Oh right, this still seems to be the BT thread, Maybe Mr Curl has something to say about his PS guidelines.
(though i suspect to be somewhat different from the Japanese philosophy of the highest PSRR and most squeemish powersupply)
(Teflon lovers, give the French Chateauroux caps a try, not sure how affordable they are outside of France and/or available through Solen. Manufactured with metalised Dupont's fluorinated FEP foil, only teflon capacitor in a range up to 10uF i've heard of, and with decent Max L/D sizes.)
(though i suspect to be somewhat different from the Japanese philosophy of the highest PSRR and most squeemish powersupply)
(Teflon lovers, give the French Chateauroux caps a try, not sure how affordable they are outside of France and/or available through Solen. Manufactured with metalised Dupont's fluorinated FEP foil, only teflon capacitor in a range up to 10uF i've heard of, and with decent Max L/D sizes.)
GRollins said:
I started this by throwing out another way of looking at power supplies based on my experience. I tripped across the seed of the ideas years ago and have experimented with the concepts in everything I've built since.
Sure, quite possibly I've built a house of cards, but then how does one know unless you throw it out there and someone else tries it and calls bull. Quite frankly I'm convinced in the results of my experimentation enough that I'm standing here p#%*ing in the wind, but curiosity is a thing of the past.
Being blunt, I'd rather draw my power off of the transformer because the caps can't keep up. This puts a whole different slant on the function of the capacitors, the supply layout, the quality of the transformer core (which I also noted years ago makes a big difference, and has not even entered the comments here) and the rectifiers.
Is my approach or the prevailing logic, so elequently summarized here by Grey, the only way? All I have to do is consider the various approaches to speakers, all with similar end results, to think that other approaches are possible.
So, unless someone has something new to add to this discussion I say we let it go before the modertors step in and shut the thread down.
Mike.
MikeBettinger said:
Mike,
The problem with that is that the transformer isn't even connected to the supply for 80% of the time or so. That's when the diodes are reversed. Only when the sec voltage gets above the (sagged) cap voltage, does the xformer get a short period to recharge those caps.
Jan Didden
G.Kleinschmidt said:
Some people cause smiles by entering a room, others do so when they leave.
anatech said:
Switching power supplies will generate more RF noise than I care to deal with. That the industry is headed that way seems clear. If nothing else, more efficient power supplies will be seen as "green." Whether they are good for sound quality is another question entirely.
As for the load placed on the rectifier, I have three observations:
1) I ran my Threshold S-500s with roughly 250000uF for several years and the (quite ordinary) bridge rectifiers that came stock in the amps never complained. The same bridges are still in those amps today.
2) If it does become a problem, put an inductor or resistor in front of the cap bank. Granted, adding resistance runs counter to my philosophy of regarding the power supply as a voltage source, so I prefer inductors, but either will work.
3) Use fast/soft rectification--which to me is a given, regardless of how the rest of the circuit is handled.
If large amounts of capacitance are indeed a fad, it's been a remarkably persistent one--I remember it as going back at least as far as the early to mid-'70s and depending on how big is big, tube folks added capacitance even earlier than that. To me, something that's been accepted that long is hardly a fad.
As I said above, I'm trying output rail regulation for the first time. I may stay with it, I may not. One of the factors I'm facing is that I wanted 25V rails for the output stage, but the raw rail was coming in at just under 30V. I made a list of options and chose regulation.
MikeBettinger said:
Everything is up for grabs. You don't even have to have a transformer. You can build a really big amp by simply rectifying the line voltage and accepting that as your rail. That way you're freed of all the constraints the transformer places on your power supply. On the other hand you're wide open to RF riding in on the AC, etc. The transformer's inductance is a double-edged sword. It's a nuisance in some ways, but it also helps cut the nasties.
Some sort of rectification is necessary, but you've got several options on devices. A lot of tube folks swear by tube rectification. If you go with solid state, you've got Schottky, fast, fast/soft, whatever floats your boat.
Filtration can be any combination of R, L, and C that gets you where you want to go.
I confess that I split cap multipliers off into a separate category from regulation. To me, a regulator has a voltage reference of some sort. A cap multiplier doesn't, so I regard it as a separate entity. It's a matter of semantics. If you go with regulators, you can go with feedback or no feedback. Quite a number of people feel that no feedback regulators sound better. Feedback regulators arguably give you somewhat better control over the voltage and the Zout of the power supply.
