[luisdent]

I'm trying to build a passive notch filter eq based on the linkwitz er4s passive eq circuit. I'd like to know if anyone here is knowledgeable with rc filters and notch filters and that sort of thing and might be able to help me design one and answer some questions for me. Please chime in!
 
I basically want to lessen the depth of his notch filters to be about -4db at 2.5khz and -4db at 7.5khz. I realize you need to change some values of the components in the notch filter part of the circuit, but I can't wrap my brain around exactly how everything interacts together. Such as the amplifier output. I will be using a JDS Labs C5 headphone amplifier. I don't want to damage anything, and I want to make sure I get the correct results...

 

[stv014]

Is the filter intended to be used on the amplifier output, or the line input of the C5 ? In the former case, it is also important to know what the headphones are.

 

[ab initio]

  I'm trying to build a passive notch filter eq based on the linkwitz er4s passive eq circuit. I'd like to know if anyone here is knowledgeable with rc filters and notch filters and that sort of thing and might be able to help me design one and answer some questions for me. Please chime in!
 
I basically want to lessen the depth of his notch filters to be about -4db at 2.5khz and -4db at 7.5khz. I realize you need to change some values of the components in the notch filter part of the circuit, but I can't wrap my brain around exactly how everything interacts together. Such as the amplifier output. I will be using a JDS Labs C5 headphone amplifier. I don't want to damage anything, and I want to make sure I get the correct results...

 

  Is the filter intended to be used on the amplifier output, or the line input of the C5 ? In the former case, it is also important to know what the headphones are.

 
@stv014 brings up an excellent point. You probably don't want to try and put an passive eq circuit at the headphone output. Putting it at the line in will allow you to use more reasonably sized (and priced!) components and you won't have to worry about each specific headphone you use changes the filter characteristics.
 
Also, because headphones are stereo, you will need to carefully match the left and right circuits, otherwise you will introduce imbalance.
 
Cheers

 

[limpidglitch]

  I'm trying to build a passive notch filter eq based on the linkwitz er4s passive eq circuit. I'd like to know if anyone here is knowledgeable with rc filters and notch filters and that sort of thing and might be able to help me design one and answer some questions for me. Please chime in!
 
I basically want to lessen the depth of his notch filters to be about -4db at 2.5khz and -4db at 7.5khz. I realize you need to change some values of the components in the notch filter part of the circuit, but I can't wrap my brain around exactly how everything interacts together. Such as the amplifier output. I will be using a JDS Labs C5 headphone amplifier. I don't want to damage anything, and I want to make sure I get the correct results...

 
Have you played around with a software EQ to find out where those notches should be placed?
You are very unlikely to have those 'pressure points' in the exact same places as Siegfried.

 

[luisdent]

Yes i have. I know my exact frequency corrections needed, 2.5khz -4db medium width 7.5khz -4db narrow width.

This is for the er4s. Following the article on linkwitz he suggests using it after the amp output. I'd only use it on the er4s.

 

[limpidglitch]

Yes i have. I know my exact frequency corrections needed, 2.5khz -4db medium width 7.5khz -4db narrow width.

This is for the er4s. Following the article on linkwitz he suggests using it after the amp output. I'd only use it on the er4s.

 
If all you want to do is replicate Linkwitz' setup, but with shallower notches, all you need to do is to increase R3 in the below schematic.
Use the equations to find the exact values. (R1 is output impedance, R2 is headphone impedance)
 

But as the others have noted, if you already have an external amplifier, having it at the input would be a much more elegant solution.

 

[stv014]

  @stv014 brings up an excellent point. You probably don't want to try and put an passive eq circuit at the headphone output. Putting it at the line in will allow you to use more reasonably sized (and priced!) components and you won't have to worry about each specific headphone you use changes the filter characteristics.

 
A higher impedance filter for the line input would need smaller capacitors, but larger inductors. For example, 4.7 uF capacitance (available as large polypropylene capacitors) at 2500 Hz is enough for ~13.5 ohms, but for 1000 ohms at the same frequency one would need a ~64 mH inductor which is also large and roughly similarly priced anyway.
 
It is probably not an issue if different headphones change the filter response, as I guess the filter is intended to equalize a specific model of headphone. Although even for a single driver it is more difficult to design a filter if it is highly reactive (e.g. a balanced armature IEM).
 

Originally Posted by ab initio /img/forum/go_quote.gif
 
Also, because headphones are stereo, you will need to carefully match the left and right circuits, otherwise you will introduce imbalance.

 
Fortunately, with reasonable components the imbalance is likely to be less than whatever is already there between the drivers. But one can always buy more than needed, and measure and find the best matching parts.

 

[stv014]

Yes i have. I know my exact frequency corrections needed, 2.5khz -4db medium width 7.5khz -4db narrow width.

 
Medium width and narrow width are not quite exact. Do you have the Q values, or a frequency response graph ?

 

[luisdent]

 

Yes i have. I know my exact frequency corrections needed, 2.5khz -4db medium width 7.5khz -4db narrow width.

 
Medium width and narrow width are not quite exact. Do you have the Q values, or a frequency response graph ?

 
I do, and I believe they are similar at least visually to his widths. Here's the current eq i'm playing with that I derived using test tones that is very impressive to my ears...
 
-4db @2500hz 1.1 BW
-11.5 @8200hz 1.1 BW
 
I might find after listening a while the 11.5 cut is drastic or something, but right now that's not the case. this is making the er4s sound extremely linear to me. It turns it into a uerm practically. (at least the demo unit i have)
 

 

Yes i have. I know my exact frequency corrections needed, 2.5khz -4db medium width 7.5khz -4db narrow width.

