PSUD is a lovely power supply simulator. That's all it does, and it does the job brilliantly, with traces for looking at voltage or current through any element of the supply. The parts list even contains the exact rectifier you want.
PSUD2
Geek-Bob says two solder-burned thumbs way up!
PSUD2
Geek-Bob says two solder-burned thumbs way up!
Last edited:
I have been watching with keen interest a few people debate cap size in power supplies. One fair observance is that large caps when depleted call for gobs of current to replenish. If you have accommodated that lust for current with sufficient L so that that L can supply the demand for current through its collapsing field, then large C is good. Further, diode isolation of large C can assist in gaining from large C's low resistance and grand ability to filter ripple without discharging them so their hunger for current comes into play.
Lundahl is trying to sell you transformers but that's OK. They have been poo pooed for being expensive and heavy but dang, they do good things when properly applied. I'd follow their (Lundahl's) advice and stay away from very large caps with the above/given caveats.
Lundahl is trying to sell you transformers but that's OK. They have been poo pooed for being expensive and heavy but dang, they do good things when properly applied. I'd follow their (Lundahl's) advice and stay away from very large caps with the above/given caveats.
jdg123,
Unfortuneately, I can't afford Lundahl. I'm just trying to pick their brains. The "mystery" power supply I picked up for a song had a Merit 436-0-436 vac PT, a large Thordarson swinging choke 5-20H 300mA, & two smaller Stancor 4.5H 200mA chokes.
I'm very interested in more explanation of applying "diode isolation". Can you point me in the right direction ? Is this similar to having a tuned resonance choke with a small bypass cap?
Frank
Unfortuneately, I can't afford Lundahl. I'm just trying to pick their brains. The "mystery" power supply I picked up for a song had a Merit 436-0-436 vac PT, a large Thordarson swinging choke 5-20H 300mA, & two smaller Stancor 4.5H 200mA chokes.
I'm very interested in more explanation of applying "diode isolation". Can you point me in the right direction ? Is this similar to having a tuned resonance choke with a small bypass cap?
Frank
I am told it was covered in the radiotron handbook in '64 but I only had a few weeks with one I got from the library and never did find the section (It was sort of like the fog that descends at a video store-forgot about any subject of interest as it was soon displaced by another!)
My experience with diode isolation is through a fellow SET12 on the audiokarma site. I have heard his rig using 811-10 triodes and diode isolation with very low DCR. It rocks! Superbdynamics!
Basically after each C in a CLCLC C power supply the last C something around (40uF) can only be recharged by current passing through the chokes and not by the large caps that are doing the filtering because after each C (save the last) there is a diode preventing the discharge from going "forward" to the next cap. The first cap is chosen by rectifier choice, the next two are very large, 1000uF and then the last "reserve" resistor about 40uF.
The tuned resonance, if I catch your meaning, is about the nature of power supplies that begin with a choke. Read some good stuff on that today under a google search of SEAmplifiertheory and Resonant-choke Power Supply. For cap size choice, in the end, you are going to have to feed the cap that feeds the plate. Caps eat current, not milliamps but Amps! Managing that cycle of discharge and charge is a trick. I'd love to see it broken down by the nanosecond but I've not met the man or the program that explains it so.
I invite you to explore SET12 threads on power supply as well as the ones mentioned above. Don't worry about finding THE answer. I've been convinced that low DCR and diode isolation is a great combination for a SET. I decided by listening and logic. Good luck!
My experience with diode isolation is through a fellow SET12 on the audiokarma site. I have heard his rig using 811-10 triodes and diode isolation with very low DCR. It rocks! Superbdynamics!
Basically after each C in a CLCLC C power supply the last C something around (40uF) can only be recharged by current passing through the chokes and not by the large caps that are doing the filtering because after each C (save the last) there is a diode preventing the discharge from going "forward" to the next cap. The first cap is chosen by rectifier choice, the next two are very large, 1000uF and then the last "reserve" resistor about 40uF.
