Wednesday, June 27, 2018

An update to the transition from IEC 60950-1 to IEC 62368-1

On a blog post back in March 2015, I mentioned that a Hazard-Based Safety Engineering (HBSE) IEC 62368-1 standard will be replacing IEC 60950-1 and IEC 60065, covering hazards and hazard prevention for ITE (Information Technology Equipment) and audio / visual equipment.

The transition date to EN 62368-1:2014 has now been extended 18 months to December 20th, 2020 for new submittals (or a significant modification of the existing 60950/60065 file) in the EU.  UL has also announced that the UL 62368-1 Edition No. 2 date is now December 20th, 2020. A number of other countries have adopted the standard, including Australia, Singapore, Malaysia, Japan and Russia.  No exact date has been published for other countries like China, Taiwan, South Korea and Argentina.

The new standard uses a Hazard Based Safety Engineering (HBSE) science discipline, and operates in four steps:

1.      Identifying energy sources in the product

2.      Classifying the energy for the potential of causing injury or harm

3.      Identify the necessary safeguards needed to protect from those energy sources

4.      Qualify the safeguards as effective

The Energy Source ES classification handles the effect on the body for a number of hazards.  See Table 1 for example of the two most relevant to power supplies – electrical and thermal energy.

Energy Source class
Effect on the Body
Effect on combustible material
Class 1
Detectible, but not painful
Unlikely ignition
Class 2
Painful, but not an injury
Possible ignition, but limited
Class 3
An injury
Likely ignition, growth rapid & swift

Table 1 Energy Source classification

In addition, a three person scenario has been adopted for 62368-1 (see Table 2).            

Description (See standard for full details)
Ordinary Person
A person who is a user or is close by to the equipment
Instructed Person
A person that is trained to identify sources of pain causing energy and avoid them. Must not be exposed to injury causing energy sources, even during a single fault condition
Skilled Person
A person who has the training to recognize & avoid energy sources that could cause pain or injury.  Must be protected against accidental contact with injury causing energy sources

Table 2: Description of types of person

Depending on the Person type and the Energy Class, Safeguards have to be in place to protect that Person.  See Figure 1.

Figure 1: Protection for an Ordinary Person with different Energy Classes

As an example, one small change power supply manufacturers may have to make to their designs to meet IEC 62369-1, concerns “capacitance discharge” of the AC input. 

Figure 2 shows the typical schematic of an EMI filter section of a power supply and the location of a resistor that discharges the Line to Neutral X capacitors after the AC has been removed.  This avoids a user unplugging an AC plug, touching the pins and getting an electric shock.

Fig 2: EMI Filter Schematic

If the line cord was removed from the AC plug at the peak of a 240V AC cycle, then the voltage would be 339Vdc.  The time for that to decay to a safe voltage depends upon the values of the X capacitor(s), the Line to Neutral resistor and any loading of the power supply converter circuitry.

IEC 60950-1 states that after one second that voltage has to be at a maximum of 42.4Vpk.  IEC 62368-1 states that for an X capacitance of 300nF or greater, after two seconds, the limit is 60Vpk in normal condition or 120Vpk in a single fault condition.

A 10W power phone charger will have a very small X capacitance and a high value discharge resistor. Under 62368-1, if that resistor failed (a single fault condition) the power supply would probably still meet the 60Vpk condition after 2 seconds.  From Figure 1, this still complies with no Basic Safeguard is needed, even for an “Ordinary Person”.

A higher wattage power supply on the other hand, will have a much larger X capacitance.  With a single fault condition, the loss of the bleed resistor may result in a voltage of greater than 120Vpk after 2 seconds.  That would then be an ES2 - Class 2 Energy Source - and from Figure 1, a Basic Safeguard would need to be in place.  Instead of just one bleed resistor, a second one would need to be added, in parallel with the first.  The values of the resistor would be sized accordingly to ensure that the loss of one, would still discharge the X capacitor to a maximum of 120Vdc after 2 seconds.

Supplementary Safeguards would be used in other parts of the power supply circuit in the same way as reinforced or double insulated is in 60950-1.

