You can not buy a proper "swinging choke" today.
Swinging chokes made sense for class B radio modulators and 'linear' amplifiers. As you have figured, pointless for class A.
ALL chokes are IMHO pretty pointless for anything I would call a preamp. Chokes are all low-impedance at audio/power frequency. A tube voltage and current preamp is a high impedance.
Swinging chokes made sense for class B radio modulators and 'linear' amplifiers. As you have figured, pointless for class A.
ALL chokes are IMHO pretty pointless for anything I would call a preamp. Chokes are all low-impedance at audio/power frequency. A tube voltage and current preamp is a high impedance.
Typical applications for a choke input filter are large transmitting or class B amps where the current varied with the signal. Nothing wrong with a regular old choke as a power supply input. I did it on my last build to drop a lot of voltage. This is a nice explanation. https://www.diyaudio.com/community/threads/swinging-choke-or-smoothing-choke.96082/#post-1132057
One way to look at it is to remember that all chokes are swinging chokes - inductance decreases with increasing DC current - and they're all designed to optimize inductance for given amounts of copper and core. Without the requirement of maintaining a "critical inductance" at lower DC currents, they'd all converge to the balance of copper/core/gap used in conventional "smoothing" chokes.
This extra requirement shifts the copper/core/gap choice towards a more copper/ less core/ less (or shelved) gap balance, improving inductance at lower DC currents at the expense of inductance at higher DC currents. They'll often come with more winding insulation, but not always (mil-surp for instance). For your use, a conventional choke will work fine, remembering that DC voltage from a choke-input filter soars to the same value as a cap-input filter when unloaded (start-up, etc.).
All good fortune,
Chris
Choke Input B+ supplies, that have equal to or more than Critical inductance have:
The following advantages and disadvantages:
Can take more current from the same power transformer secondary versus a capacitor input filter (tradeoff: choke input filter has a lower B+ voltage than a capacitor input filter B+ voltage).
Cooler power transformer secondary windings with a choke input filter, for the same B+ output current as a capacitor input filter.
Lower transient current, and lower harmonic structure for a choke input filter, versus a capacitor input filter.
Think about the reduced possibility that a B+ supply ground loop might cause you to hear hum and noise in your speakers.
Chokes are heavier, larger, and more expensive than a capacitor.
Chokes have magnetic fields that might radiate to your output transformers.
But, capacitor input filters have large transient currents that might cause a ground loop to introduce voltages in other nearby parallel wires (hum and noise introduced into the amplifier wiring).
All the input choke filters in my B+ supplies work really well.
Just my opinions.
The following advantages and disadvantages:
Can take more current from the same power transformer secondary versus a capacitor input filter (tradeoff: choke input filter has a lower B+ voltage than a capacitor input filter B+ voltage).
Cooler power transformer secondary windings with a choke input filter, for the same B+ output current as a capacitor input filter.
Lower transient current, and lower harmonic structure for a choke input filter, versus a capacitor input filter.
Think about the reduced possibility that a B+ supply ground loop might cause you to hear hum and noise in your speakers.
Chokes are heavier, larger, and more expensive than a capacitor.
Chokes have magnetic fields that might radiate to your output transformers.
But, capacitor input filters have large transient currents that might cause a ground loop to introduce voltages in other nearby parallel wires (hum and noise introduced into the amplifier wiring).
All the input choke filters in my B+ supplies work really well.
Just my opinions.
Last edited:
Thanks! its starting to make sense. In psud I did get some weird times where the voltage soared up to nearly what a capacitor input supply would be while trying different henries. I couldn't figure out why and it went through the roof unloaded when I was trying to determine what resistor gave me my expected 40ma at approx 340v for the project. The need for adequate load would mean that I've chosen too high of an MA rating for my little load here with the c-14X (200 MA 6H), it would be easier to find the sweet spot if I went back to a 100 or 125 ma choke instead of 200 MA and get more inductance and load it more fully. I was thinking I had to overbuild the non-swinging choke for MA.
