Wednesday, July 29, 2015

Using Droop Mode Current Share Power Supplies

Connecting power supplies in parallel is commonly used to increase the available output power or to provide system redundancy in the event of a power supply failure.  The correct and reliable way to connect two or more power supplies in parallel is to have them equally share the load current.

If “brute force” current share is used (connecting power supplies in parallel without regard to load sharing), it can lead to one or more of the units operating in overload and a reduced field life due to overheating.  Here is an earlier article I wrote: http://power-topics.blogspot.com/2013/09/brute-force-parallel-of-power-supplies.html

There are two main (analog) techniques for getting power supplies to share the load - “active” and “droop” mode. 

Active current share uses a signal wire interlinking two or more power supplies.  Simply put, the voltage on the wire is proportional to the total current supplied to the load.  That voltage is used to “inform” a unit that it is not contributing enough current, thus raising its output voltage to produce more current.  In telecom and datacom systems, the power supplies are usually rack mounted and plugged in to a pcb backplane.  The current share connection is therefore identical from installation to installation.

On the other hand, industrial applications tend to use hard wired DIN rail mount power supplies.  The cable routing can change dramatically between installations.  Cable lengths can be quite long causing the current share signal connection to be vulnerable to EMC interference from motors and relay switching.  One solution that does not need any current share connection is droop mode.

Droop mode is a very simple way of paralleling power supplies.  The output voltage drops (droops) in proportion to the current drawn from the power supply.  If one power supply is supplying more current than the others, the output voltage will fall and load balancing will occur.  Certain electronic loads can be sensitive to variations in the supplied voltage (3.3V or 5V for logic ICs for example), but typically 12V, 24V or 48V outputs drive relays, DC-DC converters or motors and are more tolerant to droop mode current share.

Very often this feature is standard on DIN rail power supplies 100W and above and is enabled by either a switch, or like on TDK-Lambda’s DRF series (shown below), the removal of a wire link or jumper on a connector.
TDK-Lambda DRF DIN rail power supply

When the DIN rail power supply is not operating in parallel, the switch is closed (or wire link kept in place).  The internal control circuit will compare the output voltage to a reference (Figure 1).  Any change in the output voltage due to load will be compensated for, and the output regulation will be minimal – in the order of 10mV.
Figure 1

When two or more power supplies are connected in parallel, the switches (or links) are opened (Figure 2).

Figure 2

As the output current increases, the voltage across the shunt will add to the voltage across the output, causing the control circuit to compensate and lower the power supply’s output voltage.

As an example, for the 24V 10A TDK-Lambda DRF240-24-1 power supply, the droop characteristic is at 64mV / A.  Table 1 shows the output change against load when droop mode current share is enabled.


Output Voltage
Output Current
24.00V
0A
23.84V
2.5A
23.68V
5A
23.52V
7.5A
23.36V
10A

Table 1

When using power supplies in droop mode current share, care must be taken to:

1. Set the output voltages of the power supplies to the same voltage.  The output voltage can be adjusted slightly higher if needed, to offset for the droop voltage*

2. Use the same length and same gauge of wire from the output to the load for each unit

3. Note any additional de-rating stated by the manufacturer.  This avoids tolerances from overloading a unit

4. Do not exceed the manufacturer’s recommendation for the number of power supplies that can be paralleled.

5. Make sure the parallel switch is in the right position, or the wire link is removed!
* If the power supplies are being used in redundant mode with ORing diodes, set the output voltages higher to compensate for the forward voltage drop (Vf) of the diode.  The function of these diodes is to prevent the bus voltage from being pulled down due to an internal short in a faulty power supply.  The diodes should be rated to carry the full output current of the power supply.

Redundant mode connection with ORing diodes


Power Guy

Tuesday, June 30, 2015

Comparing DC-DC Converter’s Usable Power


Power supply manufacturers rarely use the term “usable power” in their AC-DC product literature, but it is used frequently when referring to a DC-DC converter’s performance against temperature. Is it just a fancy reference to a de-rating curve? As usual, before we answer that question, we need to look a little deeper.

