When designers worry about the distance between power source and load, it’s usually related to losses due to IR drop (V = IR) or noise pickup. The IR-drop problem can be addressed by various strategies. Among these are use of heavier-gauge wire, employing four-wire Kelvin sensing at the load (although it brings its own potential problems due to creation of a feedback loop oscillation), or going to a more distributed architecture with an intermediate bus converter supplying multiple point-of-load (PoL) DC/DC converters situated in close proximity to the loads. Noise is attenuated via ferrite beads on the power lines as well as bypass capacitors placed close to the load.

However, the same laws of physics that characterize voltage drop also call out power dissipation P = I2R. This power is dissipated as heat, of course, and is not problem in most designs, where the power loss and resultant heat in the cables is negligible compared to the overall system dissipation.

This situation is changing as data cables and even DC power cables are increasingly routed alongside other heat-dissipating cables, often in locations with little or no convection cooling aspects. In many commercial and industrial settings, the numerous AC cables go through risers and plenums and so have almost no air flow.

In fact, the National Electrical Code (NEC) in the US, and similar codes around the world, define maximum free-air allowable dissipation for these cables—and then add derating factors for cables with little or no air flow. The situation gets worse when these power cables are run alongside other cables, so there are adjacent sources of heat in addition to cutting off of air flow.

Placing cables carrying high currents in enclosed spaces is not new, of course. An industrial cabinet for motor control as well as data center racks can have hundreds of amps in their supply cables. But these installations are engineered to support the associated thermal load and their operating environment is bounded. No one casually “tosses” another cable into a carefully engineered server rack handling kilowatts.

In contrast, it seems fairly harmless to push one or two Ethernet cables into a plenum already filled with AC cables. Yet the heat of those cables will degrade Ethernet cabling’s insulation and electrical performance.

The increased use of higher-power Power over Ethernet (formerly, informally known as PoE++, now with a formal designation as IEEE 802.3bt) is aggravating the situation as it allows for over 100 watts to be delivered to the load. While most of this is dissipated at the load, some fraction will be dissipated along the Ethernet cable itself. This is not a problem for free-air cables or in relatively free space; however many of these PoE cables will end up under carpets, or in tight risers and cramped cable runs alongside AC cables (Figure 1).

Figure 1 Power over Ethernet (PoE) offers flexibility in providing power to remote units, but also can allow for impromptu cable routing. (Image source: Intellinet Network Solutions)

These self-heating problems are modest compared to the issues associated with charging of electric vehicles (EVs) at high-current charging stations.

EV chargers see severe thermal stress

In the case of EV charging, even though the distance between charging station and car is relatively short, we’re looking at hundreds of amps and more being pushed through the charging cable. In fact, the risk of overheating this cable is among the many limitations that limit charging rates (Figure 2). It’s a problem that power-system designers have accommodated, but there’s more that can be done.

Figure 2 Key components of a typical DC electric vehicle charging system, with Combined Charging System Type 1 standard (J1772 AC + CCS) connector shown as an example. (Image source: Purdue University)

A team at Purdue University has analyzed, devised, and tested a way to increase the current-carrying capacity of ultra-fast electric vehicle charging cables from a present maximum of a little over 500 A to over 2400 A—a nearly five-fold increase. They developed a method for predicting the heat-transfer and pressure-drop characteristics of both laminar and turbulent flows though concentric circular annuli with uniformly heated inner wall and adiabatic outer wall. By capturing heat in both liquid and vapor forms, a liquid-to-vapor cooling system can remove at least ten times more heat than pure liquid cooling (Figure 3).

Figure 3 Schematics of annulus flow geometry and boundary conditions used by researchers for the basic thermal mode. (Image source: Purdue University)

Their scheme involves pumping highly subcooled dielectric liquid HFE-7100 though a concentric circular annulus which mimics a segment of an actual cable, with a uniformly heated 6.35-mm-diameter inner surface representing the electrical conductor and adiabatic 23.62-mm-diameter outer surface for the external conduit. At these levels of power, heat, and fluid flow, the test and measurement “plumbing” is complicated as are the various transducers needed to control and measure and the parameters of interest (Figure 4).

Figure 4 Photograph of the Purdue experimental facility, identifying its key components. (Image source: Purdue University)

You can read the full details in their two very lengthy papers: “Experimental investigation of subcooled flow boiling in annuli with reference to thermal management of ultra-fast electric vehicle charging cables” (20 pages) and “Consolidated theoretical/empirical predictive method for subcooled flow boiling in annuli with reference to thermal management of ultra-fast electric vehicle charging cables” (22 pages). Both are behind paywalls, but accessible copies are posted at the Two-Phase Flow and Thermal Management Laboratory (TPFTML) web page of the Gwangju Institute of Science and Technology (South Korea) here.

Have you ever had to accommodate power-cable self-heating in your design? Was this anticipated or an unexpected surprise?

Intellinet Network Solutions, “What is Power over Ethernet?” Smart Buildings, “Power over Ethernet (PoE): Power from the network” FS.com, “How to Avoid Overheating in PoE Cabling?” Planet Technology USA, “How to Maintain Your PoE Cable Bundles Cool” International Association of Electrical Inspectors, “Wire Temperature Ratings and Terminations” Velo Engineering, “Conductor Ampacity and Terminal Ratings” Applied Thermal Engineering, “Reduced-scale model study on cable heat dissipation and airflow distribution of power cabins”

Bill Schweber is an EE who has written three textbooks, hundreds of technical articles, opinion columns, and product features.

Oh, yes. When towing a device consuming significant power at shallow depths. There is a lot more wraps on the winch drum and the temperature can rise significantly, even if the winch is in the water. In the air, all the much worse. With deep tow, most of the cable is in the water and everything is hunky-dory.

This seems like a very complicated solution to the problem. As was mentioned, but not expanded on, it would appear that the use of a heavier gauge, very flexible wire would accomplish essentially the same goal. This type of wire is available which contains numerous strands of fine wire in order to achieve a required AWG. Am I missing something?

This appears to be an article on EV charging. I can’t think of something more opposite from EV charging than PoE, not sure why you drag PoE into this. Then pile on making inaccurate statements about it. Sure, heat can degrade Category cable, but I wouldn’t say that it WILL. Depends on the ambient temp, the temp of the heat source, and the temperature rating of the Category cable.

But the biggest misstatement is this paragraph: “The increased use of higher-power Power over Ethernet (formerly, informally known as PoE++, now with a formal designation as IEEE 802.3bt) is aggravating the situation as it allows for over 100 watts to be delivered to the load. While most of this is dissipated at the load, some fraction will be dissipated along the Ethernet cable itself. This is not a problem for free-air cables or in relatively free space; however many of these PoE cables will end up under carpets, or in tight risers and cramped cable runs alongside AC cables ”

IEEE 802.3bt clearly limits PSE power to 90W. There is no IEEE 802.3 standard that allows delivery of more than 90W. The power delivered to the load is 90W minus the cable loss. This can never exceed 100W and be IEEE 802.3 compliant. This power level was obviously a carefully selected value to allow conformance to LPS requirements which also aligns with Class 2 in the NEC. Ethernet cables (carrying PoE or not) should not be run under carpets, pretty sure there is no Category cable rated for UC. A good installer avoids AC circuits in proximity to data cables to ensure there is no interference, especially when you get to the higher data rates that are more susceptible.

But back to the article: yes EV charging have some technical challenges that smart people are trying to solve in interesting ways. This could (and should) have been written without reference to a completely unrelated topic of PoE.

Chad Jones Retired Chair, IEEE 802.3bt

You must Sign in or Register to post a comment.