It’s a well-known trope that water and electricity don’t mix, but keeping the two separated is often deceptively difficult, because the simple solution of just sealing the box is insufficient on its own. This is because of condensation, which can come from water vapor in the air at the time the box was sealed, or from years of air exchange via supposedly sealed connectors, wire pass-throughs and the like. A presentation at the Charged Virtual Conference on EV Engineering this past April by Stego, a manufacturer of active condensation control measures, touched on some of these issues, with a particular emphasis on DC fast chargers, but there are many other possible solutions which should be considered, especially if a device has to deal with vibration and shock, as will be the case when it is installed in an EV.
The first things to consider when choosing a scheme to protect against water intrusion or condensation are the cost of the device, closely followed by the consequences of it failing.
The first things to consider when choosing a scheme to protect against water intrusion or condensation are the cost of the device, closely followed by the consequences of it failing, especially while the vehicle is in operation or otherwise away from its home base. By these criteria, the traction inverter would warrant more attention (and budget allotment) to protecting it against damage from water than, say, the DC-DC converter that keeps the 12 V battery charged—the EV will likely be able to continue operating for some time on the charge remaining in its 12 V battery, just as an ICE-powered vehicle can keep going after its alternator has failed. An onboard charger occupies a middle position between the other two devices in that it is highly unlikely to fail while the vehicle is being driven, but would definitely leave a negative impression should it fail upon arriving at a remote destination without sufficient charge to make it back.
Another important consideration is whether the device will have to contend with other environmental hazards such as vibration and shock or even explosive gases (e.g. locomotives, EVs operated in mines). This obviously weighs heavily on any onboard devices, like the inverter and DC-DC converter, but is less of a concern for something like a DC fast charger. Another similar consideration is how much of a temperature swing can be expected in going from quiescence to active operation and then back to quiescence, as it is during the cool-down phase that condensation tends to form. This can be an especially insidious issue for high-power devices like traction inverters and DC fast chargers—even if they were to achieve a near-mythical efficiency of 99%, that still means they would produce 1 kW of waste heat for every 100 kW of power handled, or about the same as a typical residential space heater.
One of the most popular (and obvious) ways to protect a device from contamination by dust or water is to seal it, and the two most common rating systems for describing how well-sealed an enclosure is are those published by the National Electrical Manufacturers Association (NEMA), and the International Electrotechnical Commission (specifically, code IEC 60529). While there is some overlap between the two rating systems, they embody slightly different philosophies and therefore aren’t one-to-one analogous. A NEMA enclosure rating consists of a single number that describes resistance to contamination by physical objects and water intrusion, along with one or more optional letters that describe resistance to various other environmental hazards such as corrosion (X) or snow/ice (S). The IEC code separates resistance to penetration by physical objects of varying size (down to that of dust particles) and resistance to water intrusion into a two-number IP rating, for Ingress Protection, and in that order. One confusing aspect of both rating schemes is that a higher number generally, but not always, translates into a higher resistance to dust or water intrusion. For example, the most common NEMA ratings for enclosures used outside are 3 and 4, which means they are protected from dripping and sprayed water, respectively, whereas a NEMA 5 rating only guarantees dust resistance (that is, no water-resistance rating), while NEMA 6 is immersion-proof temporarily (or continuously, for 6P). Similarly, an IP rating of IP67 means the enclosure is both dust-tight and can withstand immersion in 1 m of water for 30 minutes, but can’t necessarily withstand jets of water sprayed directly at it (which is specified by a 6 for the second numeral), while IPn8 means no harm will come from immersion in up to 3 m of depth, but doesn’t actually require that water not make it inside. Confused already? Well, you’re not alone.
Sealing an enclosure obviously requires filling any gaps, seams or other penetrations (such as for wires, buttons, etc) with some kind of compliant material like a rubber gasket, bead of silicone caulk, etc, as such will accommodate any difference in the coefficients of thermal expansion between dissimilar materials. Sealing around wires or cables is a special headache all its own—gland nuts are specifically made to make a waterproof (or resistant, at least) pass-through for a cable, but they don’t necessarily prevent air slipping past, and air tends to carry water vapor with it.
Another thing that needs to be noted is that any sealant that cures through a chemical process (such as epoxy and silicone) might produce corrosive off-gases. For example, the two types of general-purpose silicone available at most hardware stores (and therefore likely to be used in a pinch) produce noxious, and potentially corrosive, off-gases—one emits methanol and ammonia during curing, and the other emits acetic acid. Of the two types, the acetic acid-emitting one is by far the worse, as it will corrode many metals. However, methanol will react with some plastics, and ammonia will discolor copper and some copper alloys (brasses and bronzes). Fortunately, there are “electronics- grade” silicones available which cure through different (though maddeningly proprietary) processes so as to not off-gas anything damaging to the typical materials used in electronic assemblies.
