Why use an insulated power supply for a HS gate driver?

[@sadhusuman]

Hello Professor, Does a use of an active miller clamp resolve this issue which indirectly isolates the HS from main power supply?

[@sambenyaakov]

No, Miller calmp attenuates the so called Miller spike that can cause a shoot through.

[@fabioposser2]

Hi professor, another important point is that is difficult to find a diode fast enough and not so expensive.

[@sambenyaakov]

Well, I don't think it is that bad.

[@GokhanSen-k8o]

Thank you again for this great video Prof. Ben-Yaakov. During the video in the section where you discuss "Sizing the capacitor", you took the Idr current as 5mA. But as you also encircled, it is 3mA max. I guess it is a typo right?

[@sambenyaakov]

Just rounded it to take into account possible switching losses within the drive.

[@benm4784]

Great video professor! Would there be a benefit to replace the series resistor with the diode (R6) with a ferrite bead, or otherwise filter the bootstrap input current to present a higher impedance at HF?

[@sambenyaakov]

This might cause an over voltage . I will discuss this in next video.

[@absurdengineering]

For experiments on the bench I sometimes stick a solar cell on a heatsink and shine one of those crazy bright LED flashlights on it. Heatsink is needed because the solar cell overheats otherwise. The parasitic capacitance is mostly limited by the space available in the lab, and whatever shade is needed not to be blinded by such an arrangement. In terms of common-mode current this wins hands down and is almost embarrassingly simple. Of course it’s not practical to put into a product, unless large distances are already available. I’ve used this trick when isolating instrumentation for high voltage experiments (0.5-1MV). Let’s say that a lab setup lasting longer than a semester can be called a product :)

[@femboyee]

Budget alternative to power over fiber! Quite fun. I think isolation motor-generators are competitive for common mode coupling, once the power levels justify them, and definitely beat the efficiency.

[@JacquesMartini]

Ever heard about a invention called batteries?

[@sambenyaakov]

Interesting. Thanks.

[@absurdengineering]

@@JacquesMartini Batteries in a teaching lab, left without air conditioning over the summer - spicy :)

[@JacquesMartini]

@@absurdengineering Oh yeah! They sell billions of these every year to much less qualified people. Your suffering the "not invented here syndrom" and "solar" is Zeitgeist. Any average battery has much mope power and energy density than an solar panel. And guess what! You replace them for a dollar!

[@tamaseduard5145]

👍🙏❤

[@sambenyaakov]

Thanks 😊

[@Jasfon]

Saludos !! No tienes canal en español?

[@sambenyaakov]

No, sorry

[@ganopterygon]

@josesegura24 Puedes usar la traducción autogenerada al Español, es bastante buena

[@sambenyaakov]

@@ganopterygon Puedes usar la traducción del español al inglés generada automáticamente, es bastante buena."

[@femboyee]

Hi Professor, thanks for the video. It aligns with my preference towards isolated high side supplies (also for 100% duty cycle operation and bipolar drive capability). Some of the unwanted effects from the bootstrap can be mitigated by a series switch such as in the LMG1210 and local high side drive voltage regulation. They are not perfect, but can be much smaller and cheaper than an additional isolated supply. I would be interested to hear your perspective on these approaches. The last section on power loss sounded a little off and I think there are some corrections: the calculation of (12*5e-7*10e5)/2 is 300 mW, rather than 600 mW. More relevant, the voltage across the bootstrap is delta V of the bootstrap capacitor + diode Vf (this part not divided by 2), and results in much lower charging bootstrap losses than the power dissipated in the gate drive (~ the 600 mW you calculated). Probably around 50 mW for the example. However, reverse recovery current can be a significant pulse load that produces relevant power loss as well, especially with a pn diode and in fast switching, high voltage systems.

[@sambenyaakov]

Thanks for keen eyes, yes the loss calculation is off. I was carried away, probably because I have observed in the past heating of the series resistor. I have chopped this part. As for the switch solution, experience has shown that for GaN they are problematic due to back injection of spikes.

[@femboyee]

Appreciate your reply, I will keep an eye out for these spikes if I use the LMG1210 with a bootstrap. Is there a similar problem with the synchronous GaN bootstrap approach (in addition to a higher risk of overcharging)? I assume it is predominantly a capacitive effect.

[@sambenyaakov]

I am not sure about the switch approach we could not make it work reliably. BTW you are wrong in your loss calculations. I will post an addendum video to cover the loss and will address the issue . You were right though about the arithmetic error, not dividing by 2.

[@femboyee]

Thanks for sharing your experience, I would definitely be interested in hearing more about your practical experiences in addition to the theoretical tutorials. It could save a lot of time wading through datasheets and wondering what's between the specs, and figuring out how commercially available technologies compare. However, if this involves too much detail or uncertainty, I understand. I'm looking forward to watching the update video. I realize now that the calculation I offered is not valid in most/all cases of dead time and Rbst Cbst sizing, and ignores effects like switch node ringing, but if there's a more fundamental misstep, I hope to never make it again. :)

[@sambenyaakov]

Thanks again for alerting me to the error, which was fact not just the division by 2 but fundamentally wrong. As you pointed out, I have considered charging the capacitor from zero while in fact it is already charges. This distinction is in fact discussed in some of my KZbin videos. If Ok with you, I would like to acknowledge your alerting me to the error in the follow up video (coming soon). It will be nice if you will send me your name otherwise I will identify you femboyee