One side effect of the ES1 and ES2 limits is that IEC 62368-1 deems a 48V or even a 36V output power supply in hiccup mode over current as an AC output.  The ES1 limits for DC are 60V, but for up to 1kHz that drops to 30Vrms or 42.2Vpk.  IEC 60950-1 recognizes that as SELV, but under the new standard that would be ES2 and the output should not be made accessible to an Ordinary Person.  Pins and connectors not accessible by a blunt probe are exempt.

Power Guy

Friday, April 27, 2018

Where and why is a 277Vac input used in the USA?

The AC power supplied and used by household and commercial buildings in the USA is primarily 120Vac.  This is suitable for most electrical and electronic equipment in the home or office, but for the higher power requirements of electric driers, air conditioning and electric ovens, 208Vac or 240Vac is available. 

The 208Vac is derived by connecting across two phases of a “Y” or “WYE” three phase supply as shown in Figure 1.

 Figure 1
The 240Vac is supplied from a distribution transformer as shown in Figure 2.
Figure 2
In larger industrial properties, in addition to 208 and 120Vac, 480Vac three-phase is supplied to the building in a "WYE" (or Y) configuration as shown in Figure 3.

Figure 3 480Vac three-phase
The 480Vac three-phase is used to power larger electrical equipment like fork lift battery chargers, and heavy machinery.  Less current is drawn by the load at this higher voltage, enabling the use of lighter gauge wire, and the current is balanced equally in each phase.  This improves both power distribution and generation efficiency.

Measuring across each line to neutral we see 277Vac (480Vac divided by the square root of 3)  This voltage is used in the USA for powering lighting and HVAC (Heating, Ventilation, and Air Conditioning).  Using 277Vac instead of 120Vac reduces the current drawn by lighting fixtures by over 50%, enabling smaller gauge (lighter) and lower cost wiring.  Alternately more lighting fixtures can be connected to a 277Vac feed than with a 120Vac feed.  Power losses in the wiring is also reduced (P=I2 x R).
The awareness of the existence of 277Vac feeds in commercial buildings was very limited until the introduction of LED lighting in the late 2000s.  The AC-DC power supply industry until then had focused on products that operated between 85 – 265Vac.  As LEDs require low voltage DC current, LED “drivers” came on to the market, capable of operating from input voltages of 90 to 305Vac.  This voltage range covers 100V, 120V, 230V and 277Vac nominal inputs (with a regulatory +/-10% tolerance for line variations) to enable use world-wide.

Some of the control infrastructure for smart buildings is now using 277Vac for automatic window sun filtering, intelligent thermal monitoring and security.  Often the location of the controllers and monitors are in places where 120Vac is not available and 277Vac can be easily dropped down from above the ceiling tiles.

TDK-Lambda is one manufacturer that is providing industrial power supplies capable of operating from 90-305Vac.  These are an alternative to using constant voltage LED drivers that potentially can have short product lifecycles.  Engineering resources then have to be diverted to find alternative products.  Power supplies developed for the industrial market tend to be produced for ten years or more.

Below are two such products; the 2 to 4W rated pcb mount KAS series and the 40 to 65W rated chassis / DIN rail mount CSW65 series

Tuesday, February 27, 2018

Synchronizing the i6A DC-DC Converter Switching Frequencies

When multiple DC to DC converters are used to power sensitive circuitry, input and output noise can cause system issues, particularly when measuring very low signal voltages.  The problem is compounded if the converters’ switching frequencies vary with input voltage or output load. 

Even when the converter’s operating frequency is fixed, there will be a tolerance on the switching frequency timing circuit between the converters and just a few Hertz differences between the converters can cause sub-harmonic beat frequencies.

Filtering the inputs and outputs is one solution, but this can be complex over a wide frequency bandwidth.  If the DC-DC converters had a fixed switching frequency and could be synchronized to together or tied to a master clock, any board and/or system EMI filtering would be simpler.

Although somewhat uncommon, there are some DC-DC converters that have synchronization capability.  When the “full feature” option is specified on TDK-Lambda’s low cost, fixed frequency, i6A series of 250W non-isolated DC-DC converters, multiple units can have their operating frequency synchronized.

There are four ways to connect the modules:

1. Master / Slave, with no phase shift.  One i6A module is the “master” and the other modules will operate at the master’s switching frequency.  All the modules draw input current at the same time.