Somewhere in the various permutations you can hopefully find a happy balance for what you're trying to do. At the moment I'm hip deep in amplifier-related thoughts, so I tend to reference the amp I'm working on, but the principles are pretty much the same regardless of whether you're building a phono stage or a power amp. Your focus tends to slide a bit, in that a phono stage power supply must be ultra-quiet, whereas a power amp will need much more current capability, but the tools you have available to do the job remain the same.
Grey
janneman said:
I know. I've been called on that before and I keep playing with this and still stand my ground. I'll repeat what I've said in the past. The real power is being supplied by the transformer and caps are there to prop it up during the off times. It's an imperfect system but nobody's come up with a way to supply DC without the AC as a starting point.
The caps stink at keeping up with the speed of the amps demand. Or maybe I'm hearing the minimizing of the charging currents to the caps the way I do the layout. Don't know.
I do know that even with a large bank of caps the supply is jumping all over the place (relatively speaking) because of their slow response time. So part of the stability argument of the large cap bank is questionable.
In other words I have no acceptable response to your statement.
Mike
Hi Grey,
We gain:
Weight reduction
Size reduction
Supply rail regulation (big advantage)
Lower noise on our supply lines as it is easier to filter RF frequencies out.
A cost reduction in both materials, and more importantly - shipping our product to market.
An opportunity to control shut down quickly in the event of a circuit fault. This reduces the cost of warranty and can enhance reliability. You may also be able to reduce collateral damage to speakers.
Hi Jan,
With large capacitance, I'll bet the conduction angles are a lot less than 20 % unless under very heavy load.
-Chris
True, but consider this a little further then. It's much easier to filter out the RF at the source. It's even far removed from our area of operation. We tend to completely shield normal power supplies anyway, so why not switchers?
We gain:
Weight reduction
Size reduction
Supply rail regulation (big advantage)
Lower noise on our supply lines as it is easier to filter RF frequencies out.
A cost reduction in both materials, and more importantly - shipping our product to market.
An opportunity to control shut down quickly in the event of a circuit fault. This reduces the cost of warranty and can enhance reliability. You may also be able to reduce collateral damage to speakers.
Hi Jan,
With large capacitance, I'll bet the conduction angles are a lot less than 20 % unless under very heavy load.
-Chris
I agree that the layout is quite important. It's very easy to run a charge puls through a piece of wire or PCB tracking and then mess up the signal ground so that you actually add the ripple voltage across the wire or track to the input signal.
But caps stinking for a slow response, I'm not sure what you mean, but I can't see that if the caps are on their own for 80% of the time, how a bigger xformer can 'unstink' them?
Jan Didden
But caps stinking for a slow response, I'm not sure what you mean, but I can't see that if the caps are on their own for 80% of the time, how a bigger xformer can 'unstink' them?
Jan Didden
anatech said:
Ohh yes. I've seen 10%. Thing is, the larger the caps, the shorter the charge period, and the sharper (more harmonics) the ripple. Going over the top with cap values demonstrably increases ripple harmonics on the supply, so I am a little wary of those reports that the bigger the cap, the better the sound.
Jan Didden
GRollins said:
The problem with a fad is that once they take hold they tend to get implemented without thinking, experimentation tends to be done with a pre-conceived outcome. There are so many variables to a project it's nice to have one less thing to think about.
My point is, this is one that bears looking at. I guess the Japanese experimenters have, based on the earlier comment.
I agree with your comments on the rules being the same across the board. I'm struggling (still) with a MC amp and have spent some time relooking at the raw supply and what works with the amp works here as well.
Mike.
Hi Jan,
Agreed. From what I have seen, larger caps seem to create more problems than they solve. I think there is a minimum size of capacitor for a given application and increasing the size beyond a higher point in the "good range" tends to bring about more problems.
Hi Grey,
-Chris
Agreed. From what I have seen, larger caps seem to create more problems than they solve. I think there is a minimum size of capacitor for a given application and increasing the size beyond a higher point in the "good range" tends to bring about more problems.
Hi Grey,
Lemmings. That's all I have to say. I don't think a great deal of thought goes into this other than "more is better". Not a convincing argument as we know better.
And those are the designs that didn't age well. Consider tube designs of the 50s and 60s. They aged very well. All those large capacitance designs in the 70s did not. They also often have other design errors that you would not have seen in the earlier days when real engineers who knew what they were doing designed audio equipment. I'm not saying that the earlier "classic" designs didn't have their own faults. I am saying that those earlier designs didn't commit design blunders like later ones do.
So, how long did we know that the earth was flat? Beliefs can last a very long time and still be wrong.
