1
\$\begingroup\$

Background:

I am designing a buck/boost converter to convert back and forth between my supply voltage of 3.6 V (battery) and an output voltage of VDD = 60 V to bidirectionally drive a capacitive actuator via a full H-bridge.

The main design criteria are small size and high efficiency. For this reason, I am using GaNFETs (EPC2070) because of their low RDS(on) and extremely small size for the buck, boost, and bridge transistors.

I could not find any suitable IC drivers that could operate off the only supply voltage of 3.6 V, hence I am looking at discrete high-side driver implementations for the buck transistor and high-side transistors of the bridge.

High-side driver for buck transistor:

The main criteria here are fast rise-time, high efficiency, and small size. The on-time of the main switch is less than 10 μs. I am currently considering the design below where the main trade-off between speed and efficiency is the level-shift resistor R3. Low values lead to high speed but high off-state losses and vice versa. Any suggestions to work around this? I am also not perfectly sure of what would be the best BJT choice in a small package (currently considering PBSS4130PANP,115 for the totem pole stage).

Discrete HS Driver for Buck GaNFET

High-side driver for bridge high-side transistor:

The main criteria here are high efficiency and small size, switching speed is not an issue (but should not exceed 1-10 μs). The on-time of the main switch is in the order of 50 ms. I am currently considering the design below but again the main limitation is the loss in RG during the off-state of the main switch. I could increase RG drastically or I could add another small PMOS before Rboot to only charge the bootstrap capacitor shortly before the main switch should close but this requires additional logic signals. Any other ideas?

Discrete HS Driver for Bridge GaNFET

Thank you very much for any suggestions, I look forward to your responses.

\$\endgroup\$
1
  • \$\begingroup\$ I'm not sure I have enough of an understanding of your requirements to provide a proper answer, but these two videos by Prof. Sam Ben-Yaakov might be of interest to you: youtube.com/watch?v=lKt9nA7W4ag and youtube.com/watch?v=1wQD0FSRkkQ - the asymmetric MOSFET push-pull approach shown in the latter video may well be ideal for your use-case here. \$\endgroup\$
    – Polynomial
    Sep 21, 2022 at 10:42

1 Answer 1

1
\$\begingroup\$

The following is a portion from a SPICE simulation of a buck converter driver IC that I modelled recently:

enter image description here

As you can see, the pre-driver buffer comprises two cascade CMOS inverters to reduce the dissipation further. And this pre-driver section is driven by another CMOS inverter. The frequency is about 200 kHz.

BSSxx MOSFETs are general purpose small signal MOSFETs but you can use different models that fits to your needs.

\$\endgroup\$
5
  • \$\begingroup\$ Thank you for the snippet. I have tried similar approaches with a MOSFET totem-pole replacing the BJT totem-pole but I found that the unavoidable shoot-through when switching leads to further losses which I would like to avoid. What is the benefit of cascading two MOSFET totem-poles? \$\endgroup\$
    – Son-mic
    Sep 22, 2022 at 10:34
  • \$\begingroup\$ Shoot through happens when top and bottom switches of the buck turns on at the same time. To prevent this you should introduce a dead time between them. The biggest advantage of using MOSFETs is the low consumption. But BJTs are advantageous over MOSFETs when higher switching frequencies are required because BJTs have lower input capacitances. \$\endgroup\$ Sep 22, 2022 at 10:56
  • \$\begingroup\$ I understand the shoot through but what I don't understand is how it can be avoided if the MOSFETs are arranged in a totem pole as in above. Since the gates are connected (M3/M4 and M6/M7) shoot through will always happen. I can see from datasheets that most IC drivers use MOSFET totem poles and they do have additional timing circuits but my best guess is that shoot through is just accepted? Still wondering what benefit cascading two MOSFET totem poles has? \$\endgroup\$
    – Son-mic
    Sep 23, 2022 at 11:06
  • \$\begingroup\$ @Son-mic sorry I overlooked your question. Shoot-through on a MOSFET-based totem pole is possible but negligible, because 1) the MOSFETs are made chip-level so g-s thresholds can be adjusted accordingly to decrease the risk further, and 2) There's always a non-zero resistance between the MOSFET drains which helps the shoot-through to get lower. Still wondering what benefit cascading two MOSFET totem poles has? Actually, the simplest MOSFET totem pole is an inverting one. Since the signal is inverted, there's another inverter needed to convert the polarity back into its normal. \$\endgroup\$ Oct 20, 2022 at 6:25
  • \$\begingroup\$ Thank you very much @Rohat for following up on this. Ok, yes the second MOSFET totem pole to fix the inversion makes sense. And the explanation regarding the chip-level implementation also. In my case, I had to use discrete components (no IC driver that can operate off only 3.6V which is the only voltage available) while minimizing PCB footprint and maximizing efficiency. I actually had another issue which was charging the bootstrap capacitor without pulling the switch node to ground (which is connected to the battery). For these reasons, I ended up using a PMOS for the buck stage... \$\endgroup\$
    – Son-mic
    Oct 21, 2022 at 9:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.