Kashmire said:
You need to raise the inductor to at least 2mH. I have never tried anything but air-core, so I can't comment on your exact setup, but if using an air-core 2.2mH and socalled super-soft diodes, snubbed that is, you can get fine results.
Magura 🙂
A couple of things I would try, in addition to getting chokes with higher inductance, are:
Put a cap in parrallel to the transformer secondaries, that should lower the ringing frequency (see again Jim Hagermans article) so that your filter caps are able to filter the spikes.
Try LCLC if this is possible.
Reduce the loop size of the wiring as much as possible to lower radiated noise.
Your chokes may be ringing, so try an additional snubber after the choke, or a snubber made of two caps connecting from both ends of the choke to a middle point, and a snubber resistor from there to ground ( Y ).
Play with the resistor value of the snubber, if I understand this correctly the value should be close to the impedance of the ringing circuit.
Put a cap in parrallel to the transformer secondaries, that should lower the ringing frequency (see again Jim Hagermans article) so that your filter caps are able to filter the spikes.
Try LCLC if this is possible.
Reduce the loop size of the wiring as much as possible to lower radiated noise.
Your chokes may be ringing, so try an additional snubber after the choke, or a snubber made of two caps connecting from both ends of the choke to a middle point, and a snubber resistor from there to ground ( Y ).
Play with the resistor value of the snubber, if I understand this correctly the value should be close to the impedance of the ringing circuit.
Thanks for the suggestions. Here is my setup:
Avel-Lindberg 330VA transformer. I've used standard diode bridges and IRF and IXYS fancy diodes (HexFRED, Ultrafast, Soft, etc.) The secondary is snubbed with 0.01uF (straight across the secondary) and a 0.1uf (in series with a 60 Ohm resistor). Snubbing the primary doesn't seem to be interesting, because the winding capacitance of the transformer and the effective series inductance effectively separates what's happening at the secondary from the primary (i.e. if noise occurs at the secondary, the transformer's own parallel capacitance and series inductance affects how that noise shows up at the primary, and vise-versa).
The choke is a 120mH toroidal that can handle the peak currents. This much inductance easily offers "continuous conduction" and awesome performance. The inductor is followed by about 60kuF capacitor bank. This setup rings badly. Using an oscilloscope, I can easily view the PN junction collapse of the diodes. It's obvious where the noise is coming from.
So I ordered some Jantzen 6mH air-core chokes from Partsexpress and tried the same setup without success. This didn't offer continuous conduction, but it substantially lowered peak currents and extended the capacitor charging time. It also was ringing, even with the snubbers.
Then I made and LCLC filter first using 120mH:50kuF:6mH:50kuF and then 6mH:50kuF:120mH:50kuF. Neither of these worked, because the nanosecond spike shoots through these filters. I also tried bypassing the bulk capacitors with (fast) 0.01uF capacitors.
So I still haven't found a solution.
Avel-Lindberg 330VA transformer. I've used standard diode bridges and IRF and IXYS fancy diodes (HexFRED, Ultrafast, Soft, etc.) The secondary is snubbed with 0.01uF (straight across the secondary) and a 0.1uf (in series with a 60 Ohm resistor). Snubbing the primary doesn't seem to be interesting, because the winding capacitance of the transformer and the effective series inductance effectively separates what's happening at the secondary from the primary (i.e. if noise occurs at the secondary, the transformer's own parallel capacitance and series inductance affects how that noise shows up at the primary, and vise-versa).
The choke is a 120mH toroidal that can handle the peak currents. This much inductance easily offers "continuous conduction" and awesome performance. The inductor is followed by about 60kuF capacitor bank. This setup rings badly. Using an oscilloscope, I can easily view the PN junction collapse of the diodes. It's obvious where the noise is coming from.
So I ordered some Jantzen 6mH air-core chokes from Partsexpress and tried the same setup without success. This didn't offer continuous conduction, but it substantially lowered peak currents and extended the capacitor charging time. It also was ringing, even with the snubbers.
