LpR Article | Jul 04, 2016

DC-Grids - Challenges and Chances for LED Lighting by the Fraunhofer Institute and LED professional

At the beginning of November many important aspects of DC Grids were discussed at the Fraunhofer Institute for Integrated Systems and Device Technology in Erlangen at the European Center for Power Electronics (ECPE) conference. Besides fundamental issues, very specific ones like how the introduction of DC Grids affects lighting technologies and especially LED lighting were part of the program. Below, Arno Grabher-Meyer from LED professional gives a summary of the speeches and expert discussions and gives us an idea of what the key ideas could mean to the LED lighting community.

Since the dispute between Thomas Edison and N. Tesla more than 125 years ago, the question of whether AC or DC grids are the better approach arises from time to time. By the end of the 20th century it looked like AC had clearly won the race. But it wasn’t true then and probably won’t be true in the future. Before the 1950’s AC and DC grids existed in parallel due to de-centralized electricity generation. From the 50’s onward, centralized electricity generation became the standard and AC grids started to dominate. Since the beginning of the new century new technologies have brought DC back to the attention of technicians, economists, ecologists and politicians. Today, AC and DC grid advocates discuss this topic just as passionately as Edison and Tesla did between 1885 and 1891 although the arguments may be a little different. It is generally agreed that DC grids can reduce energy consumption but economical and ecological benefits are questioned as well as safety standards.

LEDs as semiconductors are DC driven and have additional advantages when installed in a DC grid environment. However, it is necessary to know and understand the DC grid concepts in detail and to know which one, if any, should be used to replace or complement the current AC grid. The dimensions range from room level to local, regional and even continental world-wide concepts. While the latter will probably not be realized in the short or mid-term, DC grids on a smaller scale are already being demonstrated in pilot projects and could become reality in the near future. The new trend back to de-centralized energy generation, especially from sustainable sources that provide DC current, could drive the development of local DC grids.

DC grids seem to be advantageous for LED lighting. LEDs, like all semiconductors, need DC current. Several concerns about AC grids don’t seem to be relevant or are of minor importance in DC grids. But is a change to DC grids really that simple? Will it bring advantages for LED lighting systems with it? Or are there challenges that are not obvious at the first glance? To answer these questions it is necessary to have a closer look at current discussions and some technical issues.

Screen Shot 2016-07-04 at 09.01.08.pngFigure 1: History of the dominant grid current (Source: Dr. März, Fraunhofer ISB, translated)

Status Quo of DC Grids

DC grids can be found all around us - for instance - they are already standard in the automotive and aerospace industries. A battery is an integrated part of such a system and energy requirements seem to be completely different to the requirements in buildings. Distances between source and load are, in general, relatively short and are very often regarded as special situations. This is correct in many cases but they are also struggling with overcoming difficulties that are relevant in immobile applications and common in all DC grids.

DC grids are also applied in island or so-called solar home systems. Although these are very often small scale systems with low voltage levels of typically 12 V to 24 V, specialist insights for these types of systems are useful for defining the crucial parameters and finding the right concepts for an optimized solution.

In addition to the areas mentioned above, several demonstration projects for DC grids have been initiated (mostly in server plants) and are running all over the world. The server plant applications especially seem to be textbook examples for DC grids. The derived results and conclusions are often questioned and sometimes considered not to be useful for other applications. These extremes must be covered by a new DC grid concept. Apparently the change to DC grids makes sense and the effort is paying off even when prototypes for switches and other equipment still need to be designed. There are similar aspects and difficulties and therefore some knowledge can be passed on to other applications.

However, there is still a lack of well-recognized results that are based on a proper scientific work. Therefore, uncertainties regarding the pros and cons of efficiency gain, costs or safety issues, are still high. In addition, some key issues have to be solved before DC grids will have a chance to be installed on a regular basis and DC products can become mainstream. This is true for some technical issues, but even more so for standardization issues. Voltage levels, grid topology, grounding, arcing, earth leakage circuit breaker requirements, over-current detection and load balancing are just a few questions that need to be solved, defined, and standardized.

Grid Topology

The first question that needs to be answered is what else can be covered by a DC grid. Does it make sense to discard our 230 VAC grid completely? Should we consider a hybrid topology? Or would it be best just to apply a DC grid on a very small scale for low power applications? To answer these questions it has to be taken into account that in the end, it is very likely that a completely changed infrastructure with de-centralized, distributed energy resources will be the result of the change to sustainable (“green”) energy generation. At that stage, at least local DC microgrids [1] will make sense. As a consequence, the specifications for DC grids must incorporate the requirements for this type of system. That means that different voltage levels and the physical structures need to be defined.