Thanks, yes I was going to put some ohms between first and second LC stages and in series with C1 for oscillation. psud only let's you do so much.
What I think I learned about choke input and why an animal like a swinging choke was invented...
The prime directive seems to be keeping the choke loaded with enough current so that the coil maintains good coupling to the core at all times. When there isn't enough flux coming off the coil due to low load it kind of decouples magnetically from the core and the result is the voltage goes up maybe ruining the Amp. So if the load doesn't vary (class a) its simple just select a choke that is maybe 30% more ma gapped than what your draw is. It will stay coupled magnetically, not saturate and be OK. But if your load varies you need a swinger where the gap is small and you might need a low ohms bleeder in case your load drops even below the smaller gap. But if the load goes up, you don't want to always saturate, how does a smaller gap help that side of the swing? Its all a compromise of finding sweet spots, selecting the right swinger range, knowing your min, max and average draw, etc. For this reason in todays world choke input is really only viable for class a amps it seems, where you can simply select a choke that stays loaded IOW not too big of a gap. With choke input PS it seems like bad but harmless things happen if the choke saturates with rising load. But disasterous things happen if the load drops to where the coil can't couple to the core and it begins to float on its own and the Amp suddenly sees 150 more volts it didn't expect.
The prime directive seems to be keeping the choke loaded with enough current so that the coil maintains good coupling to the core at all times. When there isn't enough flux coming off the coil due to low load it kind of decouples magnetically from the core and the result is the voltage goes up maybe ruining the Amp. So if the load doesn't vary (class a) its simple just select a choke that is maybe 30% more ma gapped than what your draw is. It will stay coupled magnetically, not saturate and be OK. But if your load varies you need a swinger where the gap is small and you might need a low ohms bleeder in case your load drops even below the smaller gap. But if the load goes up, you don't want to always saturate, how does a smaller gap help that side of the swing? Its all a compromise of finding sweet spots, selecting the right swinger range, knowing your min, max and average draw, etc. For this reason in todays world choke input is really only viable for class a amps it seems, where you can simply select a choke that stays loaded IOW not too big of a gap. With choke input PS it seems like bad but harmless things happen if the choke saturates with rising load. But disasterous things happen if the load drops to where the coil can't couple to the core and it begins to float on its own and the Amp suddenly sees 150 more volts it didn't expect.
Last edited:
Thanks Chris, yesterday grasping this was hard, but today it is all clear now. Intuitively a Newby would think a choke input choke has to have a bigger gap when its just the opposite. Even the OP phrased his question in this thought. The big compromise they made it seems was to not worry about a swinger saturating once in a while in a PA amp in order to make sure the magnetics always had enough load to stay coupled to not let the voltage pin the meters.
Exactly for this reason, a swinging choke is optimal for a changing current draw purpose (class AB amplifier). What you're describing for the voltage going up is not related to any magnetic core coupling, but critical current draw for the specific inductance for the choke input filter to enter regulation.
A swinging choke does just that. It has little to no airgap, resulting into high inductance that guarantees choke input operation at low currents. When current draw ramps up, the flux density goes up too and the choke begins to enter saturation. In this state, inductance starts dropping, but the power supply still operates in choke-input duty due to the fact critical inductance vs current is a linear function and choke saturation inductance vs current draw also behaves linearly in a region, until ohmic losses take over as main choke impedance (complete choke saturation).
By Patrick Turner simplified formula for critical inductance. Lcrit = Udc_out / Ima
Choke DC flux density, B = L * Idc / T * Afe ; where T stands for turns and Afe for core area.
In practice, it takes a lot of choke saturation (almost no inductance) until the filter starts translating from choke input to capacitor input.
