I am designing the highpass and lowpass parts of an eq i am making, using 4th order butterworth filters. I have the frequencies i want to use, which gives me the C1,R1-R4 for highpass and R1,C1-C4 for lowpass. I plan to have at least 4 selectable frequencies via rotary switch for both highpass and low pass, which is a combined total of 16 filters. 4 highpass left, 4 highpass right, 4 lowpass left, 4 lowpass right.
My question is, do I make one filter, then keep duplicating it for all the other cutoffs just changing out the capacitors/resistors as needed to get the cutoffs i need? or is there a way to use less components?
If you think about it, 1 mono butterworth filter needs 4 capacitors, 4 resistors and 2 opamps. So i would need to multiply this times 16, which would be 64 capacitors, 64 resistors, 32 opamps, that just seems like a ton of components, i could be wrong though.
Thank you for any design help you can provide.
My question is, do I make one filter, then keep duplicating it for all the other cutoffs just changing out the capacitors/resistors as needed to get the cutoffs i need? or is there a way to use less components?
If you think about it, 1 mono butterworth filter needs 4 capacitors, 4 resistors and 2 opamps. So i would need to multiply this times 16, which would be 64 capacitors, 64 resistors, 32 opamps, that just seems like a ton of components, i could be wrong though.
Thank you for any design help you can provide.
You can use a subtractive filter, with perfect summed square wave response, and half
of the components and switching. One output will have just a 6dB rolloff, though.
Subtractive Crossover Networks
Last edited:
A 4pole Butterworth is made from two 2pole filters that combine to give the Butterworth Q at the frequency you desire.
The Q of each 2pole are different.
The lower Q goes first in the cascade.
Use Q=0.54118 for stage 1 and Q=1.3065 for the second stage.
These combined give the Butterworth Q= 0.707107
The Q of each 2pole are different.
The lower Q goes first in the cascade.
Use Q=0.54118 for stage 1 and Q=1.3065 for the second stage.
These combined give the Butterworth Q= 0.707107
Thank you. I am wondering how this helps me though? Is there something to what you said that would allow me to use less components? If there is, i am not picking up on it and i apologize. Any feedback would be greatly appreciated.
there are many ways to create a 2pole filter.
Read up on unity gain Sallen and Key low pass and high pass filters.
Compare to the gain settable S&K types. These are sometimes referred to as equal component value filters. This is the type I prefer, because it allows all the capacitors to be the same value. It also allows all the frequency setting resistors to be the same value. As said above you only need to switch, either the capacitors or the resistors to give you variable frequencies.
The 2pole ECVF uses 2matched capacitors. The 4pole will use 4 matched capacitors.
A combined low pass and high pass 4pole filter will use 8 matched capacitors.
For stereo you need
Read up on unity gain Sallen and Key low pass and high pass filters.
Compare to the gain settable S&K types. These are sometimes referred to as equal component value filters. This is the type I prefer, because it allows all the capacitors to be the same value. It also allows all the frequency setting resistors to be the same value. As said above you only need to switch, either the capacitors or the resistors to give you variable frequencies.
The 2pole ECVF uses 2matched capacitors. The 4pole will use 4 matched capacitors.
A combined low pass and high pass 4pole filter will use 8 matched capacitors.
For stereo you need
Once you have selected these, I suggest you don't change them. Change the resistors to change frequencies.
Last edited:
oh thats awesome. I will definitely look into this. I only chose the buttherworth because of flatter frequency response at the passband. I am designing a mastering grade eq so i wanted things to be as transparent as possible. Thank you for your input.
Awesome i will really check that out. The capacitors i wanted to use are about $100 each, so 64 of of them would put the unit cost wayyyy to high.
OTAs tend not to have the very best distortion characteristics so might not be suitable if you're after opamp-grade measurements of THD.
oh no, don't fall in to that trap!
I am also a freaky audiophile, and I have build many active crossovers, and my experience is that Polypropylene capacitors with <2% tolerance will do fine. There is no audible difference to paper-in-oil or other exotic capacitors for this exact use - precision in value is actually more important sound-wise ( and that is not even super critical).
Think of it this way: The capacitors are not simple AC-transfers / DC-traps, they are performing a filter task, so they are supposed to interact in some way.
The choice of opamps has a much, much larger effect on the sound than the capacitors has, so do rather spend your 100$ on these if you must.
Read the many threads on opamp rolling, and make sure you use them correctly (with the proper compensation-cap and stay within accepted input-impedance ranges)
Why?wfrohwein said:
I assume you are aware that 4th order Butterworth is a filter frequency/phase response, not a filter circuit? As AndrewT says, one way to implement a 4th order Butterworth response is to cascade two second order filters such as Sallen-Key - but note that neither of the two filters will itself be a Butterworth.
In my view there's precious little to gain shelling out three figures on capacitors if you have any opamps at all doing audio duty. If you're after maximum subjective transparency discrete classA stages are the solution. NP0 capacitors are a very cost-effective solution unless you need high values (say >1uF).
thats funny you say that. I had purchased 4 types of .47uF caps to test this filter with. I made audio files yesterday of each and in a blind test, i could note tell the difference. After looking, they were only 10% tolerance, so i will be looking for something tighter per your recommendation.
The only noticeable difference so far in my circuit, was when i replaced the OPA627's with a discrete opamp, the 2520 from capi, was i able to hear a difference. The sample i sent out, everyone said the discrete sounded fuller in a blind test.
I have a question about opamps if you dont mind. I am currently using INA137's and DRV134's to handle my input and output of balanced audio. Should these be discrete opamps as well? I have a schematic from a very popular mastering eq and they use a dc servo input to handle the balanced audio firstly going through an OPA2604 and a pair of MJE3055/MJE2955. So i am curious, if i went with discrete opamps for all my filters, would it be a waste to go through the ina137/drv134's ?
Thank you for your direction, i really appreciate it. I am open to any and all feedback as i am still in the design and breadboard phase.
even better, both. What better way to learn than hands on. I am a software programmer by day, went to school for circuit design (be it years ago) and I also master for my label. I am trying to bridge all these loves into one thing.
When i said design phase, i mean, i know what i want this eq to do, i know what filter cutoffs i want, i know what peaking / notch filters i want, now just to pick the individual component and put it all together.
Exactly just like that is how i currently have it. Balanced cable from my pure2 going into the INA137 chop, unbalanced output into my filters, final filter output into the DRV134 chip, then balanced cable out from there.
I have some transformers coming from cinemag as another option to test, just not sure which one i want to go with yet.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.