I see most of the power supply design for gailcone (3886, 4780) are uses discrete diodes (e.g. MUR860 or similar). Any specific reason? Why not bridge modules which are available for any current rating and also have less forward drop compared to discrete diodes bridge. Also, PCB would be simpler.
I am sure there must be good reason to use descrete diodes. Can anyone clarifies.
I am sure there must be good reason to use descrete diodes. Can anyone clarifies.
MUR860's were identified by Peter Daniel as sounding better than others he had test driven. Bridges are usually frowned upon because you have a smaller selection, and are tougher to get in fast- and ultrafast-fast recovery flavours, which impart less switching noise.
That being said, don't be afraid to shop around, there's so much out there. I just bough a bunch of 15A, isolated, ultrafasts for $1/each at Digikey, exactly what I need for my next project!
That being said, don't be afraid to shop around, there's so much out there. I just bough a bunch of 15A, isolated, ultrafasts for $1/each at Digikey, exactly what I need for my next project!
Hi
How can it have less switching noise if there is only 50Hz present, unless you will use SMPS, but even in that case, if designed good you can say it will have zero switching noise.
So I don't see why would Mur's or any fast diodes be better than bridge modules.
And please don't say you can "hear" the difference
How can it have less switching noise if there is only 50Hz present, unless you will use SMPS, but even in that case, if designed good you can say it will have zero switching noise.
So I don't see why would Mur's or any fast diodes be better than bridge modules.
And please don't say you can "hear" the difference
Did some more search, a nice writeup by Bob Pease on similar topic:
http://www.elecdesign.com/Articles/Index.cfm?AD=1&ArticleID=13166
I also plan to go with bridge but thought I would ask.
http://www.elecdesign.com/Articles/Index.cfm?AD=1&ArticleID=13166
I also plan to go with bridge but thought I would ask.
luka said:Hi
How can it have less switching noise if there is only 50Hz present, unless you will use SMPS, but even in that case, if designed good you can say it will have zero switching noise.
So I don't see why would Mur's or any fast diodes be better than bridge modules.
And please don't say you can "hear" the difference
Because you have junction capacitance in the diode and leakage inductance from the transformer which form a resonant circuit -- this is what it looks like -- the radio frequency interference likes to get rectified by all the base-emitter junctions in low level circuits.
An externally hosted image should be here but it was not working when we last tested it.
Ordinary bridge rectifiers have significant reverse diode conduction at the instant they become reverse biased. (This is not the normal reverse leakage current but real reverse current flow because the P-N junction is still in an 'on' state). When the P-N junction turns off,it does so suddenly after a few hundred nanoseconds, and there may be several hundred mA of reverse current flowing at this point. The sudden 'snap-off' of this current has harmonics that spread up to high frequencies, and can easily be measured.
The ideal diodehas fast turn-off characteristics, but also has a 'soft' turn-off so that there is not a sudden 'snap-off' in the reverse current.
The ideal diodehas fast turn-off characteristics, but also has a 'soft' turn-off so that there is not a sudden 'snap-off' in the reverse current.
So you can see parasitic and shock excitation associated with rectifier diodes. So what! If you fix it with snubbers or minimize it with diode type choices; what sonic improvements have you measured? Could you have gotten as good or better results by engineering your grounds or circuit layout more intelligently?
My experiences with SMPS for off-line commercial applications leads me to believe that after energy storage, cap-mult, or regulators you aren't going to see any of this in your audio circuit unless you have lowsy wiring practices or mount your audio circuit on top of your rectifiers! If radiated noise is the worry, think distance and Faraday cages. Think tight current loops in the supply around the rectifiers.
Snubbing of rectifier diodes is normally encountered when using Schottky's. This is to protect the diode from excessive reverse voltage mostly, and occassionally to minimize radiated EMI. In most cases conducted EMI is not a problem with minimal use of CM chokes and line caps. The down stream conducted noise after PS filtering is not a problem. Even in flybacks where snubbing is used for the switch cap/leakage inductance and several percent of the total power budget is burned; this is to protect the switch, not reduce EMI (though it may reduce EMI). Rectifier snubbing wastes energy and reduces the power factor. Use only to increase component reliability when all else fails.
Good design practice would seem to me to entail elimination of noise sources that are demonstrably degrading performance (or necessary for regulatory compliance). If your audio equipment has a very high amplification factor and/or several MHz of BW then it will be more susceptible, I have no doubt. But a power amp with a 200KHz power BW?
I would appreciate any references to real live data that show measurable sonic improvements from modifications to rectifiers in low freq power supplies for audio applications. I am even more interested if these improvements were detectable in supplies with regulators or cap multipliers or CRC/CLC filtering.
Thank you for your indulgence,
VSR
My experiences with SMPS for off-line commercial applications leads me to believe that after energy storage, cap-mult, or regulators you aren't going to see any of this in your audio circuit unless you have lowsy wiring practices or mount your audio circuit on top of your rectifiers! If radiated noise is the worry, think distance and Faraday cages. Think tight current loops in the supply around the rectifiers.
Snubbing of rectifier diodes is normally encountered when using Schottky's. This is to protect the diode from excessive reverse voltage mostly, and occassionally to minimize radiated EMI. In most cases conducted EMI is not a problem with minimal use of CM chokes and line caps. The down stream conducted noise after PS filtering is not a problem. Even in flybacks where snubbing is used for the switch cap/leakage inductance and several percent of the total power budget is burned; this is to protect the switch, not reduce EMI (though it may reduce EMI). Rectifier snubbing wastes energy and reduces the power factor. Use only to increase component reliability when all else fails.
Good design practice would seem to me to entail elimination of noise sources that are demonstrably degrading performance (or necessary for regulatory compliance). If your audio equipment has a very high amplification factor and/or several MHz of BW then it will be more susceptible, I have no doubt. But a power amp with a 200KHz power BW?
I would appreciate any references to real live data that show measurable sonic improvements from modifications to rectifiers in low freq power supplies for audio applications. I am even more interested if these improvements were detectable in supplies with regulators or cap multipliers or CRC/CLC filtering.
Thank you for your indulgence,
VSR
Those particular rectifiers have been chosen for a simple reason that they sounded better than few other that I tested in that specific application. So far nobody presented a better alternative, so probably that's why most people still use them.
http://www.diyaudio.com/forums/showthread.php?postid=102326#post102326
http://www.diyaudio.com/forums/showthread.php?postid=102326#post102326
sherelec said:
I can tell you as an absolute fact that the Super Oscillator (Linear Tech application note 67) is unhappy if there is any RFI present --
I used to do super-conductivity experiments in a Faraday shielded room -- as an undergrad lab rat over 35 years ago -- if it were possible to eliminate noise you did so. you would see the readings drift all over the place and then discover that someone had not properly affixed a cover on a power supply, etc., etc.
Ouroboros,
Do you have any evidence that this reverse recovery of the rectifier causes problems? Have you ever changed diodes to ones with less reverse recovery current and measured an improvement in a circuit connected to the supply? Power supplies with ripple reduction elements typically attenuate noise frequencies above the ripple fundamental frequency. Aren't these supplies going to have reduced noise from the frequencies injected by reverse recovery signals? In my experiences with higher frequency smps; the noise spectrum, conducted and radiated is changed by diode characteristics. However, using fast diodes or soft diodes does not always make things better. Several times the use of a cheap low freq rectifier diode has improved the noise signature. This comes with thermal penalties at higher freq, but if you need to met a noise spec, diode induced inefficiency may be a small price to pay!
rdf,
If the tank formed by the rectifier capacitance and the xformer leakage inductance is shock excited into a high freq oscillation; then adding capacitance (if the capacitor is very lossy then it may help) to the circuit topology is only likely to change the freq of oscillation. The point of a snubber is to add more resistive (dissipative) properties to the circuit so that the excitation energy is dissipated in a shorter period of time than the natural resonant period of the tank circuit. There is a link to a paper by Hagerman (on snubbers and tank circuits) elsewhere in the forums that has mathematical derivations and circuit models for a transformer secondary and rectifier. You might insert the capacitance into his circuit models and see what happens.