A swinging choke does just that. It has little to no airgap, resulting into high inductance that guarantees choke input operation at low currents. When current draw ramps up, the flux density goes up too and the choke begins to enter saturation. In this state, inductance starts dropping, but the power supply still operates in choke-input duty due to the fact critical inductance vs current is a linear function and choke saturation inductance vs current draw also behaves linearly in a region, until ohmic losses take over as main choke impedance (complete choke saturation).
By Patrick Turner simplified formula for critical inductance. Lcrit = Udc_out / Ima
Choke DC flux density, B = L * Idc / T * Afe ; where T stands for turns and Afe for core area.
In practice, it takes a lot of choke saturation (almost no inductance) until the filter starts translating from choke input to capacitor input.
Thanks, now I'm starting to get the whole regulation aspect, it makes total sense. I'll have to erase my coupling reasoning for the voltage pinning, I was thinking when flux density goes down that was the cause of the voltage pinning. What is Udc out above? So its really not a bad thing in a class A Amp situation to be running the choke close to saturation, but not more, and there I should be in critical inductance? With that rule of thumb I can do it off the mfg specs. alone right? Lets say my draw at the T1 is 50ma, then selecting a 60 to 80 ma choke will then be at critical inductance, or close right? Assuming all off the shelf buys.
Last edited:
For those of us with 60Hz power mains, the formula to calculate Critical Inductance is:
350 / Load current in mA
Example: 35mA load current: 350 / 35 mA = 10 Henry choke
For those of us with 50Hz power mains, the formula to calculate Critical Inductance is:
420 / Load current in mA
Example: 35mA load current: 420 / 35 mA = 12 Henry choke
Remember, if the manufacturer says his choke is 10H or 12H, what if it is actually 10% lower than the nominal rating.
So, for the examples above, be a little conservative in order to know that you have enough inductance to meet the Critical Inductance requirement.
For the 10 Henry choke, purchase a 12 Henry or 15 Henry choke.
For the 12 Henry choke, purchase a 15 Henry choke or a 20 Henry choke.
If your load current is 35mA, use a 50mA or higher rated choke (or a similar somewhat higher rating than the actual load current).
Then you do not have to worry that the choke is not a real 'swinging' choke.
Some choke have maximum voltage ratings.
Do not use a choke rated for 400V if the B+ is 400V (the peak voltage is about 1.4 x the secondary Vrms, or 1/2 secondary Vrms; per bridge or full wave center tapped rectification).
Choke input filters output about 0.9 x secondary Vrms (whole secondary Vrms for bridge rectification; 1/2 secondary Vrms for center tapped full wave rectification).
But . . . before the output tubes warm up, the is very low load current, so . . . the voltage may rise to about 1.4 x secondary Vrms (whole secondary Vrms for bridge rectification; 1/2 secondary Vrms for center tapped full wave rectification).
The actual B+ voltage also depends on the rectifier voltage drop (less than 1V or 2V solid state bridge or full wave rectifier circuits) typically 40 to 60V for tube rectifiers; the primary DCR, secondary DCR, and choke DCR, all versus the DC load current.
350 / Load current in mA
Example: 35mA load current: 350 / 35 mA = 10 Henry choke
For those of us with 50Hz power mains, the formula to calculate Critical Inductance is:
420 / Load current in mA
Example: 35mA load current: 420 / 35 mA = 12 Henry choke
Remember, if the manufacturer says his choke is 10H or 12H, what if it is actually 10% lower than the nominal rating.
So, for the examples above, be a little conservative in order to know that you have enough inductance to meet the Critical Inductance requirement.
For the 10 Henry choke, purchase a 12 Henry or 15 Henry choke.
For the 12 Henry choke, purchase a 15 Henry choke or a 20 Henry choke.
If your load current is 35mA, use a 50mA or higher rated choke (or a similar somewhat higher rating than the actual load current).
Then you do not have to worry that the choke is not a real 'swinging' choke.
Some choke have maximum voltage ratings.
