Isolated Power Supply world-wide AC input

Part 1


manufactorer evaluation board of PFC power supply

Der Bericht geht nicht auf Schaltungstechnik ein, er zeigt am Ende auch keine nachbaufähige Schaltung, die Bauteilenamen werden nicht genannt, unwichtig; es geht hier nur um Entwicklungsmethoden, die vielleicht eigene Anregungen bewirken.

Wer eine nachbaufähige Schaltung sucht ist auf den Seiten der IC Hersteller besser bedient und möge seine Zeit hier nicht verbrauchen.

Es soll ein kleiner Überblick entstehen, was alles zu messen und zu beachten ist, bis man eine fertige Schaltung hat. Es ist eine Demoversion und daher nicht vollständig dokumentiert. In Deutsch und English gemischt geschrieben, da doch viele mitlesen möchten.

Developing a Power Supply, input voltage range 90-230Vrms with 20 watt power, Power Factor Controlled. Application of this power supply is not important, board name and part names not important. This application note is not a instruction-guide or a part-guide to rebuild this circuit.

Better visit IC manufactorer websites when searching for a circuit, don`t use your time to search here.

The purpose of this application note is to give a short overview how many measurements are necessary to get a final circuit design.



Safety Considerations: working on 400 Volt AC is very dangerous for your life, do this electronic work only when you have the skills, safety tools and when you are professional trained to work with higher voltage. Use isolating safety transformers. Don´t work allone.

Warnung: Aufbau eines Schaltnetzteiles direkt aus der 230V Haus-Steckdose ist lebensgefährlich bei Berührung der Potentiale (Phase). Um diese Gefahr zu vermeiden muss die Versorgung der Schaltung aus einem regelbaren Schutztrenntransformator erfolgen.

Hinzu kommt, fast alle Messgeräte und insbesondere Oszilloskope haben eine Verbindung von ihrer Mess-Masse hin zum Schutzleiter, das bedeutet sie können nicht für Messungen an der galvanisch ungetrennten Netzspannung verwendet werden, sie werden dabei zerstört, da sonst die spannungsführende 230V Phase direkt über Messgeräte-Masse an den Schutzleiter gelegt werden kann, was die Netzsicherung auslösen kann und/oder wahrscheinlich das Messgerät zerstört.

Trotz Verwendung eines Schutztrenntrafos sollte man nicht vergessen, primärseitig sind Spannungen von 400V zu erwarten, diese sind bei Berührung von Hin- und Rückleiter genauso lebensgefährlich. Wer an solchen Schaltungen arbeitet muß über die notwendige Ausbildung und Werkzeugen im sicheren Umgang mit hohen Spannungen verfügen. Arbeiten Sie nicht alleine.

Oscilloscopes

Why using here an older type of digital scope? When developing an unknown evaluation board with 400V, it is a potential risk to destroy the test equipment. Especially the first hours when getting experience with the board are the worst. One known problem, accidential misuse measuring on the wrong potentials on different grounds. Using safety transformers is an additional must.

Oscilloscope here has a max. 250Vpk input voltage range, 1:10 and 1:100 probes can extend the range. For me a risk using more expensive higher bandwidth models having only 100Vpk. input voltage range - avoiding tears of pain after an accidential misuse.

This older type, year 1997 has enough performance for such a development, easy to operate and trustful measurement results, it is still an excellent oscilloscope.


Warum verwende ich hier ein Digitalspeicheroszilloskop das 1997 gebaut worden ist obwohl moderneres zur Verfügung steht? Weil es für diese Aufgabe technisch ausreichend ist und mehr kann als nötig. Der Hauptgrund sind die vorkommenden 400 Volt am Evaluation Board. Trotz Verwendung von Schutztrenn-Transformatoren und geeigneten Tastköpfen sind beim Messen schnell Fehler passiert und manche moderne Digitalspeicher Oszilloskope wären am Eingang leicht zerstörbar.  Dieses Modell hier hat wenigstens 250Vpk als Maximum, mit 1:10 und 1:100 Tastkopf noch höher. Warum soll man sich ein teueres moderneres Modell mit mehreren GHz Bandbreite riskieren, dass nur 100Vpk als Maximum aufweist? Ja, es würde mehr Spaß damit machen, aber das Risiko ist mir zu hoch - außerdem ist das gezeigte Oscilloscope leicht bedienbar und liefert absolut zuverlässige Messdaten.