The tuned resonance, if I catch your meaning, is about the nature of power supplies that begin with a choke. Read some good stuff on that today under a google search of SEAmplifiertheory and Resonant-choke Power Supply. For cap size choice, in the end, you are going to have to feed the cap that feeds the plate. Caps eat current, not milliamps but Amps! Managing that cycle of discharge and charge is a trick. I'd love to see it broken down by the nanosecond but I've not met the man or the program that explains it so.
I invite you to explore SET12 threads on power supply as well as the ones mentioned above. Don't worry about finding THE answer. I've been convinced that low DCR and diode isolation is a great combination for a SET. I decided by listening and logic. Good luck!
If I understand correctly you are saying that diodes are inserted in series with each smoothing choke. Why? It is hard to think of a scenario in which these diodes will do anything other than forward conduction under 'normal' conditions. Are they intended to suppress subsonic ringing caused by using caps which are far too big?
I'm not sure I know what this means. Maybe that is why nobody has explained it?
If the caps you use are all linked, like soldiers, when one, say the last in the series just before the plate of the output tube, discharges even a little bit, the others will lend current. Mind you I am not an EE. I am relaying to you what has been told to me...These others, the big ones, will then, due to their very low resistance call for all the current there is in your power supply until they are fully charged. It would be better to only have that last cap to charge rather than have the big ones lend it current but you need the big ones for their filtering duties. The solution is to keep them from lending their voltage by isolating them from the ones forward through diodes.
As far as looking at the rush of electrons in terms of nanoseconds it makes sense to me! They travel (even having to accelerate across free space of the vacuum tube) at 200 million miles an hour, they get around. If you want to look at the cycling of those caps you are wondering about then you'd need to know how much charge is moving and how fast (it accumulates), the when and where of the needs of each device, through the whole amplifier! Everything is linked, not at the speed of thinking or even the human heart but at the speed of (nearly) light (and of course at one end it all changes direction at approximately every 60th of a second,) everything effects everything else. Hopefully, every part is needed.
Truly, there has been little spoken about the amplifier as a complete unit save some sonic reviews... All of us on this site and others seem to be slightly confused about some little aspect of how the thing works, we follow those before us and make a few stabs and changing values or arrangements based on again, those who came before us...much, much before us.
Don't get me wrong, I am impressed how much the good men and women of science and industry have learned to manipulate electrons and chemistry but I don't think we have it all down. If we did there would have been established by now the "right" way to build...well, anything! I'd just love to see one devoted scientist develop a comprehensive series of events for a simple system like an SET as has been done for the human genome project. It may not be that complex or important but I can't help but want what I want
Cheers!
As far as looking at the rush of electrons in terms of nanoseconds it makes sense to me! They travel (even having to accelerate across free space of the vacuum tube) at 200 million miles an hour, they get around. If you want to look at the cycling of those caps you are wondering about then you'd need to know how much charge is moving and how fast (it accumulates), the when and where of the needs of each device, through the whole amplifier! Everything is linked, not at the speed of thinking or even the human heart but at the speed of (nearly) light (and of course at one end it all changes direction at approximately every 60th of a second,) everything effects everything else. Hopefully, every part is needed.
Truly, there has been little spoken about the amplifier as a complete unit save some sonic reviews... All of us on this site and others seem to be slightly confused about some little aspect of how the thing works, we follow those before us and make a few stabs and changing values or arrangements based on again, those who came before us...much, much before us.
Don't get me wrong, I am impressed how much the good men and women of science and industry have learned to manipulate electrons and chemistry but I don't think we have it all down. If we did there would have been established by now the "right" way to build...well, anything! I'd just love to see one devoted scientist develop a comprehensive series of events for a simple system like an SET as has been done for the human genome project. It may not be that complex or important but I can't help but want what I want
Cheers!