Do not use a choke rated for 400V if the B+ is 400V (the peak voltage is about 1.4 x the secondary Vrms, or 1/2 secondary Vrms; per bridge or full wave center tapped rectification).
Choke input filters output about 0.9 x secondary Vrms (whole secondary Vrms for bridge rectification; 1/2 secondary Vrms for center tapped full wave rectification).
But . . . before the output tubes warm up, the is very low load current, so . . . the voltage may rise to about 1.4 x secondary Vrms (whole secondary Vrms for bridge rectification; 1/2 secondary Vrms for center tapped full wave rectification).
The actual B+ voltage also depends on the rectifier voltage drop (less than 1V or 2V solid state bridge or full wave rectifier circuits) typically 40 to 60V for tube rectifiers; the primary DCR, secondary DCR, and choke DCR, all versus the DC load current.
Last edited:
Udc is the rectified DC voltage after the choke. Some other tips:
Vrms across the choke is close to Vrms secondary x 0.5. This is also important.
Flux density of a choke for input duty is a balance between AC flux and DC flux density.
AC flux density across the choke is calculated via B = Vrms / ( 4.44 x F x Afe x T ); where F is mains frequency x2 (100Hz and 120Hz)
In the end, a choke for input duty requires AC flux density headroom, hence it will need a larger core area considering other parameters (Idc, Rdc, Inductance) are kept the same.
Critical inductance increases together with the rectified Udc voltage for the same current, due to the reason a higher Udc to current draw. For example, critical inductance for a 10Vdc 50mA power supply will be 200mH, but for a 1000Vdc 50mA it becomes 20H.
I am not that sure about this: the induction B is going to be proportional to the current through the inductor, assuming the magnetic parameters remain constant ( which they should, more or less if the choke is designed to be well-behaved; swinging chokes are somewhat different).
The formula: B = Vrms / ( 4.44 x F x Afe x T ) is a clever mathematical shortcut avoiding to deal directly with the permeability of the core in AC devices like transformers which generally have a gapless magnetic circuit. Computing the current in those conditions would be a nightmare, which the formula elegantly avoids.
However, we should never put aside first principles, ie. proportionality between B and I, for a "perfect" inductor.
Such a choke is used to smooth (=average) the current, which means that ideally, the instantaneous current should not deviate too much from Idc, otherwise the filter doesn't do its job properly.
The current will fluctuate though (slightly), and the calculation should be based on the peak current, which can be derived from the circuit's values, and another "shortcut" formula, like B=L*I/n*Aeff
The formula: B = Vrms / ( 4.44 x F x Afe x T ) is a clever mathematical shortcut avoiding to deal directly with the permeability of the core in AC devices like transformers which generally have a gapless magnetic circuit. Computing the current in those conditions would be a nightmare, which the formula elegantly avoids.
However, we should never put aside first principles, ie. proportionality between B and I, for a "perfect" inductor.
Such a choke is used to smooth (=average) the current, which means that ideally, the instantaneous current should not deviate too much from Idc, otherwise the filter doesn't do its job properly.
The current will fluctuate though (slightly), and the calculation should be based on the peak current, which can be derived from the circuit's values, and another "shortcut" formula, like B=L*I/n*Aeff
Care is needed to separate the two limit scenarios when using a choke input filter.
The classic scenario is when there is a lot of choke inductance, and the current through the choke winding is dominated by the DC level, with the AC variation being a relatively small-signal cycling on the B-H curve. The inductance measurement for that situation is an incremental inductance measurement at one location along a BH curve (determined by the DC current), where the DC operating point is likely up towards where the BH curve is starting to roll over as it approaches saturation region (ie. as commonly shown in choke performance curves where the inductance is starting to droop at higher DC currents approaching the DC current limit for the choke).
