Hi Andre,
I was thinking along the same lines, problem is you need to charge the cap during the dead time or negative cycle. If one could use both cycles then of course a lot more energy can be transferred. E=1/2CV^2
One can maybe transfer positive half cycle to one cap, negative to the other and alternately to the output cap and twice the energy.
Mihai,
What about using super caps? Also I thing the control circuitry could easily be achieved with a little micro.
Nico
I was thinking along the same lines, problem is you need to charge the cap during the dead time or negative cycle. If one could use both cycles then of course a lot more energy can be transferred. E=1/2CV^2
One can maybe transfer positive half cycle to one cap, negative to the other and alternately to the output cap and twice the energy.
Mihai,
What about using super caps? Also I thing the control circuitry could easily be achieved with a little micro.
Nico
This looks like an interesting starting point 
it avoids the problem of selecting the correct point to switch by charging the capacitor on one half cycle, then transferring the charge to the main circuit on the second half.
http://www.diyaudio.com/forums/showthread.php?postid=1675373#post1675373
it avoids the problem of selecting the correct point to switch by charging the capacitor on one half cycle, then transferring the charge to the main circuit on the second half.
http://www.diyaudio.com/forums/showthread.php?postid=1675373#post1675373
Symon said:This looks like an interesting starting point
it avoids the problem of selecting the correct point to switch by charging the capacitor on one half cycle, then transferring the charge to the main circuit on the second half.
http://www.diyaudio.com/forums/showthread.php?postid=1675373#post1675373
Yes it is a good starting point. A lot of the work has been done.
Nico
Bidirectional switches are required because the input side must be able to float at least a few volts in both directions with respect to the output side. Switch capacitance is critical because the "never connected" principle is a fallacy. Switch capacitance is always connecting the input side to the output side at high frequencies. A mains transformer with low input-output capacitance could just do as good or better.
Remember that the point of this circuit is to allow an AC potential (hypothetical noise) between input terminals and output terminals. A combination of common-mode and differential-mode filtering can do the same in a much simpler way.
Nico:
I'm currently developing my first class D amplifier for the professional audio market (several thousand watts, efficiencies approaching 95%, small form factors...) I think there are other ways to innovate. Linear power supplies and amplifiers are already somewhat outdated.
Remember that the point of this circuit is to allow an AC potential (hypothetical noise) between input terminals and output terminals. A combination of common-mode and differential-mode filtering can do the same in a much simpler way.
Nico:
I'm currently developing my first class D amplifier for the professional audio market (several thousand watts, efficiencies approaching 95%, small form factors...) I think there are other ways to innovate. Linear power supplies and amplifiers are already somewhat outdated.
You need 4 switches for this to work properly.
The first 2 switches close and connect the cap across a rectified and smoothed DC source, while the 2nd pair of switches are open. During this phase, the flying capacitor is charged.
You then open the first two switches and close the 2nd pair, connecting the flying cap to a storage cap which supplies the load.
This approach completely isolates the input source voltage from the load.
In the arrangement described above, both the supply rail and the ground rail are switched to provide isolation.
The first 2 switches close and connect the cap across a rectified and smoothed DC source, while the 2nd pair of switches are open. During this phase, the flying capacitor is charged.
You then open the first two switches and close the 2nd pair, connecting the flying cap to a storage cap which supplies the load.
This approach completely isolates the input source voltage from the load.
In the arrangement described above, both the supply rail and the ground rail are switched to provide isolation.
You need 4 switches for this to work properly.
The first 2 switches close and connect the cap across a rectified and smoothed DC source, while the 2nd pair of switches are open. During this phase, the flying capacitor is charged.
You then open the first two switches and close the 2nd pair, connecting the flying cap to a storage cap which supplies the load.
This approach completely isolates the input source voltage from the load.
In the arrangement described above, both the supply rail and the ground rail are switched to provide isolation.
The first 2 switches close and connect the cap across a rectified and smoothed DC source, while the 2nd pair of switches are open. During this phase, the flying capacitor is charged.
You then open the first two switches and close the 2nd pair, connecting the flying cap to a storage cap which supplies the load.
This approach completely isolates the input source voltage from the load.
In the arrangement described above, both the supply rail and the ground rail are switched to provide isolation.
Exactly Charles : " but you have to be aware that the only problems you may "filter" out that way is common mode noise problems. "
This is the most important feature about this kind of schematics !
Maybe explain Eva how do you reach exact the opposite conclusion ?
I think you misjudged the issue here .
Nothing personal , but always everyone must decide what or which is the most important source of noise he want to address about a psu : is it the common mode noise or semiconductor noise from psu or the rf coming from ac lines ?
All ?
This is the most important feature about this kind of schematics !
Maybe explain Eva how do you reach exact the opposite conclusion ?
I think you misjudged the issue here .
Nothing personal , but always everyone must decide what or which is the most important source of noise he want to address about a psu : is it the common mode noise or semiconductor noise from psu or the rf coming from ac lines ?
All ?
Nico Ras said:
Nico:
Unfortunately switching will generate its own noise and power losses, and to address this the switching action ideally should incorporate some type of "soft switching" technique.
However, you don't need to change the switching devices into linear mode to accomplish this. You can modify the configuration of the primary energy storage device from a simple capacitor into a partially resonant network, and this will take care of the major problems.
regards and hth, jonathan carr
Here is something to start playing with.
Firstly in the circuit from above thread there is a timing overlap when charging the caps thus you will still get AC noises coming through. I used two op-amps with thresholds set so that the time to charge the next cap is slightly less than one half cycle, thus charging of the output cap is clean.
I used power transistors to switch and the circuit is good for amps, not milli amps, however the op-amp is driven of the input circuit and thus transformer voltage must be selected not to kill the op-amp, else some plans need be put into place.
The final stage consists of a capacitance multiplier to get rid of all the ripple and it is followed by my favorite virtual earth which splits the supply, in this case into +12 and -12 rails.
I have not built anything yet but will do this when I get back to work and have PCB facilities available and can have transformers wound etc.
Anyway this is my contribution so far, you are welcome to comment, make changes or whatever. This is only a concept and could well work very nicely, it is based on the previous designers thread (I think it was John Carr).
Kind regards
Nico
Firstly in the circuit from above thread there is a timing overlap when charging the caps thus you will still get AC noises coming through. I used two op-amps with thresholds set so that the time to charge the next cap is slightly less than one half cycle, thus charging of the output cap is clean.
I used power transistors to switch and the circuit is good for amps, not milli amps, however the op-amp is driven of the input circuit and thus transformer voltage must be selected not to kill the op-amp, else some plans need be put into place.
The final stage consists of a capacitance multiplier to get rid of all the ripple and it is followed by my favorite virtual earth which splits the supply, in this case into +12 and -12 rails.
I have not built anything yet but will do this when I get back to work and have PCB facilities available and can have transformers wound etc.
Anyway this is my contribution so far, you are welcome to comment, make changes or whatever. This is only a concept and could well work very nicely, it is based on the previous designers thread (I think it was John Carr).
Kind regards
Nico
BTW this is not my design and I am not particularly interested in it for a practical application. I was only curious about the idea and I can see merit in it.
It would probably have the same effect as running from an SMPS. I currently run my listening amps off +-50V 20 Amp @ 100 kHz SMPS and I will never go back.
It would probably have the same effect as running from an SMPS. I currently run my listening amps off +-50V 20 Amp @ 100 kHz SMPS and I will never go back.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.