Power Integrations 12 V 6 W Flyback reference design - Home Appliance and Industrial Power Reference Design Kit

Table of contents

RoadTest: Power Integrations Home Appliance and Industrial Power Reference Design Kit

Author: Jan Cumps

Creation date:

Evaluation Type: Power Supplies

Did you receive all parts the manufacturer stated would be included in the package?: True

What other parts do you consider comparable to this product?: a set of wall warts with similar input range and comparable output specifications

What were the biggest problems encountered?: It's a reference design. Focus is on showing a perfect design rather than on hackability. That makes it hard to experiment. But not impossible!

Detailed Review:

TL;DR: I score the Power Integrations (P.I.) 12 V 6 W power supply reference design Eight spoked asteriskEight spoked asteriskEight spoked asteriskEight spoked asteriskEight spoked asterisk.
The core IC is a capable and flexible controller with integrated MOSFET package. The PCB is a good showcase of a realistic design. The documentation is excellent.

In my road test report, I'll show my hands-on experience while I use the design within its specs.

Power Integrations 12 V 6 W reference design overview

The road test kit is a mains-to-12V isolated flyback design. It offers:

  • a circuit, built around a P.I. LinkSwitch IC. The circuit focus is on low standby consumption, energy availability at standby, good efficiency and low BOM count
  • A small size example layout. A single layer board with components mounted on both sides. This is a realistic real-world PCB that shows a very condensed lay-out.
  • A reference design report that digs into all aspects of this circuit. 

image: the PCB on the workbench.

The power supply is intended for devices that are mains connected at all times. Things like home appliances.
During stand-by, the supply should use minimal power. But still be able to run some functions (e.g. a display, remote control receiver, communication listener, ...).
Under full load conditions, it should have a decent efficiency.
These constraints are usually prescribed in consumer goods (and other) efficiency requirement standards.

Design characteristics:

  • 12 VDC output,
  • 500 mA max current
  • low stand-by consumption, 30 mW
  • 80% efficiency at full load
  • isolated
  • input 90 - 305 VAC
  • 47 - 64 Hz
  • input surge and over-voltage protection
  • output over-voltage, over-current and over-temperature 
  • zero-crossing detection, can be used for timing and synchronisation purposes.
  • size: 3.5 x 5 cm
  • EMI, standby, efficiency, safety compliant with current standards and directives.

P.I. IC LNK3306D

The switch and regulation IC is part of P.I.'s LinkSwitch-TNZ Family. They integrate the power MOSFET and all control logic to build switch mode converters. P.I. shows example circuits for these configurations

  • Buck (Application Note AN-98)
  • Buck-Boost (Application Note AN-98)
  • Flyback, isolated (Reference Design Report RDR-877) and non-isolated (Application Note AN-98)

image source: P.I. datasheet LNK33x2-7D. I coloured the safety related blocks

Oscillator specific functionality:

The IC provides soft start. During the initial cycles, the clock will run half speed, to limit the stress at startup. The system switches to 66 kHz when output level is reached, or after 256 cycles.
It's used to generate the timeouts and auto-restart for some fault conditions. And its main purpose: It generates the switch cycle start, and the maximum duty cycle signal. 

The clock has intentional jitter. A range between 2 kHz under and 2 kHz above the core frequency.

image source: P.I. datasheet LNK33x2-7D. I highlighted the jitter.

This jitter with a modulation frequency of 1 kHz, lowers the EMI emission peaks at clock frequency by spreading out switching artifacts across that 4 kHz spectrum.

Over-voltage, Over-current, Over-temperature

The IC supports over-voltage protection at input and output. Both are retried via the auto-restart block.

Over-current works differently. When that occurs. the FET is switched off for the remainder of one cycle. Over-temperature switches off the FET once kicking in, until the temperature drops below the threshold.