The other limit scenario is where there is a low level of choke inductance, and the choke is operating with a large current swing, with one portion of the current cycle where choke current starts to reach zero (and the choke voltage is then influenced by parasitic effects). The choke inductance varies throughout a mains cycle as the BH operating point cycles from down near the BH origin (choke current reaches zero), to up in to likely the saturation region (where DC + AC current is maximum). The BH operating locus is not a straight line, so incremental inductance varies along the locus, and that variation will depend on the core being used, which is one reason to be cautious when doing simulations.
The 'critical' inductance value that influences whether the choke current reaches zero is better assessed by measurement of incremental choke inductance when DC current is quite low. The benefit of the swinging choke is that it provides a high inductance at a low DC current level, but the inductance doesn't rapidly collapse when there is high AC+DC current. When using an E-I lamination core, that can be allowed for by alternating E's and I's in batches, rather than stacking E then I then E ... as in a typical power transformer - this is a similar situation to what is commonly done in output transformer cores to allow them to handle some level of push-pull current imbalance. Other plausible core construction methods are a little more complex, like the commonly seen stacking of all E's on one side and all I's on the other side and butting them together with specific gap separation, but with a variation such as using some percentage of E's that have shortened legs, or displacing some of the I's so that their gap to the E's are different, or the simplest method of adjusting the stack of I's so that it forms a gap with an angle (smaller gap at one end, and larger gap at the other). So lots of opportunities for DIY experimentation.
And of course the practical benefit (if a choke input filter is a requirement) is that a swinging choke can be physically a lot smaller than a standard choke.
PS. I note my post #32 makes an incorrect comment about diode commutation when choke current reaches zero, but too late to edit that now.
The classic scenario is when there is a lot of choke inductance, and the current through the choke winding is dominated by the DC level, with the AC variation being a relatively small-signal cycling on the B-H curve. The inductance measurement for that situation is an incremental inductance measurement at one location along a BH curve (determined by the DC current), where the DC operating point is likely up towards where the BH curve is starting to roll over as it approaches saturation region (ie. as commonly shown in choke performance curves where the inductance is starting to droop at higher DC currents approaching the DC current limit for the choke).
The other limit scenario is where there is a low level of choke inductance, and the choke is operating with a large current swing, with one portion of the current cycle where choke current starts to reach zero (and the choke voltage is then influenced by parasitic effects). The choke inductance varies throughout a mains cycle as the BH operating point cycles from down near the BH origin (choke current reaches zero), to up in to likely the saturation region (where DC + AC current is maximum). The BH operating locus is not a straight line, so incremental inductance varies along the locus, and that variation will depend on the core being used, which is one reason to be cautious when doing simulations.
The 'critical' inductance value that influences whether the choke current reaches zero is better assessed by measurement of incremental choke inductance when DC current is quite low. The benefit of the swinging choke is that it provides a high inductance at a low DC current level, but the inductance doesn't rapidly collapse when there is high AC+DC current. When using an E-I lamination core, that can be allowed for by alternating E's and I's in batches, rather than stacking E then I then E ... as in a typical power transformer - this is a similar situation to what is commonly done in output transformer cores to allow them to handle some level of push-pull current imbalance. Other plausible core construction methods are a little more complex, like the commonly seen stacking of all E's on one side and all I's on the other side and butting them together with specific gap separation, but with a variation such as using some percentage of E's that have shortened legs, or displacing some of the I's so that their gap to the E's are different, or the simplest method of adjusting the stack of I's so that it forms a gap with an angle (smaller gap at one end, and larger gap at the other). So lots of opportunities for DIY experimentation.
And of course the practical benefit (if a choke input filter is a requirement) is that a swinging choke can be physically a lot smaller than a standard choke.
PS. I note my post #32 makes an incorrect comment about diode commutation when choke current reaches zero, but too late to edit that now.
Last edited:
I believe there is a kind of feedback going on that can be explained via these simple formulas. When an inductor enters saturation threshold (knee of magnetizing curve) via DC, permeability drops, hence inductance as well.
The moment inductance decreases, flux density does that too, due to the formula B = Idc * L / T * Afe. Not necessarily 100% linear (at least taking ohmic losses into account), but this feedback mechanism keeps the core from reaching full saturation.
