Power Supply Efficiency – How Important is it?

ENERGY STAR®’s decision to “sunset” programs for EPSs (External Power Supplies), and applications using them, has raised the question “Does this mean that power supply efficiency is no longer a concern?”

In a mainly cost driven power supply market, product marketing is often challenged during a new product business plan review, regarding the demand for high efficiency on the product specification.

The biggest requestor of high efficiency power supplies are the manufacturers of large data centers, primarily because of the huge amount of electricity they consume.  In 2007 the EPA estimated that the national annual electricity cost for servers and data centers could be a staggering $7.4 billion in 2011.

Power supply manufacturers have been told by data center producers that they are willing to pay (a little) more for higher efficiency products, because the operators are aware that the ROI is quite short.  TDK-Lambda also is participating in the evaluation and use of DC power systems. 

DC power systems for data centers use up to 15% less power by removing one or more conversion stage by supplying ~380VDC to the servers rather than high voltage AC.

Current UPS Backed-up System

DC Bus System

The US government has had the 80 Plus program in place for servers and computers for a number of years with a variety of efficiency levels ranging from a basic level to “platinum”.

In consumer electronics where although the power draw is quite small (particularly in standby), there are 100s of millions of devices deployed (laptop power supplies, phone & camera chargers & LCD TVs to mention a few) and the overall energy usage is considerable.  The main drivers for efficiency improvement are the state and utility companies, because of the cost and time to build new power stations.  The average consumer is not that concerned as it will not dramatically affect their electricity bill.

As an industrial power supply manufacturer, TDK-Lambda is being asked by some customers for higher efficiency power supplies.  These customers are selling to end users like hospitals or large retailers who are going "green".  Where the purchaser is not being mandated to reduce energy consumption or is driven by price, the efficiency “sell” usually fails.

TDK-Lambda's R&D is very focused on developing higher efficiency power supplies, driven by what we see as a future market and TDK's initiative on environmental consciousness.  Current high efficiency products include the new EFE series of power supplies that feature 90% efficiency using digital control.  http://us.tdk-lambda.com/lp/products/efe-series.htm.

TDK-Lambda also supplies two of the major automotive manufacturers the DC-DC converters for hybrid electric vehicles and is developing new products for a host of upcoming electric cars.  Efficiency standards of automobiles is something that the purchaser understands, primarily because of the cost of running a vehicle is significant.

Using the Inhibit or Enable function on Power Supplies

The inhibit or enable function allows the user to electronically turn on or off the output voltage of a power supply without having to interrupt the input AC or DC voltage with a relay or switch. This is useful during initial set up of the system, during maintenance or for saving energy during periods of non operation.

An easy way of remembering the difference between the two types is that “Inhibit” requires that the user has to do an action to turn off the output voltage, where as “Enable”, the user has to do an action to turn on the output voltage.

It is standard with most DC-DC converters for example, to utilize an enable type function that requires the remote on / off pin to be connected to the negative input (primary side) to activate the output voltage. This is often referred to as “negative logic”. First time users of DC-DC converters often forget to pull that pin low and call Tech Support to complain about a non functioning power supply.

AC-DC power supplies usually have a remote on / off referenced to the secondary side for safety reasons. An “Inhibit” type function, requiring an external voltage, is usually more popular on simple power supplies because once the output voltage is turned off, any auxiliary voltages driving the secondary control circuit is also turned off.

One way power supply designers overcome this is to have an integrated independent “stand-by” auxiliary output like the one used on pc power supplies. This is also used to power the secondary control circuit and allow a closed contact to the 0V terminal to enable the output.

If the system uses several power supplies, the remote on /off can be used to sequence the voltages. I know of thermal printer applications where if the 5V supply driving the control circuit goes faulty, they require the 24V motor drive be inhibited to avoid embarrassing amounts of paper shooting out!

Where several different voltages are used to drive processors, sequencing the output voltages is often critical to avoid damaging those devices. The 3.3V output is rarely allowed to be applied before the 5V is present.

The industry standard I²C based PMBus is now gaining popularity. Power supplies such as TDK-Lambda’s HFE series can be remotely turned on or off using the PMBus software, either as a group or individually for load shedding to save energy.

Damaging Power Supplies with Repetitive Peak Current Draws

There are many devices that require peak currents when first turned on including print heads, motors, disk drives and pumps.