Noch ältere analoge Oszilloskope haben oft noch höhere maximal zulässige Eingangsspannungen, wäre natürlich klasse sie zu verwenden, nur ein wenig dämlich in der Verwendung bei single shot Signalen, wie sie hier bei einer PSU häufig zu messen sind. Momentan kann ich nicht auf einer großen Werkbank entwickeln und entsprechend ausbreiten, geht zur Zeit nur auf kleinstem Tisch, da ist der Platz zu haushalten.

Power ON without load

power on without load
Figure 1

Power ON without load, voltage should be 24V.
Increase slowly to 50V ! without load, danger by overvoltage.

Beim Einschalten ohne Belastung am Ausgang läuft die Spannung ganz kurz auf die Sollspannung von 24V, um danach langsam auf 50V anzusteigen. Bei leichter Last kann man sich dadurch ganz schnell die angeschlossene Last zerstören.

Power ON with a higher 3k6 ohm load


inrush current and output
Figure 2
Ch2 = output voltage
Ch1 = primary current
3k6 ohm load

Starts fine. Power Factor Controller reduce inrush currents.

Settled Output Voltage vs. minimum Load Resistance

minimum load eval board
Figure 3
Doing an experiment, starting the power supply with different minimum output load resistors. Evaluation Board needs at least a minimum load of 8 kohm. Min. Output Power (24V)²/8k2 = 70mW.


Output Load 16 Watt @230Vrms - settled output


16 Watt Load
Figure 4
Load 16 Watt, 24VDC @ 230VACrms
Channel 2 = Input Current
Channel 3 = Output Voltage
User decides if output ripple +/-2V acceptable, depending on application.

Load 16 Watt @230Vrms - Power-ON

power on 16 Watt at 230VACrms
Figure 5
Fine Power-ON procedure
Channel 2 = Input Current
Channel 3 = Output Voltage
Current peak at t=0 comes from small filter foil capacitors in the primary EMC filter. During start-up input current remains low. Output voltage settle to a stable value, overshooting to 26V.

Load 16 Watt @230Vrms - Power-ON Input Current Peak

input current peak
Figure 6
Channel 2 = Input Current of EMC foil capacitor

First current peak at t=0 comes from filter foil capacitors in the primary EMC filter. Input current should be considered when driven by small semiconductors, should always remain within the drivers safe operating area.

Peak current depends on phase angle of the input sine wave. Catching a trigger at >1A, ten times Power-ON applied. Finding the maximum possible peak current requires a equipment, switching ON under defined phase angle.

Load 16 Watt @230Vrms - settled output

output 16W at 90VACrms
Figure 7
Load 16 Watt, 24VDC @ input 90VACrms
Channel 2 = Input Current
Channel 3 = Output Voltage

User decides if output ripple +/-2V acceptable, depending on application. Output looks exactly the same as with 230VACrms. No remarkable heat in any component.

Load 16 Watt @230Vrms - Power-ON AC-line

power-ON 16watt 90VACrms
Figure 8
Fine Power-ON procedure
Channel 2 = Input Current
Channel 3 = Output Voltage

During start-up input current remains low and settle to the nominal level after start. Output voltage settle to a stable value, overshooting to 26V.

Results of First Tests

  • works under a wide input voltage range.
  • voltage ripple depends on application.
  • no remarkable heat in any component.
  • fine Power-ON, if driven by relais contacts - no problem.

Das Board läuft schön. Man sollte anfangs immer erst alles grob durchtesten, bevor man an die Details geht. Es geht auch darum, auch Schwachstellen kennenzulernen.



