Efficiency. It is very easy to obtain soft switching (low switching losses - and emi). Conduction losses are minimized by the nearly 100 percent duty cycle. Since there is no output inductor and no duty cycle modulation what is given up is regulation, but such a supply can be made to serve as a crude inrush current limiter for the boost stage - if most of the energy storage capacitance is on the output side.
Regards -- analog(spiceman)
So how do you start it up safely without using NTCs or similar measures? After all, you switch into a dead short when the output is discharged.analogspiceman said:
An inactive PFC still has the boost diode between the mains and the reservoir cap.phase_accurate said:
Hi,
I have just found a 200VA, 30V sec toroidal transformer. Quick measurement shows 16.5uH secondary leakage inductance and 233mOhm secondary resistance. Extrapolated values for dual 33V secondaries give 20uH secondary leakage inductance and 500mOhm secondary resistance. Withn the differential resistance of typical MUR860 rectifier somewhere around 100mOhm and typical 4700uF capacitor with 70mOhm ESR, you get attenuation of 18.8dB at 1kHz and 26.1db at 10kHz. So the coupling is not 100% at audio frequencies.
BTW, primary to secondary capacitance is 1.05nF.
Best regards,
Jaka Racman
I don't know how good the regulators want the PFC to be and what supression of mains harmonics is desired (just keep in mind that convetional amp PSUs are the worst culprits in this regard).
But it should also be possible to rectify and dynamically downconvert the input voltage to a voltage below say 100 Volts DC. Such a circuit could be made with soft start. It would of course only make use of a portion of every half-wave but would still be significantely better than a conventional PSU in terms of generation of mains harmonics.
Or the diode in an ordinary PFC circuit could be substituted by a mosfet (synchronous rectifier) so that the PFC can be disabled during startup and an additional (low - power) downconverter could be used for slow startup. Or maybe the synchronous rectifier could perform the latter as well in this case.
Regards
Charles
But it should also be possible to rectify and dynamically downconvert the input voltage to a voltage below say 100 Volts DC. Such a circuit could be made with soft start. It would of course only make use of a portion of every half-wave but would still be significantely better than a conventional PSU in terms of generation of mains harmonics.
Or the diode in an ordinary PFC circuit could be substituted by a mosfet (synchronous rectifier) so that the PFC can be disabled during startup and an additional (low - power) downconverter could be used for slow startup. Or maybe the synchronous rectifier could perform the latter as well in this case.
Regards
Charles
One possibility would be to greatly increase the operating frequency (while still keeping a small deadtime to allow soft switching) so that the leakage inductance served as a kind of non dissipative resistance. With the transformer designed for 375 watts average throughput, the leakage inductance might be sized to drop about 15 volts (referred to the 375 volt boost side) at full load. This amount of droop would indicate a non dissipative resistance equivalent to 15 ohms. Increasing the frequency 20 fold would up this equivalent resistance to 300 ohms. This would limit initial charging current to about the same as normal full load current. Of course, the amplifier would have to be inhibited until the output rails reached normal levels. Also, winding proximity losses would go way up with a frequency increase from 25kHz to 500kHz, but this would only occur briefly during the initial turn on surge. Operating frequency could be arranged to decrease in step with the ramping output voltage so that it would ramp linearly while charging. Sustained short circuits would have to be handled by a hiccup mode.
Regards -- analog(spiceman)
This is an interesting idea that is used in some sinewave dc-ac inverters. In principle the pfc stage could be made with the same control scheme, mosfets, inductors, etc. as the class d output stage. And the transformer could be about half the normal size since its rms current would be less than half that of a standard rectifer-capacitor design.
Regards -- analog(spiceman)
Some years back I made a simulation of basically the same idea as Jaka mentioned.
Although I "used" SD - modulation instead of PWM. The SD-modulator is the thingie in the box which could be replaced by a PWM modulator (though the control loop is of course also a part of the modulator, to be exact).
The simulation model also used soft-start (V2, R& and C4). The control loop however was never optimised, it was just a quick and simple (i.e. dirty ?) try.
