TDK Lambda Spotlight - Why use a modular power supply with wide range output voltage adjustment?

Modular multiple output power supplies are usually factory assembled using pre-manufactured and pre-tested output modules, converters and subassemblies. The user only needs to determine their required output voltages and currents, signaling and overall power level. This type of product is primarily used as a centralized power supply to deliver multiple output voltages throughout a system.

One challenge that Engineers face is finding a power supply that can provide all the different voltages necessary for these complex systems. Often voltages outside of the nominal 3.3V, 5V, 12V, 15V, 24V and 48V are required for the optimum operation of sensors, additional circuit cards and motors. Output voltages may also need to be set higher to offset voltage drops in the cabling or series FETs for system redundancy. TDK-Lambda has been designing and manufacturing modular power supplies since 1979 and has fulfilled hundreds of requests for these “non-standard” voltages, often just a few volts above what is considered “nominal”.

If a modular power supply series has a narrow adjustment range or a small offering of modules, there is the possibility during the system development that a particular output voltage cannot be provided. This requires either the development of a modified standard part, a DC-DC converter fitted on a circuit board or an alternative power supply chosen to provide these required voltages.

TDK-Lambda’s Vega series addressed this challenge by having a significantly large choice of modules available, coupled with high technology and lean manufacturing processes for rapid delivery. The output modules use amorphous core (magnetic amplifiers) technology to provide a wide range adjustment.

The QM modular series was developed to address the requirement for low audible noise. Medical technology complexity has risen sharply in recent years, requiring more power and a need for forced air cooling. Audible noise can delay or complicate patient recovery and cause fatigue for the operator. A smaller number of modules were developed addressing the popular voltages and power levels while minimizing cost.

Figure 1: QM 550W-2000W module power supplies

TDK-Lambda’s most recent modular series, the MU4, is 41mm high (to fit inside 1U high enclosures), to meet industry requirements for lower profiles. The topology used directly addressed the issue of output voltage coverage by providing a seamless output voltage range from 3.3V to 104V (Figure 2).

Figure 2: MU4 adjustment ranges

1U high power supplies often generate higher levels of audible noise, as the fan has to rotate faster to move the same amount of air as a product that uses a larger diameter fan. The MU series uses an intelligent microcontroller algorithm (patent pending) to monitor the temperature of the primary converter and each output module, allowing the fan speed to be adjusted for optimum cooling, hence reducing audible noise.

Figure 3: MU4 600-800W 7-year warranty modular power supply

The voltage regulation specifications of the outputs are also important. Wide voltage variations can lead to errors, malfunctioning, or performance degradation in electrical devices. The MU series regulation is < 1% for 0-100% load changes and < 0.1% for 90-264Vac input line changes, to maintain stable output voltages to ensure consistent operation.

In summary, providing modular wide-range output voltage adjustment caters to the diverse requirements of electrical devices and ensures reliable operation across numerous applications. Other benefits of modular power supplies include lower system cost, a smaller system footprint, and the elimination of multiple power supplies and the additional noise and EMI generated that needs to be filtered.

To facilitate easy configuration of modular part numbers, TDK-Lambda has developed the on-line Quick Product Finder or contact us at sales@odonnell.com

Why use a modular power supply?

TDK-Lambda’s history with the design and manufacture of modular power supplies began in 1979 with the launch of the ML series - a world first. Over the next 40+ years new modular products have been developed and launched addressing medical applications, the need for higher power, communication options and more recently, lower audible noise. The latest product, the MU4 series, was announced in November 2022 featuring a low 1U profile and very low acoustic noise.

What is a modular power supply?

As the name implies, modular power supplies are factory assembled using pre-manufactured and pre-tested output modules, converters and subassemblies. The user only needs to determine their required output voltages and currents, signaling and overall power level. TDK-Lambda utilizes an on-line configurator (Quick Product Finder) to optimize the module selection and provide a unique part number for quoting and ordering purposes. Once a configuration is determined and an order placed, the assemblies are mounted into a chassis, given a final test and shipped.

Modular products are available from around 400W to upwards of 2,000W and offer 1 to 18 outputs.