Zero cross detection

This functionality can be used to get a safe isolated pulse at each zero-crossing of the high-voltage AC input voltage. The signal can be used to synchronise your design, or as a utility-line-frequency based timer. In one of the ICs in the LinkSwitch-TNZ family (LNK3317D), this can also be used to discharge the X capacitor(s). To comply with some regulations (see X Capacitor Discharge Must Satisfy Both Safety and Energy Efficiency Rules), it can be a requirement to have X capacitor discharged to a safe level within a given time.

On/off control

The output is regulated by sampling the output voltage, and suppressing switching cycles as needed. This is different to other topologies, that use concepts like variable duty cycles, variable on or off time, ... By default it switches the MOSFET on for max duty cycle time, +- 70%. The on/off control will cut off chunks of that on-time, as required to keep the output voltage at the desired level.


The switch FET and its driver are integrated in the package. It's an n-channel MOSFET. Some of the specs:

source: P.I. datasheet LNK33x2-7D.

Isolated Flyback design overview

Time to dig into the reference design. It's an isolated flyback, where the flyback transformer has four duties:

  • isolation from mains (primary side)
  • store energy for the switch mode power generation
  • convert voltages, based on winding ratio of the individual coils.
  • generate primary side bias for the switcher operation

image source: Reference Design Report RDR-877

If we follow the design from top left to right: the power transfer part: 

  • mains passes through protection circuit, then is rectified in the diode bridge.
  • the inductor and 2 bulk capacitors filter the ripple.
  • the switcher will connect the inductor to ground during the on-time. This allows current to run through it. The magnetic field builds up in the flyback transformer core. Diode D1 on the secondary prevents that the energy is transferred via the secondary winding.
  • during the off-time, the energy stored in the transformer core can be released. The current in the secondary winding wants to flow in the opposite way, and it's allowed to do that, because D1 is forward biased and will pass the current. The bulk capacitor at secondary is charged, and an LC filter smoothens the ripple. The voltage is available at the output.

Then the feedback part, the middle part of the circuit from right to left:

  • the LED of opto-coupler U2 is emitting light in relation to the output voltage. Zener VR2 and the voltage divider underneath make that happen. The light dictates if collector-emitter current can run through the opto-coupler's transistor part.
  • that current flows into the controller IC, as its feedback control signal. If that current is above a given threshold when the IC samples it, it will suppress the next switching cycle.
  • this keeps the circuit controlling itself so that the output is at (near) the set voltage.

The lower part of the circuit is the zero-cross detection. This runs from left to right. 

  • during positive cycle, Q1 conducts and turns the LED of opto-coupler U3 on. This makes the transistor of that opto-coupler pull signal ZCD low.
  • during negative cycle, Q1 is high impedance. the LED of the opto-coupler is off and its transistor will also be high impedance. R11 will try to pull ZCD up to 12 V, but VR4 keeps it at 5 V level. 
  • the result is a TTL signal that has a transition at each zero crossing of the mains input.

Road test: efficiency

What did I do during the road test?

  • behaviour across the full spec range, stand-by to maximum load, in stable state
  • sample some interesting circuit nodes
  • measure the efficiency and power factor over the output power range
  • output voltage stability

What did I not do:

  • ripple measurement
  • test fault conditions and auto-recovery
  • sample the soft startup
  • check startup and close-down waveforms
  • check regulation for fast changing loads

For the efficiency test, I attached a Programmable Electronic Load to the output. I used two DMMs to measure voltage and current. At the input, I also measured the power with two DMMs, just behind the rectifier bridge, where it connects to the bulk capacitor. I used the procedure from P.I. University's Course Notes: Techniques for Measuring Efficiency. This is fairly precise, but doesn't take in account the loss in the protection and full bridge rectifier.

Conditions: mains 230 VAC, 50Hz. Room temperature 20°C. 

