4 Pole Butterworth high pass filter using TL072

[jimladladlooklike]

Hi folks

I'm messing around with high pass filters for a friend who wants one and I'm attempting to breadboard a 4th order filter with a roll off frequency of ~150Hz. Here's the schematic I'm using:



I'm getting a very small amount of signal at the output, so this is obviously not ideal. Wondering if anyone can tell me anything from the TL072 voltages?

The voltages on the TL072 are as follows:

1. 8.82
2. 8.82
3. 0.027
4. 0
5. 0.003
6. 8.82
7. 8.82
8. 9.36

Cheers folks

[idy]

Do you understand the schematic is for a bipolar power supply? And so you will need to send those two Rs (2.7k and 330) to bias voltage and not ground?

Do you think the values you have chosen (super high C and super low R) are desirable for a filter? Don't we usually opt for larger R and smaller C? I'm not an expert....

[bean]

Just to piggyback on idy's comment, you want something more like this:



Using an online calculator, you can get to 150Hz with a mix of larger resistors/smaller caps like he suggested:
http://www.calculatoredge.com/electronics/sk%20high%20pass.htm
Just play with Ra and Rb values if you want to use a standard cap value like 68n or 82n.

[jimladladlooklike]

Thanks for the replies.

I did wonder about the bipolar thing, thank you for clarifying. Having connected the 2.7k and 330r resistors to ground there was a bit more volume, but not much.

The reason I have attempted with the current component values is because I would like quite a drastic slope at the roll off point. Having used the filter simulator here - http://www.falstad.com/afilter/ - I adjusted the sliders accordingly to get 150Hz cut off and a steep roll off, then rounded off values slightly and got to breadboarding. Is there a reason I shouldn't use such values?

Another thing is when I accidentally knocked the 470r resistor out of the circuit, it created much more volume, pretty much at unity. (Although I was also testing different values at this point, which I can't remember.)

Thanks!

[antonis]

Is there a reason I shouldn't use such values?

The impedance thing..

[Rob Strand]

Is there a reason I shouldn't use such values?

The impedance thing..

More than 2k2 for most opamps.   Around 10k is a good start.   Use larger resistor values to make the caps smaller.
For resistors over 1M resistors and over you might get noise issues.

[bean]

And just to add a bit more-

You generally don't want to go lower than 10k on an bias resistor for the non-inverted input to an op-amp. And, that's assuming that you've got some sort of active stage proceeding it (IOW, a buffer or small gain stage with low impedance output which is pretty much a must). The larger resistor values afford smaller cap values (as mentioned) in this particular setting which also means lower cost and smaller foot-print. And, then there's the whole reactance portion of the caps too (there are people here a lot smarter than me that can clarify that point so I won't put my foot in my mouth here).

[PRR]

Use 10k nominal filter resistors with '741 '4558 '5532 990 opamps. Easy load, low hiss (for most uses), and saves several pennies on caps (those values you found are fairly expensive as audio filter grade caps).

Tubes, JFETs, and TL072 open up the 100k zone which saves another few pennies.

http://sim.okawa-denshi.jp/en/OPstool.php

[Rob Strand]

If you make the resistorsห too low you can also get problem with filter not attenuating well at high frequencies.

This example shows the effect,

https://electronics.stackexchange.com/questions/301834/opamp-performance-for-sallen-key-filter

The post talks differences between opamps. 

However, it's not just the opamp.   It's both the opamp *and* the resistor values.

The opamp determines what frequencies things go bad and the resistor + opamp
determines how bad it gets.   Low resistors cause more badness  .
(The root cause is the opamp's output impedance.)

[antonis]

resistorห

Mixed English & Russian алфави́т (а́збука)..??

[Rob Strand]

resistorห

Mixed English & Russian алфави́т (а́збука)..??

My keyboard must have switched to Thai mode when I edited the post. (s is on the same key as ห). 

[ElectricDruid]

If you make the resistorsห too low you can also get problem with filter not attenuating well at high frequencies.

This example shows the effect,

https://electronics.stackexchange.com/questions/301834/opamp-performance-for-sallen-key-filter

The post talks differences between opamps. 

However, it's not just the opamp.   It's both the opamp *and* the resistor values.

The opamp determines what frequencies things go bad and the resistor + opamp
determines how bad it gets.   Low resistors cause more badness  .
(The root cause is the opamp's output impedance.)

The multiple feedback filter is another solution to this performance-degraded-at-high-frequencies problem. The first version of my DigiDelay pedal used SK filters, but for filtering digital noise out of an audio signal, high frequency performance is probably pretty vital, so later on I switched it to MFB filters. It costs two resistors more, and the filter stage is inverting (not that that would matter for a 4th order filter like we have here - It flips it, then flips it back) but there are no other gotchas.

[ElectricDruid]

Also this is a very handy link for butterworth filters:

https://www.researchgate.net/profile/Muhammad-Khan-82/post/Which_method_is_better_in_order_to_design_a_4th_order_low_pass_filter/attachment/59d622056cda7b8083a1b848/AS%3A273803895083009%401442291343591/download/Lab8_ButterworthS13%281%29.pdf

One relevant section from that paper:

In the case of a fourth order Butterworth filter (n = 4), the correct response can be achieved by cascading two biquad stages. Each stage must be characterized by a specific value of gain (or Q) that achieves both a flat response and stability. For a fourth order Butterworth filter, the Q factors of the two stages must be Q = 0.541 and Q = 1.307. From (2) it follows that one stage must have a gain of 1.152 and the other stage a gain of 2.235. As a result the overall DC gain of the fourth-order Butterworth realized with two biquad stages will be equal to the product of the DC gain of the two stages, i.e., 2.575.