Figure 1: 4th generation Vega series launched in 2002 showing the modular concept

The advantages of using a modular power supply are numerous compared to a modified standard power supply or a complete custom design. I shall address these and include some examples.

No production set-up charges

Most power supply companies offering modular power supplies do not charge a set-up or configuration fee. Sophistication in a product selector can be tied to the manufacturing software to electronically configure a bill of materials and issue a work order to production. The pre-built assemblies stocking levels can be integrated in a just-in-time Kan-Ban system.

Automated Test Equipment (ATE) is flexible enough to fully test any combination of modules, signals and output power. The configurator software can be used to automatically program the ATE, often by reading a barcode on the production work order. The product rating label printer can be tied to the ATE to ensure a product has been tested thoroughly and successfully; a label cannot be printed unless the supply 100% meets the specification

No minimum order requirements

A modified standard power supply will often require modified printed wiring boards, custom transformers and magnetics, which in turn will likely require a MOQ from the component manufacturer. The use of modular power supplies with standard assemblies removes this requirement.

No engineering charges and flexibility

Modifying a standard product will often require non-recurring engineering charges. Using a modular power supply avoids this. If the system load requirements change during development, requiring more current or a different output voltage, another configuration can be easily created.

Safety certification

Most modular products have certification to the medical (IEC 60601-1) and industrial and I.C.T. (Information and Communication Technology) (IEC 62368-1) standards. The safety reports cover all the possible configurations, which takes a considerable effort and money by the manufacturer, but full series compliance ensures that any unit ordered will automatically have safety logos applied to the rating label.

Wide range output adjustment

Depending on the power supply topology that the manufacturer uses, the output voltage may be capable of adjustment across a wide range. Figure 2 shows the adjustment range of TDK-Lambda’s new MU4 series, this also provides seamless availability of any voltage from 3.3 to 104V in the configurator.

Figure 2: MU4 output voltage ranges

Figure 3: MU4 1U high modular power supply

No minimum loads, improved regulation and isolated outputs

A standard multiple output (non-modular) power supply will have a lower cost basis, due to its single board construction and one power transformer with multiple windings. The control circuit will use the highest power output for control and regulation purposes (via an optocoupler) to the converter control IC. The lower power outputs will often have increased load and cross regulation, requiring a minimum load on the main output. All the outputs may have a common 0V (return) connection which pre-determines their polarity – either positive or negative with respect to the 0V.

Modular products typically have independent control for each output voltage, this results in tighter regulation and allows for isolation between each output. This avoids the need for any minimum load requirements and allows the user flexibility to determine if the output is connected as a positive, a negative voltage or remain completely isolated from the other voltages in their system.

Signals and remote on/off functions

If the manufacturer offers signals or the ability to turn outputs on or off, this further increases flexibility. For example, individual outputs or the entire power supply can be turned off if the DC Good signal on a module displays an overload or fault condition.

Utilizing multiple single output power supplies

Another alternative to using a modular power supply is to use several single output power supplies. This may reduce the BOM cost but does come with some disadvantages and an increase in assembly time.

Each individual power supply will have its own leakage current, which is additive. It may not impact a system that is hard wired to the AC source, but certainly will impact the low leakage current requirements for medical and even some systems using a plug in AC line cord.

Similarly with meeting EMI and EMC. Multiple power supplies each have their own electromagnetic footprint, whereas the performance of a modular power supply is predetermined.

Cooling is simplified as the modular product can have an integral fan or fans. Modular products like the MU4 and QM series have been designed for low audible noise with sophisticated cooling techniques and algorithms. Mechanical assembly and wiring are easier with one power supply.

In summary, if you require more than one mid-power output in a system, it is certainly worth considering a modular power supply. Try the TDK-Lambda on-line configurator for yourself.

For more information on TDK Lambda’s products please contact O'Donnell Associates North Inc. - sales@odonnell.com

Your golden ticket to liquid cooling – enter through the rear door

Could rear door heat exchangers be the missing link for retrofit?

A common fallacy has begun to emerge in the data center industry. As we move towards ever-increasing workloads, fueled by the rise of artificial intelligence (AI), there has been a march towards liquid cooling as the zenith for keeping your precious multi-million dollar racks running as they should. However, an assumption persists that retrofitting liquid cooling involves expense, construction work, and possibly even downtime.