On that, I think we can all agree. Nothing will be perfect. The trick is to identify when we have gone far enough in a certain direction, and then change our approach to a more effective direction before losing too much. Engineering, intelligent mixture of compromises depending on your goals. Having realistic goals help to, so half the job of an engineer is to define what is a reasonable goal to begin with. 😉
-Chris
janneman said:
Then, why can the differences between transformers be heard?
I use the words speed, fast and stink because they make sense to me. It's a visual thing and I hate sounding like I'm quoting a textbook, been called on this before
Mike.
MikeBettinger said:
Can they? Between, say, a 300VA and a 800VA for a 2*100W amp? Or is this another "everybody knows that..." ?
Jan Didden
MikeBettinger said:
Actually I'm making a big assumption there. This is a stupid path to turn down.
Nevermind.
Hi Jan,
You know, I've been holding my tongue on that one. I figure the only way you can tell is if one transformer has a problem, or the PSRR of the amp is very poor. Wiring problems included.
The entire purpose of a filtering section is to eliminate effects upstream, like transformers and AC supply noise. The amplifier designs normally further this rejection.
-Chris
You know, I've been holding my tongue on that one. I figure the only way you can tell is if one transformer has a problem, or the PSRR of the amp is very poor. Wiring problems included.
The entire purpose of a filtering section is to eliminate effects upstream, like transformers and AC supply noise. The amplifier designs normally further this rejection.
-Chris
janneman said:
Like I said, stupid direction to turn. Speaking personally, I can hear the difference, but I would be speculating and hypothesizing as to what the mechanism is that I’m hearing. It’s the same reasoning for not responding to the slow caps question.
If one was to try to hear the difference it would be heard more readily by comparing similar rated transformers with different laminations and construction approaches. Try listening to a toroid and a transformer from a mid 80's Japanese power amp, especially in a preamp application (80-100 watt). One of the ones with the real thin grain oriented silicon steel laminations (Bando is a great mfr). Or worse yet try a the same vintage off the shelf Stancor or Triad.
More to the point, out in the rest of the world I hear recent reports of R-core transformers sounding better for audio as one example. Another would be reports of sound quality differences between toroids and EI cores. These come from somewhere. But then just like double blind testing whether or not you hear differences comes down to how resolving the rest of the system is.
There is only madness in this direction.
Mike.
MikeBettinger said:
No, not at all. There are sound 😉 reasons for audible differences in xformers. For instance, toroids are very wideband and let almost anything through that is on the mains, noise, EMI etc. Depending on the amplifier it feeds, that junk *may* get into the amp and signal. Thats the reason why I like to avoid toroids in low-level preamp/DAC type equipment.
Then there is the mechanical noise that EI and other laminated cores often emit (the toroid has the advantage here).
I am no expert on the R-cores but apparently they combine the best of the two - quite good isolation from mains borne junk, low mech noise and good efficiency - and not too expensive.
But saying that an upgrade from xVA to 2xVA improves the sound, there I am sceptical, to put it mildly.
Jan Didden
As for why a larger transformer might sound better, I've already stated at least once or twice--and rdf has also posted to the same effect--lower impedance. Given that a power supply should theoretically approach a perfect voltage source, that should be a relatively easy thing to follow from a theoretical standpoint. Whether that's the actual reason or not (I suppose it's possible that there are other contributing factors) is irrelevant. It provides a very realistic starting point for those who are prone to demanding why before they will even consider trying something new.
Or you could simply try it and listen rather than sit and theorize.
The problem I have with this is that there is no way I can see that a capacitor can slow a transformer down. As a semi-reasonable analogy, consider two resistors in parallel. Suppose you have R1= 5 Ohms and you want to decrease that. You could parallel it with another 5 Ohm resistor and you'd end up with an effective resistance of 2.5 Ohms. If you were to use a 1 Ohm resistor, you'd end up with an effective resistance of .83 Ohms. But here's the kicker--if you paralleled R1 with a 1M resistor, you'd end up with 4.999975 Ohms. In other words, even with an enormous resistor, it still leads to a decrease in the net resistance. No matter how slow you regard the caps as being, there will always be some improvement in speed, even if it's not as much as you might wish.