Part 2

Increasing the Output Voltage to 28-29VDC, application changed


Output Load 24 Watt @230Vrms - settled output

24W 28VDC 230V
Figure 9

ok

24V 28V 230V power on
Figure 10

fine Power-ON, check if overshooting OK for your application

Power Transformer and secondary side rectifier diode runs too HOT under 24W constant load, these parts can not be used for a permanent 24W load, possible for a short-time.

What is short time? Can not answer this question, depends on environmental temperature, thermal mass of transformer and diode heat sink - not a question of today.


power off 24w 28v 230v
Figure 11
Switching off line power, fine Power-Off


removing load, 24V, 28v, 230Vrms
Figure 12
Remove 24W load, output jumps up to 32V, settle later to 30V.
Depends on application if this voltage jump is acceptable.

24W Load needs at least a 150VAC rms input, lower voltage require another current limit or less output power.


EMC Problem with this Evaluation Board


Figure 13

When running the Evaluation with 24W load there is an EMC problem. The radio on the photo is a digital radio with DAB+ standard. One station "xxx" stops receiving when the Evalboard is loaded with overpowered 24W.

Transformer has been exchanged, additional using too long transformer wires, can be a reason.

When increasing the distance between radio and Evaluation Board (antenna up, like in the photo), the radio program received clearly. Before the antenna was hanging down, this was too near. 

Checking Emitted Air Spectrum

spectrum-analyzer
Figure 14

Signal Analyzer
  • Span 9kHz to 13GHz, very sensitive instrument
  • Broadband antenna, usable as indicator from 50MHz to 2GHz, linear range from 100MHz to 1GHz
  • Gain on analyzer 50 ohm Input: 0dB (without antenna), approx. +30 to +35dB (with antenna) @ linear range
  • Cable max. 18GHz, low-loss cable, highly flexible mechanical structure

Environmental Air Spectrum - AC/DC Power Supply OFF

air spectrum - acdc converter powered OFF
Figure 15
Environmental spectrum with antenna, Evaluation board switched OFF. Left side 9kHz, right side 300MHz, 30MHz/Division, sweeptime 10ms.

87-108 MHz, UKW radio
170-220 MHz, DAB+ radio


Environmental Air Spectrum - AC/DC Power Supply ON

environmental air spectrum Evalboard powered ON 24W
Figure 16
Evaloard powered-ON (24W, 28VDC, 150VAC rms)

Antenna position and analyzer setting same as figure 15. 
Antenna distance 1 meter to Evalboard

100-300 MHz disturbed under a short distance. When moving the antenna away, EMC level decrease rapidly. Above 300 MHz less radiation (not shown here), no harmonics in the mobil phone frequency range observed.

Performance of the Evalboard:
  • layout single layer
  • small ground plane (single layer)
  • capacitors can not be placed better on a single layer, parasitic higher capacitor ESL
  • unshielded power transformer construction
  • unshielded EMC filter construction
  • some PCB traces with high du/dt and di/dt are longer size
  • free wires on the power transformer due an power transformer replacement
  • Evalboard free air
 
"good power efficiency + single layer layout + unshielded parts = fine BOM, difficult EMC"

Always a demanding development task:
BOM, space, heat-management, different loads, EMC.

When developing a circuit with fast changing currents und voltages always use a spectrum analyzer to observe the circuit, beginning from the first hour of the development.


Part 3

Shorting the 5V Auxiliary Winding

shorting aux. 5V
Figure 17
Board has a 28V/24W power winding and 5V/4W auxiliary winding. In Figure 17 there is no load on the power winding and a short in the 4W aux. winding. Question, does the controller detect the short cut? Yes, controller detects a short-cut via the second primary winding and shut off. Results in an ON-OFF oscillation. Transformer windung runs not hot, no other part runs hot. Controller has a fine short-cut behaviour.

overload on 5V aux. winding
Figure 18
5V aux. winding with a constant applied 3R3 (7.5W) winding overload condition.

Controller can not detect this 7.5W overloaded auxiliary winding  (overcurrent detection starts at 24W).