An AC-DC power supply, like the TDK-Lambda’s LS50 series, has a de-rating “curve” as shown below. It can deliver full power at 50C ambient and it de-rates linearly to 70% load at 70C. (The knee points of the chart vary from product to product but not normally dramatically between competitors of like products.)





The chart is very simple because the LS50 does not require any forced air, and has a metal case that is used as a heat sink and to provide a level of physical protection.

Looking at TDK-Lambda’s iQG ¼ brick DC-DC converter, we can see a much more complex set of de-rating curves.




To be fair, the industry standard ¼ brick has migrated from a product where 50W output power was leading edge, to products that are fast approaching 1000W. The emphasis for DC-DC converters has been put on package size. Even when fitted with an integral baseplate, like TDK-Lambda’s iQG ¼ brick (shown below), the iQG’s volume is 1.6 cubic inches, compared to the LS50’s 21 cubic inches. That is 10 times the output power in less than a 1/10th of the volume.






The “brick” style DC-DC converters are designed to be either conduction cooled (to a cold plate), or forced air cooled, often without an external heat sink. The rate of airflow available will depend on the user’s application, and so a number of performance curves are provided. It can be noticed that in some cases for low airflow conditions, de-rating has already occurred already at 30C ambient.

Usable power really refers to the slope and start point of the de-rating curve. Often Engineers will focus on the output current of the converter, and choose a higher power, more expensive product, expecting to significantly better performance. This is where “usable power” comes into play.

Below is a simplified pair of curves for 2m/s airflow. The blue line is for a 12V 67A (800W) DC-DC converter, and the green line for the TDK-Lambda 12V 42A (500W) converter. The 800W model is 1.6 times more powerful at low ambient temperatures, but in the yellow area at higher ambient the ratio drops to 1.35 times at 70C and 1.24 times at 75C. (Typically customers operate DC-DC converters in the 65 to 80C range.)


Although the 800W converter has more available power, the 500W unit has more usable power, demonstrated by a much less steep de-rating curve. It can be seen that at higher ambient temperatures, it would be more cost effective to use the 500W converter.

Wednesday, May 27, 2015

Understanding convection cooled power supplies

There are a number of commonly used terms to describe cooling methods in the power supply industry:
  • Fan cooled - Unit has an internal fan
  • Convection cooled - Unit requires no fan cooling
  • Forced air cooled - Unit requires external airflow
  • Conduction cooled - Unit relies on a cold plate to remove the waste heat
The most misunderstood and hence most misapplied is probably convection cooled. Many Engineers assume that a convection cooled power supply is one that does not need any airflow to operate.

One definition of convection is “The transfer of heat by the circulation or movement of the heated parts of a liquid or gas”. In our case – the circulation or movement of hot air.

Open frame power supplies, for example, are typically mounted on a flat surface upon standoffs, and below, we can see how the air behaves.
 

As the hot air rises, cooler air is drawn in from the sides. Although the airspeed is quite low, just 0.3m/s, it is sufficient to reduce internal temperatures. During the safety certification process for the power supply, this is taken into account during thermal testing.

It is very important to ensure that there is adequate space for the air to be drawn in from the sides and allowed to exit above the power supply. A distance of 50mm is considered adequate.

Orientation of the product is also very important. Most manufacturers will state a recommended mounting orientation and any de-rating associated if that is not followed. Mounting the product upside down for example can severely reduce field life unless heavy de-rating is applied, and is often forbidden.


The ramifications of mounting the power supply vertically should also be studied. Ideally the input (bulk) and output electrolytic capacitors should be located at the bottom, where the temperature will be the coolest.


If in doubt, consult the manufacturer’s installation manual. For high density products, recommended maximum component temperatures will be advised for critical parts.