It is linear enough however to keep a swinging choke into a choke input critical region. Tested that in practice on a class AB amplifier with a gapless 600mH choke from 100mA idle current ramping up to 2A.
I think the easiest way to reach absolute saturation is to be done via alternating voltage amplitude.
This is what I mean, based on this example:
The inductor is linear, and the peak current is 141mA for Idc=100mA. The ratio of B^ to Bdc is thus 1.41.
You can make the calculation of the peak induction by adding Bdc and Bac. I am not going to write down all the equations, because it would be fastidious without a formula editor, but both have the common factor n*Aeff.
When you make the ratio of Btot to Bdc, this factor disappears, and you arrive at a value of 1.94, much higher than the actual 1.41.
The situation would probably be different with a highly non-linear inductor, like a swinging choke
The inductor is linear, and the peak current is 141mA for Idc=100mA. The ratio of B^ to Bdc is thus 1.41.
You can make the calculation of the peak induction by adding Bdc and Bac. I am not going to write down all the equations, because it would be fastidious without a formula editor, but both have the common factor n*Aeff.
When you make the ratio of Btot to Bdc, this factor disappears, and you arrive at a value of 1.94, much higher than the actual 1.41.
The situation would probably be different with a highly non-linear inductor, like a swinging choke
There has been just a little too much math for me.
Neglecting voltage drops due to primary DCR, secondary DCR, rectifier(s) voltage drop, and choke DCR, for a power mains that is a pure sine wave (*) then I simply know that:
Capacitor input filter B+ = Root(2) x Vrms = 1.414 x Vrms
and -
Choke input filter B+ = 0.9 x Vrms
The difference in B+ between capacitor input filter and choke input filter is 1.414/0.9 = 1.57 more B+ voltage from the capacitor input filter, versus the choke input filter.
(*) most power mains are not pure sine waves, which changes the Vpeak versus the Vrms ratio to something other than Root(2),
Vpeak is no longer 1.414 times Vrms.
If the B+ load current is the same for the capacitor input filter and the choke input filter . . .
When you consider the DCR of the primary and the DCR of the secondary, another factor is that there are larger transient currents into the capacitor input filter, versus for the lower and smoother current of the choke input filter.
So, there is more voltage drop in the primary DCR and secondary DCR when using a capacitor input filter, versus when using the Choke input filter.
1.414 and 0.9 factors.
Short lived atomic particles disappear.
Factors do not just disappear, even if they are cancelled and/or if they drop to zero, they are still factors or we would not even mention them.
Just my opinions, until you prove them all to be right.
Neglecting voltage drops due to primary DCR, secondary DCR, rectifier(s) voltage drop, and choke DCR, for a power mains that is a pure sine wave (*) then I simply know that:
Capacitor input filter B+ = Root(2) x Vrms = 1.414 x Vrms
and -
Choke input filter B+ = 0.9 x Vrms
The difference in B+ between capacitor input filter and choke input filter is 1.414/0.9 = 1.57 more B+ voltage from the capacitor input filter, versus the choke input filter.
(*) most power mains are not pure sine waves, which changes the Vpeak versus the Vrms ratio to something other than Root(2),
Vpeak is no longer 1.414 times Vrms.
If the B+ load current is the same for the capacitor input filter and the choke input filter . . .
When you consider the DCR of the primary and the DCR of the secondary, another factor is that there are larger transient currents into the capacitor input filter, versus for the lower and smoother current of the choke input filter.
So, there is more voltage drop in the primary DCR and secondary DCR when using a capacitor input filter, versus when using the Choke input filter.
1.414 and 0.9 factors.
Short lived atomic particles disappear.
Factors do not just disappear, even if they are cancelled and/or if they drop to zero, they are still factors or we would not even mention them.
Just my opinions, until you prove them all to be right.
Last edited: