Q factor and SRF - how are these related?

[Plasmateur]

I'm looking for some inductors with a particular to serve as a notch filter.

When looking through Digi-Key or Mouser I see that their are SRF values and Q factor values.

When picking out the SRF values, let's say 13MHz, the spec sheets usually list a Q value at a much lower frequency than the SRF frequency.

Should I pick out the inductor with a high Q as close the the SRF to achieve what I'm looking for?

[TimFox]

Are you going to use the naked inductor to give a notch at its SRF (the frequency at which it parallel-resonates without additional capacitance)?
The manufacturer's data are measured using standard frequencies (based on traditional Q-meters, q.v.) which are always below self-resonance, since they must be measured where the reactance is inductive.
Normally, you would use an inductor at a frequency below its SRF and add a capacitor (probably adjustable) in parallel to get a useful notch.

[Plasmateur]

Are you going to use the naked inductor to give a notch at its SRF (the frequency at which it parallel-resonates without additional capacitance)?
The manufacturer's data are measured using standard frequencies (based on traditional Q-meters, q.v.) which are always below self-resonance, since they must be measured where the reactance is inductive.
Normally, you would use an inductor at a frequency below its SRF and add a capacitor (probably adjustable) in parallel to get a useful notch.

I've used some before without looking to deep into the Q factor for the purposes of a RF compensated Langmuir Probe and they seemed to have worked for the application. Its time to buy some more and after reading further into why these inductors are used, one of the papers cited a high Q value.

Just out of curiosity I hooked my sig-gen up to one of my previously purchased inductors with a set SRF and then into my scope and noticed attenuation close the SRF frequency.

Then I did the same with a network analyzer and it appears the attenuation was offset to the SRF by a few MHz.

I can try building a notch with an adjustable cap....so long as the adjustable cap is very tiny. I'll have to look into those. Thanks!

[Plasmateur]

Every packaged component element whether R or C or L is physically in some package and physically has some kind of construction using wires, metal sheets, conductive compounds, insulating compounds.
Every insulator or conductor can affect the capacitance of nearby conductors.
Every wire or conductor when current travels through it has some associated inductance.
So the point is regardless of your PCB, just the packaged component itself will not be a pure R, L, C or whatever due to these parasitic resistances, inductance, capacitance values not intended but inevitably present to a degree.

So what might 'look like' an inductor at low to medium frequencies also possesses stray parasitic capacitances which can act as if they are in shunt to the inductance.  Also the inductive materials themselves like ferrites or whatever can have a different permeability with frequency causing the inductance of the inductor to decrease with frequency and maybe decrease with increased current flow as well.

So the SRF is the self resonant frequency where the phase of current of the measured impedance of the element looks resistive or "in phase" because over the span of frequencies from low frequency up to the SRF the measured inductance of the packaged device gradually keeps decreasing while the capacitive reactance of the shunt parasitic capacitances keep decreasing.  In a pure inductor the phase of the current lags the phase of the voltage by 90 degrees.  In a pure capacitor the phase of the current leads the phase of the voltage by 90 degrees.  In a pure resistance the phase of the voltage is the same as the phase of the current.  So as frequency increases an inductor that has a 90 degree leading phase of voltage to current will have shunt capacitive reactances which cause the combined L, C, R (with C, R being parasitic) circuit starts to creep in phase from 90 degree voltage leading current (perfectly inductive) toward 0 degrees curent versus voltage (basically resistive at that specific frequency since the inductive reactance component is canceled by a capacitive reactive component) and ultimately it will switch to looking more like a capacitive phase than inductive one as you go up past that frequency.  More complex L, C, R series / shunt circuits are possible by the time you account for all the possible parasitic R, L, C components of the package and your circuit wiring / PCB etc.

So basically you can't "meaningfully" expect to use the inductor very near, at, or above the SRF since it isn't specified to "look" inductive in phase at or above that frequency.  Usually you want to design your circuits so the highest frequency you care much about is well under the inductor's rated SRF.  So if your circuit operates at 1 MHz and the inductor SRF is, say, 10 MHz, 100 MHz, 1 GHz, great.  If you operate at 1 MHz and the SRF is only specified to be 2 MHz, well. you'll probably 'see' less inductive phase than you expect and less impedance than you expect because you're getting to the area where the inductive reactance is not overwhemling the other parasitics and other circumstances.

