DF,
I am going to go back and re-read the various filter papers on my HDD.
There is obviously something that I have misunderstood.
At the moment, I see it as: all active filters can choose a Q value and that applies to all odd order and all even order filters, of the Bessel, Butterworth, Linkwitz type.
I don't use any of the ripple type filters and have not studied them.
I am going to go back and re-read the various filter papers on my HDD.
There is obviously something that I have misunderstood.
At the moment, I see it as: all active filters can choose a Q value and that applies to all odd order and all even order filters, of the Bessel, Butterworth, Linkwitz type.
I don't use any of the ripple type filters and have not studied them.
Andrew, to the best of my knowledge you are correct, eg a sallen key stage can be overdamped to get Q 0.7 Butterworth, or 0.5 Q Bessel 1st order. Df is correct that a simple RC has no Xi damping and has a simple TF but a quadratic TF can be made to perform identically. Essentially he is mathematically correct but misses the point entirely, and hasnt contributed anything, except a perpetuation of a semantic arguement. Once again Andrew thank you, you have (as usual) helped a great deal.
I'm not certain that a second-order filter can perform identically to a first-order filter. Perhaps in the limit of vanishingly small Q a second-order can behave like two cascaded first-orders?mondogenerator said:Df is correct that a simple RC has no Xi damping and has a simple TF but a quadratic TF can be made to perform identically.
Fortunately I would rather be mathematically correct and 'miss the point' than the converse. Assigning a fictitious value to a non-existent parameter might be a useful engineering trick, but it is best if people realise when they are doing this so they don't attach meaning to it.Essentially he is mathematically correct but misses the point entirely, and hasnt contributed anything, except a perpetuation of a semantic arguement.
What is the Q of two cascaded 1st order filters?
Or does this second order filter not have a Q?
What if we cascaded three first order passive filters each interposed with a buffer to ensure the near zero source impedance and near infinite load impedance, so that we have a third order passive filter? Would this too have no Q value?
Or does this second order filter not have a Q?
What if we cascaded three first order passive filters each interposed with a buffer to ensure the near zero source impedance and near infinite load impedance, so that we have a third order passive filter? Would this too have no Q value?
I believe the Q of two cascaded 1st order filters of the same corner frequency is 0.5. Different corners freqs give a different Q value. All second order filters have a Q.
A first order is 1+s - no other parameter so no Q.
Second order is 1+as+s^2. Q=1/a.
Third order is 1+as+bs^2+s^3. What is Q? It may be possible to define a Q-like parameter from some combination of a and b, but this will not fully characterise the filter as there are two parameters. As a minimum you would need Q1 and Q2.
A first order is 1+s - no other parameter so no Q.
Second order is 1+as+s^2. Q=1/a.
Third order is 1+as+bs^2+s^3. What is Q? It may be possible to define a Q-like parameter from some combination of a and b, but this will not fully characterise the filter as there are two parameters. As a minimum you would need Q1 and Q2.
LOL CHILL DF
EDIT:
scratch that, just tried on filter modeller and im wrong Megalols 😀
phase maxima would still be 180 deg anyway. doh!
A single reactor cannot resonate, and such has no resonance magnification (Q)
Like a pendulum in a vacuum, which cannot exceed a max freq due to its mass/inertia, and unlike a 2nd order spring/damper suspension system which can have a resonant Q
EDIT:
scratch that, just tried on filter modeller and im wrong Megalols 😀
phase maxima would still be 180 deg anyway. doh!
A single reactor cannot resonate, and such has no resonance magnification (Q)
Like a pendulum in a vacuum, which cannot exceed a max freq due to its mass/inertia, and unlike a 2nd order spring/damper suspension system which can have a resonant Q
Last edited:
Just realised that an hour later I answered my own question:DF96 said:Perhaps in the limit of vanishingly small Q a second-order can behave like two cascaded first-orders?
I believe the Q of two cascaded 1st order filters of the same corner frequency is 0.5.
Really? Really really?
Excuse me BUT.
2 RCs in series will add their Cs in the regular way (1/Ctot = 1/c+1/c) and the turnover point will change.*
*Disclaimer: I think that is correct, at least partially
Like you said, there is no Q in first order. You were right there.
There is no resonance with a single reactive component, only attenuation, since there is no negative reactance to counter the positive reactance. It is not a tuned circuit.
It will shift Cos Phi in a power engineering sense, and phase angle, but you need an inductor (or simulated inductance) to achieve resonance and Q magnification (either current resonance magnification in shunt reactors, or Voltage resonance magnification in series reactors)
I believe that the Q of 0.7 assigned to 1st order systems is due to the rate of attenuation, and half power -3dB / half voltage -6 dB point in the roll off (have to look this up)
Excuse me BUT.
2 RCs in series will add their Cs in the regular way (1/Ctot = 1/c+1/c) and the turnover point will change.*
*Disclaimer: I think that is correct, at least partially
Like you said, there is no Q in first order. You were right there.
There is no resonance with a single reactive component, only attenuation, since there is no negative reactance to counter the positive reactance. It is not a tuned circuit.
It will shift Cos Phi in a power engineering sense, and phase angle, but you need an inductor (or simulated inductance) to achieve resonance and Q magnification (either current resonance magnification in shunt reactors, or Voltage resonance magnification in series reactors)
I believe that the Q of 0.7 assigned to 1st order systems is due to the rate of attenuation, and half power -3dB / half voltage -6 dB point in the roll off (have to look this up)
Last edited:
It is not that simple. Two options:2 RCs in series will add their Cs in the regular way (1/Ctot = 1/c+1/c) and the turnover point will change.
1. use a buffer amplifier between two first-order filters, then you can just multiply the responses;
2. directly connect the two CR, then you get interaction between them so you need to do some algebra to find the total response.
Last edited:
I see your point,but the ~1msec in your example doesn't phase me much tbh.
who cares about that? its only to illustrate a point I am making. the delay can be anything depending on your filter freq and Q.
-RM
Anything? Within bounds im sure,depending on phase rotation id imagine. Imnot sure I follow you entirely. I don't have the ability to model in the way you have demonstrated, any help here?
My LP for eg is 5th order, RC @ 500Hz, 3.8kHz @ Q=0.618, 2.8kHz @ Q = 1.618.
My LP for eg is 5th order, RC @ 500Hz, 3.8kHz @ Q=0.618, 2.8kHz @ Q = 1.618.
Imnot sure I follow you entirely. I don't have the ability to model in the way you have demonstrated, any help here?
My LP for eg is 5th order, RC @ 500Hz, 3.8kHz @ Q=0.618, 2.8kHz @ Q = 1.618.
The plot was from an active filter... passive would do same (speaker cross-over etc.) Do you have access to MLSSA or MLS software?
Thx-RNMarsh
- Status
- Not open for further replies.
- Home
- Source & Line
- Analog Line Level
- filter Q Question