2. Master / Slave, with 180 degree phase shift.  One i6A module is the “master” and other modules will operate at the master’s switching frequency, but directly out of phase.  This can reduce the peak input ripple current from the supply, requiring less input capacitance.  Any module with a jumper between pins 4 and 33 will have the phase shift function activated.

3. An external clock is used with no phase shift.  All the i6A modules operate at the same frequency as the external clock (no master / slave). All the modules draw input current at the same time.
4. An external clock is used with a 180 degrees phase shift.  All the i6A modules operate at the same frequency as the external clock (no master / slave), but any module with a jumper between pins 4 and 33 will have the phase shift function activated.  This again reduces the peak input ripple current from the supply, requiring less input capacitance.


The i6A non-isolated DC-DC converters are a series of step down converters (the input voltage has to be higher than the output) in the industry standard 1/16th brick footprint.  All models feature wide range input voltages, as high as 9 to 53V and have wide range output adjustment from 3.3 – 15V to 3.3 – 40V.  Operating efficiencies can be as high as 98%.

Monday, January 8, 2018

Factors to consider when using 1U high power supplies in industrial environments

The height of rack mounted equipment is specified in multiples of “U”, where 1U is defined as 1.752” or 44.5mm in accordance with the Electronic Industries Association EIA-310-D standard.  The actual rack, or shelf as it is sometimes referred to, is slightly lower in height to allow easy insertion and removal without binding on adjacent shelves.  Commonly 1.721 inches (43.7 mm) tall is used.  The most common rack width is 19”, but is also available with a 23” width.  A wide range of standard electronic and electrical equipment is designed to be mounted into a 19” enclosure, including instrumentation, computers and power supplies.
Embedded (chassis mount) power supply height is often referred to as being “suitable for 1U applications” even though it is not offered with a rack enclosure.  In this case the product height will normally be less than 41mm (1.61”) to allow for the rack sheet metal thickness.

The communications industry has standardized on 1U high power supply racks to power datacenters and other datacom related equipment for networking.  These power supplies are frequently configured for redundancy so if one fails, it can be replaced without disturbing equipment operation.   The amount of power that can be delivered from a 1U rack can be 10kW to 14kW with four plug-in power supplies.
Why have these high density, cost driven products not found a great deal of popularity in the industrial segment?  There are several reasons for this:

Audible noise

To cool a 2kW to 3kW 1U power supply requires fan cooling in the form of two, high speed 40mm fans.  These emit very high audible noise levels at frequencies quite annoying to the human ear.  As datacenters rarely have human operators present for long periods of time, this is not an issue.  For industrial use where operators are constantly monitoring and using equipment, like test and measurement and analyzers, such levels of noise would not be considered acceptable.
Industrial mid power (600-2000W) power supplies tend to be 63mm in height for single phase inputs and use lower speed 60mm fan(s).  Three phase mid to high power (3000W+) 2U high units may be fitted with a low speed 80mm fan.  Temperature controlled fan speeds extend field life and dramatically reduce audible noise levels.
Fan life and contaminants

The faster a fan rotates, the quicker the life of the fan is reduced due to mechanical bearing wear.
The higher the airflow speed, the more contaminants (dust and dirt) will be drawn into the power supply.  Eventually this will block airflow and/or cause product failure due to circuit shorts if the contaminant is conductive.  Datacom equipment is usually situated in fairly sterile conditions, unlike industrial equipment particularly in harsh environments.
Operating environment

Datacenters have a controlled operating environment, usually air-conditioned.  This is not always the case in the industrial sector.
Redundant configuration of the power supply is usually mandated with datacom products.  In this mode, the power supplies are not run at 100% load in normal operation.  For example, with in a 3+1 2500W power supply configuration, where the maximum load drawn is 7500W, each of the four power supplies will provide 1875W and be running at 75% capacity.  Only if a unit fails will they operate at full load until the faulty unit is replaced.  Such derating of a power supply reduces internal temperatures, particularly those of the electrolytic capacitors, and improves its operating life.
The ability to operate power supplies in a redundant mode can be very important to some industrial users, to minimize production downtime.
The input supply to a datacenter is usually two dedicated feeds (again for redundancy), protected from AC line transients by switchgear and backed up by uninterruptible power supplies.  Industrial power supplies are subjected to input transients when large neighboring inductive equipment is switched on and off.
Datacom equipment power supply loading is usually well defined and relatively static with few severe load changes.  Industrial power supplies can be driving inductive DC motors, relays and capacitive loads, all of which can stress a power supply.
Product availability

