C1 has very little effect on the way that the choke L3 acts - that's the intention.
It is added to clean up the noise from the rectifier stage, and to reduce the peak voltage across the inductor.
you can omit C1, and perhaps the dc output will decrease a little ( you would then have a pure choke-input filter), but the noise will probably be higher.
A high-value choke is not necessarily a good filter at HF, not only because of its parasitic shunt capacitance, which allows noise through, but it may also radiate somewhat: choke + long feed cables = base-loaded antenna!
It is added to clean up the noise from the rectifier stage, and to reduce the peak voltage across the inductor.
you can omit C1, and perhaps the dc output will decrease a little ( you would then have a pure choke-input filter), but the noise will probably be higher.
A high-value choke is not necessarily a good filter at HF, not only because of its parasitic shunt capacitance, which allows noise through, but it may also radiate somewhat: choke + long feed cables = base-loaded antenna!
Ratios are not easily helpful: the values depend upon load current and L value.
Suggested approach:
Use PSUD2. Set up the load as constant current, and choose a high value of C2. then add a convenient choke, fairly high value.
Add C1 and increase its value until you get the dc output voltage, and conduction angle you want. Monitor the C1 voltage to check conduction angle, and peak inductor voltage. The 'ripple' on C1 should be almost equal to the rectifier output voltage - say 90-98%.
you have a 'choke-input' supply when Vout (dc) = about 90% of Vin (rms). Increasing C1 a little will elevate the dc output to somewhere between choke- and cap-input. but beware, this is load dependent.
You can optimise the supply by looping through lower values of L, and reoptimising C1, to get the ripple voltage you need, versus the cost of the choke.
Suggested approach:
Use PSUD2. Set up the load as constant current, and choose a high value of C2. then add a convenient choke, fairly high value.
Add C1 and increase its value until you get the dc output voltage, and conduction angle you want. Monitor the C1 voltage to check conduction angle, and peak inductor voltage. The 'ripple' on C1 should be almost equal to the rectifier output voltage - say 90-98%.
you have a 'choke-input' supply when Vout (dc) = about 90% of Vin (rms). Increasing C1 a little will elevate the dc output to somewhere between choke- and cap-input. but beware, this is load dependent.
You can optimise the supply by looping through lower values of L, and reoptimising C1, to get the ripple voltage you need, versus the cost of the choke.
I see I could go further and add RC Snubbers or just parallel cap to each Diode? I see/hear two different things thought that it will mess up UF4007...or that it will not....
further clean up any junk on mains maybe??? could very well be an issue where I am at.....
isn't that like asking to blow up the transformer though? haha
further clean up any junk on mains maybe??? could very well be an issue where I am at.....
isn't that like asking to blow up the transformer though? haha
Rod makes a good point that applies "in spades" to high value cap. I/P filters, such as well functioning voltage multiplier PSUs. I have consistently advocated, and will continue to advocate, the incorporation of a "hash" filter LC section made from a high current RF choke and mica or NPO ceramic cap. of modest value between the I/P cap. setup and the PSU filter's choke. Applying Fourier's Theorem to the highly "triangular" ripple waveform present shows how much RF energy is present. That garbage will pass through the shunt capacitance of "normal" PSU filter chokes. The LC section made from "RF parts" attenuates the trash, before it can reach the choke.
Last edited:
These capacitors will have small values and their impedance isn't likely to be low enough to be significant at the mains frequency. Besides, when using the RC snubber mentioned above, R will be chosen to suit and not to overload.
Capacitors across the diodes tends to be more of a quick and dirty solution.
Capacitors across the diodes tends to be more of a quick and dirty solution.
A high WVDC 10 nF. cap. in parallel with UFnnnn diodes does help. However, the discussions on this site and over on AA point in the direction of a series combination of resistance and capacitance in parallel with each diode as providing maximized snubbing action.
Yet another option to investigate is the reverse recovery spike filter (RRSF) devised by John Camille, AKA "Buddha". Hop over to AA and scan the archives for RRSF, should you be interested.
Rod,
Yet another question, since you have practical experience!