This is for the er4s. Following the article on linkwitz he suggests using it after the amp output. I'd only use it on the er4s.

 
If all you want to do is replicate Linkwitz' setup, but with shallower notches, all you need to do is to increase R3 in the below schematic.
Use the equations to find the exact values. (R1 is output impedance, R2 is headphone impedance)
 

But as the others have noted, if you already have an external amplifier, having it at the input would be a much more elegant solution.

 
I don't understand how it differs whether it is before or after my amp? If my player is a sansa fuze for instance, the output impedance is 1ohm roughly. The c5 amp is 2.2 ohms, but that itself would have no effect on the er4s itself. So why would it matter it the passive filter were before or after the amp? What if I wasn't even using the amp? I just figured I'd use it because there would be a gain loss. But if it isn't too much of a loss I may not even need the amp. I'm not sure what to expect. I use a -6db cut on the fuze EQ and I'm fine with that.
 
Maybe I'm missing something, but I've read the page over and over, and I'm just not grasping two things. First, what values generate what cut amount and width. I see the formulas but they go in one eye and out the other.
 
And second, how the amp/source effect the circuit. I was told that i need to use the "alternate" values on the page with my low impedance amp output. I guess I don't get why that's the case. I had electronics courses back in high school quite a long time ago. I understand the basics and then some, but my brain just isn't following all of this. I get the concepts, but not the literal stuff I need for implementation.

 

[stv014]

By 1.1 BW, do you mean Q = 1 / 1.1 ? If yes, the frequency response would look like this:

Is the above graph correct ?

 

[stv014]

 
This still needs some tweaking, especially if you actually want a different frequency response from what is shown in the previous post. R1, L1, and everything else to the right is the approximate model of the ER4S (based on goldenears.net and innerfidelity.com measurements), and obviously should not be built into the circuit. V1 is the source (amplifier), if it has a known significant non-zero impedance, then that should be subtracted from R3. The same applies to L4/R4 and L5/R5, the resistors may need to be adjusted according to the serial resistance of the inductors. The 1.22 uF capacitor (C2) would in practice be parallel 1 uF + 220 nF capacitors.

 

[stv014]

By the way, if power efficiency is an issue (due to the < 0 dB gain, and lower impedance load presented to the amplifier), an alternate solution could be to connect parallel RLC circuits in series with the ER4S.
 

Originally Posted by luisdent /img/forum/go_quote.gif
 
I don't understand how it differs whether it is before or after my amp? If my player is a sansa fuze for instance, the output impedance is 1ohm roughly. The c5 amp is 2.2 ohms, but that itself would have no effect on the er4s itself. So why would it matter it the passive filter were before or after the amp?

 
It can matter because the impedance of both the load (headphones/IEMs, or amplifier line input) and the source (amplifier or line output) affects the frequency response. Also, although it might not be an issue with the Sansa Fuze, a line output may not like driving a low impedance and reactive passive filter circuit.
 

Originally Posted by luisdent /img/forum/go_quote.gif
 
I just figured I'd use it because there would be a gain loss. But if it isn't too much of a loss I may not even need the amp. I'm not sure what to expect. I use a -6db cut on the fuze EQ and I'm fine with that.

 
In the above example, the filter has about -2.4 dB gain. It also reduces the load impedance to the amplifier to about 40 ohms at 8.2 kHz, but that is likely still acceptable.
 

Originally Posted by luisdent /img/forum/go_quote.gif
 
And second, how the amp/source effect the circuit. I was told that i need to use the "alternate" values on the page with my low impedance amp output. I guess I don't get why that's the case. I had electronics courses back in high school quite a long time ago. I understand the basics and then some, but my brain just isn't following all of this. I get the concepts, but not the literal stuff I need for implementation.

 
The output impedance of the source (amplifier or Sansa Fuze) is added to the first, serial resistor (R3 in my circuit). Increasing it makes the notches deeper than originally intended, so R3 needs to be reduced by the same amount for the correct frequency response.
 
If you want to modify the filter:
- a lower R4 makes the 2.5 kHz notch deeper, while reducing R5 does the same for the 8.2 kHz notch
- the center frequency of the first notch is 1 / (2 * PI * sqrt(L4 * C2)) Hz; the second can be calculated similarly with L5 and C3
- increasing L4 and reducing C2 by the same ratio to keep L4*C2 unchanged makes the first notch narrower (again, the second one is controlled by L5 and C3)
 
For the simulated frequency response, I used LTspiceIV.

 

[luisdent]

  By 1.1 BW, do you mean Q = 1 / 1.1 ? If yes, the frequency response would look like this:

Is the above graph correct ?

 
BW not Q. Would look similar to that graph, but you need to convert BW to Q values.
 
http://www.sengpielaudio.com/calculator-bandwidth.htm
 
1.1 BW = 1.3 Q roughly

 

[luisdent]

 
 
This still needs some tweaking, especially if you actually want a different frequency response from what is shown in the previous post. R1, L1, and everything else to the right is the approximate model of the ER4S (based on goldenears.net and innerfidelity.com measurements), and obviously should not be built into the circuit. V1 is the source (amplifier), if it has a known significant non-zero impedance, then that should be subtracted from R3. The same applies to L4/R4 and L5/R5, the resistors may need to be adjusted according to the serial resistance of the inductors. The 1.22 uF capacitor (C2) would in practice be parallel 1 uF + 220 nF capacitors.

 
What app is that? Is there a mac equivalent?

 

[jiiteepee]

   
What app is that? Is there a mac equivalent?

http://www.linear.com/designtools/software/