We take AC and transform to DC .
We add capacitors to smooth out the high points (ripple).
We add a wazoo of cap’s to further reduce the ripple (?)
My terminology is probably wrong but I think it’s called filtering?
My dumb question is: Can the ripple be eliminated? So called perfect DC.
Can the filtered DC be filtered again and again,,,, etc. to become ,,,,,,,,,,,a battery(perfect DC,( zero ripple?)).
Is straight line DC the ideal world?
To make this happen, will it be the size of a suit case before the amp circuit.
Is this a dumb question?
Thanks, Les
We add capacitors to smooth out the high points (ripple).
We add a wazoo of cap’s to further reduce the ripple (?)
My terminology is probably wrong but I think it’s called filtering?
My dumb question is: Can the ripple be eliminated? So called perfect DC.
Can the filtered DC be filtered again and again,,,, etc. to become ,,,,,,,,,,,a battery(perfect DC,( zero ripple?)).
Is straight line DC the ideal world?
To make this happen, will it be the size of a suit case before the amp circuit.
Is this a dumb question?
Thanks, Les
Yes, you can have multiple stages of filtering for increasingly smooth DC. I doubt you ever get quite ripple free, but you can get ripple low enough it isn't really an issue. The simplest is to keep adding additional capacitance, infinite capacitance would get you zero ripple in theory. You can have more than simple cap filters by combining them with inductive coils. You can have series of cap-coil combos. You can use voltage regulation to actively filter out ripple. This without mentioning switching supplies etc.
Sounds like you need to read a good book on basic power supplies. Wish I had a current one in mind. But someone else will likely suggest one here.
Sounds like you need to read a good book on basic power supplies. Wish I had a current one in mind. But someone else will likely suggest one here.
Perfect DC. Well nothings perfect, but you can get close.
You didn't say what the output of this perfect power supply would be so its a little tough to give you an example.
What I do rather than spend a bucket load of cash on the cap of the week is to build regulated power supplies.
Lets say for example you want to build a 24VDC supply with 100ma current. For this I would go for a simple LM317 series regulator and follow the manufacturers suggestions. You could use a 24VAC output transformer @ 1 amp, a bridge rectifer and a 1000uf 50vdc cap. That would get you about 35 volts of dc with some ripple on it, maybe a volt or so. Now what the regulator does is to take that dirty 35 volts and cut off the dirty part and just pass one the clean 24VDC under the ripple. Another advantage of using a regulator is that the output impedance of the regulator is much lower than just a cap filtered supply and your amp circuit will like that a lot better. A lower impedance power supply gives the amp a much more stable source to operate from.
Regulater power supplies can be made for just about anything including high voltage, high current or both it just gets a little more complicated.
Any time I have compaired and measured an amp with first a C or CLC power supply to the same amp with a regulated supply the regulated supply always wins. The amp is more stable, lower hum, better distortion, less noise and with power amps more power.
I hope that helps.
You didn't say what the output of this perfect power supply would be so its a little tough to give you an example.
What I do rather than spend a bucket load of cash on the cap of the week is to build regulated power supplies.
Lets say for example you want to build a 24VDC supply with 100ma current. For this I would go for a simple LM317 series regulator and follow the manufacturers suggestions. You could use a 24VAC output transformer @ 1 amp, a bridge rectifer and a 1000uf 50vdc cap. That would get you about 35 volts of dc with some ripple on it, maybe a volt or so. Now what the regulator does is to take that dirty 35 volts and cut off the dirty part and just pass one the clean 24VDC under the ripple. Another advantage of using a regulator is that the output impedance of the regulator is much lower than just a cap filtered supply and your amp circuit will like that a lot better. A lower impedance power supply gives the amp a much more stable source to operate from.
Regulater power supplies can be made for just about anything including high voltage, high current or both it just gets a little more complicated.
Any time I have compaired and measured an amp with first a C or CLC power supply to the same amp with a regulated supply the regulated supply always wins. The amp is more stable, lower hum, better distortion, less noise and with power amps more power.
I hope that helps.
The regulator usually is after the filter section except with switching regulators which usually have a filter before and after the regulator.
There is no advantage to having a filter after the regulator for hum reduction but one can supply transient demand when a slow or underpowered regulator cannot supply the transient load.
There is no advantage to having a filter after the regulator for hum reduction but one can supply transient demand when a slow or underpowered regulator cannot supply the transient load.
If one wants to build a regulated PS based on a linear supply for power amps which needs around 40-50VDC and around 10A peak, what are the possibilities? Any currently available boards for the DIY community and any in particular to recommend ? -I´ve seen that SKA has introduced such a board: GB150D Power Amplifier
Any other possibilities, or experiences on this subject out there?
cheers
Any other possibilities, or experiences on this subject out there?
cheers
Certainly good questions.
AC to rectified AC = pulsed DC.
Filter the pulsed DC to attenuate the High Frequency (HF) components in the supply and you get closer to DC. But it is not DC. It has some HF components remaining.
Each additional filter will attenuate the HF even more, but can never remove it in entirety.
Now apply a DC (constant) load on that filtered supply. One will find that the output current varies very slightly due to the non DC components in the filtered output. These can simply be noise and/or ripple. They will always be there.
Taking that DC load and changing it to a different DC load. What does the supply current look like? Is it a an instant step change from one DC current level to the new DC current level. No it won't be an instant step change. There will be some variation over time. These variations are HF components getting into the DC output due to load changes. These cannot be eliminated.
Both these HF aberrations (due to input variations and due to load variations) can be reduced. The trick/skill of the designer is to reduce both HF effects to levels that do not impinge on the overall performance of the load.
AC to rectified AC = pulsed DC.
Filter the pulsed DC to attenuate the High Frequency (HF) components in the supply and you get closer to DC. But it is not DC. It has some HF components remaining.
Each additional filter will attenuate the HF even more, but can never remove it in entirety.
Now apply a DC (constant) load on that filtered supply. One will find that the output current varies very slightly due to the non DC components in the filtered output. These can simply be noise and/or ripple. They will always be there.
Taking that DC load and changing it to a different DC load. What does the supply current look like? Is it a an instant step change from one DC current level to the new DC current level. No it won't be an instant step change. There will be some variation over time. These variations are HF components getting into the DC output due to load changes. These cannot be eliminated.
Both these HF aberrations (due to input variations and due to load variations) can be reduced. The trick/skill of the designer is to reduce both HF effects to levels that do not impinge on the overall performance of the load.
Nothing is perfect and there is no perfect DC supply. Even a battery has internal resistance, which means that its' voltage will vary with changing current load.
Ripple and HF noise can be filtered to very low values, but not eliminated completely.
The degree of filtering applied is a question of cost, size and the requirements of the circuit. That is, all else being equal (like PSRR), an RIAA pre-amp (phono-stage) will call for smoother DC than line level pre-amp.
Excellent practical DC supply can be attained by a CLC filter, followed by capacitance multiplier, followed by a shunt regulator. Better still could be different shunt regulator for each channel. This may give better DC than most batteries. Anything more than that will probably have no practical benefit.
Ripple and HF noise can be filtered to very low values, but not eliminated completely.
The degree of filtering applied is a question of cost, size and the requirements of the circuit. That is, all else being equal (like PSRR), an RIAA pre-amp (phono-stage) will call for smoother DC than line level pre-amp.
Excellent practical DC supply can be attained by a CLC filter, followed by capacitance multiplier, followed by a shunt regulator. Better still could be different shunt regulator for each channel. This may give better DC than most batteries. Anything more than that will probably have no practical benefit.
The LM317 can also regulate high voltages. You just have to keep the input-output differential within limits (I think it's 40V).
"Straight-line DC" is indeed the ideal, but like the other "ideal" components in electronics they aren't the reality. There is always compromises of one type or another. As someone said previous, the power supply is designed to reliably operate the circuit. That would include a little bit of headroom. Beyond this the gains made aren't worth the extra parts, size, cost, etc.
The simple capacitor is the most common filter, due to cost I'm sure, but resistors and/or inductors can be added for increased rejection at the ripple frequency. That frequency is 60Hz or 120Hz in North America and 50Hz or 100Hz in the rest of the world. That doubling depends on the type of rectification used, full-wave or half-wave.
Relatively speaking, there's not many books devoted solely to power supplies, Any decent general electronic book should cover it, and there ought to be some worthwhile application notes with more details. TI and National are my first choices for that sort of stuff.
"Straight-line DC" is indeed the ideal, but like the other "ideal" components in electronics they aren't the reality. There is always compromises of one type or another. As someone said previous, the power supply is designed to reliably operate the circuit. That would include a little bit of headroom. Beyond this the gains made aren't worth the extra parts, size, cost, etc.
The simple capacitor is the most common filter, due to cost I'm sure, but resistors and/or inductors can be added for increased rejection at the ripple frequency. That frequency is 60Hz or 120Hz in North America and 50Hz or 100Hz in the rest of the world. That doubling depends on the type of rectification used, full-wave or half-wave.
Relatively speaking, there's not many books devoted solely to power supplies, Any decent general electronic book should cover it, and there ought to be some worthwhile application notes with more details. TI and National are my first choices for that sort of stuff.