Then I made and LCLC filter first using 120mH:50kuF:6mH:50kuF and then 6mH:50kuF:120mH:50kuF. Neither of these worked, because the nanosecond spike shoots through these filters. I also tried bypassing the bulk capacitors with (fast) 0.01uF capacitors.
So I still haven't found a solution.
I can only say I do not have these problems, using 300VA Toroids and LCLC with 300mH (too high, I know), 33mF, 1.5mh foil inductor, 30mF. My PS voltage looks perfectly clean but I will measure again over the weekend.
Also, note that with normal snubbers you cannot completely eliminate the ringing, you just dampen it so that only the first half wave is still pronounced but the energy is absorbed quickly afterwards (you can of course increase the damping / Q).
When I did my original calculations I noted that 10 nF was too low as a parallel capacitance and would still leave the ringing frequency in the range of several 100 kHz. I suggest you at least try a higher value cap, say 100nF or even more.
Also, I still believe you may suffer from RF coupling which is also dependant on the frequency, the coupling increases by 6dB per octave of the frequency, so the lower the ringing frequency the lower the coupling (again, that's what I gathered from other texts).
P.S. When I said "so try an additional snubber after the choke" - normally the snubber goes accross the snubbed device, so the Y configuration might be better.
Also, note that with normal snubbers you cannot completely eliminate the ringing, you just dampen it so that only the first half wave is still pronounced but the energy is absorbed quickly afterwards (you can of course increase the damping / Q).
When I did my original calculations I noted that 10 nF was too low as a parallel capacitance and would still leave the ringing frequency in the range of several 100 kHz. I suggest you at least try a higher value cap, say 100nF or even more.
Also, I still believe you may suffer from RF coupling which is also dependant on the frequency, the coupling increases by 6dB per octave of the frequency, so the lower the ringing frequency the lower the coupling (again, that's what I gathered from other texts).
P.S. When I said "so try an additional snubber after the choke" - normally the snubber goes accross the snubbed device, so the Y configuration might be better.
MRupp said:
did you mean this article: http://www.hagtech.com/pdf/snubber.pdf?
Thank you very much 🙂
This is an old thread, but reading it made me go back and make some power supply measurements in the LC power supply feeding my aleph 30. Earlier in this thread, there was concern about the diodes slamming closed because of the choke input, and causing a nanosecond spike which cannot be filtered out..
My supplies were made using some heavy Signal transformers that I had on hand...56 volts center tapped. To cut down the voltage to 26 volt rails after rectification, I employed a choke input type filter. The chokes are 3 mh 14guage air core coils, which then go to 200,000 uf of capacitors.
The Aleph 30 draws 5 amps per rail. Putting an oscilliscope on the rails show ripple of 50 mv. I did not see any anomolies in the signal trace, just a nice round sine wave.
I then compared this to the output of another power supply I have, currently sitting idle. This supply is 38 volts per rail...I loaded it down with 2.7 amps of resistive load and put an oscilliscope across the supply and measured .5 volt of ripple. The sine wave had more of a V shape to it.
So...The LC supply seems to do a better job of filtering, if you don't mind wasting a little extra power...
The Alephs really do sound good with this type of supply...I think the choke input filters out the power line junk that rides the rails.
My supplies were made using some heavy Signal transformers that I had on hand...56 volts center tapped. To cut down the voltage to 26 volt rails after rectification, I employed a choke input type filter. The chokes are 3 mh 14guage air core coils, which then go to 200,000 uf of capacitors.
The Aleph 30 draws 5 amps per rail. Putting an oscilliscope on the rails show ripple of 50 mv. I did not see any anomolies in the signal trace, just a nice round sine wave.
I then compared this to the output of another power supply I have, currently sitting idle. This supply is 38 volts per rail...I loaded it down with 2.7 amps of resistive load and put an oscilliscope across the supply and measured .5 volt of ripple. The sine wave had more of a V shape to it.
So...The LC supply seems to do a better job of filtering, if you don't mind wasting a little extra power...
The Alephs really do sound good with this type of supply...I think the choke input filters out the power line junk that rides the rails.
- Status
- Not open for further replies.