SiC MOSFET gate driver design and functionality

M

mike buba

Member level 3
Joined
Nov 17, 2013
Messages
56
Helped
0
Reputation
0
Reaction score
1
Trophy points
1,288
Visit site
Activity points
1,765
Hi All,

I have a general (basic) question about the gate drivers. I am looking into using SiC MOSFETs for my project, and I understand I need to use gate drivers to convert the PWM signal (0-3Vdc) to the high-current drive input (e.g., 0-12Vdc).

I came across this gate driver, CRD-001 Wolfspeed (datasheet), with the schematics shown below. The price is >$50. Even if I calculate the BoM price and try to do it myself, the price comes in at over $30.


On the other hand, MAX22701E (datasheet) is approximately $5, including the additional recommended circuit:


Both are isolated gate drivers, where MAX22701E has (proprietary) integrated digital galvanic isolation, and the CRD-001 (a more expensive one) uses an opto-isolator component. MAX22701E uses the board's Vcc, while CRD-001 uses two DC/DC converters (each costing approximately $15).
There is also the TI TIDA-01605 (web and schematics), which has +15Vdc for high and -4Vdc for low, where "negative driving voltages allow for safer operation and improved noise immunity", while in the previous two, I think, low is 0Vdc.

Basically, the question is which one is 'better', i.e., how to define what is 'better'.
Also, if I am to design my own gate driver, should I be leaning more towards the CRD-001 or the MAX22701E design or something else?

Thanks, MB
 

Sort by date Sort by votes
SiC MOSFETs may need negative drive for best leakage.
They are crappy like early '80s power MOSFETs used to be.
So you might like drivers from that vintage. A couple of my
old (long gone along with the fab) driver designs were that
way, 0-8V in and "- whatever" to "+ whatever" output swing.

UC37xx drivers, if you can still find them, look carefully at
whether "logic ground" and "power ground" are kept
independent or bonded inside the package / on die.

If you are for flying high side drive then you want those
isolated drivers. A low side switch can use a cheaper simpler
driver. But keep negative gate rail in mind, in evaluating /
selecting SiC MOSFETs as depending on leakage and
turn-off speed, you may need that.
 

Upvote 0 Downvote
FvM said:
There's no secondary power supply in the shown MAX22700 circuit. In so far it's no fair comparison.
Click to expand...
Pricewise, yes. Probably another $10 to $20 more if I go with e.g., DCR010505U.
But leaving the price aside for a moment... I am more interested in terms of functionality; which would be more advisable to use?
 

Upvote 0 Downvote
you have not mentioned high side voltage, or frequency - but the IR2110 / 2113 accepts 3v3 inputs and can go to 600V ( nearly ) high side, as well as the low side drive - needs buffering for best results.
 

Upvote 0 Downvote
First step is identifying driver requirements:
- gate voltage
- gate current/charge
- isolation working voltage
- isolation dv/dt (particularly for SiC high side drivers)

Driver module CRD-001 is rather old and apparently designed for high gate currents. We usually don't have so many parallel gate resistors, except with highest power IGBT drivers.
 

Upvote 1 Downvote
Hi All,

Following up on the previous discussion, I have opted for UCC21738-Q1 gate drive for SiC MOSFET C3M0075120D (Drain - Source Voltage 1,200 V, Operational Gate-Source Voltage -4/15V). Rough calculation for the gate driver current Ig, based on the turn-on time (56 ns) and required gate electric charge (54 nC) give Ig= dQ/dt= 56/54= 1A (so 10A should be plenty of magine).
The Gate Driver will be on a separate PCB board, something like CRD-001 from Wolfspeed.

The schematic description is below:
  • Connector J1 is for +/-5V, +15V, GND, PWM signal from the controller, and Logical Enable and Ready also from the controller. The controller output is +3.3V, but the Vcc is +5V.
  • At the end, I am not using any additional features of the UCC21738-Q1 like overcurrent detection, Miller clamp control, or Active High during faults, so the input pins are either floating or connected to COM.
  • I liked the idea from CRD-001 to have adjustable Rgon and Rgoff. Hence, I added several DNM resistors in parallel. Not sure what the optimal Rgon/Rgoff will be.

However, I have several questions:
  1. Unfortunately, the IN+, IN-, and RST/EN min high level is 0.7xVCC (0.7x5=3.5V). But the max low level is 0.3xVCC (0.3x5=1.5V). Will feeding 3.3V be okay?
  2. I added decoupling capacitors (10uF, 0.1uF and 100nF between +15V and Source and -5V and Source), as well as 10nF and 0.1uF between GND and +5V/-5V/+15V. Is that okay, or should I modify?
  3. Sicne I am using CRD-001 as an inspiration; the CRD-001 uses optocouplers (ACPL-4800-300E), but I don't think UCC21738-Q1 uses any. The datasheet just says: "The input side is isolated with the output side with a reinforced isolation barrier based on capacitive isolation technology" and "...as well as providing low part-to-part skew, and >150-V/ns common-mode transient immunity (CMTI)". Do I need (i.e., is it recommended) to add any optocouplers between the input PWM signal and UCC21738-Q1? How about the additional DC/DC converter for +15/-5V (RP-1212D)?
This is my first time doing any gate driver boards. Previously I used off-the-shelf CRD-001 (product is no longer manufactured and not sure if it can be used with newer SiC MOSFETs), so any suggestions and advice would be much appreciated.

 

Upvote 0 Downvote
1. If your driving logic level is 3.3 V, input side of UCC21738 must be supplied with 3.3V, datasheet is pretty clear about this point.
2. Looks o.k.
3. UCC21738 is isolated gate driver, why would you use a second isolator level?
You need an isolated driver power supply, e.g. +15/-5V for most gate drivers, except those with inbuilt DC/DC converter.
4. If you don't use any of the extended UCC21738 features, I would think about an isolated basic 8-pin driver like UCC53xx.

Generally, moving the driver away from MOSFET involves additional circuit inductance and possible gate voltage overshoot, particularly at SiC speed.
Shown schematic misses a Roff resistor.
 

Upvote 1 Downvote
Easy peasy said:
but the IR2110 / 2113 accepts 3v3 inputs
Click to expand...
IR2113 datasheet says:
V_IH minimum is 9.5V, this is for VDD = 15V (test condition)

Since the datasheet (Infineon PD60147 rev.V) does not specify the range for VDD in it´s "Operating conditions" .. it´s a bit hard to find.
A search for "VDD" just gives some tables where VDD goes down to about 3V.

A further search in the datasheet showed on first page "Logic power supply can be 3.3V ... 20V" ... but this is a bit hard to find (Because not in the expected section nor with the expected abbreviation "VDD") --> Infineon should do a better job in writing datasheets.


So for logic levels coming from a 3.3V powered device ... the IR2113 should be 3.3V, too. Even 5V gives not reliable function.
Currently I don´t see a schematic - including power supply voltage information - for an IR2113 in this thread.

Klaus
 

Upvote 0 Downvote
I have made some changes:
  • added -5/20V DC/DC converter for gate power supply.
  • added few Zener diodes.. copy/paste from Fig. 7 from here.
  • replaced the UCC21738-Q1 gate drive with a much more simple one, the UCC5350-Q1 (UCC5350SBQDRQ1).
  • I will increase the 3.3V DPS PWM output to 5V using MAX4427, but this will be done outside, on the DSP board.
Hope this looks okay? ..any other suggestions I can implement? I am still not sure on the final Rgon and Rgoff values; hence, I am having several parallel DNM resistors, just in case. Although I might not need six in parallel.
The next step is doing layout and placement.


MB
 
Last edited:

Upvote 0 Downvote
You must log in or register to reply here.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…