Screen Shot 2016-07-04 at 09.01.18.pngFigure 2: Evaluation of pros and cons of different DC grid topologies for an office lighting application (Source: Prof. Waffenschmidt, Cologne University of Applied Science)

Voltage levels
Several voltage levels are being discussed. For low power applications of up to 100 W per line, for example, eMerge Alliance favors 24 V, and for high energy distribution 380 / 400 V. While the 380 / 400 V specification is already favored by many specialists and organizations, the 24 V approach may be questioned. Meanwhile, even the automotive industry is promoting a switch to a 48 V system.

P. Brueckmann, specialist for DC island systems, demonstrated in numerous projects the feasibility of 24 VDC grids [6]. A maximum of 1.5 kW can be realized in smaller grids. Mr. Brueckmann strictly advises using a star-topology for distribution. Arcing is no problem for most 230 VAC switches at the nominal current. Because this is already voltage standard, numerous components, especially semiconductors like Mosfets or voltage regulators, are available. If buffer batteries are part of the system, cell reversal is still detectable.

J. Schönberger from Tridonic [5] demonstrated that even when using efficient LED lighting systems, several lighting applications, like office lighting, may become impossible to be achieved with good performance and reasonable costs when applying the 24 V approach, and a challenge with 48 V DC grids (Figure 3).

Screen Shot 2016-07-04 at 09.01.23.pngFigure 3: In a 48 VDC grid the losses exceed an acceptable amount of 5% after a certain distance, dependant on the power requirements of the luminaire. This is the critical length (Source: John Schoenberger, Tridonic)

These two voltage levels are of special interest due to the SELV (Safety Extra Low Voltage) classification. Above the 60 V SELV limit no other voltage 380 / 400 V is considered from specialists to be a relevant alternative to become a future standard.

Physical topology and grounding
Independent of the chosen voltage level, a DC grid can be set up as two-wire grid (+/-) or three-wire grid (+/0/-). Both systems have advantages and disadvantages and differ clearly regarding the topology and construction of safety measures.

The choice of the topology is relevant for the grounding type and its consequences. The relevance for a low voltage level may not be that big or clearly visible, but for the discussed 380 / 400 V grid the different grounding options have relevant consequences (Figure 4).

Screen Shot 2016-07-04 at 09.01.32.pngFigure 4: 380 / 400 VDC grids are the most likely option for replacing 230 VAC grids. The various topologies have relevant consequences regarding safety issues and necessary protection measures (Source: Prof. Waffenschmidt, Cologne University of Applied Science)

For the two wire solution, there is just a choice between a floating high ohmic (<1MΩ grounding or a single ended low ohmic (<1Ω) grounding). The single ended system bears the safety risk that a maximum current of 380 mA at 380 V can seriously endanger life.

For the three wire system there are more options possible; symmetrical low or high ohmic grounded or AC grounded. While the AC grounded solution also leads to a maximum current of 380 mA at 380 V, the low ohmic symmetrical grounded system has a maximum 190 mA at 190 V which is just a quarter of the energy, but still a lethal value. The high ohmic symmetrically grounded system limits the current to a maximum of 17 mA. Under current regulations this is currently recognized to be a safe value for DC systems.

Besides safety issues, there are also other aspects of grounding to be issued; for instance EMI. A floating system is less prone to have common mode currents. These aspects and some other arguments may have different relevance for different applications or projects. In the end, the best overall solution should be selected.

For lighting applications, Prof. Eberhard Waffenschmidt from the Cologne University of Applied Science evaluated the presented options and came to the conclusion that a three wire system with +/- 190 V would be the preferred choice in most cases. This approach, like the 380 V approach, is sufficient for higher loads and therefore could replace the existing 230 VAC grid [17].

Following that approach, one could also imagine a +/-24 V or a +/-48 V DC system, or maybe even other voltages. The main question that appears is: What for, where and when does it make sense to use one of these voltages, or does it in general make sense?

Arcing – One of the Biggest Challenges

The problem of arcing has several issues. Arcing can be caused because of damaged isolation which leads to a short circuit. Serious risks and damages should be prevented by using an appropriate fuse. Improper contacts or a break in the line is more critical and has to be detected. Otherwise, arcing will ignite a fire with the tremendous amount of heat produced within a very short time. In addition, plugging and unplugging loads that are accidentally switched on involves a very high risk.

One might think that this would only be critical at very high voltages; maybe those beyond the SELV voltage. But this is a misapprehension. At very high currents, 24 V, and under certain conditions even 12 V [13], may cause and sustain arcing that leads to serious damage and risk.

Detection of arcing is one of the biggest issues. While there are required arcs, for instance in discharge lamps, the undesired arcs have to be distinguished properly. Especially in-series arcs are critical and can be reduced by using an appropriate driver concept and detection method.

