Design Article
Comment
nicolas.mokhoff
Thanks, Robert. Links in story have been corrected.
Robert K
Power Tip 52: Making over the wall wart
Robert Kollman, Texas Instruments
9/28/2012 4:27 PM EDT
With worldwide sales of cell phones closing in on two billion per year, the cell phone charger’s size, cost and efficiency is under scrutiny. For example, Amazon and Apple have established new benchmarks of small size and aesthetics, while reducing circuit losses and cutting the cost of low-power chargers. This has been accomplished with the use of an advanced topology and clever control methods.
[Get a 10% discount on ARM TechCon 2012 conference passes by using promo code EDIT. Click here to learn about the show and register.]
The most popular topology for a charger at the 5-10 watt power level is the discontinuous flyback. However, it has evolved into the quasi-resonant flyback, which reduces some switching losses. In the traditional discontinuous flyback, the switching frequency is fixed and the control IC simply sets the peak current in the power transformer. This represents a fixed amount of energy that is delivered to the load during a switching cycle. The power switch’s drain waveform is shown in Figure 1. During the charge interval (portion of switching cycle when waveform is zero volts), energy is stored in the primary inductance.
When the power switch is turned off, energy flows into the secondary where it is stored in the output capacitor and delivered to the load. Once the power transformer is demagnetized, the FET drain voltage collapses and rings around the input voltage. In the traditional approach, the FET is turned on at the next switching interval, regardless of the FET drain voltage. It can be at a minimum, maximum or somewhere in between. The losses associated with switching this voltage can be appreciable, often resulting in a two to three percent loss of efficiency. Quasi-resonant flybacks minimize the switching loss by only turning the FET on when the drain voltage is at a valley.
Click on image to enlarge.
Recent control methods do more than just valley switching. Figure 2 shows how two parameters (switching frequency and peak primary current) are varied to control the output voltage. At full load, the power supply is operated at maximum peak current and maximum frequency. As the load is reduced, the switching frequency is reduced. Since both output power and switching losses are directly related to the power supply’s frequency, this results in an almost flat efficiency in this mode of operation. Note that valley switching is still occurring, so the power supply switching frequency is not fixed.
The turn on of the power FET hops from one valley to another, with an average switching frequency as shown in Figure 1. Audible noise considerations limit how low the switching frequency can be as switching the power supply may induce audible noise in magnetics and ceramic capacitors. Many times, you may not want to allow the power supply switching frequency to drop below 10-20 kHz and an alternate control scheme is employed. In this case, once the minimum allowable operating frequency is reached, the peak current in the primary FET is controlled to regulate the output voltage at low-power levels.
Click on image to enlarge.
Figure 3 presents a typical power supply schematic of a universal input, 5 watt output charger. The schematic is very simple; it does not require a reference or optocoupler to regulate the output voltage. It uses the reflected output voltage on the primary bias winding for feedback. Referring to Figure 1, which shows the FET drain voltage, this waveform is an analog of both the bias and output voltage. When the drain voltage flies up, the drain voltage is related to the output voltage plus diode and resistive drops in the secondary. The drain voltage decreases linearly as the reflected inductance is discharged through the output diode. At the end of the diode conduction, this voltage and the voltage on the bias winding is a reflection of the output voltage plus a diode junction voltage. A feedback loop is closed around this reflected voltage and gives a reasonable regulation tolerance (three to five percent).
There is an additional challenge when improving the voltage regulation of these types of power supplies. The device being charged is at the end of a cable, which can have significant voltage drop at full load. In this particular implementation, the controller estimates the output power from the power switch peak current, allowing the output voltage to be adjusted to compensate for the voltage drop across the cable.
Click on image to enlarge.
Figure 4 shows the physical embodiment of the power supply. High-frequency switching and advanced control methods provide significant improvements over previous wall adapter designs. Input voltage capability is increased from a single voltage to universal input. The no-load dissipation is reduced from the 1 watt range to less than 30 mW. Full-load efficiency is improved from the 50 percent range to about 80 percent with diode rectification, and 85+ percent with synchronous rectification. Finally, size and aesthetics have been significantly improved.
Click on image to enlarge.
Advanced circuit techniques have remade the lowly cell phone charger from a clunky wall wart that consumes significant amounts of power to an innocuous device that is not much larger than a wall plug. The power savings from this improvement is significant. With two billion new cell phones being put into service yearly, the savings is measured in tens of power plants that will not be needed worldwide. For more detail on some recent charger designs, check out these three sites:
www.ti.com/pmp4335-ca
www.ti.com/pmp7389-ca
www.ti.com/pmp8286-ca.
For more information about this and other power solutions, visit:
www.ti.com/power-ca
Robert Kollman is a Senior Applications Manager and Distinguished Member of Technical Staff at Texas Instruments with more than 30 years of experience in the power electronics business offers another in a series of Power Tips.Please join us next month when we will discuss how to simply simulate a power supply control loop.
Past entries in the PowerTip Series are available here.
[Get a 10% discount on ARM TechCon 2012 conference passes by using promo code EDIT. Click here to learn about the show and register.]
The most popular topology for a charger at the 5-10 watt power level is the discontinuous flyback. However, it has evolved into the quasi-resonant flyback, which reduces some switching losses. In the traditional discontinuous flyback, the switching frequency is fixed and the control IC simply sets the peak current in the power transformer. This represents a fixed amount of energy that is delivered to the load during a switching cycle. The power switch’s drain waveform is shown in Figure 1. During the charge interval (portion of switching cycle when waveform is zero volts), energy is stored in the primary inductance.
