| A man in OEM manufacturing here. Not an electronics engineer professionally, but more or less involuntary having to deal with it. First, I will say that GaN is not a make or break deal yet now. There are very high performance silicon switches that can produce equal gains for use cases in consumer electronic. In absolute comparison GaN is of course winning over Si, but use case wise, not so much. GaN's main appeal is that you can retrofit it into simplest buck topologies, and get instant gains without any extra hassle. GaN works best for low current, low voltage applications you normally see in consumer electronics. Everywhere else besides RF, it's not a clear improvement over silicon. For long time GaN existed in a narrow niche of high frequency circuits, where people were OK with n channel depletion mode device with a lot of tricky sides. Its use as a power switch is relatively novel. Enhancement mode devices are now on the market, but there is a catch. Because p-GaN is nowhere near as good as n-GaN, pure GaN enhancement mode devices are nowhere near as good as depletion mode devices. For this reason, some companies are trying a an approach with hybridised GaN-Si device to overcome that. Again, that is way more tricky and expensive than a single material device. Now, back to chargers. It is easily possible to make more compact chargers without any exotic switches GaNfets included for as long as you put just a little bit more brains to engineering by going to better topologies. What a lot of charger makers do these days to go along with the trend for smaller chargers is to turn up their switching frequencies as much as possible without melting the charger. They still do miscalculations about that, and you now have a lot of molten/burnt chargers as a result. They are being misguided by the switching speed narrative, and completely miss that point that size of passive components has to do with way more things than just switching speed. First, they can throw away the standalone rectification, gaining some efficiency in the process. Second, they should either go with some advanced flat transformer setup with appropriate resonant topology. Or they can throw away transformer altogether and use piezo transformer, or capacitive isolation. Third, a separate buck for last stage DC-DC can also be thrown away if you use multitap transformer and some electronic switching. Transformers are by far biggest contributors to charger volume, and weight, and are bigger than capacitors in some cases. Throwing them away will provide way more space savings, along with other extraneous components. So think twice about going the GaN route before considering things above first. Switching performance in a typical AC-DC is by far not the biggest of your problems. |