Another popular way of keeping water and other contaminants out is to protectively coat the internal surfaces.
Another popular way of keeping water (and other contaminants) out is to protectively coat the internal surfaces. This can be a conformal coating that is strategically applied to printed circuit boards and other areas with exposed solder, conductors, etc, by spraying or dipping, or the rather more drastic approach of filling the entire internal volume of an enclosure, which is called potting. There are a wide variety of coating and potting compounds available, and every last one has advantages and disadvantages relative to the rest. Note also that some compounds can be used for both potting and conformal coating—for example, silicones and epoxies—while others are really only used as coatings—such as acrylics and polyurethanes—and others still are only used for potting—the most notable example being asphalt.
Parker LORD says that its CoolTherm® potting and encapsulants will improve performance by optimizing heat dissipation with high thermal conductivity and low viscosity. They also protect components from dust and moisture and help reduce vibration. CoolTherm materials are available in a variety of chemistries to fit many application needs.
As is the case with sealing the enclosure itself, a prime consideration in selecting a coating or potting material is how much compliance is needed to accommodate thermal expansion/contraction. However, additional factors that might complicate the decision process are whether the coating will be exposed to large potential differences (such as between the pins of a MOSFET) and/or high frequency and amplitude voltage changes (which can cause dielectric heating). Needless to say, conformal coating or potting has to be done closer to the end of the assembly process—and definitely after all interconnects are made—but a less obvious consideration is how much more difficult it will be to effect a repair later on; some coatings are relatively easy to remove either chemically or mechanically, like acrylic, while others can only be mechanically removed, like silicone, and some can’t be removed non-destructively without a heaping dose of luck, such as epoxy and polyurethane. Most of the time, sealing the enclosure and applying an acrylic-type conformal coating to the circuit boards will provide many years of service under harsh conditions, but if a higher dielectric strength is needed, then a spot application of silicone is usually a good choice. Potting should only be a last resort, and it only tends to make sense in very cost-sensitive devices that aren’t considered worth repairing (such as a golf-cart motor controller, which is where I’ve seen it used the most in the EV field).
The low price of silica gel is matched by relatively low performance, particularly in the absolute level of relative humidity that can be achieved (typically down to 40% RH or so), especially as ambient temperature exceeds 35° C.
Even with a belt-and-suspenders approach to keeping water vapor (or water) out of a device, there still might be a need to make sure condensation does not occur, and that is where passive or even active means of water removal come into play. By far the most common example of passive humidity control is the humble packet of silica gel, which is so cheap it can be included in a disposable bag of beef jerky. That low price is matched by relatively low performance, however, particularly in the absolute level of relative humidity that can be achieved (typically down to 40% RH or so), especially as ambient temperature exceeds 35° C. A desiccant which operates by a similar mechanism to silica gel—adsorption—but which can achieve much lower levels of relative humidity, even at higher ambient temperatures, is a molecular sieve, which is made from a clay-like material called zeolite, which is covered in tiny pores of just the right size to trap water molecules (“sieving” the water out of the air, then). One downside to molecular sieves is that they are much more friable than silica gel (that is, easier to pulverize into dust), so they might not be the best choice for onboard devices. Both silica gel and molecular sieves can be regenerated by baking at a high temperature for several hours, which drives off the adsorbed water, and both can be treated with a chemical that indicates when they need to be replaced or regenerated (a common indicator is cobalt chloride, which is blue when dry and pink when wet).
Finally, there are active humidity control methods, which could be a better choice when outside air is likely to make it into the equipment cabinet (for cooling, servicing, etc). The first approach is simply to heat the air inside the cabinet so that the temperature is always above the dew point. This could be done with a heater running all the time—maybe using a PTC (positive temperature coefficient) element so it semi-regulates its temperature—or by controlling it with either a thermostat, a humidistat (switches based on humidity level, rather than temperature), or both, so that the heater doesn’t run unnecessarily. The latter approach has been used with success on a DC fast charger, according to Stego, though it would be a tough sell to put any kind of active humidity control system inside any of the power electronics on an EV, and the additional battery drain would be unwelcome.
For completeness, it should be mentioned that there are two other methods of actively controlling humidity—condensing it with a refrigeration system, or periodically regenerating one of the desiccants mentioned above, using externally-supplied hot air—and while these approaches are incredibly effective, they also take up way too much space (and are too expensive) to consider using on even a large piece of equipment like a DC fast charger, much less on an EV itself. Still, if you need to bring the dew point down to where dry ice can’t precipitate water out of the air, then those last two methods are the only way to go.