🫶 One of the things I enjoy most about writing these Newsletters is the feedback I get from people and the discussions that ensue. I’m always on a path for self-improvement and I love it when people are willing to give me honest criticism, it’s the best and only way!
In this Newsletter I’m going to embrace that philosophy, and firstly I would like to acknowledge some feedback I had that the narrative of my stories can be lost in my excitement.
🎯 So, here’s a brief introduction with the key points I want you to take away from this week’s Newsletter:
My analysis of a DC Inductor with increased operating frequency in the last article was a little misleading.
WBG devices are not readily equipped to tackle the challenge of increased operating frequencies all by themselves.
As I further explore the limitations of other Magnetic topologies, I will try harder to address the full picture.
📃 In my article, I broke down a few quotes from an article written by Cuk back in 2018. The overarching principle for me, and a subject I don’t think I’ll ever get bored of talking about, is the optimal frequency selection for a Converter design. I appreciate that this wasn’t quite the narrative of Cuk’s article, but it served as a nice platform from which to articulate my many thoughts on the topic.
I discussed that, for the DC Inductor, the AC losses which limit us so much in other Magnetic topologies are negligible for a wide frequency of operation, so we are much less restricted and therefore stand to gain a lot by increasing the frequency.
🔎 As I continue my quest, I would like to look at other Magnetic architypes to see how a comparable increase in operating frequency is received i.e., what are the limitations, and what gains in terms of losses and size should we expect to see. If you note, last time I concluded that, for the DC Inductor, an increase from 20 kHz to 2 MHz realized a size reduction of about x50 and a loss reduction of about x5.
However, another bit of feedback I received has provided me with a nice moment to talk about Devices… So, can we get serious now?
☝️ My characterization of the effect of frequency on the simple DC Inductor, though not wrong, may have been somewhat misleading, and after some pertinent feedback (thank you to Mike Tommasi, who I would like to credit) I think this is the perfect time to address the reality of things.
🤔 YES, you will stand to make those volume and loss reductions in the DC Inductor if you move up to 2 MHz. But is that actually possible?
Let’s take any WBG Semiconductor for example, and note this is quite arbitrary, this is the G3R60MT07K (but the same point can be made with any other device on the market). In the figure below you can see the switching energies for the operating conditions shown at the bottom of the graph.
Figure 1 - Switching energies for G3R60MT07K under specific conditions
🤯I should make a point now by saying that the multitude of different and inconsistent testing conditions from one device to the next makes the task of comparing devices objectively a very difficult one, and is often a main reason we run double pulse tests with our hardware. So, I’m not going to carry out any sort of comparison here, for my health.
Let’s just take this very specific example and these operating conditions from the graph above to make things easier. Now, I have had generally good experiences keeping my TO-247 packages at a maximum of 40W dissipation with modest heatsinking. Of course, this is very contextual, let’s just say 40W, it doesn’t change what I’m about to say.
📈 We´ll go straight for 2 MHz, after all, the marketing of WBG is pushing us up to these sorts of frequencies.
At 30A my switching losses are going to be:
🔥 I get around 204W.
Oh dear, that’s before I add the conduction losses, which I get to be around another 65W. So, at around 270W total we are dissipating almost 7 times our maximum target!
If we wanted to operate at 2MHz, we would actually need to operate with a Drain-Source current of about 10A, and that is about 30%of the maximum stated current on the datasheet headline. So, we can see that, as we increase the operating frequency, the utility of these devices is considerably hampered, as Cuk discusses.
⚠️ My point here is that these WBG devices for high frequency, although they have many vast improvements over traditional silicon, and can indeed switch on and off much faster, they are not readily equipped to tackle the challenge of high frequency alone. We find that we must adopt techniques and topologies that allow us to switch on at zero voltage (ZVS). Because the Eon losses are much greater than the Eofflosses, our switch utility will be more feasible once again by not eating so much into the room we have left for conduction losses.
My last article may have been a little misleading, because now I want you to think about what we would need to do to get the Buck converter (as was my example) up to the mighty frequencies of 2 MHz and beyond, if we’re not in ZVS operation.
💡 Let’s assume we don’t want to parallel multiple devices and run with a distasteful economy of space. Another option would be to operate the DC Inductor in CrCM or with slightly negative current on each cycle to allow for ZVS.
Figure 2 - CCM vs CrCM "DC Inductor"
😱 BUT… Not only does this take away all the nice DC-ness of our DC Inductor, and therefore brings a load (A LOAD) of AC losses. It would also require us to run with a variable operating frequency to ensure ZVS across the whole load range, making our Inductor design much more complex.
It aligns with Cuk’s core statement about the obsolescence of this Converter in a world of increasing frequencies. So, I ask again, would it be possible to operate the Buck at 2 MHz? Was my analysis pointless?
🙌 Well, when I go forwards to look at different Magnetic topologies and their limitations, I do think it will be useful to look at them objectively like I did with the DC Inductor (imagine devices continue to improve to a point they really can hard switch at these frequencies). But I think it would be inattentive of me to not elaborate on the reality of the situation also, so I will try my best.
📬 And if you disagree or see flaws in my analysis then please give me feedback, I love it!
😎 I hope you've enjoyed the read! See you in the next one.