Stefan Djuranec, nVent’s global product manager, DNS cooling, has a better idea. We caught up with him to talk about the advantages of rear door heat exchangers (RDHX) as a near plug-and-play alternative that brings the advantages of liquid cooling to new builds but can also be easily incorporated into existing facilities.

At its simplest, and as the name suggests, an RDHX, or rear door cooler, sucks in warm air generated by equipment, cools it, and expels the chilled air out the rear door. The main difference is that, rather than being processed in a specially equipped area of the data center, the cooling is done right next to the rack itself. Djuranec explains why this is an advantage:

“With traditional CRAH or CRAC (Computer Room Air Handler/Conditioner) units, the cooling is far from where it’s needed, so you must duct the air to and from the cooling equipment. With rear door cooling units, they’re right there, in the row of racks – no long pathways, no air ducting.”

Rear-door coolers use elements containing water-glycol solutions. Whereas air is a natural thermal insulator, water is a more effective transport medium for removing heat. Djuranec tells us:

“We’ve used air in the past because air is a good transport medium in that it’s practical, non-conductive, and chemically stable, but water is a much more effective means to transport heat. You don't need to transport the heat with air over large distances. You transfer the heat from the air directly, at the rack level, into a better medium, which comes with efficiency benefits and energy savings.”

Efficient synergy

One of the efficiencies is that the chilled air exiting the RDHX acts as a natural chiller for the rest of the room, meaning that as well as the concentrated cooling for each rack, it lowers the ambient temperature of the data hall for lower-intensity workloads. With the right guidance, the implementation of RDHX solutions for intense workloads can create a potential saving on conventional cooling.

While opinions vary in terms of exactly who needs to switch to liquid cooling, Djuranec suggests that its efficiencies could make a switch to an RDHX advantageous in any scenario as it offers futureproofing for a time when traffic inevitably increases in the years to come:

“It can be a high-performance computing solution, but it can also be a high-efficiency cooling solution. Especially when, as in recent years, energy costs are rising throughout Europe and the world. The rear door cooler can run with very high water temperatures (compared with CRAH units or in-row coolers) and has a high gain in cooling efficiency. There are two major applications – either you require high performance for cooling or the efficiency gains that rear door coolers can bring.”

This has opened the advantages of rear door cooling to a much wider range of data center clients. Djuranec continues:

“Up to now, high-performance racks have been more utilized in science computing in industrial applications where they are needed to run high-performance simulations. It's a niche environment compared with the broader market of the average consumer and user AI.

“Things have changed drastically because AI is now going into many data centers. Many universities and companies are playing with this technology, and AI requires high-performance racks. It has a lot of attention now in the industry, and these racks require cooling environments where the rear door cooler fits very nicely.”

Retrofitting Liquid

This goes some way to busting the myth that a liquid-cooled data center needs to be built from the ground up. nVent’s range of RDHX products measures a mere 281mm in breadth, meaning they can be placed on both sides of the data center aisle without preventing access for maintenance or compromising health and safety.

With both active (fan-assisted) and passive units available, there are a range of options for making RDHX units work in any environment. This also makes them an excellent consideration for facilities looking to consolidate into a more densely populated data hall. More importantly, they can be installed in a live environment, as Djuranec explains:

“We’ve seen many retrofit applications, and it's not tricky to apply. There might be challenges to overcome, but they have always been solvable. With traditional cooling, it takes some effort to allow liquid plumbing directly to the vicinity of the racks, but if you have a raised floor, you would need to run this pipework through the raised floor.

“If it's underneath the ceiling, you can install the pipework in a live environment, and that’s a common use case. For racks where you want to apply rear door coolers, typically you would start with freeing up space for those high-density racks. You would not usually go with racks from a system that is already up-and-running, but you can work on a live data center running in parallel. It's also simple to add more equipment for rear door coolers.”