If the diodes are in series with the chokes then they won't do what you seem to be describing. If they are in series with the caps then they might do what you are describing but this seems to me to be a bad idea. Why include big fat smoothing caps, then stop them from doing their job by isolating them with diodes? The caps would simply charge up to the highest voltage ever applied, then the diodes switch off and you might as well remove the caps altogether as they no longer do anything. I hesitate to dismiss an idea without investigating it, but it sounds completely daft to me. Maybe I have misunderstood your description, or maybe you have been sold snake oil. Perhaps you could post a diagram or a link to where this idea is discussed?
Talk of nanoseconds is merely adding confusion. Components isolated behind diodes clearly are not needed. Plenty of science has been applied to audio, but some people either don't understand it or don't believe it. More science would help, but mostly would still be misunderstood or ignored.
Talk of nanoseconds is merely adding confusion. Components isolated behind diodes clearly are not needed. Plenty of science has been applied to audio, but some people either don't understand it or don't believe it. More science would help, but mostly would still be misunderstood or ignored.
I'm not an expert on this subject, but I would say that there are three key concerns to sizing an LC, CLC, etc. power supply filter.
1. Desired input ripple (120Hz or 100Hz for a full-wave rectified supply) attenuation level. The inductor(s) and the capacitor(s) of the main supply filtering form an A.C. voltage-divider network. Having both Ls and Cs makes a resonant multi-pole filter. Multi-pole filters are more effective, but the resonant nature of a LC type filter needs to be accounted for as mentioned in the next concern.
2. The characteristic impedance of a resonant LC type filter should be determined, as you will need to ensure that there is enough series resistance within the filter so that the capacitors and inductors don't have undamped ringing at their resonant frequency, and potential low impedance loading of the power transformer. Or, ensure that their resonant frequency is below any which the filter might be presented with. Usually, but not always, the series inductor has enough winding resistance to satisfy this concern.
3. Desired maximum A.C. output impedance of the supply at the lowest frequency which the load might demand current, usually taken to be 20Hz for audio applications. Remember, the D.C. output impedance of the supply will include the series inductor's winding resistance, while the shunt capacitor will present an infinite impedance. As the frequency of the load's current demand rises the inductor presents an increasing output impedance while the shunt capacitor presents a decreasing one. In effect, the audio band output impedance is the parallel combination of those two impedances at a given frequency.
Said another way, the A.C. output impedance at a given frequency, along with the current demanded by the load at that frequency, will determine the ripple voltage on the D.C. supply rail. So, you would want to set a maximum D.C. supply rail ripple voltage depending on the power-supply rejection ratio of of any following regulator and of the amplification circuit itself.
There are any number of resources for the relevant formulas which to calculate such values . Here's one that google found right off.
Elliott Sound Products - Linear Power Supply Design
1. Desired input ripple (120Hz or 100Hz for a full-wave rectified supply) attenuation level. The inductor(s) and the capacitor(s) of the main supply filtering form an A.C. voltage-divider network. Having both Ls and Cs makes a resonant multi-pole filter. Multi-pole filters are more effective, but the resonant nature of a LC type filter needs to be accounted for as mentioned in the next concern.
2. The characteristic impedance of a resonant LC type filter should be determined, as you will need to ensure that there is enough series resistance within the filter so that the capacitors and inductors don't have undamped ringing at their resonant frequency, and potential low impedance loading of the power transformer. Or, ensure that their resonant frequency is below any which the filter might be presented with. Usually, but not always, the series inductor has enough winding resistance to satisfy this concern.
3. Desired maximum A.C. output impedance of the supply at the lowest frequency which the load might demand current, usually taken to be 20Hz for audio applications. Remember, the D.C. output impedance of the supply will include the series inductor's winding resistance, while the shunt capacitor will present an infinite impedance. As the frequency of the load's current demand rises the inductor presents an increasing output impedance while the shunt capacitor presents a decreasing one. In effect, the audio band output impedance is the parallel combination of those two impedances at a given frequency.
Said another way, the A.C. output impedance at a given frequency, along with the current demanded by the load at that frequency, will determine the ripple voltage on the D.C. supply rail. So, you would want to set a maximum D.C. supply rail ripple voltage depending on the power-supply rejection ratio of of any following regulator and of the amplification circuit itself.