Switching DC and Safety Circuit Breakers

The switching of DC currents, especially high currents and high voltages, is known to be very critical due to arcing. Different switching concepts are possible. They are also relevant for automatic safety circuit breakers. Currently, in many countries, only a mechanical solution that provides physical isolation fulfills the regulations. D. Leber from Systemtechnik Leber [15], P. Meckler and C. Strobl from E-T-A (Elektrotechnische Apparate GmbH) gave a comprehensive, up to date, overview in this field [12,13].

A very common method is switching with arc chutes and a magnetic blow field (Figure 5). The principle is similar to the principle of an AC circuit breaker and is relatively simple but effective. The contacts of a switch open and the arc builds up. The magnetic field causes the arc to expand into the arc chutes where the path again increases and is split up until the voltage that is necessary to keep the current flow over the given distance becomes higher than the provided voltage. The arc extinguishes. This would be physical isolation.

Screen Shot 2016-07-04 at 09.01.38.pngFigure 5: Structure of a mechanical circuit breaker for DC currents (Source: Peter Meckler, E-T-A Elektrotechnische Apparate GmbH)

Electronic circuit breakers are an elegant method with which to switch DC loads. They are fast and no arcing appears. In addition, they can provide increased functionality with some integrated intelligence. They can act as a current limiter and, by that, withstand and reduce inrush current. They can also react on lower overload limits without failing because of tolerable spikes. One disadvantage is that no physical isolation appears; another is the energy losses of the electronic components.

In typical Hybrid switches an electronic component takes over the current when arcing starts and a defined voltage difference is exceeded. This causes the arc to extinguish. The electronic is deactivated and additional mechanical switches will be opened to provide physical isolation. Main disadvantages are the relatively high costs.

However, especially when having in mind a worldwide application of a safety switch, the design criteria are very challenging due to different and changing standards.

Personal Protection

Current knowledge regarding the health risks of electricity and DC voltage and current limits is based on research done during the 50’s. These results are questioned by some specialists. We have to accept this data as being relevant until new research is done, either confirming or disproving it.

According to the existing data and regulations, there is clear evidence that DC currents may be up to three times higher than AC currents before they become lethal. Therefore limits are different for AC and DC currents. To avoid injuries, DC grids above 48 V need residual current protective devices (RCD) with sensitivity below 10 mA. Both AC and DC currents have to be detected. Unfortunately, no standard product is available today. While being costly, C. Loef from the RWTH – ISEA sees the best currently available solution in combining type B RCDs with a DC switch in series [14].

TN-S grid structures are the best solution due to a low contact voltage with accordingly low residual currents in case of a fault.

DC Grids and Efficiency

In combination with de-centralized energy generation DC grids apparently have a clear advantage over AC electricity distribution. They are more efficient because a lower number of inverter stages are necessary. Costly and space consuming PFC’s can be eliminated, batteries can be directly integrated to store energy. - It looks like a perfect system!

Looking at the mentioned issues, it would be perfect if we could provide a constant voltage. But there are also reasons to allow a floating voltage between, for instance, 260 V to the nominal 400 V for minutes and up to 600 V for milliseconds. This standard is in discussion for datacenters and telecom facilities. It’s possible that it could also be taken over for other types of DC grids. This means that every converter has to deal with varying voltages and again needs some energy buffer, usually a capacitor, to compensate a part of these fluctuations. Therefore efficiency is sacrificed and costs and space savings are reduced. In an example, U. Liess from OSRAM GmbH showed that a DC version of their AC FL lamp ballast gains less than 1% overall efficiency [18].

On the contrary, M. März from Fraunhofer IISB showed that DC/DC converters with efficiencies of up to 99% are possible and assumes that converters with more than 95% efficiency should be economically makeable [2]. Compared to most of today’s AC/DC converters, which often have efficiencies below 70-80%, this would mean a tremendous energy saving.

However, when comparing state-of-the-art AC/DC and DC/DC converters, the difference is much smaller and the example from Mr. Liess certainly shows the technical “truth” for mass production of well-designed products. The drawback of AC/DC converters is not technical feasibility of more efficient products. It is the cost pressure on small devices that use cheap designs and components resulting in a poor efficiency of sometimes below 60-70%. DC could help to avoid that by displacing the costly structures from a converter. But we won’t get rid of the AC/DC converter; it can be designed for best efficiency. But there is still one issue that remains: Even this converter follows a typical efficiency over load curve and will produce permanent losses 365 days a year. A DC grid design will be the key for system efficiency. Poor design may probably lead to comparable or even higher losses than today’s conventional AC grids.

Regulations and Standards

The experts’ discussion as well as the presentation from Mr. D. Barthel from VDE/DKE disclosed that there are numerous details that need to be standardized [23]. Sometimes there is even information missing to setup a standard on a proper scientific or technical basis. In addition, standardization should be aligned on a greater basis to avoid too many different regulations in different countries. Preferably, a global solution should be targeted.