The technology in the communications industry is constantly evolving and that includes the power supply architecture.  Datacom systems initially required 48V output power supplies and migrated to 12V to drive non isolated DC-DC converters.  Now datacenters are being run using high voltage 380VDC outputs.  Due to the vast amount of electricity that is consumed by these centers, an increase in efficiency warrants a change in power architecture.  This can result in early power supply obsolescence.
Industrial equipment is expected to be in service for 10 or more years.  Having to rework a system when a spare power supply is not available can lead to long downtimes.  Industrial power supplies are often produced for 15-20 years.

1U high datacom power supplies are used in some industrial applications where space is extremely limited, audible noise is not an issue and the environment is controlled.  Industrial power supplies are more likely to be subjected to uncontrolled environments and noisy AC inputs.  Product lifecycle and field reliability are key considerations to industrial users, who value long term availability and maintenance free operation.

Tuesday, November 7, 2017

Power Supply Overvoltage Category (OVC)

The IEC standards class Overvoltage Categories, which are sometimes referred to as Installation Categories, are as follows (most stringent to the least stringent):

Category IV
Used at the origin of the installation.  Examples are utility transformers, electricity meters, fusing and distribution panels.  High transient voltages are very likely.
Category III
Used in fixed installations and for cases where the reliability and the availability of the equipment is subject to special requirements.  These installations will have permanent connection to the distribution panel (hard wired).  Wiring impedance, fuses and circuit breakers somewhat reduce the level of voltage transients.
Category II
This covers energy-consuming equipment supplied from the fixed installation.  These would be items normally plugged in to a regular wall outlet or other plug in fixture requiring 115 or 230Vac.  The impedance of wiring circuits further reduces transient voltages to a lower level. Outlets, lighting switches and building connections more than 10m from a Category III source are classed at a Category II.
Category I
These are circuits requiring low voltage, which limit over voltage conditions to the appropriate level, i.e. protected electronic circuits.
Each category has to withstand a different level of voltage transient depending upon the nominal input voltage as shown in Table 1
Table 1: Tolerated transient voltage

The standard IEC 60204-1:2016 governs the safety of machinery. The general requirements apply to electrical, electronic and programmable electronic equipment that are fixed in their location.  The equipment covered by this part of the standard commences at the point of connection of the supply to the electrical equipment of the machine.

An industrial robot, or material forming machine, wired to the distribution panel would need to comply with Overvoltage Category III, as shown in Figure 1. If the AC-DC power supply inside of the robot controller was only OVC II, then an isolating transformer (not necessarily a step down type) would have to be fitted inside or between the controller and the distribution panel.  The impedance of the transformer would be enough to reduce the transient voltage level.

Figure 1: Levels of Overvoltage Category inside of a factory or facility

Some OVC III industrial AC-DC power supplies are available, like TDK-Lambda’s 240W rated 24V output ZWS240RC-24 power supply.
Although based on TDK-Lambda’s ZWS300BAF-24 which is OVC II, it has increased spacing from the Line and Neutral to ground, including additional spacing on the input connector.  It is also certified to EN 62477-1 - safety requirements for power electronic converter systems and equipment.  It is said to be a more suitable specification than EN 50178 which covers electronic equipment for use in power installations.

The use of such a power supply can eliminate the isolation transformer, saving both cost and space.

Friday, September 1, 2017

Power supply fan noise reduction

Unwanted audible noise is now part of modern human life, whether it is produced by people or machinery.  Although low level noise is not necessarily harmful, prolonged exposure can cause fatigue and stress related health issues.  This can also apply to the workplace where the use of fans to cool electronic equipment is becoming more widespread in order to make product size smaller.

Industrial, communications and medical equipment frequently use fan cooled AC-DC power supplies because of their small size.  Many equipment manufacturers though are being asked by their customers to reduce the amount of audible noise generated by their products.  Fans in medical equipment used in the proximity of the patient can delay or complicate recovery.  Engineers using laboratory test equipment or technicians operating analyzers do not want to be distracted by the noise of an irritating fan.

Some applications, like datacenters, have little human presence and equipment size takes priority. Here it is preferred to use 1U high (44.4mm) power supply racks and one or two high speed 40mm fans to keep the power supplies cool.  Although these fans have high airflow, they emit a great deal of acoustic noise at frequencies very annoying to the human ear.