Result:
  • 5V rectifier diode runs in thermal dead, changed to a higher power type, still hot, can not touched with fingers
  • 5V winding becomes warm
  • 5V rectifier electrolytic 220µF capacitor became warm under terrible output voltage ripple, here changed to 2200µF.

  • 5V winding is short-cut protected but not overlaod protected. This transformer and circuit can not be used for the target application with the requirement of safe permanent overload conditions.
Application requires another type of transformer with a single secondary winding only. 5V auxiliary voltage must generated by a DCDC converter with internal overload protection, converter driven by 24V.

Testing another Power Supply for EMC Comparison


Figure 19
Simple EMC test of a worldwide input 120W power-supply from a flatscreen television, made year 2014. 



spectrum local radio stations
Figure 20
Air spectrum with powered-OFF TV power supply, line cable removed from wall socket. Visible the local FM stations and DAB+ stations. Span 9kHz-300MHz, sweeptime 10ms.


spectrum TV power supply
Figure 21
Powered-ON TV power supply. Span 9kHz-300MHz, sweeptime 10ms.

Primary Current in the Transformer

primary current
Figure 22
Primary winding current

Off-resonant controller, fast current change. During MOSFET Off current stops fast, resulting in LC-oscillation of primary winding leakage inductivity with MOSFET capacitance.

MOSFET Drain-Source capacitance + 220pF SMD capacitor. LC resonant period 500ns.

MOSFET switching period 15µs (90VAC rms), decrease to 12µs (230VAC rms).


primary winding current
Figure 23
Primary winding current follows the sinewave envelope.


Power Factor Quality

line current and line voltage under 20W load
Figure 24
Line current and line voltage under full load. Power Controller ensures that the load behaviours like an ohmic resistor, line current and line voltage should follow each other in phase, the waveform should look the same.

squared waveforms
Figure 25
Line current and line voltage multiplied to power. With an ideal PFC controller and ideal sinewave voltage the power signal should be a 100 Hz x²signal. Power Factor is distorted in the maximum power area.
  • Parameter measured on Waveform A:
  • pkpk(A) 58W
  • mean (A) 24W
  • rms (A) 32W
Measuring the efficiency of this PSU, 8-bit scope a wrong tool, requires a power-analyzer or rms precision multimeters, Two DMM on primary and two DMM on secondary side.

power spectrum
Figure 26

Power Spectrum of the 100Hz x² signal.
Dominant 100Hz frequency, 200Hz harmonic approx. -25dbc.

This scope software is still great, even with a monochrome display - cleary channel calculations.
A modern >=12bit resolution scope would be a better tool to do this calculation.

PS: hoping, soon I will get a self-made power analyzer with a higher resolution.


Part 4

Changing Output Silicon Diode to an SMD 200V, 3A Type
Changing Output Capacitor to 50V/2200µF
Adding a 1µF/50V X7R Output Ceramic Capacitor
Changing transformer, two secondary power windings in series, no auxiliary secondary winding.


Figure 27


Changing to a smaller SMD Power MOSFET


Figure 28

too small SMD MOSFET installed, will become hot under 20W load, PCB heatsink required.




Figure 29

Power Transistor changed with another SMD 800V/1A MOSFET (cheapest price). A small piece of copper-foil acts here as heatsink. Copper becomes very warm under 20W load and low input voltages, my finger burns when touching, too hot!

For a final PCB version this small SMD transistor would require a PCB-layout heatsink area large as possible. Enough thermal vias to the inner-layers highly recommended.

The heat sink is connected to the Drain potential. Drain-to-Source Capacitance reacts with the parasitic transformer incductance as LC oscillator used in this ZCS Zero-Current-Switching system. In the first "valley" switch the transistor.


drain voltage drain current
Figure 30

Fly-Back, results in high Drain-Source voltages. When driving the Evalboard with 280VAC rms the Drain voltage increase up to 700V. MOSFET transistor has a 800V specification.

700V is large and a dangerous number. When doing the layout this high voltage has to be considered in surface-leakage and isolating distance.