Board preparation: I cut the trace just after the bridge:

image image

images: where I cut the PCB trace, and how DMMs were wired in place

image: test setup for efficiency measurement

Then I made the programmable load step through all settings, from standby to full load. I let it settle and recorded volts and amperes at input and output.


image: spreadsheet with the measurement

The results are in line with the reference design report:

  • 30 mW stand-by (load disconnected)
  • 50% efficiency at 1% load
  • 72% efficiency at 10% load
  • 81% efficiency at full load


image: graph of efficiency vs load. top: 1 - 10%, bottom 10 - 100% load. Note that the x-scale isn't fully linear. I increased the increment along the way

Let's compare with the graph from the design report:


image source: Reference Design Report RDR-877

The spreadsheet, with graphs, is attached to this review.

While I was running the exercise, I took some samples of the switch node. Where MOSFET connects to the primary winding. Below you can see my capture of that. Can you spot an occurrence of the on/off output voltage regulation?

image: comparing a capture at the drain of the MOSFET (bottom left) with the Operation at Moderately Heavy Loading (Flyback) signal (bottom right) from AN-98. Schema part taken from RDR-877.

Road test: output voltage vs load

The same data used for efficiency, can also be used to graph the output voltage. From stand-by to full load.


image: graph of output voltage (in V) vs load. The horizontal scale is not linear.

Road test: documentation

P.I. has provided a rich set of documents. I will elaborate here on the Reference Design Report RDR-877. It touches all steps of the design. That includes a full calculation of the flyback transformer. And how to correctly manufacture one. P.I. has a spreadsheet for these calculations. In the document they show the exact parameters that were used for the RDR-877 design.

image: snippet of the flyback transformer parameters, from Reference Design Report RDR-877

The spreadsheet calculated this transformer configuration:

image: snippet of the flyback transformer parameters, from Reference Design Report RDR-877

The winding instructions show step by step how to construct it:

image: snippet of the flyback transformer manufacture instructions, from Reference Design Report RDR-877

Other highlights: the principles are shown. How the P.I. LNK3306D regulates the output. What value and type of resistors and capacitors to use to set the output voltage, maximum  current, ...

And a full set of characterisation graphs. A few of them I repeated in this road test. But they also show startup and shutdown behaviour. Load transient response. Ripple. Brown-in and -out. Thermal performance, And the waveforms at the interesting points in the circuit. Almost 50 pages of test data.

And that brings me back to the TL;DR section at the top of this road test review: the combination of a performant IC, a good reference design that reflects a real-world solution, and the rich documentation, resulted in a 5 star score. 

  • The overall review looks good. The waveform from the Rigol oscilloscope was very informative.

  • And the excerpts that I used from P.I. training document on Efficiency measurement: 


  • A few unused photos and images that are sitting in my road test folder:


    image image


















  • Really nice seeing that GIF!

    It's a great, textbook-quality waveform shape to learn from. 

  • A capture of two intentional switch artifacts:

    • the clock jitter to spread EMS across a spectrum, making the wave-form "jittery"
    • the on-off regulation mechanism to keep output at 12 V. This causes the glitches in the capture.


    Tested with 240 V, 50 Hz, 250 mA load.

  • Probably, their field engineers found that a good set of the customers need a design like this:

    • compliant with the usual consumer goods regulations:
      • standby consumption regulations
      • efficiency-at-load regulations
      • EMI
      • inrush and overpower protection
    • and with manufacturer requests:
      • small BOM
      • small PCB footprint
      • some power available during standby
      • can take the full range of mains voltages/frequencies 

    If a field engineer (or technology fair rep) can hand over such a reference design to their prospective customers, that may help sales. The design is pre-characterised, but a customer's lab can validate it.  It 'll give fast feedback if it's fit for purpose...

  • I captured part of the preparation work:

  • That looks very useful, having a known good power supply design to explore, created by the experts. I wonder if it was a complete product they were planning to mass manufacture, and then decided to share their whole internal documentation (or the bulk of it!) for their customers.

    Impressive they went to lengths to show what kind of transformer they expect to be used, down to precisely how many layers of insulation are recommended! That's the kind of level of detail that would be supplied to production staff on the shop floor, assembling transformers (or supplied to a wound component manufacturer I guess!).

    Very useful, clear-to-follow review!