Many users often do not measure the actual peak current and rely on an empirical method whereby they try a power supply in the application to see if it will work.  If the power supply cannot provide enough peak current, capacitors are added to the output.  Those capacitors will act as temporary energy storage, enough to deliver load current for a few hundred micro seconds.  If that works then the power system solution is deemed as working, and the Engineer moves on to the next phase of their project.

Many power supplies have the ability to supply high peak currents, even though the datasheet does not mention it.  In fact some of the cheapest power supplies on the market can deliver very large currents for a short period of time because the output current limit is very crude, and is primarily there to protect the unit against a short circuit on the output.

In discussion with TDK-Lambda Engineering, I learned that this can lead to field failures.  Let me explain further.

Below is a schematic of a forward converter, the power FET is shown as a switch for simplicity.  That “switch” operates at a rate usually in the hundreds of kHz, energy is transferred from the secondary side to the output rectifiers and then is smoothed by the output LC filter.

When a pulsed (peak) load is applied to the power supply in excess of its rated current, the energy is first drawn from the output capacitor.  This can add to the capacitor ripple current, raising the temperature and reducing the component’s life.  Heat, as I explained in earlier blogs, dries out the capacitor’s electrolyte.

When the energy stored in the capacitor starts to deplete, the power supply will then try to continue to provide the peak current from the main switching circuit.  This in turn leads to repetitive surge currents in both the output diodes which is then reflected by the transformer to the power FET.  Often this peak current exceeds the maximum rating of the semiconductors leading to latent and erratic field failures.

Additional heating in the transformer, inductors and printed circuit board traces is also experienced because, although the average power drawn from the power supply is less than the continuous rating, we are dealing with the formula I2R and the peak current is now squared.

TDK-Lambda recommends using a power supply that has a specified peak power rating like our HWS-P series or working closely with the power supply manufacturer to determine if the product is suitable for the application.

How does the new UL Mark affect me?

UL recently announced their first Transcontinental Mark which denotes product compliance with European EN safety standards as well as the CSA/UL/US Mark for the North American markets.  Basically, UL will be providing an “international safety certification”.  And, this may affect European safety houses like TUV and others.  Here is what the new international UL mark it looks like:

Although, at first glance, this mark looks like a “UL Listed” power supply certification with its encircled UL mark (as used for External Power Supplies), UL-EU confirmed that it also applies to stand-alone component power supplies like those installed in all electronic equipment.  Prior to this new mark, the standard “component power supply” UL mark for the US and Canada looked like this:

Some power supply manufacturers have been reviewing use of UL to display the existing C-UL-US mark along with the self-declared CE mark.  The CE mark would be backed up with an IEC60950-1 CB report as documented proof of compliance.

It will be interesting to see how the European test houses (e.g., TUV) react and if CSA responds with their own version of this new UL mark.

Power Supply Leakage Current Testing to IEC60990

A customer recently asked me why we specify leakage current on a Class II power supply, when a Class II power supply has no ground terminal.  A good question, but first some background.

As part of the testing for IEC60950, power supply manufacturers measure leakage current to the IEC60990 standard.

To be more accurate, the terms "Touch Current" and "Protective Conductor Current" replace the term "Leakage Current".

Protective Conductor Current (PCC)

Is the current that flows through the protective conductor; commonly referred to as the ground connection.

As a note, the withstand voltage and insulation resistance tests measure the current flowing through the insulation of the unit under test.

 Touch Current (TC)
Is the current that flows when a human body touches the equipment, simulated by a body impedance network.

The switches are used to simulate a line, neutral or ground fault, referred to as a single fault condition (S.F.C.).  Usually there is a polarity reversal switch to reverse the line and neutral connections to the power supply.

So back to the original customer question, if a Class II power supply is used, there will be current that flows through a human body upon touching conductive parts in a system (like a USB port or a conductive product case).  That measured current is usually listed on the power supply datasheet.

Here is an excerpt from a CB report showing the test, input voltage, frequency and the measured touch current.  Note half the tests conducted were with the simulated human body touching the “output connector” or pins of the power supply. 