Also when measuring such a high voltage use a 100:1 probe, the used probe has a 1,5kV specifiaction. Don`t use 10:1 probes, many of them are not specified for 700V, can damage oscilloscope input.

  • Drain voltage oscillates with approx. 450ns, 2.2MHz
  • Transformer winding leakage inductance 35µH
  • Drain source capacitance, external smd 220pF
  • LC resonant frequency: 1/2*PI*sqrt(L*C) approx. 2 MHz
Theorie and measurement - same result.



Figure 31

Waveform with respect to the Gate voltage (Ch.4)



fast current probe
Figure 32

Before in Figure 30 the drain current is not correct measured, the bandwidth of the power current probe (max. 150A, 120kHz) is too low. Therefore for the fast changing drain current a 50MHz type.


fast current change in drain
Figure 33
Fast drain current changes measured with 50MHz probe. Drain current (Ch.1) and Drain voltage (Ch.3) are now exactly in phase.

step 1.----------------------Gate ON-------------------------------------
When switching ON the Gate, current in the transformer starts with a ramp. Current increase linear, a good message, the inductance of the transformer remains constant even under higher currents, core is not overloaded. An overloaded core would show an non-linear ramp with a fast slewing ramp under high currents.

step 2.---------------------Gate OFF-------------------------------------
After reaching a maximum current level (detected by the controller) the Gate switch OFF and stops continuing charging the core with magnetic energy. Current in the winding stops very fast, Gate goes low with fastest slew-rate possible by the controller.

For the gate discharge there is no smd resistor in the gate, only a small signal diode, fast OFF.
For the gate charge there is a 220ohm resistor in the gate, resulting in an slower ON.

Gate OFF must be done as fast as possible to keep the MOSFET switching loss to a moderate level. Consider when switching OFF the MOSFET the Drain voltage is low, but the Drain current is at a maximum.

OFF switching-power = draincurrent*drainvoltage

Disadvantage for the fast OFF switching, increased EMC emission:
Drain current changes very fast, large high H-field emission.  
Drain voltage changes very fast, large E-field emission.

step 3.---------------------Oscillation-----------------------------------
Transformer leakage inductance and MOSFET drain capacitance starts LC resonance

step 4.---------------------Demagnization-------------------------------
stored magnetic energy in the transformer core runs empty (energy transfer to secondary side).
When energy transfer stops (stored magnetic energy empty), voltage at the primary winding increase with u=L*di/dt. An increasing reversed voltage across the primary winding reduces automatically the drain voltage. This automatically decreased drain voltage can be observed as the slowly falling drain voltage ramp. The controller does the same, measure and search for a moderate -du/dt drain voltage. After detecting this demagnization point the controller will ON the MOSFET again.

step 5. ------------------------ON--------------------------------------
exactly after core demagnizited (empty magnetic energy) it is a nice moment to power ON the core again. At this moment no current is flowing in the primary winding and no drain voltage (voltage across coil), it is an ideal point, less MOSFET switching losses. Because drain voltage and current is low, the gate can be switched ON slowly, thats why there is a  Gate-ON 220 Ohm resistor, reducing the MOSFET ON time. Swichting ON creates not much MOSFET switching losses and not much EMC-emissions, this is an advantage.

MOSFET Power OFF creates much EMC-emission and much thermal Switching-losses.
MOSFET Power ON creates less EMC-emission and less thermal Switching-losses.

ZCS (Zero Current Switching) method:
approx. 50% of the EMC-emission and thermal Switching-loss could be reduced under MOSFET ON.
Remaining 50% EMC-emission and thermal Switching-loss during MOSFET OFF can not be reduced.

E-field and H-field emission can be only reduced, using a larger slower switching MOSFET with a larger thermal heatsink and less power efficiency.

It is a question of the application what is more important: less EMC-emission or a better power efficiency.

This is Part 1
Go to Part 2
Go to Part 3
Go to Part 4
Go to Part 5
Go to Part 6
Go to Part 7

Technische Berichte
www.amplifier.cd.

Impressum und Haftungsausschluss