So the best way to design is:
(a) choose an inductor with a rated SRF is at least significantly higher than your circuit's maximum frequency where it needs to see something that looks close to a 'pure' inductance.  50% margin?  100% margin?  1000% margin?  Depends on your circuit priorities and the available inductors and cost / size / performance.

(b) Choose a inductor rated for the maximum DC or low to medium frequency current you expect to see in your circuit because saturation can occur in materials like ferrites or ferromagnets so that the inductance drops with increasing current.  Also the current will cause a temperature rise so sometimes inductance decrease (saturation) limits your peak current, other times the temperature heating will cause a limit of the peak current. 

(c) Choose an inductor with a stable enough value of inductance over the lowest to the highest frequency range your circuit needs to 'look inductive' with a consistent value of inductance.  So if it operates from say 10kHz up to 1MHz and you need 10uH then pick a SRF above 1MHz and inductor where the inductance over that frequency range is consistent e.g. within 20% or 50% or whatever your tolerance is relative to the low frequency value of inductance.  As you get closer to the SRF the effective inductance will start to dip appreciably.

(d) Choose a Q high enough over the circuit range of frequencies you care about.  Q may be high at 1kHz but less high at say 10kHz or 1MHz.  So consider that in your filter's performance.  The unloaded Q is a figure of merit for the inductor component itself but the LOADED Q is what affect your circuit's R, L, C topology has in composite so if you had, say, a 1 ohm resistor in series with an 1mH inductor and 1uF capacitor which by themselves had unloaded 1kHz Q values of 1000, you'd now have a much lower loaded Q due to the other resistance in your circuit.  Or combining a L with Q unloaded = 100 with a capacitor Q unloaded = 20 is not going to give a very high loaded tank Q due to the capacitor.  Model your filter response with best and worst case parasitics and inductance / capacitance / resistance / Q values and see what you get is acceptable over frequency.  A higher 'Q' component is always 'better' at the SPECIFIC frequency they measure it at under the specific signal current / magnitude used.  But it may not be so good at a higher frequency.  So maybe for a circuit needing a good characteristic over say 10kHz to 1MHz your best Q value inductor at 10kHz may not perform so well at 1MHz so maybe optimizing for the 1MHz or 500 kHz performance might overall get you the best results depending on your needs.

The higher SRF materials will have more consistent performance of inductance from near DC to XX % of the SRF, so they are more broadband.  But it is not practically possible to make large values of inductance that ALSO have high SRF values since the SRF is dictated in part by whether they make the inductor of iron laminations, iron poweder, ferrite, amorphous ferromagnet, air core, whatever.  So you don't find 10mH air core inductors that are very practical (they would be huge).  But if you need less than maybe 100uH it is not impossible to get a (large) air core inductor, or a much smaller ferrite one, or an even smaller iron core one, but the SRF and Q would be very different among the choice of materials as well as the DC current rating for a given physical size and construction of inductor.

So choosing cost vs frequency range vs physical size vs DC current handling capacity vs Q vs SRF is often a balancing act between choosing core materials, shapes/sizes, cost, SRF, tolerance, temperature range, etc.

Get the best Q, tolerance, current/power capability, SRF you can afford / accept of what is available but understand what is 'good enough' and what is unacceptable based on your filter overall performance / response 'pass / fail' criteria.
There isn't a hard answer as to best only "this is possible" or "this is not possible" and then degrees of "better" or "worse" in performance, cost, size, etc/.

There is alot to take in here and I thank you for taking the time to go over this with me. Now I'm rethinking one of my particular applications given the information I've gained in this thread. Thank you very much!

[G0HZU]

Every packaged component element whether R or C or L is physically in some package and physically has some kind of construction using wires, metal sheets, conductive compounds, insulating compounds.
Every insulator or conductor can affect the capacitance of nearby conductors.
Every wire or conductor when current travels through it has some associated inductance.
So the point is regardless of your PCB, just the packaged component itself will not be a pure R, L, C or whatever due to these parasitic resistances, inductance, capacitance values not intended but inevitably present to a degree.

So what might 'look like' an inductor at low to medium frequencies also possesses stray parasitic capacitances which can act as if they are in shunt to the inductance.  Also the inductive materials themselves like ferrites or whatever can have a different permeability with frequency causing the inductance of the inductor to decrease with frequency and maybe decrease with increased current flow as well.