My two cents:
Large banks of capacitors, chokes, and even large transformers are inefficiently utilized for filtering the power supplies of low-level audio components. This is inefficient because we don't need all of the energy storage they provide simply to power our preamps and DACs. We use them because we're trying to filter a low frequency (usually 120Hz) signal within a low impedance circuit context, using passive components. Such passive components will be large in value to meet the filtered circuit impedance terms, but not because we need them to store energy. I know, most equipment manufactures have adopted such a supply filtering brute force approach. While there are legitimate filtering benefits of such an brute force passive approach, I also believe there is also a marketing perceptions game being played on the audio consumer as well. Most of us have probably played it before. You know how it goes, if we want to see where our money went to in that new phono stage we just purchased, all we has to do is ooh and ahh at all those rows of pretty capacitors, or feel the iron weight of the power choke or high-wattage power transformer.
Do all of these passive filtering components work? Yes, they do. Are they a rather crude, bulky, and costly solution? Yes, they are. Is there a more sophisticated, compact, inexpensive, and more effective solution? Yes, there is. Use of active power supply filtering would perform much better and without the suitcase full of large capacitors and inductors. Active current sources would replace chokes, and shunt regulators would replace large capacitors. I believe that Raleigh Audio has adopted this more sophisticated approach. The joules of energy stored in most high-end preamp power supplies, for example, far far exceed the actual energy needs of the amplifying circuit and load. Audio power amps, on the other hand, do require the actual energy storage provided by a large amount of capacitance and inductance. Nearly any other audio device does not, they only require the noise filtering.
Large banks of capacitors, chokes, and even large transformers are inefficiently utilized for filtering the power supplies of low-level audio components. This is inefficient because we don't need all of the energy storage they provide simply to power our preamps and DACs. We use them because we're trying to filter a low frequency (usually 120Hz) signal within a low impedance circuit context, using passive components. Such passive components will be large in value to meet the filtered circuit impedance terms, but not because we need them to store energy. I know, most equipment manufactures have adopted such a supply filtering brute force approach. While there are legitimate filtering benefits of such an brute force passive approach, I also believe there is also a marketing perceptions game being played on the audio consumer as well. Most of us have probably played it before. You know how it goes, if we want to see where our money went to in that new phono stage we just purchased, all we has to do is ooh and ahh at all those rows of pretty capacitors, or feel the iron weight of the power choke or high-wattage power transformer.
Do all of these passive filtering components work? Yes, they do. Are they a rather crude, bulky, and costly solution? Yes, they are. Is there a more sophisticated, compact, inexpensive, and more effective solution? Yes, there is. Use of active power supply filtering would perform much better and without the suitcase full of large capacitors and inductors. Active current sources would replace chokes, and shunt regulators would replace large capacitors. I believe that Raleigh Audio has adopted this more sophisticated approach. The joules of energy stored in most high-end preamp power supplies, for example, far far exceed the actual energy needs of the amplifying circuit and load. Audio power amps, on the other hand, do require the actual energy storage provided by a large amount of capacitance and inductance. Nearly any other audio device does not, they only require the noise filtering.
Last edited:
I just wanted to add...
This smoothing of DC with ripple is just the action of a low-pass filter, not unlike those found in amplifiers and loudspeakers. It's the application that's different.
Hopefully this helps some readers to gain a "bigger picture" of the science, principles, and objectives.
This smoothing of DC with ripple is just the action of a low-pass filter, not unlike those found in amplifiers and loudspeakers. It's the application that's different.
Hopefully this helps some readers to gain a "bigger picture" of the science, principles, and objectives.
I think some of you guys are often chasing something which is not required, well designed solid state amplifiers have 60 dB + supply rejection (essentially they are voltage controlled 4 quadrant regulated power supplies anyway) with that supply rejection and full load ripple less than 2% being attained without a great deal of effort then -35 db ripple and 60 db ripple rejection gives an output ripple at the onset of clipping of -95 dB. If anybody can hear that then I am sure some audio codec testers would love to hear from you.
What I am trying to say is understand the load and what you are trying to achieve. A power supply for a single ended amplifier without feedback has entirely different requirements for inaudible hum than a modern high gain transistor amplifier with feedback.
What I am trying to say is understand the load and what you are trying to achieve. A power supply for a single ended amplifier without feedback has entirely different requirements for inaudible hum than a modern high gain transistor amplifier with feedback.
That's a bit of an oversimplification. We are often not just dealing with 120Hz ripple. There can be line harmonics and other noises reaching up to the RF region. In addition, the power transformer secondary forms an LC tank circuit with the rectifier diode capacitance that can ring at ultrasonic frequencies, excited by the diode's commutation. All of this occurs after the A.C. Line input itself. The regulator circuit and the amplifier PSRR are left to deal linearly with all of this high-frequency noise and distortion. If they don't, audible intermodulation sidebands will result.
I have yet to observe such things, when I have put a spectrum analyser on the DC rails of the power supplies at work even the big phase controlled units manage to keep the fundamental below -40dB with harmonics rolling off into the noise floor around -100dB. The fundamental is by far the biggest peak on the power supplies I work with. If there is RF getting past the filters then the construction practices need to be examined closely.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.