As for a rail that's 'jumping all over the place,' that's definitely not a symptom of too much capacitance. Assuming that the circuit is held constant, it's either too small a transformer or too little capacitance. Another analogy: Consider various bodies of water. If you take an 8oz glass of water out of a bucket it makes a very noticeable difference in the level of the remaining water. If you take a glass of water out of a swimming pool, you could probably measure it if you had sensitive enough instrumentation, but it certainly wouldn't be something you could see with the naked eye. If you took the glass of water out of the ocean, there aren't enough zeros in the world to measure the difference it would make in sea level, even on a local basis.
It's clear that you've never heard an Audio Research D-79 or Conrad Johnson Premier One. Both of which have held their value very nicely while designs with smaller (J/W) designs continue to fall.
I find it curious that people who say that there's such a thing as too much capacitance accuse others of being 'lemmings' or saying that they're following a 'fad,' only to turn and congratulate one another on being on the right side of the discussion, all the while saying others are hearing what they 'expect' to hear. But they have no such preconceived notions...naturally.
The only two arguments that hold water as far as I'm concerned are turn-on surges and lowered conduction times in the rectifier. Both are easily cured. In the real world, you reach a limit either monetarily or in space required, but that's not to say that there's no benefit at all. It is a diminishing returns curve, though--the difference between a swimming pool and an ocean is enormous from a financial and space-required point of view compared to the benefit. The designer, whether DIY or commercial, decides how far up the curve they want to go and proceeds from there.
Sidestepping into a higher performing cap while holding the total capacitance the same changes the rules of the game. It tilts the perceived spectrum towards the mids and high end since they improve most dramatically. Then you have to begin the add-more-capacitance run all over again...after robbing a bank, of course. Over time, I'm hoping that film cap prices will fall in real world dollars, even if it's only that their price remains relatively constant as inflation takes a toll on everything else. One percent metal film resistors used to be rare and expensive. Now they're the rule rather than the exception. With luck something similar will happen with film caps.
Grey
Or you could simply try it and listen rather than sit and theorize.
MikeBettinger said:
The problem I have with this is that there is no way I can see that a capacitor can slow a transformer down. As a semi-reasonable analogy, consider two resistors in parallel. Suppose you have R1= 5 Ohms and you want to decrease that. You could parallel it with another 5 Ohm resistor and you'd end up with an effective resistance of 2.5 Ohms. If you were to use a 1 Ohm resistor, you'd end up with an effective resistance of .83 Ohms. But here's the kicker--if you paralleled R1 with a 1M resistor, you'd end up with 4.999975 Ohms. In other words, even with an enormous resistor, it still leads to a decrease in the net resistance. No matter how slow you regard the caps as being, there will always be some improvement in speed, even if it's not as much as you might wish.
As for a rail that's 'jumping all over the place,' that's definitely not a symptom of too much capacitance. Assuming that the circuit is held constant, it's either too small a transformer or too little capacitance. Another analogy: Consider various bodies of water. If you take an 8oz glass of water out of a bucket it makes a very noticeable difference in the level of the remaining water. If you take a glass of water out of a swimming pool, you could probably measure it if you had sensitive enough instrumentation, but it certainly wouldn't be something you could see with the naked eye. If you took the glass of water out of the ocean, there aren't enough zeros in the world to measure the difference it would make in sea level, even on a local basis.
anatech said:
It's clear that you've never heard an Audio Research D-79 or Conrad Johnson Premier One. Both of which have held their value very nicely while designs with smaller (J/W) designs continue to fall.
I find it curious that people who say that there's such a thing as too much capacitance accuse others of being 'lemmings' or saying that they're following a 'fad,' only to turn and congratulate one another on being on the right side of the discussion, all the while saying others are hearing what they 'expect' to hear. But they have no such preconceived notions...naturally.
The only two arguments that hold water as far as I'm concerned are turn-on surges and lowered conduction times in the rectifier. Both are easily cured. In the real world, you reach a limit either monetarily or in space required, but that's not to say that there's no benefit at all. It is a diminishing returns curve, though--the difference between a swimming pool and an ocean is enormous from a financial and space-required point of view compared to the benefit. The designer, whether DIY or commercial, decides how far up the curve they want to go and proceeds from there.
Sidestepping into a higher performing cap while holding the total capacitance the same changes the rules of the game. It tilts the perceived spectrum towards the mids and high end since they improve most dramatically. Then you have to begin the add-more-capacitance run all over again...after robbing a bank, of course. Over time, I'm hoping that film cap prices will fall in real world dollars, even if it's only that their price remains relatively constant as inflation takes a toll on everything else. One percent metal film resistors used to be rare and expensive. Now they're the rule rather than the exception. With luck something similar will happen with film caps.
Grey
- Status
- Not open for further replies.