D3 and C7 are an ordinary rectifier section fed with the down-scaled input voltage so one can compare the behaviour of the circuits. Both circuits are loaded by a resistor (R1, R7) drawing 10 Amperes continuous. What I never simulated is the dynamic load behaviour. But that would only make sense after the optimisation of the dynamic properties of the control loop.
We can see an inner loop controlling the current through the converter. The current itself is set by the outer voltage control loop and is made to be proportional to the input voltage by the use of a multiplier (thats where a CA3080 could come into play).
The pulsed voltage coming from a transformer/full-wave rectifier combination is simulated by the use of a sinusoidal voltage source, followed by the "absolute" function. The final DC voltage is about half the peak input voltage and therefore current will be drawn for a much larger part of time compared to an ordinary rectifier.
In the simulation results one can clearly see the differences in peak-current drawn, conduction angle, inrush current and also the much "gentler" ripple riding on top of the PFC's output voltage.
Regards
Charles
Although I "used" SD - modulation instead of PWM. The SD-modulator is the thingie in the box which could be replaced by a PWM modulator (though the control loop is of course also a part of the modulator, to be exact).
The simulation model also used soft-start (V2, R& and C4). The control loop however was never optimised, it was just a quick and simple (i.e. dirty ?) try.
D3 and C7 are an ordinary rectifier section fed with the down-scaled input voltage so one can compare the behaviour of the circuits. Both circuits are loaded by a resistor (R1, R7) drawing 10 Amperes continuous. What I never simulated is the dynamic load behaviour. But that would only make sense after the optimisation of the dynamic properties of the control loop.
We can see an inner loop controlling the current through the converter. The current itself is set by the outer voltage control loop and is made to be proportional to the input voltage by the use of a multiplier (thats where a CA3080 could come into play).
The pulsed voltage coming from a transformer/full-wave rectifier combination is simulated by the use of a sinusoidal voltage source, followed by the "absolute" function. The final DC voltage is about half the peak input voltage and therefore current will be drawn for a much larger part of time compared to an ordinary rectifier.
In the simulation results one can clearly see the differences in peak-current drawn, conduction angle, inrush current and also the much "gentler" ripple riding on top of the PFC's output voltage.
Regards
Charles
Here's an LTspice model of an offline PFC stage. The bits and pieces mimic whats in a typical PFC control IC
Hi,
My idea is a little different than Charles's one. I think it is basically what analog proposed, but used on the primary output voltage is +/- 400V which is a little high. I actually built a working model of this circuit. Separate small windings on the transformer can be used for curent reference and as power supply for gate drivers and control logic. Control was very simple. Outer loop consisted of voltage error amplifier driving CA3080 used as a multiplier multiplying error voltage and 50Hz sine. Resulting output was used as a reference for hysteretic comparator sensing current through L.
Best regards,
Jaka Racman
My idea is a little different than Charles's one. I think it is basically what analog proposed, but used on the primary output voltage is +/- 400V which is a little high. I actually built a working model of this circuit. Separate small windings on the transformer can be used for curent reference and as power supply for gate drivers and control logic. Control was very simple. Outer loop consisted of voltage error amplifier driving CA3080 used as a multiplier multiplying error voltage and 50Hz sine. Resulting output was used as a reference for hysteretic comparator sensing current through L.
Best regards,
Jaka Racman
Attachments
This is pie in the sky........ Something I drewed earlier based on primary side control using the UCC285XX, could be time to bend it.
http://focus.ti.com/docs/prod/folders/print/ucc28510.html
http://focus.ti.com/docs/prod/folders/print/ucc28510.html
Yes, this very close to what I had in mind. Coming from a SMPS background (as you and I both do, I think) it seems, at least to me, that a lot of audio people have an unwarranted fear of direct, off-line circuitry.
Many act as if it were dangerous voodoo black magic to be avoided at all costs.
This reason alone may give the edge to a transformer isolated pfc power supply (at least for the diy-er class d crowd). And, if as I suspect, it turns out that one could accomplish this with a little bit of extra control circuitry driving a second, completely unmodified class d module, then Bruno and Company might find this quite appealing as well. 
