Flyback with 5mw no-load Power Consumption
Power supply with zero current in no load power consumption?
Is it possible?
Just imagine how many small wall power supplies stay in wall outlets all the time. Laptops, coffee machines, monitors, and, of course, smart phones chargers being connected to AC power for years use a lot of energy.
Some smart smartphone manufacturers decided to rate power supplies according to their no-power consumption level. The best power supply they say Is the one using less than 5 milliwatt in no-power condition. That says, no smartphone connected. And such charger would be rated zero current one.
At maximum AC voltage input 240vac, 5milliwatt would mean less than 21 microamperes!
Is it really possible?
It is definitely not with an old type linear 50hz transformer based power supply. Do you remember them? Heavy, bulky and very inefficient. Linear type power supplies nowadays have found a small niche in scientific laboratories to meet demands of particularly “clean” power supply.
Switched Mode Power Supplies which are used everywhere now are, certainly, produce some electromagnetic noise, EMI, but, as you probably already know, no ideal exist. We live in the world of compromise. So, we just contain EMI noise to some acceptable level.
At one time I was charged with task to design a power supply with the following characteristics:
Power out max – 20watt
Output voltage: 13volt dc
Input voltage – 85 to 26vac rms
Efficiency: 80% plus
Nothing special, right?
There you go:
No load power consumption: 5milliwat or less.
And, as usual: cheap and fast!
I have already designed several small power supplies before and knew that the only topology satisfying all requirements, except no load consumption, is a Flyback.
Flyback topology is one of the oldest types of switch mode power supplies. Old folks should remember them being used in old vacuum tube tv sets. These TVs required voltages up to 20kv and flyback did the job.
Below id schematic of a flyback converter:
1
Flyback
It is simple: there is a switch and a transformer. Transformer in flyback is actually an inductor split in 2 winding to provide isolation and to get any voltage output you need by changing turn ration between two windings. When the switch is on, energy from external source, usually rectified ac voltage, flows in the inductor/ transformer. Diode connected to secondary winding is and energy has no other way to go as to accumulate in the inductor/ transformer. When switch is off, accumulated energy flies back through the secondary winding, voltage is changing direction forcing diode to conduct and goes to load. All this happens with pretty high frequency usually limited by 150khz due to EMI standards limits. Output capacitor removes high frequency noise and ripple.
And waveforms look like that:
Inductor is getting fully discharged here, so-called discontinuous conduction mode. After inductor is discharged, charged output capacitor keeps supplying energy to the load. During this time inductor/ transformer is disconnected from its power source (switch is off) and from the load (the output capacitor voltage forbids diode to conduct). The inductor and capacitance of the primary circuit resonate, till the switch turns on again. New cycle starts. Output voltage regulates by sending output voltage and providing feedback to the switch controller. The controller changes duration of time during which the switch is closed, duty cycle.
2
Quasi-Resonant Flyback
The conventional flyback described above is very well known topology, used exclusively for small power supplies. It is simple, chip and reliable.
It has 2 problems: it is inefficient and it is noisy. The switch namely can turn on any time when voltage on the switch is still high. This causes high EMI noise and high switching losses. Now the old flyback topology stopped to satisfy increasing efficiency demands.
So some smart engineers came up with modification of this old topology: quasi-resonant flyback.
What it does? It senses voltage on the switch (drain of the mosfet) and switches it on when the voltage is at lowest point during resonating period. It is called value switching. It employs variable frequency. As power output is high it is always switching on the first valley point of the resonating wave. The higher output power is needed the higher the switching frequency. By lighter load it starts switching on second or third low valley point. You can see it in the end of the waveform above.
3
Control Mode and Primary Side Regulation
There is another thing though: how do you control the output voltage so that it stays the same?
One has to provide feedback: information about output voltage to switch controller so that switch stays on/ supplies more energy when output voltage goes down and other way around.
Usually it is done with help of a voltage reference IC 431 and an optocoupler:
431 works here as a comparator with reference voltage inside. It compares output voltage (5v in the drawing above) to its reference voltage and regulates current through optocoupler LED so that it says equal. Optocoupler provides galvanic isolation between primary and secondary circuits.
Altogether it looks like that:
This design works very well.
It is not good for our particular design. Why? Because of no load power consumption requirement! Optocoupler namely is nonlinear device and requires current of several milliampers to function properly. Lets say it is 5milliamper. Then we loose 5 x 13v (our output voltage) = 65 milliwat already! I knew people trying to lower the current through optocoupler. There are 431 ICs working by current of 50 microampers. But because of optocoupler instability they were forced to pull up the current in their designs. Even by 1 ma current we are going to loose 13 milliwatt.
There is a solution for this problem also: primary side regulation.
As you can see on the functional schematic below, the primary side regulated flyback has transformer with third, auxilliary winding. Usually it has 2 functions: it provides power to the controller and it also provides output voltage feedback information. The controller measures information during discharge time, when power is supplied to the load.
4
5mw no-load consumption Flyback Schematic
After some search I found an interesting flyback controller: UCC28730 from TI.
It has all functioned I mentioned before: it is quasi resonant flybac, it has primary side regulation, and, in addition, it has a remarcable quality: by no load condition it falls asleep! The controler switching frequency falls as low as 32 HZ and its supply current in wait mode is only 75uA.
Failling asleep means the power supply has to wake up when load is connected. And it has to do it quickly to be able to support the output voltage on the same level (load transient response).
TI came up with smart solution: They placed a small wake-up IC on output of the secondary winding. When load current increases the wake-up IC issues a wake up pulse which is sensed by feedback controller through auxilliary winding.
This is the schematic:
