I am designing a gate drive for a high-side P-channel mosfet. The high rail is at 30V, logic is at 12V, and some sort of level shifter is needed to avoid exceeding the maximum Vgs of +/- 20V (See: Block Diagram).

Here are the options I am considering with my comments attached to each:

Option A: Simple voltage divider, but turn-on is slowed by the relatively large resistor (R_on) needed to limit current through the transistor. This design is mentioned, but then pretty much dismissed in most of the application notes I have seen. Why?

Option B: Faster turn-on, but turn off is slowed by the need to have R_off large enough not to overstress the zener.

Option C: Based on a Vishay application note (AN804). Turn-on is slower by the relatively large R_on. Biggest advantage is that the gate drive is relative to V+, not to ground.

Option D: Used R_on to bypass some of the stress on the zener in Option B. Main disadvantage is that gate voltage is related to ground, so if V+ were reduced a lot, there might be insufficient gate voltage. Fast turn-off, but greater ringing.


I am leaning toward Option C or D, as each has only one additional part (zener) compared to a simple voltage divider (Option A). I am swayed by the prestige of Vishay toward Option C, but don’t like the slow turn-on and don’t understand why Option A seems to be so neglected.

Please share freely your comments and experience related to the four options. I apologize in advance for the length of this question and the large number of attachments. I have screen shots of an oscilloscope trace for each, but left out Option B because of the limit on attachments. It is available, if ayone wants to see it. Thanks. John

Edit: I increased the resolution on the Composite schematic.

 

Ask yourself what would happen if the zener failed short circuit.

 

Nothing catastrophic as the highest current through the 2N3904 would be less than 100 mA. In B and D the drive would continue to work, but the mosfet might fail at 30Vgs. (The Fairchild version of the same mosfet is rated at 30 V. That may also be a consideration.) In C, the drive would stop working. But, why is Option A so frowned upon, except for the slow turn-on? How frequently does one expect a zener to fail like that?

As an aside, the schematic was quite readable in the original png but now, I find it unreadable. I going to try and post a readbale version.

John

 

i'll go with the c, and add a "booster" capacitor (in series with a small resistor), parallel to the R_ON_C.

 

i'll go with the c, and add a "booster" capacitor (in series with a small resistor), parallel to the R_ON_C.

You get a gold star for that! It sure speeds up the turn on. I found a value of 150R and 470 pF were just about right. Gate charge at 15 V is about 30 nC so I shot for something around a nF. Turn-on time (estimate) is reduced from about 620 nS to 140 nS.

Still looking for comments on Options A and D. Presumably the same high-pass filter trick would work on Option A as well, right?

John

 

Why re-invent the Mosfet Hi-side driver IC?

 

Why re-invent the Mosfet Hi-side driver IC?

Is there one you can recommend to drive a P-mosfet at 30V? I looked up several, and they seemed either limited in voltage to less than 30V or were ground referenced, which brings the question back to level shifting.

At the outset, I thought it would be a fairly simple level shifter. It worked on a prototype board. So, I etched the PCB. Ran into some wild overshoot and ringing, which has been resolved. But in the process came up with the options presented here.

I guess at this point it is a combination of intellectual curiosity, challenge, and the desire not to re-design completely the already-made PCB. I can fit a few components with solder and bridges, but not a complex IC. Nevertheless, any advice for an IC would be appreciated for the future.

John

 

Still looking for comments on Options A and D. Presumably the same high-pass filter trick would work on Option A as well, right?

Using the capacitor on option A is not recommended as the gate voltage would, for a brief moment, reach the supply voltage.

Option A, as you say, is slower than other options however it works and it is simple and cheap. I've used A, B and C before. B worked wonders with gate turn-off circuit and produced clean switching. C is a good and solid design.

If you don't need the fastest switching, then A is a good option. B coupled with gate turn-off circuit would be my choice for the fastest switching and C fills the rests.