As you noted, one might be able to take the sinewave reference from an additional small winding on the transformer (although this might not be stiff enough for the control scheme I have in mind). This reference would then have its size adjusted via a 3080 driven by a voltage error amp as you noted. The error signal would be taken from differencing a reference against the voltage between the positive and negative rails (not just to ground). The resulting variable size sinewave would then drive the positive pin of the d-amp's differential input and the "output" of the d-amp (scaled, of course) would drive its negative input pin. Note that this retains the natural voltage source characteristic of the d-amp. Current would only be indirectly controlled as a result of the power transformer's impedance. Unless I'm mistaken (quite possible) this scheme, unlike with standard pfc current control, would both automatically eliminate the power supply pumping problem and eliminate the need for a voltage feedforward divider in the control. I will have to run a simulation of all this over the weekend.
To recap, the advantages would be:
- AC transformer about half normal size since only its thermal performance and not its stiffness would be of concern.
- No dangerous line voltages to confront the diy-er.
- Second unmodified d-amp module serves as regulated, power factor corrected power supply for output d-amp. 🙂
- Power supply pumping problem is eliminated.
- Bruno sells twice as many amp modules.
Apart from the fact that I don't sell amplifier modules (JP does) I'd like to note that the idea of trying to face the mains with a voltage source (with only the transformer's resistance to moderate the fight) is going to result in a complicated and hard to stabilise loop. Try to think of the problem as similar to connecting a power plant to the grid, making sure they all add power. They don't rely on the resistance of the grid to ease the control problem, take that from me.
A voltage source is best approached with a current load and vice versa (should sound familiar). This brings us closer (if not back) to standard pfc control schemes.
Personally I don't see offline switching as black art. I'm just afraid of loud bangs (=afraid of being scared).
Funnily enough, when something does go bang I don't get started at all.
A voltage source is best approached with a current load and vice versa (should sound familiar). This brings us closer (if not back) to standard pfc control schemes.
Personally I don't see offline switching as black art. I'm just afraid of loud bangs (=afraid of being scared).
Funnily enough, when something does go bang I don't get started at all.
Hi analog,
your idea is interesting. But like Bruno said current control seems a better idea to me. But convering a voltage amplifier with one pole roll off and well damped filter resonance (UcD) into a current source is no big deal. I would also like to point out that proposed schematic allows for bidirectional power flow so it eliminates pumping. But this is only for purists that might like to use UcD for motor drive.
One possible drawback of proposed control is multipier offset which could cause imbalance between positive and negative output voltage. AC coupling of multiplier output would be probably sufficient. I originally used proposed circuit in a three phase system which was loaded only between rails and it performed satisfactory.
Bruno, I look forward for your design, particularly to the design of the transformer. I really like some of your unorthodox ideas. It may look strange, but for me the best part of the UcD is low closed loop gain of the switching stage.
Regarding banging, you are still young so you will probably get used to it. I started my career with melting all the fuses in the distribution panel up to the main 3x250A ones. But I started with welding inverters and in those times state of the art was center tapped push pull running on 500V rectified mains with 2 x 6 BU209 TV deflection transistors running in parallel.
The most recent bang I had when I was called in a hurry to check out supply voltage on 800V 10000uF capacitor bank. I was blind for approximately 5s and deaf for 30s. The moral is: never check voltage with multimeter connected as ampmeter.
Best regards,
Jaka Racman
your idea is interesting. But like Bruno said current control seems a better idea to me. But convering a voltage amplifier with one pole roll off and well damped filter resonance (UcD) into a current source is no big deal. I would also like to point out that proposed schematic allows for bidirectional power flow so it eliminates pumping. But this is only for purists that might like to use UcD for motor drive.
One possible drawback of proposed control is multipier offset which could cause imbalance between positive and negative output voltage. AC coupling of multiplier output would be probably sufficient. I originally used proposed circuit in a three phase system which was loaded only between rails and it performed satisfactory.