I have not used C1 (will consider in future instead of across-rectifier-snubbers). But rather regarding choke turn-off spikes, I have found a snubber (series R-C pair) across the choke some help in calming choke turn-off spikes. (Again spikes there which could reach kilovolts). I guess I look at the choke as an inductor in parallel with (C1 plus the output capacitor) in series, as possibly generating a pulse increasing circuit in parallel with an 'un-snubbed' choke; several possibilities here.
Any views/tests regarding this? (Also from other members?)
Thanks
Yet another question, since you have practical experience!
I have not used C1 (will consider in future instead of across-rectifier-snubbers). But rather regarding choke turn-off spikes, I have found a snubber (series R-C pair) across the choke some help in calming choke turn-off spikes. (Again spikes there which could reach kilovolts). I guess I look at the choke as an inductor in parallel with (C1 plus the output capacitor) in series, as possibly generating a pulse increasing circuit in parallel with an 'un-snubbed' choke; several possibilities here.
Any views/tests regarding this? (Also from other members?)
Thanks
Johan, thanks, yes, I should have mentioned this action in the choke.
If a (non-zero) choke current is abruptly switched off by the rectifier, a large voltage will appear, due to the collapsing magnetic field. You can dissipate this energy with a snubber across the choke, but that's a shame to degrade the filtering properties of the choke.
But actually, when C1 is present (1uF in my drawing) directly after the rectifier, the high voltage will be mostly suppressed.
The suppression circuit looks like this:
C1's current will swing sharply negative, as the choke tries to maintain current in the same direction. (If there were no C1, the choke would swing a negative voltage). This current-pulse flows through the choke and into the output filters, as normal. If there were any high frequency content in the current it would be dissipated in the multi-turn ferrite bead (which may be modelled as a frequency-dependent resistor, rather than an inductance).
In this way, C1 is behaving somewhat like the diode in a buck converter (dc-dc downward voltage converter). For entertainment, you can probably show that a fast diode (anode to supply negative) can suppress these voltage spikes just as well as a snubber across the choke. 2x UF4007 series should suffice.
But with C1 present, at even a fairly small value, the spikes should be under control.
If a (non-zero) choke current is abruptly switched off by the rectifier, a large voltage will appear, due to the collapsing magnetic field. You can dissipate this energy with a snubber across the choke, but that's a shame to degrade the filtering properties of the choke.
But actually, when C1 is present (1uF in my drawing) directly after the rectifier, the high voltage will be mostly suppressed.
The suppression circuit looks like this:
C1's current will swing sharply negative, as the choke tries to maintain current in the same direction. (If there were no C1, the choke would swing a negative voltage). This current-pulse flows through the choke and into the output filters, as normal. If there were any high frequency content in the current it would be dissipated in the multi-turn ferrite bead (which may be modelled as a frequency-dependent resistor, rather than an inductance).
In this way, C1 is behaving somewhat like the diode in a buck converter (dc-dc downward voltage converter). For entertainment, you can probably show that a fast diode (anode to supply negative) can suppress these voltage spikes just as well as a snubber across the choke. 2x UF4007 series should suffice.
But with C1 present, at even a fairly small value, the spikes should be under control.
You can wire a choke to the negative or positive line.
I showed the negative case, really to register that it is possible!
It is probaly worth connecting the rectifier end of the choke to the 'inside' end of the winding. C1 should be added to any old choke input supply, to improve noise performance, and should be close to diodes and PT to minimise loop. If diodes have heatsinks, then care is needed to keep parasitic capacitance loop local to ES of PT. C1 may need a good transient voltage rating, to handle power turn-off induced transient. I sometimes use mid-range value of C1 to lower the idle B+.
Hi Rod
Referring back to your HT supply circuit-
What sort of 100R snubber resistor would you recommend?
I use a 500-0-500V secondary feeding a GZ37. Would I better to use one snubber (R+C) per phase, each connected the centre tap, rather than one directly across the phases?
I assume it’s permissible to interchange L3 and L4- i.e. leave my 30H choke in the positive line, and put the ferrite bead in the ground path?