So the SRF is the self resonant frequency where the phase of current of the measured impedance of the element looks resistive or "in phase" because over the span of frequencies from low frequency up to the SRF the measured inductance of the packaged device gradually keeps decreasing while the capacitive reactance of the shunt parasitic capacitances keep decreasing.  In a pure inductor the phase of the current lags the phase of the voltage by 90 degrees.  In a pure capacitor the phase of the current leads the phase of the voltage by 90 degrees.  In a pure resistance the phase of the voltage is the same as the phase of the current.  So as frequency increases an inductor that has a 90 degree leading phase of voltage to current will have shunt capacitive reactances which cause the combined L, C, R (with C, R being parasitic) circuit starts to creep in phase from 90 degree voltage leading current (perfectly inductive) toward 0 degrees curent versus voltage (basically resistive at that specific frequency since the inductive reactance component is canceled by a capacitive reactive component) and ultimately it will switch to looking more like a capacitive phase than inductive one as you go up past that frequency.  More complex L, C, R series / shunt circuits are possible by the time you account for all the possible parasitic R, L, C components of the package and your circuit wiring / PCB etc.

So basically you can't "meaningfully" expect to use the inductor very near, at, or above the SRF since it isn't specified to "look" inductive in phase at or above that frequency.  Usually you want to design your circuits so the highest frequency you care much about is well under the inductor's rated SRF.  So if your circuit operates at 1 MHz and the inductor SRF is, say, 10 MHz, 100 MHz, 1 GHz, great.  If you operate at 1 MHz and the SRF is only specified to be 2 MHz, well. you'll probably 'see' less inductive phase than you expect and less impedance than you expect because you're getting to the area where the inductive reactance is not overwhemling the other parasitics and other circumstances.

So the best way to design is:
(a) choose an inductor with a rated SRF is at least significantly higher than your circuit's maximum frequency where it needs to see something that looks close to a 'pure' inductance.  50% margin?  100% margin?  1000% margin?  Depends on your circuit priorities and the available inductors and cost / size / performance.

(b) Choose a inductor rated for the maximum DC or low to medium frequency current you expect to see in your circuit because saturation can occur in materials like ferrites or ferromagnets so that the inductance drops with increasing current.  Also the current will cause a temperature rise so sometimes inductance decrease (saturation) limits your peak current, other times the temperature heating will cause a limit of the peak current. 

(c) Choose an inductor with a stable enough value of inductance over the lowest to the highest frequency range your circuit needs to 'look inductive' with a consistent value of inductance.  So if it operates from say 10kHz up to 1MHz and you need 10uH then pick a SRF above 1MHz and inductor where the inductance over that frequency range is consistent e.g. within 20% or 50% or whatever your tolerance is relative to the low frequency value of inductance.  As you get closer to the SRF the effective inductance will start to dip appreciably.

(d) Choose a Q high enough over the circuit range of frequencies you care about.  Q may be high at 1kHz but less high at say 10kHz or 1MHz.  So consider that in your filter's performance.  The unloaded Q is a figure of merit for the inductor component itself but the LOADED Q is what affect your circuit's R, L, C topology has in composite so if you had, say, a 1 ohm resistor in series with an 1mH inductor and 1uF capacitor which by themselves had unloaded 1kHz Q values of 1000, you'd now have a much lower loaded Q due to the other resistance in your circuit.  Or combining a L with Q unloaded = 100 with a capacitor Q unloaded = 20 is not going to give a very high loaded tank Q due to the capacitor.  Model your filter response with best and worst case parasitics and inductance / capacitance / resistance / Q values and see what you get is acceptable over frequency.  A higher 'Q' component is always 'better' at the SPECIFIC frequency they measure it at under the specific signal current / magnitude used.  But it may not be so good at a higher frequency.  So maybe for a circuit needing a good characteristic over say 10kHz to 1MHz your best Q value inductor at 10kHz may not perform so well at 1MHz so maybe optimizing for the 1MHz or 500 kHz performance might overall get you the best results depending on your needs.

The higher SRF materials will have more consistent performance of inductance from near DC to XX % of the SRF, so they are more broadband.  But it is not practically possible to make large values of inductance that ALSO have high SRF values since the SRF is dictated in part by whether they make the inductor of iron laminations, iron poweder, ferrite, amorphous ferromagnet, air core, whatever.  So you don't find 10mH air core inductors that are very practical (they would be huge).  But if you need less than maybe 100uH it is not impossible to get a (large) air core inductor, or a much smaller ferrite one, or an even smaller iron core one, but the SRF and Q would be very different among the choice of materials as well as the DC current rating for a given physical size and construction of inductor.