Peter Daniel,
I followed the trail you pointed to in the forums, and I do find that several people have tried to measure the tank ringing with little or no success. I find references to 2 or 3 "guru's" who are quoted as claiming one diode or approach is superior to another. They do not agree with each other. There are some replies by the "guru's" themselves stating some preferences, which are not always in agreement with themselves, over time, or each other. In no case did I find anyone with data or references to data that show improved sonic performance from any of their pet devices. The reverse recovery was also discussed, but here again no data to support one diode type is sonically superior to another. Lots of opinion, no measurement data.
Jackinnj,
I see you have raised the issue of rectifier ringing and applying snubbers in several threads. I had hoped that you had some real data to share for, say, a power amplifier which had rectifier ringing, had a distortion or noise problem, and in which you snubbed the rectifier which led to a sonic (measurable) improvement. I believe I have acknowledged that a very sensitive circuit might be affected by rectifier artifacts. The question was: Do the audio circuits being built or used by forum members exhibit improvements (measurable) by mitigating these artifacts, by snubbing or by choosing diodes with different reverse recovery characteristics?
Careful examination of any circuit with sensitive enough instruments will show examples of nonideal behavior. Does a competent designer try to address every anomaly without first determining which, if any, are causing a functional inadequacy?
Your scope pictures earlier in the thread show ringing superimposed on a portion of the rectifier waveform (conducting, reverse recovery, nonconducting?). The fact that a low freq signal and a high freq signal are coresident in a circuit does not provide proof of mixing, or if mixing is present at what level the products are manifest. If there is no mixing or very little, them AM demodulation downstream is likely not present or undetectable. Then removing or attenuating one of the signals is not productive. If your point is that high freq oscilation at the rectifier is getting downstream to the audio circuitry; did you make a measurement? Was the signal detectable on the power supply rails? If you did, what was the magnitude and what sort of problem does it present sonically? If radiated, could you find an induced signal at your low-level pn junctions?
>I can tell you as an absolute fact that the Super Oscillator (Linear Tech application note 67) is unhappy if there is any RFI present --
Aren't you saying in the above statement that this particular circuit is always unhappy? It would be very difficult (impossible) to find any place with no RFI. However, you might find places with RFI less than 1nW from 0 to 1GHz, or some other limit. Also, choosing an example with 180dB open loop gain at 10KHz seems a teeny tiny bit on the extreme end of things audio! Or stated another way an oscillator with distortion ~ -180dB down reference the fundamental is going to be degraded by everything in its vicinity. I believe that putting this circuit on a table, sitting down beside it and breathing will impair its performance. So you should snub the power supply rectifiers and stop breathing (or leave the room)?
Actually I am thankful that you present some measurement or factual support for your statements.
I was, however, hoping to provoke someone to present data, case studies, or the like to support their choices. From information like that I might increase my own engineering and design skills. Opinion by itself is not very useful. Opinion based on measurement and/or analysis is worthy of consideration.
Sorry for the too long reply. I hope it has, at the very least, entertainment value.