I already have some ferrite beads- FAIR-RITE 2944666671, Farnell 1191416. Would one of these do for L4?
At the moment I don’t use C1- i.e. I use a true choke input supply. I’ve modelled it in psud2, and found that 470n is the highest value I can use here before the output B+ voltage begins to rise unacceptably. Is it worth fitting this value, and what sort of cap would you recommend?
Many thanks, yet again
Paul N
Referring back to your HT supply circuit-
What sort of 100R snubber resistor would you recommend?
I use a 500-0-500V secondary feeding a GZ37. Would I better to use one snubber (R+C) per phase, each connected the centre tap, rather than one directly across the phases?
I assume it’s permissible to interchange L3 and L4- i.e. leave my 30H choke in the positive line, and put the ferrite bead in the ground path?
I already have some ferrite beads- FAIR-RITE 2944666671, Farnell 1191416. Would one of these do for L4?
At the moment I don’t use C1- i.e. I use a true choke input supply. I’ve modelled it in psud2, and found that 470n is the highest value I can use here before the output B+ voltage begins to rise unacceptably. Is it worth fitting this value, and what sort of cap would you recommend?
Many thanks, yet again
Paul N
hi Paul,
Carbon Composition 0.5W for the snubber resistor - for minimal inductance in a circuit with fast rise-time .
One snubber per phase is probably better, then the active phase is not in danger of being polluted by the other half-winding as it enters "flyback" phase.
Yes, swap chokes in + and - lines, if desired.
The 300B SE can sound really splendid, and is very sensitive to the quality of the power supplies - so adding the cap to tame wild voltage swings on the choke is likely to be audible. Even 22nF is worth adding after the rectifier. Any FKP, MKP or MKT with stacked construction and preferably 1.2 - 1.6kV voltage rating should be useful. Wima, and EPCOS certainly make good examples of these.
Enjoy those 300Bs (all working OK now?)
Carbon Composition 0.5W for the snubber resistor - for minimal inductance in a circuit with fast rise-time .
One snubber per phase is probably better, then the active phase is not in danger of being polluted by the other half-winding as it enters "flyback" phase.
Yes, swap chokes in + and - lines, if desired.
The 300B SE can sound really splendid, and is very sensitive to the quality of the power supplies - so adding the cap to tame wild voltage swings on the choke is likely to be audible. Even 22nF is worth adding after the rectifier. Any FKP, MKP or MKT with stacked construction and preferably 1.2 - 1.6kV voltage rating should be useful. Wima, and EPCOS certainly make good examples of these.
Enjoy those 300Bs (all working OK now?)
Few additional remarks:
A choke in the negative line eases voltage between itself and the surroundings. One side is at common potential. Then, using the inside connection as the 'live' (raw input) side provides some shielding by the rest of the choke and the core against its pulses - provided the former and inside insulation is good enough to handle this!
I have noticed that some designs use a relatively small filter input capacitor to boost h.t. to a desired level, but it must be warned that this results in a rather poor regulation. (See RDH4 and other relevant publications regarding load current vs. filter input and output capacitor.) One should make a difference between a C1 value (referring to Rod's earleir schematic) sufficient to remove noise, and one large enough to start towards a C-input filter.
I must also point to Eli's note (post #26) regarding a small {C - ferrite-L} where it is positioned. I got this advice from him years ago and found it quite effective for the low component cost.
A choke in the negative line eases voltage between itself and the surroundings. One side is at common potential. Then, using the inside connection as the 'live' (raw input) side provides some shielding by the rest of the choke and the core against its pulses - provided the former and inside insulation is good enough to handle this!
I have noticed that some designs use a relatively small filter input capacitor to boost h.t. to a desired level, but it must be warned that this results in a rather poor regulation. (See RDH4 and other relevant publications regarding load current vs. filter input and output capacitor.) One should make a difference between a C1 value (referring to Rod's earleir schematic) sufficient to remove noise, and one large enough to start towards a C-input filter.
I must also point to Eli's note (post #26) regarding a small {C - ferrite-L} where it is positioned. I got this advice from him years ago and found it quite effective for the low component cost.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.