So choosing cost vs frequency range vs physical size vs DC current handling capacity vs Q vs SRF is often a balancing act between choosing core materials, shapes/sizes, cost, SRF, tolerance, temperature range, etc.

Get the best Q, tolerance, current/power capability, SRF you can afford / accept of what is available but understand what is 'good enough' and what is unacceptable based on your filter overall performance / response 'pass / fail' criteria.
There isn't a hard answer as to best only "this is possible" or "this is not possible" and then degrees of "better" or "worse" in performance, cost, size, etc/.

 

I'm looking for some inductors with a particular to serve as a notch filter.

When looking through Digi-Key or Mouser I see that their are SRF values and Q factor values.

When picking out the SRF values, let's say 13MHz, the spec sheets usually list a Q value at a much lower frequency than the SRF frequency.

Should I pick out the inductor with a high Q as close the the SRF to achieve what I'm looking for?

For a typical (medium to high unloaded Q) wound inductor component with typical self capacitance the measured inductance of the component will tend to increase as you go up in frequency towards the SRF. As you get close to SRF the measured inductance will climb rapidly and the impedance will also climb up very rapidly near SRF. Just beyond SRF the inductor will look like a tiny capacitor.

[G0HZU]

evb149: (c) Choose an inductor with a stable enough value of inductance over the lowest to the highest frequency range your circuit needs to 'look inductive' with a consistent value of inductance.  So if it operates from say 10kHz up to 1MHz and you need 10uH then pick a SRF above 1MHz and inductor where the inductance over that frequency range is consistent e.g. within 20% or 50% or whatever your tolerance is relative to the low frequency value of inductance.  As you get closer to the SRF the effective inductance will start to dip appreciably.

If you wound an air cored solenoid type inductor or used a powdered iron (toroid?) core the measured inductance would tend to go UP sharply as you approach SRF. This increase is due to the self capacitance of this type of inductor. These types of inductor are very common in RF bandpass, bandstop, lowpass or highpass filters. Some very lossy inductors using a lossy ferrite core can have an inductance profile that appears to show reduced inductance with increasing frequency but these aren't generally suitable for use in high Q notch filters.

[alexo]

Every packaged component element whether R or C or L is physically in some package and physically has some kind of construction using wires, metal sheets, conductive compounds, insulating compounds.
Every insulator or conductor can affect the capacitance of nearby conductors.
Every wire or conductor when current travels through it has some associated inductance.
So the point is regardless of your PCB, just the packaged component itself will not be a pure R, L, C or whatever due to these parasitic resistances, inductance, capacitance values not intended but inevitably present to a degree.

So what might 'look like' an inductor at low to medium frequencies also possesses stray parasitic capacitances which can act as if they are in shunt to the inductance.  Also the inductive materials themselves like ferrites or whatever can have a different permeability with frequency causing the inductance of the inductor to decrease with frequency and maybe decrease with increased current flow as well.

So the SRF is the self resonant frequency where the phase of current of the measured impedance of the element looks resistive or "in phase" because over the span of frequencies from low frequency up to the SRF the measured inductance of the packaged device gradually keeps decreasing while the capacitive reactance of the shunt parasitic capacitances keep decreasing.  In a pure inductor the phase of the current lags the phase of the voltage by 90 degrees.  In a pure capacitor the phase of the current leads the phase of the voltage by 90 degrees.  In a pure resistance the phase of the voltage is the same as the phase of the current.  So as frequency increases an inductor that has a 90 degree leading phase of voltage to current will have shunt capacitive reactances which cause the combined L, C, R (with C, R being parasitic) circuit starts to creep in phase from 90 degree voltage leading current (perfectly inductive) toward 0 degrees curent versus voltage (basically resistive at that specific frequency since the inductive reactance component is canceled by a capacitive reactive component) and ultimately it will switch to looking more like a capacitive phase than inductive one as you go up past that frequency.  More complex L, C, R series / shunt circuits are possible by the time you account for all the possible parasitic R, L, C components of the package and your circuit wiring / PCB etc.

