welcome to the forum !

the values for the resistors are wrong , check the pic i attached theres the formulas u need

[Swedish Chef]

I guess you are trying to build a 2nd order Butterworth HP. Then you should double the value of some resistors and capacitors - can never remember which ones but you can either do the math or just look it up on any site. The Texas Instruments (http://www.ti.com) site has a Filter Wizard that relieves you of the work.

For most filter characteristics the cutoff point is specified as the -3 dB point. Not so for Linkwitz-Riley (two cascaded Butterworth stages, hence the name Butterworth-squared) where it is at -6 dB.

/Magnus

[D_GR8_1]

I used the formulas in the Texas instruments document "OPAMPS for everyone".

Are you saying those formulas are wrong?

[Swedish Chef]

No, those formulas are right. Recheck your math.

/Magnus

[D_GR8_1]

Which formulas would I need to use to design a 4th order Butterworth then?

Do I use the coefficients from the Filter Table ?

[Swedish Chef]

It was a while since I read that document, but for a 4th order Butterworth - yes I think the coefficients are in the table IIRC.
But why would you like to use a 4th order Butterworth for audio?

[D_GR8_1]

I like sharp cut-off curves. And a 24db/oct will provide that.

Or will it?

For example if I want to cut off my Subwoofer at 150Hz, I dont want to still play faint tunes at 500Hz.

What do you suggest?

[Ouroboros]

Download FILTERLAB (free) from the Microchip web site. It'll show you all you need.

[Swedish Chef]

Yes, sharp cutoffs are good because they reduce the comb-filtering effect that results when more than one driver are playing the same signal. And they keep each driver in its best range (out of cone-breakup) and thus reduces distortion.
24 dB/oct is pretty much the gold standard for active filters.

But as good as the Butterworth response is for many engineering applications it is not ideally suited for audio active filters. If you add up the resultant acoustic output from an LP and HP Butterworth filter you will have a 3 dB bump in response at the crossover point. That is why you usually use a Linkwitz-Riley response which sums up flat and as a side bonus has a linear phaseshift between the LP and HP section. Thus they are also sometimes called "phaselinear" filters.

So there is no surprise that 99.9% of the (analog) active filters for audio on the market are LR...

The LR transfer function is very easy to obtain as it is just composed of cascaded Butterworth stages. So a 4th order (24 dB/oct) LR is made by cascading two 2nd order Butterworth's with exactly the same component values. A dual opamp like the NE5532, AD8620, OPA2604, TL072 or whatever will do the trick for one channel.

Once you have gone active you will probably never look back to passive filters. There is a reason why active filters are the standard in all pro sound applications today with passive crossovers rapidly becoming obsolete. And there is a reason why the "audiophile" industry sticks to passive crossovers and exotic components. Feel free to choose.

/Magnus