So your two two-pole stages need Qs of 1.307 and 0.541. For noise reasons, it's generally thought best to put the higher Q section first followed by the lower Q section. I second PRR's recommendation of this tool for the filter design:

http://sim.okawa-denshi.jp/en/OPstool.php

HTH

[jimladladlooklike]

Thanks all for the great info!

Following your collective advice and using the Sallen-Key HPF tool I came up with what now seem like more appropriate values (at least to me at this point). I then used the tool I used previously to input said values (or the closest ones I physically had on hand and came up with what looks like a nice result:



I think it needs a coupling cap on the output?

Regarding the low output impedance buffer stage before the actual filter circuit: Are there any things I should consider as this circuit will be used by a bass guitarist? Could I just search tagboardeffects for something that already exists?

Thank you again, I'm enjoying learning about this filter business

EDIT - I should mention that I've breadboarded this (resistors going to 4.5v as opposed to ground as per schem) and it does indeed cut quite a lot of low end, so that's good! It sounded marginally "weaker", which I've interpreted as both the loss of low end and also maybe a general loss of overall volume, but I'm hoping the input buffer will help with this issue. If not, then maybe a stage with a clean boost after the filter stage.

Cheers

[Rob Strand]

The multiple feedback filter is another solution to this performance-degraded-at-high-frequencies problem.

Yep, they have something to offer.

Thanks all for the great info!

You did well.

I think it needs a coupling cap on the output?

It's a good idea, and also add a resistor to ground to stop pops when you connect things.   If the output is going to drive a cable then its a good idea to add a resistor in series with the output.

Regarding the low output impedance buffer stage before the actual filter circuit: Are there any things I should consider as this circuit will be used by a bass guitarist? Could I just search tagboardeffects for something that already exists?

Thank you again, I'm enjoying learning about this filter business

EDIT - I should mention that I've breadboarded this (resistors going to 4.5v as opposed to ground as per schem) and it does indeed cut quite a lot of low end, so that's good! It sounded marginally "weaker", which I've interpreted as both the loss of low end and also maybe a general loss of overall volume, but I'm hoping the input buffer will help with this issue. If not, then maybe a stage with a clean boost after the filter stage.

It's a good idea to add a buffer.  I'm not sure how you are testing the filter but connecting the bass directly to that filter will load down the pickups if it is a passive bass.    That's where an input buffer helps.

Here's a generic circuit which shows both the buffer and output circuit.   Ignore the part between U1 and C2, place your filter in there.



If you increase the resistor values on the first-stage of your filter it is possible make the first filter stage input impedance high enough that you don't need an extra buffer stage. (I should add, in this case backing off the volume control on a passive bass can shift the filter frequency a bit.)

You should have the 47uF cap on the VREF line.  Some of the circuits early in the thread had this missing.

[ElectricDruid]

+1 for Rob's circuit, including the 47uF on the Vref. Although I think the 10uF output cap is overkill! Even if you were feeding a 10K input impedance on the next thing you plugged in, you'd still have a roll-off at 1.6Hz!! Given that 10K is *very* low for an input impedance and  you could easily go up to 16Hz (x10 up) or 32Hz (x20 up) I think you could get away with 470n or 220n with no problems and not need an electrolytic cap.

[Rob Strand]

+1 for Rob's circuit, including the 47uF won the Vref. Although I think the 10uF output cap is overkill! Even if you were feeding a 10K input impedance on the next thing you plugged in, you'd still have a roll-off at 1.6Hz!! Given that 10K is *very* low for an input impedance and  you could easily go up to 16Hz (x10 up) or 32Hz (x20 up) I think you could get away with 470n or 220n with no problems and not need an electrolytic cap.

It wouldn't hurt to make it smaller.    ATM you can drive it into a short or headphones with very little roll-off  .

[Elektrojänis]

It wouldn't hurt to make it smaller.    ATM you can drive it into a short or headphones with very little roll-off  .

If steep highpass is what is wanted here and the impedance of the input to be driven is known, he could have another pole for the filtering just by sizing the coupling cap.

[antonis]

the impedance of the input to be driven is known,

The above only stands for effect incorporated HPF..

[Rob Strand]

If steep highpass is what is wanted here and the impedance of the input to be driven is known, he could have another pole for the filtering just by sizing the coupling cap.

There's many ways to spin it.

I think a 4th order Butterworth is a good place to start for a 4th order version.   Play with the cut-off frequency to get the best sound - that's probably where the work really is.

In the past I've used second and third orders for the same purpose.   You will see second order high-pass filters on many modern bass amps.  The idea is to prevent excessive speaker excursions but you can also think of it as a tone tweak.  The 3-band preamp on the Music-Man Stingrays had a permanent high-pass filter at around 50/60 Hz.

There's an old trick were you replace the Q=0.541 section from the Butterworth with a Q=0.5 filter, which is just two cascaded first-order RC filters tuned to the same frequency.   You then tweak the f0 and Q on the Q=1.037 stage to approximate a 4th order Butterworth response.  The input or output cap can firm two RC filters.   Another way is only use the output RC as one of the filters and use a 3rd order single opamp filter,

http://sim.okawa-denshi.jp/en/Sallenkey3Hikeisan.htm

With the last scheme the first stage doesn't roll-off so you can use the first stage buffer as a buffered bypass.

I'm not sure of the OP is going to bypass this thing with a switch.