Power Guy


Thursday, April 16, 2015

Reducing noise on open frame power supplies

We get a lot of questions on how to reduce noise, both output & EMI (ElectroMagnetic Interference), on open frame power supplies.  Usually it is a result of a failure to ground the product correctly.  With an enclosed power supply, encased in a metal box, it is simple as all the connections are made for the user by the chassis.  Connect up the input and output wiring and everything works fine.  With an open frame (pcb type) it is a little different.

TDK-Lambda’s ZPSA open frame power supply

First a look at what we are aiming to do.  Below is a simplified diagram of the noise decoupling capacitors in a typical power supply.  The Y capacitors on the left provide a low impedance path for high frequency noise to ground.  This avoids electrical noise (EMI) exiting the power supply and interfering with other devices on the AC input.   The capacitors on the right have the same function, but in this case stops electrical noise from appearing on the output of the power supply and interfering with the load.  In some cases, just one capacitor is sufficient.


You can see two blue Y capacitors on the ZPSA photograph, close to one of the mounting holes.


Looking at the underside of the ZPSA pcb, we can see the locations of those capacitors.

The red arrow shows the Y capacitors CY2 and CY3 are connected to a common trace that leads to the bottom left mounting hole.  This hole is in fact a plated through hole and the mounting screw and standoff will make a connection with that trace.
The blue arrow shows the output to ground capacitor CY1, again connected to a copper trace leading to the bottom right mounting hole.

Note that in the case of the ZPSA, there is no pcb trace between those two holes.  (The top two mounting holes do not have any traces going to them, so we can ignore them electrically.)

Looking at our schematic again, we need a connection from chassis ground to both the input and output capacitor traces to reduce the electrical noise.  This we do by mounting the power supply on a grounded metal plate, with metal standoffs and screws.


Follow these guidelines and the open frame power supply will meet the EMI and output noise specifications.

Power Guy

Monday, March 9, 2015

Should power supply manufacturers adopt IEC 62368-1:2014 early to replace IEC 60950-1 and IEC 60065?


In 2014 the IEC introduced edition 2 of the Hazard-Based Safety Engineering (HBSE) IEC 62368-1 standard.  Intended to replace new submittals for IEC 60950-1 and IEC 60065 in June 2019, it covers hazards and hazard prevention for ITE (Information Technology Equipment) and audio and visual equipment.
UL & CSA published the 2nd edition of the IEC 62368-1 standard on December 1st, 2014, following Europe’s publication in September 2014.

The current standards are known as “prescriptive” and closely control product design.  The new standard retains some of those design rules, but offers performance options that might allow some design flexibility.  It must be stressed that unlike IEC 60601-1, a formal risk analysis is not required.  It is also not a simpler merger of 60950-1 and 60065.
No date has been set yet for when “grandfathering” of the 60950-1 and 60065 standards ends.  It has been recommended by some test authorities that if a product is going to be removed from the market by 2018, then do not proceed with certifying to IEC 62368.  As there may be some regions that have not accepted the new standard, products that will be on the market after 2018 should look at obtaining certification to IEC 62368, but keep the existing 60950-1 and 60065 certifications.

Does this mean that power supply manufacturers should start transitioning over now?  No, the IEC committee recognizes this would impose a huge burden on everyone’s resources, including the certification bodies.  Clause 4.1.1 states that “Components and subassemblies that comply with IEC 60950-1 or IEC 60065 are acceptable as part of equipment covered by this standard without further evaluation other than to give consideration to the appropriate use of the component or sub-assembly in the end product.”
One snippet of information I found this week; ever wonder where the numbering convention of standards comes from?  If a standard is IEC 50xxx it is based on a standard of CENELEC (European Committee for Electrotechnical Standardization) origin.  If it is IEC 55xxx, it is based on CISPR (Special international committee on radio interference) origins.  For standards IEC 6xxxx, they originate from the IEC (International Electrotechnical Commission).
Power Guy
There was an error in this gadget

Popular Posts

TDK Corporation - Americas | Press Release