The amount of noise a fan generates is related to their size, rotation speed, construction and how the air is travelling over or through the components it is cooling.  For the same airflow requirement, a smaller fan is much noisier than a larger one because it has to rotate at higher speeds to produce the same cooling effect.  Table 1 below compares a 40mm and a 60mm fan.  It shows that the 40mm version has 66% of the cross sectional area blocked due to the fan hub, whereas the 60mm has only 51%.

Table 1

The fan bearing type also will affect acoustic noise.  The quietest is the sleeve bearing, but is less reliable long term than a ball bearing type, especially at higher temperatures and care has to be taken with the mounting orientation.  Bearings do create vibration which can affect the system performance of say a digital microscope or scanner.  A lower speed fan can reduce this.

Fan blades are often designed to create turbulence in the airflow, which increases acoustic noise as it passes over the components and heat sinks inside the power supply. When high velocity turbulent air comes into contact with a physical object, it can create a highly annoying audible tone and increase the noise by up to 10 dBa which equates to a doubling of the perceived loudness.

To protect operators and service technicians, a finger guard is usually fitted to the fan.  As the guard is in the path of the airflow, it will also create acoustical noise.  Table 2 shows typical levels of noise for two fan guard types when used with a low speed fan.  The wire grill (with circular wire) offers the best balance of protection and acoustic noise, but is more costly and involves manual labor to assemble compared to a machine punched pattern in the power supply enclosure.  Of course, the noise levels rise as the speed of the fan is increased.

Table 2

Careful consideration should be taken when designing the internal construction and layout of the power supply.  Obstructions to airflow will reduce its cooling effect, resulting in the need for a higher speed fan and hence higher noise.

Variable speed fans are growing in popularity, as quite often a power supply is not running in a hot environment, or running at full load. Not only will audible noise be reduced, the fan will last longer. Sensors are used to measure a heatsink temperature or other hot component to determine if more or less air is required. The circuit must have sufficient hysteresis to avoid constantly changing fan speeds, as this can be more annoying that a fixed speed, higher noise fan.  TDK-Lambda’s recently launched QM series of modular power supplies senses the incoming air temperature.  This enables the fan to run slower and quieter at room temperature, but faster at higher temperatures, where humans would not normally be present.

The amount of wasted heat is determined by the efficiency of the power supply. The graph in Figure 1 shows that with a 90% efficient 600W output power supply, only 67W is generated, compared to 115W unit with an efficiency of 85%.  Increasing efficiency enables the use of quieter, slower speed cooling fans.

Figure 1

TDK-Lambda’s 91% efficient 1200 to 1500W rated QM7 series utilizes two slow running 60mm fans to further reduce audible noise.

The QM series design team performed extensive audible noise testing on current TDK-Lambda products, the QM7 and competitive models using the BS EN ISO 3744:2010 standard (Acoustics: Determination of sound power levels and sound energy levels of noise sources using sound pressure).

The results proved that the steps taken in the QM7 design had significantly reduced fan noise, with the 1500W rated QM7 measuring a low 44.3 dBa. Other products measured were as high as 58 dBa. Note again, a 10 dBa which equates to a doubling of the perceived loudness.

Twenty one employees were asked to take part in a “blind” acoustic study, listening to the individual models and rating them on loudness and how annoying they were.  The tests again revealed that the QM7 came out as the best model.

The QM series is the latest in a 37-year legacy of modular power supplies, beginning with the invention of the ML series, a world first in 1979.  Having both medical and industrial safety certifications, the QM is very suitable for a wide range of applications, including BF rated medical equipment, test and measurement, broadcast, communications and renewable energy applications. With a wide range 90-264Vac, 47-440Hz input the QM7 can deliver 1200W, and 1500W with a 150-264Vac high line input.  Up to 16 outputs can be provided, with voltages ranging from 2.8V to 52.8V, with additional higher voltages of up to 105.6V in development.

