G.Kleinschmidt said:
Bingo!
Got it in one.
And I suppose you expect us to believe that this constituted a proper listening session? Oh, and don't forget the front end and speakers...they were of good quality also, right?
You were a careful listener when you were a teenager...right?
You weren't?
Oh.
Well, that pretty much explains things, doesn't it?
Let's see...presumptions, eh? So you're saying that increasing the power supply capacitance doesn't lower the AC impedance to ground? Ooops! Wrong. Or were you saying that a larger bank of capacitance doesn't store more charge? Ooops! Wrong again. Or perhaps you meant that you don't believe that the signal can modulate the rails? Ooops! Wrong again.
You're long on attitude, but short on practical knowledge. Or as I like to put it: Reality beats theory seven days a week.
Your post is its own refutation.
As usual.
Tony said:
Yes. Unfortunately, electrolytic capacitors have pretty poor high frequency response. They're good at getting a lot of fairly economical capacitance into a smallish space, but they're far from the ideal cap. A good film cap--meaning polypropylene, polystyrene, or (gulp! rob a bank first) Teflon--comes about as close to an ideal capacitor as present technology allows. There's still a lot to be desired, but the high frequency response of a film and foil cap is far superior to that of an electrolytic.
The analogy is hardly exact, but you can think of an electrolytic as a woofer; rather slow-ish, but good at moving air/electrons. If you can live with that metaphor, then picture the film cap as a tweeter; fast--blindingly so compared to an electrolytic--but not so good at heavy hauling because film caps are, of necessity, small in value.
Okay, so what happens when you bypass an electrolytic with a film cap? The high frequencies now have a lower impedance path from the rail to ground. Or, if you prefer, the energy stored in the film cap is more quickly accessed by the circuit. Both points of view are correct. Neither should be held as a religion to the exclusion of the other.
Grey
GRollins said:
ROFLMAO.
Yes, the circuits were crappy. That's why they needed abnormally huge amounts of filter capacitance (which you prescribe for your circuits) principally due to lousy PSR. I'm glad you've finally got it.
GRollins said:
Yep, presumptions.........
GRollins said:
No
GRollins said:
Sorry, wrong agian.
GRollins said:
Yep, you're wrong again.
You have got to be kidding. But, thanks for such a productive post.
I wondered how you guys managed to ruin even this thread. A nice mess you've made of it. Thanks admin, there is Ignore List here.
Tony said:
That is usually the intention, but it is rarely executed with any real degree of success (as far as the main supply rails of power amplifiers go). Modern electrolytic capacitors have rather low inductance and it is more than easy to circumvent any potential benefit provided by a heap of film capacitors with several inches of PCB trace or connecting wire.
The bypass caps must be PCB mounted and big fat tracks (for low inductance) are mandatory.
There are several reasons for film bypasses across electrolytic caps. First, electrolytic caps are usually pretty lousy at high frequencies, so a film cap can provide a bypass path for RFI, either self generated by the rectifiers or from the power line. However, too much of a good thing can be silly too. IF you see rows and rows of caps in a power supply, usually this is more for the 'eye' than for the ear.
When I first worked with Parasound, many extra caps were added to the power supply that were unnecessary and even detrimental to the sound. We cut them out, and found that the amp actually performed better.
It must also be understood that 'ground return' is just as important as DC supply bypassing. This is another good reason for quality film bypass caps.
When I first worked with Parasound, many extra caps were added to the power supply that were unnecessary and even detrimental to the sound. We cut them out, and found that the amp actually performed better.
It must also be understood that 'ground return' is just as important as DC supply bypassing. This is another good reason for quality film bypass caps.
mr john curl,
some time ago you came up with the value of optimum voltage drop across emitter resistors...
have you investigated the ratio of power transformer impedance versus power supply capacitance impedance for optimum ratio?
thank you...
some time ago you came up with the value of optimum voltage drop across emitter resistors...
have you investigated the ratio of power transformer impedance versus power supply capacitance impedance for optimum ratio?
thank you...
john curl said:
The bypass film capacitor sometimes provides parasitic resonances with the pretty lousy electrolytic cap, those who have no tools to verify should not use it as a rule.
I don't think this conclusion follows.GRollins said:
Vpk into a reactive load will determine Ipk. If both the low frequency and high frequency loads are equally reactive (or resistive) then the Ipk will be equal.
I reckon this applies particularly to similar sensitivity drivers. The transient peak current into a treble driver can be just as large as the sustained peak current into a bass driver.
& yes, the smoothing/decoupling requirements for supplying these different signals will/are very different.
While your argument is of course true, there are two points which support Grey; first bass drivers show at least at one point usually far lower impedance than trebles and need therefore more current for certain spl compared to most trebles*. 2nd most energy in music is stored in rather low frequency content (below 5kHz, easily seen in FFT-plots of your favourite cds).
All the best, Hannes
*ribbons are a big exception with their nice 2Ohms.
All the best, Hannes
*ribbons are a big exception with their nice 2Ohms.
now you are asking us to compare drivers of a different impedance and using that to support the erroneous conclusion that bass requires more peak current than treble.h_a said:
Well, it's nothing more than a general observation. I don't see as to which this contradicts anything you wrote nor how it should complicate things.
Maybe Grey could have been more precise and stating that with common drivers and common music material, more than average current is asked for the bass frequencies. Agreed?
Have fun, Hannes
Maybe Grey could have been more precise and stating that with common drivers and common music material, more than average current is asked for the bass frequencies. Agreed?
Have fun, Hannes
AndrewT said:
This is true, but the bass peak has much longer duration resulting in more stored energy to be consumed.
The treble problem is the rise time of the peak not the energy consumed.
Uli
Once you start down that path you get into the ratio of the duration of the low frequency signal vs. the duration of the charging cycle which will in turn depend on the impedance of the transformer in addition to the capacitor bank. If the transformer is big enough it will have large gauge windings, hence low impedance, and will be able to recharge the capacitor bank faster. The best possible outcome is to have the transformer be able to supply sufficient current that the capacitors almost become superfluous--at least in terms of charge storage. Unfortunately, like an all film cap filter, this can get impractically expensive for an amplifier. It's easy for preamps, though.
This argues for transformers that are rated for more current than pencil and paper would first indicate--simply to get a lower impedance. I generally try for at least twice what the "regular" transformer "should" be. Again, not so bad for low-level circuits, but it's a booger when you get to power amps. Sometimes you can't even find a transformer in the size you're looking for if you're building a class A amp. To get the current rating you would have to accept a lower rail voltage. Class AB or B amps are somewhat easier in that the current demands are the same as far as the signal goes, but the transformer doesn't have to endure the unrelenting load presented by a class A circuit and it doesn't get as hot.
If you can get a circuit that draws something pretty close to DC, then things are better still. The inductance of the transformer begins to fade in importance and you begin looking more at the DC performance as a stand-alone feature.
G.Kleinschmidt,
You're a lost ball in high weeds. I pity you.
Grey
This argues for transformers that are rated for more current than pencil and paper would first indicate--simply to get a lower impedance. I generally try for at least twice what the "regular" transformer "should" be. Again, not so bad for low-level circuits, but it's a booger when you get to power amps. Sometimes you can't even find a transformer in the size you're looking for if you're building a class A amp. To get the current rating you would have to accept a lower rail voltage. Class AB or B amps are somewhat easier in that the current demands are the same as far as the signal goes, but the transformer doesn't have to endure the unrelenting load presented by a class A circuit and it doesn't get as hot.
If you can get a circuit that draws something pretty close to DC, then things are better still. The inductance of the transformer begins to fade in importance and you begin looking more at the DC performance as a stand-alone feature.
G.Kleinschmidt,
You're a lost ball in high weeds. I pity you.
Grey
GRollins said:
That's strange. I've got a 1kW into 2 ohms per channnel, 200A peak class A rail tracking amp and a 200W into 4 ohms, 100A peak conventional class A amp under construction and all the transformers were readily avaliable off the shelf.
GRollins said:
Riiiiiiight - and you still haven't given the slightest indication that you are capable of grasping the important concept of designing an amplifier with a high PSRR, which, despite your inability to draw the distinction, is a seperate issue from providing a power supply capable of delivering an adequate amount of current for a given load impedance and power output.
I think this is a red herring, this discharge time vs period.
The only time this would affect the sound would be when you are clipping, or close to clipping. Amplifiers with poor supply rejection are not considered, as they have enough known problems to begin with.
So, can someone explain why this transformer / capacitor size issue is important as long as you have not depleted the stored energy? Personally, I can't see it.
-Chris
The only time this would affect the sound would be when you are clipping, or close to clipping. Amplifiers with poor supply rejection are not considered, as they have enough known problems to begin with.
So, can someone explain why this transformer / capacitor size issue is important as long as you have not depleted the stored energy? Personally, I can't see it.
-Chris
anatech,
Depends on what you mean by 'depleted.' Is it when the cap hits zero volts or when it drops by more than some arbitrary percentage from the nominal rail voltage?
I look at capacitors as part of an attempt to build a perfect voltage source, meaning one that will maintain voltage, no matter what the current draw. From that perspective any drop in the rail voltage means you've still got an imperfect power supply. In the real world that's inevitable, but it's still the goal. Mine, anyway. I'm sure G.Kleinschmidt finds 50% of nominal rail voltage to be quite acceptable, given his attitude. From his point of view the NFB will take care of things, right?
Is PSRR useful? You bet. But it's only part of the battle. To the extent that something like a differential can shrug off ripple on the rail, that's a good start. But if the rail sags enough that the front end clips on a peak--and a peak is exactly when it would happen--then that's a transient distortion mechanism that won't show up on a THD test. (And people still defend THD as the be-all, end-all? Jeez, get real, folks. It's a decent first approximation tool, no more than that.) While a pulse test of some description would presumably show it, how many standardized pulse tests do you find on spec sheets? Nada.
So what do you do? Try to make the rail as stiff as possible. Throw as big a transformer as possible into the fray and back it up with lotsa capacitance. No, I don't see it as only a problem at clipping or something close to full power. To me it's a problem if the rail sags at any power. The power supply, which is supposed to act like a voltage source, is falling down on the job. It astonishes me that people who will go to any length to winkle out the tiniest imperfection in the circuit itself will then turn around and treat the power supply like an afterthought. But it takes all kinds to make the world go 'round.
The amp I'm working on at the moment has a fully regulated rail for the output stage--the first time I've gone to that length. Will I go to the trouble again in the future? I have no idea. It's early days yet. I nearly always regulate amp front ends, preamps and the like, but regulating the output stage is another ballgame. We'll see how it goes. And yes, there's a fair amount of capacitance after the regulation. Some argue that the regulator should see the circuit directly. I feel that even the 'fastest' regulator is going to lag somewhat behind the signal, and the caps can help supply transient current. It can also be argued that the cap bank slows the regulator's response in that it takes the regulator longer to recharge the caps than it would a smaller amount of capacitance or even no capacitance at all. This point of view ends up looking at the regulator as a synthetic capacitor. I wrestled with this for a while and decided to go with capacitance. I may do it differently next time.
Grey
Depends on what you mean by 'depleted.' Is it when the cap hits zero volts or when it drops by more than some arbitrary percentage from the nominal rail voltage?
I look at capacitors as part of an attempt to build a perfect voltage source, meaning one that will maintain voltage, no matter what the current draw. From that perspective any drop in the rail voltage means you've still got an imperfect power supply. In the real world that's inevitable, but it's still the goal. Mine, anyway. I'm sure G.Kleinschmidt finds 50% of nominal rail voltage to be quite acceptable, given his attitude. From his point of view the NFB will take care of things, right?
Is PSRR useful? You bet. But it's only part of the battle. To the extent that something like a differential can shrug off ripple on the rail, that's a good start. But if the rail sags enough that the front end clips on a peak--and a peak is exactly when it would happen--then that's a transient distortion mechanism that won't show up on a THD test. (And people still defend THD as the be-all, end-all? Jeez, get real, folks. It's a decent first approximation tool, no more than that.) While a pulse test of some description would presumably show it, how many standardized pulse tests do you find on spec sheets? Nada.
So what do you do? Try to make the rail as stiff as possible. Throw as big a transformer as possible into the fray and back it up with lotsa capacitance. No, I don't see it as only a problem at clipping or something close to full power. To me it's a problem if the rail sags at any power. The power supply, which is supposed to act like a voltage source, is falling down on the job. It astonishes me that people who will go to any length to winkle out the tiniest imperfection in the circuit itself will then turn around and treat the power supply like an afterthought. But it takes all kinds to make the world go 'round.
The amp I'm working on at the moment has a fully regulated rail for the output stage--the first time I've gone to that length. Will I go to the trouble again in the future? I have no idea. It's early days yet. I nearly always regulate amp front ends, preamps and the like, but regulating the output stage is another ballgame. We'll see how it goes. And yes, there's a fair amount of capacitance after the regulation. Some argue that the regulator should see the circuit directly. I feel that even the 'fastest' regulator is going to lag somewhat behind the signal, and the caps can help supply transient current. It can also be argued that the cap bank slows the regulator's response in that it takes the regulator longer to recharge the caps than it would a smaller amount of capacitance or even no capacitance at all. This point of view ends up looking at the regulator as a synthetic capacitor. I wrestled with this for a while and decided to go with capacitance. I may do it differently next time.
Grey
GRollins said:
To be blunt, that is inane. You’re steadfastly sure of things that make no sense whatsoever and are confident in your ignorance to a degree that beggars belief. I’m through with trying to drum a modicum of sense into this discussion.
Goodbye!
Hi Grey,
The better power amps I've seen use a different supply that is regulated to run the voltage amp stage. Therefore, some sag in the supplies for the current end will not cause any trouble. That is until you run out of volts and the gain drops in your output stage.
It comes down to accepting that your supply will always have noise riding on it. It will always have some resistance and therefore sag no matter how much capacitance you jam in there. If you want to attack that issue, why not simply build a high frequency regulated switching supply to generate your rail voltages? Then you can achieve the same thing with lower amounts of capacitance. Think Carver PM 2.0t as one example, and a good one as far as I can see.
My main concern is the amount of hash you will generate as the conduction angle of your rectifiers. I think that extremely high levels of filter capacitance are a fad and not that effective. The problems this brings exceeds the benefits. Turn on surge, high frequency noise, excessive weight and size. Now the chassis must be built even heavier to support the extra bulk until it's next to impossible to move and very difficult to set up in a home.
I can't see too much advantage to regulating the supplies for the output stage. Your problem is in the voltage amp stages, not in the output stages. Just my thoughts on the matter.
-Chris
The better power amps I've seen use a different supply that is regulated to run the voltage amp stage. Therefore, some sag in the supplies for the current end will not cause any trouble. That is until you run out of volts and the gain drops in your output stage.
It comes down to accepting that your supply will always have noise riding on it. It will always have some resistance and therefore sag no matter how much capacitance you jam in there. If you want to attack that issue, why not simply build a high frequency regulated switching supply to generate your rail voltages? Then you can achieve the same thing with lower amounts of capacitance. Think Carver PM 2.0t as one example, and a good one as far as I can see.
My main concern is the amount of hash you will generate as the conduction angle of your rectifiers. I think that extremely high levels of filter capacitance are a fad and not that effective. The problems this brings exceeds the benefits. Turn on surge, high frequency noise, excessive weight and size. Now the chassis must be built even heavier to support the extra bulk until it's next to impossible to move and very difficult to set up in a home.
I can't see too much advantage to regulating the supplies for the output stage. Your problem is in the voltage amp stages, not in the output stages. Just my thoughts on the matter.
-Chris
- Status
- Not open for further replies.