So basically you can't "meaningfully" expect to use the inductor very near, at, or above the SRF since it isn't specified to "look" inductive in phase at or above that frequency.  Usually you want to design your circuits so the highest frequency you care much about is well under the inductor's rated SRF.  So if your circuit operates at 1 MHz and the inductor SRF is, say, 10 MHz, 100 MHz, 1 GHz, great.  If you operate at 1 MHz and the SRF is only specified to be 2 MHz, well. you'll probably 'see' less inductive phase than you expect and less impedance than you expect because you're getting to the area where the inductive reactance is not overwhemling the other parasitics and other circumstances.

So the best way to design is:
(a) choose an inductor with a rated SRF is at least significantly higher than your circuit's maximum frequency where it needs to see something that looks close to a 'pure' inductance.  50% margin?  100% margin?  1000% margin?  Depends on your circuit priorities and the available inductors and cost / size / performance.

(b) Choose a inductor rated for the maximum DC or low to medium frequency current you expect to see in your circuit because saturation can occur in materials like ferrites or ferromagnets so that the inductance drops with increasing current.  Also the current will cause a temperature rise so sometimes inductance decrease (saturation) limits your peak current, other times the temperature heating will cause a limit of the peak current. 

(c) Choose an inductor with a stable enough value of inductance over the lowest to the highest frequency range your circuit needs to 'look inductive' with a consistent value of inductance.  So if it operates from say 10kHz up to 1MHz and you need 10uH then pick a SRF above 1MHz and inductor where the inductance over that frequency range is consistent e.g. within 20% or 50% or whatever your tolerance is relative to the low frequency value of inductance.  As you get closer to the SRF the effective inductance will start to dip appreciably.

(d) Choose a Q high enough over the circuit range of frequencies you care about.  Q may be high at 1kHz but less high at say 10kHz or 1MHz.  So consider that in your filter's performance.  The unloaded Q is a figure of merit for the inductor component itself but the LOADED Q is what affect your circuit's R, L, C topology has in composite so if you had, say, a 1 ohm resistor in series with an 1mH inductor and 1uF capacitor which by themselves had unloaded 1kHz Q values of 1000, you'd now have a much lower loaded Q due to the other resistance in your circuit.  Or combining a L with Q unloaded = 100 with a capacitor Q unloaded = 20 is not going to give a very high loaded tank Q due to the capacitor.  Model your filter response with best and worst case parasitics and inductance / capacitance / resistance / Q values and see what you get is acceptable over frequency.  A higher 'Q' component is always 'better' at the SPECIFIC frequency they measure it at under the specific signal current / magnitude used.  But it may not be so good at a higher frequency.  So maybe for a circuit needing a good characteristic over say 10kHz to 1MHz your best Q value inductor at 10kHz may not perform so well at 1MHz so maybe optimizing for the 1MHz or 500 kHz performance might overall get you the best results depending on your needs.

The higher SRF materials will have more consistent performance of inductance from near DC to XX % of the SRF, so they are more broadband.  But it is not practically possible to make large values of inductance that ALSO have high SRF values since the SRF is dictated in part by whether they make the inductor of iron laminations, iron poweder, ferrite, amorphous ferromagnet, air core, whatever.  So you don't find 10mH air core inductors that are very practical (they would be huge).  But if you need less than maybe 100uH it is not impossible to get a (large) air core inductor, or a much smaller ferrite one, or an even smaller iron core one, but the SRF and Q would be very different among the choice of materials as well as the DC current rating for a given physical size and construction of inductor.

So choosing cost vs frequency range vs physical size vs DC current handling capacity vs Q vs SRF is often a balancing act between choosing core materials, shapes/sizes, cost, SRF, tolerance, temperature range, etc.

Get the best Q, tolerance, current/power capability, SRF you can afford / accept of what is available but understand what is 'good enough' and what is unacceptable based on your filter overall performance / response 'pass / fail' criteria.
There isn't a hard answer as to best only "this is possible" or "this is not possible" and then degrees of "better" or "worse" in performance, cost, size, etc/.

 

I'm looking for some inductors with a particular to serve as a notch filter.

When looking through Digi-Key or Mouser I see that their are SRF values and Q factor values.

When picking out the SRF values, let's say 13MHz, the spec sheets usually list a Q value at a much lower frequency than the SRF frequency.

Should I pick out the inductor with a high Q as close the the SRF to achieve what I'm looking for?

I think you might want to re-word this. I can't digest it.

[T3sl4co1l]

Note that Q drops when the test frequency is near the SRF, so it's not always a meaningful measurement.

At low frequencies, DCR is dominant; at high frequencies, EPR is dominant (mostly core and eddy current losses).

Tim