There are any number of resources for the relevant formulas which to calculate such values . Here's one that google found right off.
Elliott Sound Products - Linear Power Supply Design
Last edited:
jdg123: you seem to completely disregard emotion and preference in your argument above, with these taken into account, no matter how good and complete the science is (and its pretty bloody complete in our arena, but not totally) it can never happen; we will never agree on one best and final design, because it is different things to different people, thus your argument holds no water imo.
the articles by pass, self, jung, borbely (sorry you can tell i'm a SS guy, dont know any tube aficionados of the same calibre) all do a pretty damn good job of talking about the amplifier as a whole, pass especially as many of his designs are so paired down that the power supply and amplifier are literally one unit, for example many of the x'd designs feed the output so very close to the power supply return that you would be daft to have it any other way.
the articles by pass, self, jung, borbely (sorry you can tell i'm a SS guy, dont know any tube aficionados of the same calibre) all do a pretty damn good job of talking about the amplifier as a whole, pass especially as many of his designs are so paired down that the power supply and amplifier are literally one unit, for example many of the x'd designs feed the output so very close to the power supply return that you would be daft to have it any other way.
I cannot find "Pictures and albums" under my control panel...If I sort that out I'll put up a hand drawn schematic of the system of very large and much smaller caps in a power supply, the one I listen to and felt had superior dynamics. I am no EE. I only tried to offer some option to a fellow who had questions about cap sizing. try looking into this thread:
Power supply for 811A SE, part deux... - Page 2 - AudioKarma.org Home Audio Stereo Discussion Forums
Set12 is the man to see for diode isolation of large capacitors...
I don't think our knowledge is even "complete" in audio, I think it is pretty darn good, a big step up from the gramophone for sure considering what one can do with just a pencil, some common knowledge and a box of old parts. I'm impressed by and in fact love my seriously flawed stereo as it sits! I just can buy we know it all, not by far. There are serious problems with transducers, serious problems with magnetic fields, durations and timing limitations heck, we cannot agree of what we want from our illusion producing devices! To say that we know all there is to know about electrons and what they do is pure hubris. Anyone can say "its good enough for me", that is fine. I'd just like to know a lot more and in turn use it to make more people smile, more broadly.
Power supply for 811A SE, part deux... - Page 2 - AudioKarma.org Home Audio Stereo Discussion Forums
Set12 is the man to see for diode isolation of large capacitors...
I don't think our knowledge is even "complete" in audio, I think it is pretty darn good, a big step up from the gramophone for sure considering what one can do with just a pencil, some common knowledge and a box of old parts. I'm impressed by and in fact love my seriously flawed stereo as it sits! I just can buy we know it all, not by far. There are serious problems with transducers, serious problems with magnetic fields, durations and timing limitations heck, we cannot agree of what we want from our illusion producing devices! To say that we know all there is to know about electrons and what they do is pure hubris. Anyone can say "its good enough for me", that is fine. I'd just like to know a lot more and in turn use it to make more people smile, more broadly.
I could not see the diagram, as I am not registered on that site. The discussion about stored energy seems to be based on misunderstandings. They talk about a transient lasting 10 seconds - if it lasts for 10 secs then it isn't a transient! They calculate the energy needed to provide for that, then provide caps to store that much energy. This seems to ignore two facts: you can't completely discharge the caps to recover all that energy, and the caps are being recharged on each mains cycle. I would not base my designs on stuff like that.
I've found something, that for me, is a Holy Grail of information:
The Radio Amateur's Handbook, dated 1936 & written by Hiram Percy Maxim (the inventor of the gun silencer & the 30.06 cartridge)
In chapter 15, he methodically guides the reader step by step to design a two stage choke input filter (LC LC)
I've now settled upon:
parallel 5R4GYA rectifiers
(1) L1- 5-20H swing choke
(1) C1- 4uF 1200v PIO cap (2x 8uF 600v in series)
(2) L2- 4.5H smoothing choke. One each for the Left & Right amp.