Maintenance Mythbusting

Djuranec busts another myth as he explains that, contrary to the popular perception of liquid cooling, using an RDHX requires very little extra maintenance compared with traditional air cooling systems:

“Rear door coolers require similar amounts of maintenance as CRAC units and CRAH units. You need to check the water quality regularly, you need to check for leaks, you need to check that things are still fitting tightly, and so on. The difference is that it’s close to the rack, which allows for swapping components quickly and easily.

“Our rear door coolers work with hot-swappable fans that can be changed without tools or training technicians. You can even swap power supplies when they fail without a trained electrician or maintenance personnel. Take the old, failed power supply out, and put the new one in. We call it ‘true-hot-swappable’ because the process can be completed in around 30 seconds.”

“We are unique in the market because we even have a hot-swappable controller. If it fails, the cooler will go into emergency cooling mode. The fans will go on into a default fan speed and the valve will open. It will not provide 100% cooling, but it will provide a minimum level to keep the IT gear running until the problem is solved.”

The other angle to maintenance is the learning curve. Without the talents of well-trained maintenance technicians, no data center will maintain its uptime. So is it difficult to learn how to keep an RDHX running smoothly?

“There are specific things that need to be learned – what about condensation? How does the system react to fan speed control, valve control, and so on? In the first couple of weeks, while technicians get familiar with the device, you will get more questions, but later on, it's silent, especially when a controller is defined or designed nicely and intuitively.”

Plan early, plan with partners

Regular readers of DataCenterDynamics will know that one of the most common pieces of advice we’re given is that the earlier you can engage with partners like nVent, the better the outcomes. We’ve spoken mostly about retrofitting thus far, but that doesn’t mean a rear door cooler doesn’t represent the best option at the blueprint stage of a greenfield build, whether for futureproofing or as part of a hybrid array. Djuranec tells us why maximum benefit can be gained from carefully considering your cooling requirements at the design stage:

“If you plan with rear door coolers from the start, you get instant benefits even from smaller workloads. As they can run at high temperatures, you need fewer or smaller chillers, utilize dry coolers, or have operative cooling as much as possible.

“You can already plan with higher power per rack and plan your electricity accordingly, meaning you don't need to upgrade to a higher service later. As with many other applications, it is much easier to plan for higher loads, and then work with lower loads. If you do it the other way around, it's trickier.”

Djuranec warns of a common trap – it’s impossible to work with multiple workloads using liquid alone, and the dangers of retrofitting air cooling are as manifest as not planning for liquid. A modern data center needs a hybrid solution, planned from the outset, for maximum efficiency.

Granular Control

The secret to controlling cooling to those varying workloads lies in an ability to change settings at a granular level. nVent’s rear door cooling units are ‘smart’ by design, with sensors for measuring air and water temperatures, fan and valve health, differential pressure, and fan speed, amongst many others:

“Everything can be monitored, fully remotely. In addition, we provide integration into our Sunbird data center monitoring environment. You can easily make our coolers work in those environments. The standard languages we support are easy to integrate into the end user’s environment. It's very intuitive to operate and work with.” says Djuranec.

In finishing, we look at the thorny issue of waste heat and sustainability. With more countries limiting power available to data centers, RDHX is, according to Djuranec, a good decision for power-conscious operators:

“In some countries, they request a maximum power usage effectiveness (PUE) value for a data center to approve it. Some territories are already limiting the power available for data centers. If you're restricted, you want to go as efficiently on the cooling side as possible to use your space and power for the computer gear.”

Heat reuse gateway

He also points out the value of RDHX as a means of providing efficient heat reuse, another increasing requirement for newly built data centers:

“The value of heat is defined by its temperature. You can’t do much with low-temperature air or water. If you increase the temperature, you can heat water for people to wash their hands or bathe. If you increase the temperature of the return water, then you can drive chemical processes with it.

“Rear door coolers have a high heat exchange surface, high performance, and work at comparatively high return temperatures, which lifts the excess heat to a level usable for district heating, household heating, or industrial processes.”

When all is said and done, Djuranec makes a powerful case for considering RDHX in your plans, whether for retrofit or greenfield. It isn’t the be-all-and-end-all of liquid cooling, but for many situations, it could prove to be the holy grail for a market that continues to evolve at breakneck speed.

For more information on nVent Schroff products, please contact O'Donnell Associates North, Inc. - sales@odonnell.com