When the power switch is turned off, energy flows into the secondary where it is stored in the output capacitor and delivered to the load. Once the power transformer is demagnetized, the FET drain voltage collapses and rings around the input voltage. In the traditional approach, the FET is turned on at the next switching interval, regardless of the FET drain voltage. It can be at a minimum, maximum or somewhere in between. The losses associated with switching this voltage can be appreciable, often resulting in a two to three percent loss of efficiency. Quasi-resonant flybacks minimize the switching loss by only turning the FET on when the drain voltage is at a valley.
Click on image to enlarge.
Figure 1: The quasi-resonant flyback turns the FET on at its minimum drain voltage to reduce CV2 losses.
Recent control methods do more than just valley switching. Figure 2 shows how two parameters (switching frequency and peak primary current) are varied to control the output voltage. At full load, the power supply is operated at maximum peak current and maximum frequency. As the load is reduced, the switching frequency is reduced. Since both output power and switching losses are directly related to the power supply’s frequency, this results in an almost flat efficiency in this mode of operation. Note that valley switching is still occurring, so the power supply switching frequency is not fixed.
The turn on of the power FET hops from one valley to another, with an average switching frequency as shown in Figure 1. Audible noise considerations limit how low the switching frequency can be as switching the power supply may induce audible noise in magnetics and ceramic capacitors. Many times, you may not want to allow the power supply switching frequency to drop below 10-20 kHz and an alternate control scheme is employed. In this case, once the minimum allowable operating frequency is reached, the peak current in the primary FET is controlled to regulate the output voltage at low-power levels.
Click on image to enlarge.
Figure 2: Primary peak current control and frequency modulation enhance efficiency over load.
Figure 3 presents a typical power supply schematic of a universal input, 5 watt output charger. The schematic is very simple; it does not require a reference or optocoupler to regulate the output voltage. It uses the reflected output voltage on the primary bias winding for feedback. Referring to Figure 1, which shows the FET drain voltage, this waveform is an analog of both the bias and output voltage. When the drain voltage flies up, the drain voltage is related to the output voltage plus diode and resistive drops in the secondary. The drain voltage decreases linearly as the reflected inductance is discharged through the output diode. At the end of the diode conduction, this voltage and the voltage on the bias winding is a reflection of the output voltage plus a diode junction voltage. A feedback loop is closed around this reflected voltage and gives a reasonable regulation tolerance (three to five percent).
There is an additional challenge when improving the voltage regulation of these types of power supplies. The device being charged is at the end of a cable, which can have significant voltage drop at full load. In this particular implementation, the controller estimates the output power from the power switch peak current, allowing the output voltage to be adjusted to compensate for the voltage drop across the cable.
Click on image to enlarge.
Figure 3: Advanced control ICs eliminate the optocoupler with primary voltage sensing.
Figure 4 shows the physical embodiment of the power supply. High-frequency switching and advanced control methods provide significant improvements over previous wall adapter designs. Input voltage capability is increased from a single voltage to universal input. The no-load dissipation is reduced from the 1 watt range to less than 30 mW. Full-load efficiency is improved from the 50 percent range to about 80 percent with diode rectification, and 85+ percent with synchronous rectification. Finally, size and aesthetics have been significantly improved.
Click on image to enlarge.
Figure 4: High-frequency switching and advanced control techniques minimize size and loss in this offline cell phone charger.
Advanced circuit techniques have remade the lowly cell phone charger from a clunky wall wart that consumes significant amounts of power to an innocuous device that is not much larger than a wall plug. The power savings from this improvement is significant. With two billion new cell phones being put into service yearly, the savings is measured in tens of power plants that will not be needed worldwide. For more detail on some recent charger designs, check out these three sites:
www.ti.com/pmp4335-ca
www.ti.com/pmp7389-ca
www.ti.com/pmp8286-ca.
For more information about this and other power solutions, visit:
www.ti.com/power-ca
Robert Kollman is a Senior Applications Manager and Distinguished Member of Technical Staff at Texas Instruments with more than 30 years of experience in the power electronics business offers another in a series of Power Tips.Please join us next month when we will discuss how to simply simulate a power supply control loop.
Past entries in the PowerTip Series are available here.
|
|
|
|
|
Navigate to related information
GREAT-Terry
10/1/2012 7:08 AM EDT
It is a good idea to use "quasi-resonant flyback" in low power wallwart design. It circuit is so simple! How much can be further reduced?
Sign in to Reply
Hughston
10/4/2012 11:27 AM EDT
This doesn't look much more expensive than a linear wall wart. A linear would have the input fuse, transformer, rectifier on the secondary, large filter cap and a load resistor to drain the cap charge.
The rectifier is not always in the wall wart; it can be in the charged product. A more efficient way to use a linear wall wart is to give it a higher source impedance and put a switch in the product being charged and pulse charge with the input.
Sign in to Reply
WKetel
10/3/2012 8:37 PM EDT
The one more easy way to cut losses, at this point, is to add a switch so that the AC input can be switched off. That should have been mandated 40 years ago when the wall warts were starting to appear. Of course it would have been a serious hardship and it would have raised the cost of each unit by about 15 cents, but consider how much wasted power it would have saved. At some point the folks who make all of the laws need to tell the whiners and complainers to "shut up, because we don't care about the profit that you lose." If it is OK to mandate all kinds of efficiency increases in cars, which requires all manner of new innovations, it should certainly have been OK to mandate on/off switches, which were nothing new, and very well understood.
Sign in to Reply
Robert K
10/5/2012 9:48 AM EDT
Try these links, the last two in the article are wrong.
http://www.ti.com/tool/pmp7389
http://www.ti.com/tool/pmp8286
Sign in to Reply
nicolas.mokhoff
10/24/2012 9:17 AM EDT
Thanks, Robert. Links in story have been corrected.
Sign in to Reply