Bruno, I look forward for your design, particularly to the design of the transformer. I really like some of your unorthodox ideas. It may look strange, but for me the best part of the UcD is low closed loop gain of the switching stage.
Regarding banging, you are still young so you will probably get used to it. I started my career with melting all the fuses in the distribution panel up to the main 3x250A ones. But I started with welding inverters and in those times state of the art was center tapped push pull running on 500V rectified mains with 2 x 6 BU209 TV deflection transistors running in parallel.
The most recent bang I had when I was called in a hurry to check out supply voltage on 800V 10000uF capacitor bank. I was blind for approximately 5s and deaf for 30s. The moral is: never check voltage with multimeter connected as ampmeter.
Best regards,
Jaka Racman
'zero ripple' isolated PFCs
Hi all,
The following paper describes Sepic and Cuk single stage isolated PFC converters where input current (_and_ output current in the case of the Cuk topology, at the price of some more components) feature near-zero ripple at the switching frequency (-> near-zero diff mode conducted EMI!!), with just one transistor and one magnetic component (integrated magnetics), very impressive!
The control scheme is very interesting too: their non linear ramp for PWM generation gives inherent PFC property to the converters, so no current control loop is required in spite of the continuous conduction mode.
"POWER FACTOR CORRECTORS BASED ON COUPLED-INDUCTOR SEPIC AND CUK CONVERTERS WITH NONLINEAR-CARRIER CONTROL".
http://ece-www.colorado.edu/~maksimov/sepic.pdf
"Abstract: This paper presents design of high performance power factor correctors based on coupled-inductor Sepic and Cuk converters operating in the continuous conduction mode. All inductive components are realized on the same magnetic core and designed so that levels of switching-frequency ripples and conducted EMI are greatly reduced. Input-current shaping and output voltage regulation are accomplished using the nonlinear-carrier control technique. A Spice-compatible large-signal averaged model is introduced to facilitate computer-aided design verification. The design and the models are validated by experimental results on two power factor correctors based on the coupled-inductor Sepic and Cuk converters."
Anyone can think of any drawbacks in these topologies? They look too good somehow, it must hide something 🙂
Hi all,
The following paper describes Sepic and Cuk single stage isolated PFC converters where input current (_and_ output current in the case of the Cuk topology, at the price of some more components) feature near-zero ripple at the switching frequency (-> near-zero diff mode conducted EMI!!), with just one transistor and one magnetic component (integrated magnetics), very impressive!
The control scheme is very interesting too: their non linear ramp for PWM generation gives inherent PFC property to the converters, so no current control loop is required in spite of the continuous conduction mode.
"POWER FACTOR CORRECTORS BASED ON COUPLED-INDUCTOR SEPIC AND CUK CONVERTERS WITH NONLINEAR-CARRIER CONTROL".
http://ece-www.colorado.edu/~maksimov/sepic.pdf
"Abstract: This paper presents design of high performance power factor correctors based on coupled-inductor Sepic and Cuk converters operating in the continuous conduction mode. All inductive components are realized on the same magnetic core and designed so that levels of switching-frequency ripples and conducted EMI are greatly reduced. Input-current shaping and output voltage regulation are accomplished using the nonlinear-carrier control technique. A Spice-compatible large-signal averaged model is introduced to facilitate computer-aided design verification. The design and the models are validated by experimental results on two power factor correctors based on the coupled-inductor Sepic and Cuk converters."
Anyone can think of any drawbacks in these topologies? They look too good somehow, it must hide something 🙂
Re: 'zero ripple' isolated PFCs
Regards -- analog(spiceman)
Ripple cancellation depends on large capacitors to work. It is a misnomer and would be better known as ripple reduction. By adding an extra winding or two in the right places almost any convertor topology can be made zero ripple. Cuk (rhymes with kook) converters are not unique in this regard and do not do better than any other four filter element topology. Please reread Jaka's post about power converters and the laws of physics.
Regards -- analog(spiceman)
- Status
- Not open for further replies.