Enclosure leakage current (normal conditions, normal polarity)	264 V~	63 Hz	5,3 µA
Enclosure leakage current (normal conditions, reverse polarity)	264 V~	63 Hz	4,1 µA
Enclosure leakage current measured on output connector (normal conditions, normal polarity)	264 V~	63 Hz	89,0 µA
Enclosure leakage current measured on output connector (normal conditions, reverse polarity)	264 V~	63 Hz	87,0 µA
Enclosure leakage current (single fault conditions, neutral open, normal polarity)	264 V~	63 Hz	4,1 µA
Enclosure leakage current (single fault conditions, neutral open, reverse polarity)	264 V~	63 Hz	6,0 µA
Enclosure leakage current measured on output connector (single fault conditions, neutral open, normal polarity)	264 V~	63 Hz	3,0 µA
Enclosure leakage current measured on output connector (single fault conditions, neutral open, reverse polarity)	264 V~	63 Hz	129,0 µA

The ammeter used is a specialized meter; do not use a regular hand-held multi-meter!

For more details, including the limits of the measured currents, please consult a professional safety engineer.

Get the product brochure from TDK-Lambda Americas for product descriptions.

Did my power supply fail or just wear out?

A true power supply failure is a rare occurrence provided the following occurs:

1. The manufacturer has taken the appropriate steps on component evaluation, component derating and has utilized sound design techniques.
2. The user is operating the product in accordance with the manufacturer’s instructions.

TDK-Lambda, for example, puts many components through a myriad of stress tests including voltage testing under extreme humidity and atmospheric pressures, beyond the manufacturer’s specified maximum ratings.  Only upon passing those tests is the component supplier added to the approved vendor’s list.

Monitoring primary switching currents at high line and high ambient temperatures during transient loading can reveal how much design margin the power supply has.  Below you can see that our competitor’s power transformer is inadequately sized and is drawing a huge increase in current as it saturates.

So, when I hear “My power supply failed after just three years in the field” from potential customers, I review their application and often have to deliver the bad news.  Their power supply has just worn out.

Recently a manufacturer of semiconductor fabrication equipment called me.  They were using a one-year warranty power supply and were running it at the full rated power level. 

I asked the Engineer how long was their equipment typically operated in the field.  “Our equipment is usually running 24/7 (24 hours a day, 7 days a week)” was the reply and “our customers expect to use the machine for at least ten years.”

In a power supply, the most frequent wear-out component is the electrolytic capacitors.  Capacitor life has improved greatly with reasonably priced electrolytics, now typically rated for 10,000 hours at 105°C.  Even then, that is only 1.14 years when used 24 hours a day in a very harsh environment.

The solution is quite simple, choose a suitable grade of power supply and apply sufficient deratings.  I always ask about the application, the expected field life, the cost impact of their customer’s equipment being out of service, and the cost impact of having to service the equipment.  If the application is in a remote application or would require a service person to drive or fly out to the location, paying an extra $50 or even $100 more for an rugged, industrial grade power supply will be more economical than paying for a $500 field service call, perhaps 3 years times.

I often equate power supply life to that of buying a cheap set of brake pads for one’s car.  Yes, it costs less initially, but you will probably be back for another set of pads (and spending time in the repair shop) in less than half the time that a higher quality set of pads would last.

Common safety standards used with power supplies

A topic that comes up very regularly is what do the various safety standards pertain to. Here is a brief list of commonly used standards, the equipment that uses them, typical products and their purpose.