VSR
Do you have any evidence that this reverse recovery of the rectifier causes problems? Have you ever changed diodes to ones with less reverse recovery current and measured an improvement in a circuit connected to the supply? Power supplies with ripple reduction elements typically attenuate noise frequencies above the ripple fundamental frequency. Aren't these supplies going to have reduced noise from the frequencies injected by reverse recovery signals? In my experiences with higher frequency smps; the noise spectrum, conducted and radiated is changed by diode characteristics. However, using fast diodes or soft diodes does not always make things better. Several times the use of a cheap low freq rectifier diode has improved the noise signature. This comes with thermal penalties at higher freq, but if you need to met a noise spec, diode induced inefficiency may be a small price to pay!
rdf,
If the tank formed by the rectifier capacitance and the xformer leakage inductance is shock excited into a high freq oscillation; then adding capacitance (if the capacitor is very lossy then it may help) to the circuit topology is only likely to change the freq of oscillation. The point of a snubber is to add more resistive (dissipative) properties to the circuit so that the excitation energy is dissipated in a shorter period of time than the natural resonant period of the tank circuit. There is a link to a paper by Hagerman (on snubbers and tank circuits) elsewhere in the forums that has mathematical derivations and circuit models for a transformer secondary and rectifier. You might insert the capacitance into his circuit models and see what happens.
Peter Daniel,
I followed the trail you pointed to in the forums, and I do find that several people have tried to measure the tank ringing with little or no success. I find references to 2 or 3 "guru's" who are quoted as claiming one diode or approach is superior to another. They do not agree with each other. There are some replies by the "guru's" themselves stating some preferences, which are not always in agreement with themselves, over time, or each other. In no case did I find anyone with data or references to data that show improved sonic performance from any of their pet devices. The reverse recovery was also discussed, but here again no data to support one diode type is sonically superior to another. Lots of opinion, no measurement data.
Jackinnj,
I see you have raised the issue of rectifier ringing and applying snubbers in several threads. I had hoped that you had some real data to share for, say, a power amplifier which had rectifier ringing, had a distortion or noise problem, and in which you snubbed the rectifier which led to a sonic (measurable) improvement. I believe I have acknowledged that a very sensitive circuit might be affected by rectifier artifacts. The question was: Do the audio circuits being built or used by forum members exhibit improvements (measurable) by mitigating these artifacts, by snubbing or by choosing diodes with different reverse recovery characteristics?
Careful examination of any circuit with sensitive enough instruments will show examples of nonideal behavior. Does a competent designer try to address every anomaly without first determining which, if any, are causing a functional inadequacy?
Your scope pictures earlier in the thread show ringing superimposed on a portion of the rectifier waveform (conducting, reverse recovery, nonconducting?). The fact that a low freq signal and a high freq signal are coresident in a circuit does not provide proof of mixing, or if mixing is present at what level the products are manifest. If there is no mixing or very little, them AM demodulation downstream is likely not present or undetectable. Then removing or attenuating one of the signals is not productive. If your point is that high freq oscilation at the rectifier is getting downstream to the audio circuitry; did you make a measurement? Was the signal detectable on the power supply rails? If you did, what was the magnitude and what sort of problem does it present sonically? If radiated, could you find an induced signal at your low-level pn junctions?
>I can tell you as an absolute fact that the Super Oscillator (Linear Tech application note 67) is unhappy if there is any RFI present --
Aren't you saying in the above statement that this particular circuit is always unhappy? It would be very difficult (impossible) to find any place with no RFI. However, you might find places with RFI less than 1nW from 0 to 1GHz, or some other limit. Also, choosing an example with 180dB open loop gain at 10KHz seems a teeny tiny bit on the extreme end of things audio! Or stated another way an oscillator with distortion ~ -180dB down reference the fundamental is going to be degraded by everything in its vicinity. I believe that putting this circuit on a table, sitting down beside it and breathing will impair its performance. So you should snub the power supply rectifiers and stop breathing (or leave the room)?
Actually I am thankful that you present some measurement or factual support for your statements.
I was, however, hoping to provoke someone to present data, case studies, or the like to support their choices. From information like that I might increase my own engineering and design skills. Opinion by itself is not very useful. Opinion based on measurement and/or analysis is worthy of consideration.
Sorry for the too long reply. I hope it has, at the very least, entertainment value.
VSR
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.