(2) C2- 25uF 1200v PIO cap (2x 50uF 600v in series). One each for the Left & Right amp.
Individual (separate from the HV transformer) filament transformers as per H.P. Maxim.
The idea for separate stage 2 LC's for each amp is from Lundahl.
I thank all of the respondents that have given opinions on this subject.
I'm still perplexed by those that advocate large caps but feel comfortable with my final selection of components.
Frank
The Radio Amateur's Handbook, dated 1936 & written by Hiram Percy Maxim (the inventor of the gun silencer & the 30.06 cartridge)
In chapter 15, he methodically guides the reader step by step to design a two stage choke input filter (LC LC)
I've now settled upon:
parallel 5R4GYA rectifiers
(1) L1- 5-20H swing choke
(1) C1- 4uF 1200v PIO cap (2x 8uF 600v in series)
(2) L2- 4.5H smoothing choke. One each for the Left & Right amp.
(2) C2- 25uF 1200v PIO cap (2x 50uF 600v in series). One each for the Left & Right amp.
Individual (separate from the HV transformer) filament transformers as per H.P. Maxim.
The idea for separate stage 2 LC's for each amp is from Lundahl.
I thank all of the respondents that have given opinions on this subject.
I'm still perplexed by those that advocate large caps but feel comfortable with my final selection of components.
Frank
My curiosity got the best of me and I registered to have a look. I may be missing some subtlety, but the circuit looks to be a simple cascade of diodes separating the individual capacitors of a parallel storage array. Picture a cascade of LC 'pi' filters, but with each choke replaced by a diode. Each capacitor in the array would charge to one forward-diode-drop lower than the capacitor immediately preceding it in the array. All capacitors in the array are electrolytic except for the final output cap., which would be a quality film type.
I suppose, the idea is that, for the majority of signals, the supply will take on the sonic character of that film type cap. situated at the output of the filter. The load then drawing on the mostly for large current demanding signals. Meanwhile, the large electrolytic bank well smooths the rectified A.C.. I think it's an attempt produce an asymmetric filter behavior, seeking to address the differing needs of low-frequency input-side filtering, and wide-band load-side current delivery/quality. At least, that's my take, I'm not quite sure. If so, it's an interesting notion, but I don't quite see that the final (film type) capacitor is isolated from the preceding electrolytics by a diode. As soon as the load draws current from that film cap. it will also forward bias the diode and draw current from the electrolytic as well. Perhaps, even that, makes for an audible difference.
Last edited:
You're welcome, but, I would suggest that you have selected the wrong hobby if you are adverse to critically exploring what may seem like daft ideas.
Last edited:
Ken, Thank you for taking a look. SET12 is not an EE but he IS adventurous in his builds. I have heard his one 811-10 amp driving his Klipsch Fortes and it changed my ideas about what SET could do (and about what that relatively humble speaker could do). SET12 would say that the crossovers have as much to do with his successful sound as his amp but every link in the chain must be good to get a good final result. The sound of drums on his system was so dynamic it was startling while retaining the delicate information often traded in for "slam". I am no EE either (you sound as though you might be but as it is explained to me each diode points backward (toward the rectifier) so that nothing from caps "behind" that last good cap can come forward. My ears tell me it works. I was trying to pass it on but fell down in explaining it. SET12 would be open to any info you might add or ask about. He is a fine fellow and passionate about making good things happen through well thought out, thoroughly discussed (and actually executed) plans! He is at this moment experimenting with the use of low H, high current chokes just behind that final cap to keep it well fed, never hungry. I am intrigued...
but as it is explained to me each diode points backward (toward the rectifier) so that nothing from caps "behind" that last good cap can come forward. My ears tell me it works. I was trying to pass it on but fell down in explaining it. SET12 would be open to any info you might add or ask about. He is a fine fellow and passionate about making good things happen through well thought out, thoroughly discussed (and actually executed) plans! He is at this moment experimenting with the use of low H, high current chokes just behind that final cap to keep it well fed, never hungry. I am intrigued...- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.