Base Standard	Country (ies)	Equipment Type	Generic Product Types	Purpose
CCC	China	Many	Many	Safety & Quality Mark
CE	EU	Voluntary declaration based on miscellaneous standards	Many	Indicate conformity to standards
CSA 60950	CSA	Information Technology	Office machines, data & telecom networks, IT, Kiosks	Protect against fire, electric shock, injury
CSA60065	CSA	Audio, Video and Similar	Video, audio, projectors	Protect against fire, electric shock, injury
CSA60601	CSA	Medical Electrical	Surgical, monitoring, hospital equipment	Safety
CSA61010-1	CSA	Measurement, Control and Laboratory	Meters, Oscilloscopes	Protect against fire, electric shock, injury
EN50178	EU	Power	Power generation, power installations	Safety
EN55011	EU	Industrial, scientific, and medical r-f	Monitoring equipment, automation controls, measuring	Limits and measurement of radio disturbance
EN55015	EU	Lighting	Lighting, LED lighting, street lighting	Limits and measurement of radio disturbance
EN55022	EU	Information Technology	Office machines, data & telecom networks, IT, Kiosks	Emission Limits (radiated and conducted)
EN60065	EU	Audio, Video and Similar	Video, audio, projectors	Protect against fire, electric shock, injury
EN60601	EU	Medical Electrical	Surgical, monitoring, hospital equipment	Safety
EN60950-1	EU	Information Technology	Office machines, data & telecom networks, IT, Kiosks	Protect against fire, electric shock, injury
EN61010-1	EU	Measurement, Control and Laboratory	Meters, Oscilloscopes	Protect against fire, electric shock, injury
Factory Mutual	US	Use in hazardous locations	Oil refinery, petrochemical	Protect against explosion
FCC Part 15	US	Many	Many	Emission Limits (radiated and conducted)
ISA 12-12	US	Use in hazardous locations	Oil refinery, petrochemical	Protect against explosion
REACH	EU	Many	Many	Registration, Evaluation, Authorization and restriction of Chemicals
SEMIF47	US	Semiconductor fabrication	Die manufacturing, wafer fabs	Withstand dips in AC input
UL 60950	UL	Information Technology	Office machines, data & telecom networks, IT, Kiosks	Protect against fire, electric shock, injury
UL1310	UL	Power Supplies	DIN Rail, LED lighting, access controls, building automation	Protect against fire
UL508	UL	Industrial Control	Process control, factory automation	Safety
UL60065	UL	Audio, Video and Similar	Video, audio, projectors	Protect against fire, electric shock, injury
UL60601	UL	Medical Electrical	Surgical, monitoring, hospital equipment	Safety
UL61010-1	UL	Measurement, Control and Laboratory	Meters, Oscilloscopes	Protect against fire, electric shock, injury
VCCI	Japan	Information Technology	Office machines, data & telecom networks, IT, Kiosks	Limits and measurement of RF emissions
VDE0805	Germany	Information Technology	Office machines, data & telecom networks, IT, Kiosks	Protect against fire, electric shock, injury

Let me know if this table is helpful.

Cold Temperature Start Up of Low Cost Power Supplies with Inrush Thermistors

I often get asked the question: "Regarding the TDK-Lambda low cost power supply that is rated from -25°C to 70°C, will it start up at -40°C?" I usually reply, "It depends."

Most low cost, low wattage power supplies avoid large surges of current being drawn when the AC input is first applied by using a thermistor in series with the AC line (see figure 1). This device is a type of resistor that when cold has a much higher resistance than when warm.

Figure 1: Thermistor

When the AC is first applied, the thermistor limits the amount of inrush current that charges the bulk storage capacitor. Once the power supply starts-up and delivers power, the thermistor self heats and decreases in resistance to improve the power supply efficiency and operation.

At ambient temperatures below freezing, these thermistors have very high resistances, and if the supply is “rated” to start-up at a cold temperature, it should have been tested during the design stage to ensure correct start-up at full load and at minimum AC input. If the power supply is not specifically rated for a cold temperature start-up, there is a possibility that the power supply will turn on, try to deliver power, but the thermistor will not have self-heated due to the very cold ambient and hence will have a large voltage drop across it, causing the power supply to switch off again. The power supply will try to restart again, causing “blips” on the output. (Fig. 2, top trace). In some circumstances, the power supply might not start up at all. These attempts to restart can cause system problems.

Figure 2: Cold Temperature Start Up

If the output load is light, however, the power supply may be able to start up correctly.

So back to my reply to the initial question! "It depends on what the loading will be at -40°C. If the application has say 20% loading, then usually the answer is yes." There are other issues with cold temperature operation, but I shall cover that in the future.

Hipot or Dielectric Strength Testing

One area of confusion in production safety testing is the dielectric strength test, sometimes known the as dielectric withstand test or “hipot” test.

This test is usually applied between the secondary output and chassis ground and then between the AC connection (primary) and ground / secondary. This test can identify any assembly errors such as a pinched wire.


It is important to ensure to short the line and neutral together during the test, and when making the primary to secondary test, connect the secondary side to chassis. Short the output terminals together if testing a standalone power supply. Failure to do this can result in damage to the power supply.

A routine question is “should the test voltage be AC or DC?” The majority of power supply manufacturers use a DC voltage because the leakage current through the “Y” capacitors can mask another fault.

The “Y” capacitors are identified in a very simplified diagram below are used to reduce EMI and electrical noise.


As can be seen, applying an AC input to chassis hipot test would result in mill-Amps of current flowing through the capacitors

The majority of safety standards allow DC hipot voltages. Instead of applying 1500VAC, one would use the peak voltage of the AC, √2 x 1500 = 2121VDC. Add 10% to reduce the test time from 1 minute to 1 – 2 seconds.

Apply the DC voltage slowly to allow the capacitors to charge up without tripping the current limit of the test equipment. Remember to discharge the capacitors after the test.