For that reason the Strategic Group (SG) 4 “Low Voltage Direct Current (LVDC) distribution systems up to 1500 V DC” was established in 2009. Some very specific standards or quasi-standards have been realized for IT and server plants and mobile products or battery driven vehicles. But they do not cover all technical aspects. However, regulations should also incorporate these existing standards, or at least be aligned to them.

For a successful, wide spread, introduction of DC grids standards are necessary but they will not be realized immediately. This will still take some time.

Screen Shot 2016-07-04 at 09.01.44.pngFigure 6: In 2006 M. Jovanovich showed that a volume of between 15% and 60% less space is needed for DC supplies


DC grids will not be a standard installation in the near future, but it could be a growing niche for which LED/OLED lighting could be the preferred method. Mainly two voltage levels may be defined. The Low Voltage domain of 380/400 V is likely to be the first choice. In the Very Low Voltage domain it is less clear if 48 V or 24 V is the favored voltage level.

DC grids definitely have an advantage when combined with alternative energy generation and local energy storage systems. Therefore, the speed of adoption also depends on political decisions. How willing are governments to support de-centralization of power generation? Or will they stick to the traditional centralized system of large power plants?

DC grids make driver design simpler to a certain degree. Therefore system costs for LED lighting are lowered and can come closer to conventional lighting. Nevertheless, simplification potential is limited. Drivers need to be able to withstand and handle voltage fluctuations. For example, a 400 VDC grid of at least 260 VDC to 400 VDC.

Currently, lighting is responsible for about 20% of the overall electricity consumption worldwide and has one of the largest energy saving potentials. To optimize this potential further, it would be desirable to match the grid specification to the requirements of future lighting systems.

All in all, at the current time, there are many uncertainties about DC grids and no clear direction is given. While most modern AC drivers can also be powered by DC current, a dedicated DC design is necessary to gain full advantage. The lighting industry should be prepared and active collaboration of lighting organizations in standardization bodies for DC grids should be intended.

[A] A microgrid consists of interconnected distributed energy resources capable of providing sufficient and continuous energy to a significant portion of internal load demand.

[B] A microgrid possesses independent controls, and intentional islanding takes place with minimal service interruption (seamless transition from grid-parallel to islanded operation).

The following presentations from the ECPE seminar “Niederspannungs-Gleichstromnetze - LV DC Grids” from November 7–8, 2012 in Erlangen were used as main information sources.
[1] ECPE, Cluster Leistungselektronik, E. Petri, ECPE e.V.

[2] Thematische Einführung und Übersicht, M. März, Fraunhofer IISB, Erlangen

[3] DC-Bordnetze in Kraftfahrzeugen, D. Grohmann, Daimler, Sindelfingen

[4] DC-Bordnetze in Flugzeugen, P. Jänker EADS, München

[5] DC-Netze für Beleuchtungssysteme, J. Schönberger, Tridonic, Ennenda

[6] DC-Hausnetze, P. Brückmann, Ingenieurbüro Brückmann, Davos

[7] DC-Stromversorgung für Rechenzentren, Server, K. Bittinger, ABB Automations Products, Ladenburg

[8] Solare Inselsysteme in DC, M. Vetter, Fraunhofer ISE, Freiburg

[9] DC-Gerätetechnik – eine Übersicht, D. Gutzeit, Phaesun, Memmingen

[10] Elektrische Maschinen in DC-Netzen, A. Möckel, Technische Universität Ilmenau

[11] Spannungsniveaus für Niederspannungs-Gleichstromnetze

[12] Schalten von DC-Strömen, P. Meckler, E-T-A Elektrotechnische Apparate, Altdorf

[13] Erkennung von Störlichtbögen, C. Strobl, E-T-A Elektrotechnische Apparate, Altdorf

[14] Personenschutz, Fehlerstromschutzschalter für DC-Netze, C. Loef, RWTH – ISEA, Aachen

[15] Konventionelle und elektronische Absicherung in DC Netzen, D. Leber, Systemtechnik Leber, Schwaig

[16] EMI / EMV in DC-Netzen, J. Kirchhof, Fraunhofer IWES, Kassel

[17] Erdung von DC-Netzen, E. Waffenschmidt, Fachhochschule Köln

[18] DC-Beleuchtungstechnik, U. Liess, OSRAM, München

[19] Stromspeicher für DC-Systeme, G. Bopp, Fraunhofer ISE, Freiburg

[20] Leistungselektronik für DC-Netze, S. Zeltner, Fraunhofer IISB, Erlangen

[21] Kommunikation in DC-Netzen, J. Wachtel, Fraunhofer, ISE, Freiburg

[22] Kommunikation in DC-Netzen (Feldbus, Powerline), A. Oeder, Fraunhofer IIS, Nürnberg

[23] Normen für DC-Technik / DC-Netze, D. Barthel, VDE DKE, Frankfurt