FWIW. On power supply noise (RF) I have found that using a capacitor across secondary windings (up to 2uF) often stops RF noise detected by an AM transistor radio held against wiring to bridge rectifier. On a scope the sharp glitch seen is changed to a series of low (audio) freq damped sine waves, too low in freq to radiate and easily filtered out by the filter caps.
I am just lowering the freq at which the LC circuit oscillates when shocked by the diodes turning off. This idea depends on the wiring, transformert used etc etc.
I have not experimented with fancy diodes.
I am just lowering the freq at which the LC circuit oscillates when shocked by the diodes turning off. This idea depends on the wiring, transformert used etc etc.
I have not experimented with fancy diodes.
Fredex, that sounds like a good thing. I have a power line filter that uses 2uF caps across an isolation transformer, and it cuts off at about 20KHz. I also use a 10UF polypropylene across the power line of both my audio and my video system. That works pretty well too.
Anatech, I apologize if I insulted you in some way, but I think that I was insulted first by you, and that is why I reacted. IF you or ANYONE knows how to make normal solid state diodes work very, very well, I would be interested in any input. If, however, you just dismiss my efforts to use a better behaved diode in my better circuits, I certainly recommend that you stick with my cheaper stuff, because they use cheaper diode assemblies. They work OK.
Anatech, I apologize if I insulted you in some way, but I think that I was insulted first by you, and that is why I reacted. IF you or ANYONE knows how to make normal solid state diodes work very, very well, I would be interested in any input. If, however, you just dismiss my efforts to use a better behaved diode in my better circuits, I certainly recommend that you stick with my cheaper stuff, because they use cheaper diode assemblies. They work OK.
john curl said:
Most solutions to this problem are to dissipate the RF energy in a resistor using snubber networks (with small value caps). Your big plain cap across the mains could be working by lowering the freq of a tank circuit formed by the mains wiring L and C and whatever is connected to it. Possible??
It creates a limited bandwith from the power line, or even the output of the transformer. This is why it is used with isolation transformers. Almost something for nothing.
I agree that snubbers could work, but they also add complexity and cost. We just go with the faster diodes, as it keeps the parts count down. Still, with an existing high current rectifier bridge, snubbers might make a good upgrade.
I agree that snubbers could work, but they also add complexity and cost. We just go with the faster diodes, as it keeps the parts count down. Still, with an existing high current rectifier bridge, snubbers might make a good upgrade.
On a slightly different slant: a new integrated CCS from LTC, the LT3092.
Based on that 'innovative' LT3080 which in itself was based on the obsolete ML1488. ...
http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1003,C1794,P88400
Jan Didden
Based on that 'innovative' LT3080 which in itself was based on the obsolete ML1488. ...
http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1003,C1794,P88400
Jan Didden
PMA, those are both very general questions. It is almost impossible to answer them in any useful way. As you must know, separating the power supply with a separate box helps to contain the line interference by physical separation of the AC line from the audio amps and switches, minimizing the magnetic component of the AC field, and the separate shield around the power supply removes the electrostatic component of the AC field.
Of course, about 2 MHz or so, all bets are off, except for added AC filtering in the DC line. However the range from 20KHz to 2 Meg Hz can be wonderfully attenuated by a 10uf capacitor across the power line or 2uf across an isolation transformer.
We have found, over years of experience with real products, that adding much of anything else across the cap, often does as much damage as it potentially fixes, due to secondary resonances generated by the second cap, even if it is only 100 times less in value, a typical 'rule of thumb' for cap bypassing.
Of course, about 2 MHz or so, all bets are off, except for added AC filtering in the DC line. However the range from 20KHz to 2 Meg Hz can be wonderfully attenuated by a 10uf capacitor across the power line or 2uf across an isolation transformer.
We have found, over years of experience with real products, that adding much of anything else across the cap, often does as much damage as it potentially fixes, due to secondary resonances generated by the second cap, even if it is only 100 times less in value, a typical 'rule of thumb' for cap bypassing.
john curl said:
Do you use bypass caps much? I know that they are necessary/helpful near opamps power supply pins and such. Being a DIYer I would go overkill and add large bypass caps paralleled with electrolytic caps in power supplies sometimes. I guess the problem with bypass caps is that the electrolytic caps have a certain amount of inductance and adding a parallel film cap creates a resonant circuit?
The caps that I use, self-resonate at 250KHZ or so, but the inductive impedance keeps rising on the line, (I think) and the attenuation, ideally, just flattens out for a range that I would think, at least a decade.
The two uf cap across each isolation transformer was designed in by CRC, the filter specialists, for super line purification. I measured the 20KHz breakpoint, myself.
The two uf cap across each isolation transformer was designed in by CRC, the filter specialists, for super line purification. I measured the 20KHz breakpoint, myself.
I would like to return briefly to 'diode reverse recovery voltage' as I think that it is still not well understood, and often ignored effect.
First, most common diode bridges have standard diodes. These diodes have many qualities, but they do not turn off very quickly and leave a 'glitch' at every zero crossing of the voltage across them. They still work fine for many, many applications such a battery chargers, industrial equipment, etc. , but for high end audio, they should be addressed. This is because this 'glitch' might be thought, at least partially, as an AC signal that gets past the diode array itself. Then, the filter caps become coupling caps, and where do they couple? The ground! Not the 'earth' but an arbitrary point of reference inside the electronic circuit. Can we measure it? It is difficult because the 'ground' has no reference to measure itself. However, when you couple the 'grounds' of 2 pieces of equipment, then maybe some extra flow of energy due to these 'glitches' generated by the diodes, might be possible. Still, difficult to measure, but I consider it important enough to be considered.
First, most common diode bridges have standard diodes. These diodes have many qualities, but they do not turn off very quickly and leave a 'glitch' at every zero crossing of the voltage across them. They still work fine for many, many applications such a battery chargers, industrial equipment, etc. , but for high end audio, they should be addressed. This is because this 'glitch' might be thought, at least partially, as an AC signal that gets past the diode array itself. Then, the filter caps become coupling caps, and where do they couple? The ground! Not the 'earth' but an arbitrary point of reference inside the electronic circuit. Can we measure it? It is difficult because the 'ground' has no reference to measure itself. However, when you couple the 'grounds' of 2 pieces of equipment, then maybe some extra flow of energy due to these 'glitches' generated by the diodes, might be possible. Still, difficult to measure, but I consider it important enough to be considered.
No. If it is "the ground point".
Yes. If it is a real peace of metal that has resistance and inductance, no matter how small they are.
Running for better and better parts as the single proper way to make better and better design, you create more and more problems. The higher are filter capacitances, the lower is their ESR, the better is your toroid, the better are diodes, the higher are current peaks. Not only peaks of currents discharging charges in diodes, but peak currents that charge your tanks. Both of them create huge current peaks from transformer to capacitor that cause voltage spikes on wires. Diode charge discharging currents may be addressed by ceramic caps (that you discarded as bad devices for audio) soldered as close as possible to diodes. Charging currents can't be addressed such a way.
Yes. If it is a real peace of metal that has resistance and inductance, no matter how small they are.
Running for better and better parts as the single proper way to make better and better design, you create more and more problems. The higher are filter capacitances, the lower is their ESR, the better is your toroid, the better are diodes, the higher are current peaks. Not only peaks of currents discharging charges in diodes, but peak currents that charge your tanks. Both of them create huge current peaks from transformer to capacitor that cause voltage spikes on wires. Diode charge discharging currents may be addressed by ceramic caps (that you discarded as bad devices for audio) soldered as close as possible to diodes. Charging currents can't be addressed such a way.
- Status
- Not open for further replies.