Wednesday, May 24, 2017

Medical power supplies meeting IEC 60601-1-2 4th edition voltage dips and interruptions

Customers call TDK-Lambda wanting their medical product to meet the strict IEC 60601-1-2:2015 4th edition immunity standard, and ask us if our medically certified power supplies fully comply.  In particular, concerns are raised about meeting the section dealing with voltage dips and short interruptions to the AC supply.
IEC 60601-1-2 is derived from the IEC 61000-4 standard, which covers Electromagnetic compatibility (EMC).  The testing and measurement methods are very similar, but some of the test levels for dips and interruptions in the section based on IEC 61000-4-11 are much tougher.  The interruption test of removing the AC supply for 5 seconds, without the loss of the output, is almost impossible without a custom solution with some form of battery back-up.  One may well question why are standard medical power-supplies being sold if they do not meet that standard.
Firstly, power supplies are not classified as medical devices, it is the customer’s product or system that is the medical device.
Secondly the term “essential performance” used in the standard has to be examined.  In the 3rd edition of IEC 60601-1 it is defined as “the performance necessary to achieve freedom from unacceptable risk”.  To clarify, the designer/manufacturer has to determine if a loss of performance or functionality of their medical device product or system will result in an acceptable risk or an unacceptable risk.  That risk is the potential to harm a patient, operator or the environment.  Analysis must be made of the probability or the frequency of an event happening compared to the severity of that event.
Let’s give a simple example.  Diabetics check their blood glucose level on a regular basis and most use a handheld battery-operated meter that accepts disposable test strips.  If that meter was to stop working, say due to a faulty display, it would be classified as an acceptable risk.  Replacement meters are readily available from supermarkets and pharmacies and a short delay in testing would not normally cause harm.  An unacceptable risk would be if the internal sensor measuring the blood glucose level was to produce incorrect readings and the diabetic administered too much or too little insulin.
Power supplies, although not classified as medical devices, can have an impact on the IEC 60601-1-2 immunity performance of the device they are powering.  For the voltage dips and interruptions section of the standard, there are five tests performed.  Table 1 below shows the input voltage dip and the duration.  100Vac input and 50Hz conditions are shown as they could represent the worst case.
Test results are judged against four performance criteria levels:
Performance Criteria A – ‘Performance within specification limits’
This is the best result.  A very slight drop in output of a few milli-volts (within the regulation limits) should not cause the end device to malfunction.
Performance Criteria B – ‘Temporary degradation which is self-recoverable’
Criteria B is usually acceptable in the majority of cases.
Performance Criteria C – ‘Temporary degradation which requires operator intervention’
This would be classified as unacceptable from a user point of view, without even considering a risk analysis.  If the AC power was interrupted and the power supply had to be reset by a patient or operator, it would be much too inconvenient.
Performance Criteria D – ‘Loss of function which is not recoverable’
Criteria D is really a “fail” test result.  If a power supply is damaged and needs replacing after the test, it is very unlikely that a product with this performance level would be placed on the market.
AC Input Voltage
Actual Voltage Dip for 100Vac nominal
Voltage Dip by AC Input Cycle
Voltage Dip Time Period for 50Hz
Suggested Performance Criteria Level
Dip down to 0%
0.5 of a cycle
Dip down to 0%
1 cycle
Dip down to 40%
10/12 cycles
Dip down to 70%
25/30 cycles
Dip down to 0%
250/300 cycles
5000ms (5s)
Table 1: Test Levels
Referring to Table 1, most power supplies will pass the first two tests with a Performance Criteria level A with some output derating to increase the hold-up time.
The third and fourth tests requires the power supply to continue to operate for 200ms when the input drops to 40% of nominal or for 500ms at 70% of nominal.  Criteria A could be achieved by having the power supply’s low voltage input protection circuitry modified to allow the power supply to operate at the lower input voltage for a short time.  As the AC input current will be higher, it is best to ensure that the power supply is not operated at full load.  As hold-up time is related to the actual output power drawn, operating the power supply at 50% load will result in a significant “ride through” capability during the interruption.
The fifth test of a 5 second interruption to the AC supply is usually met with the installation of battery back-up or a UPS (Uninterruptible Power Supply).  Adding sufficiently large energy storage inside the power supply would result in a significant increase in size.
In summary, the medical device designer/manufacturer must decide which performance criteria is needed, based on their risk analysis to meet IEC 60601-1.  Unless continuous performance is critical, most manufacturers will opt for the criteria in Table 1.

Popular Posts

TDK Corporation - Americas | Press Release