Oscilations on half-bridge upper switch gate

Thread Starter

GeorgiyKlimov

Joined Nov 24, 2020
4
Hello, dear collegues! I'm currently designing H-bridge converter for DC-motor drive.

I decided to use bootstrap driver for driving converter MOSFETs. IRF8721PBF were choosen as switches, IR2184 was chosen as a gate driver.
To quickly test their performance I soldered bread board with Half-bridge converter. You can see the circuitry on the figure. There is also decoupling capacitors on inputs of L7815 and IR2184, and on Vcc pin of IR2184
osc3.jpgCircuit.png

The problem is the gate voltage drop on upper switch during switching on or it better to say oscilation on upper switch gate.
The oscillograms was pictured and you can see voltages on load, on upper and lower gates, and on Vboot and Vcc pins of IR2184.
osc1.jpgosc2.jpgosc3.jpg
I have tried different valiues of Cboot from 6.8nF to 1uF but there is no significant difference between results.

It would be great to get any suggestions concerning the problem.

Thanks
 
Where on earth did you get the component values for the boost circuit from, they are all clearly inadequate!! Also you dont mention what sort of capacitor you are using for C1 nor EXACTLY what decoupling you have for U1. Switching circuits will not work with poor design, layout is critical also. Might I suggest that you start with the manufacturers application notes. You should also realize that all boost type circuits are duty cycle dependant and will not work in DC drive scenarios.
 

Thread Starter

GeorgiyKlimov

Joined Nov 24, 2020
4
Where on earth did you get the component values for the boost circuit from, they are all clearly inadequate!!
Where on earth did you get the component values for the boost circuit from, they are all clearly inadequate!! Also you dont mention what sort of capacitor you are using for C1 nor EXACTLY what decoupling you have for U1. Switching circuits will not work with poor design, layout is critical also. Might I suggest that you start with the manufacturers application notes. You should also realize that all boost type circuits are duty cycle dependant and will not work in DC drive scenarios.
I calculated Cboot according to equation from Design Tips "Using monolithic high voltage gate drivers" from Infeneon site.
Cboot=Qtot/dVbs=6.3nF;
Qtot=Qg+Qls+(Ilk_gs+Iqbs+Ilk+Ilk_diode+Ilk_cap+Ids)*Thon=26nC
dVbs=Vcc-Vf-Vgs_min-Vds=4.14V

Bootstrap diode was also choosen accoding to the DT mntioned below so that diode breakdown voltage was lower that DC bus voltage and it had low reverse recovery time (<100ns)

Bootstrap diode was also choosen basing on bootstrap circuit time constant Tboot. Tboot should be five times lower than the minimum time, when lower switch is on.

5*Tboot<1/f_sw*(1-DC); (f_sw - switching frequency, DC-maximum duty cycle)
Tboot<1/20000*(1-0.8);
Tboot<2us;

See all parametrs in the end of this message
Tboot=Rboot*Cboot=>
=>Rboot<2us/6.8nC=295 Ohm plus I decided to add some margin, so I have R=235 Ohm
Also you dont mention what sort of capacitor you are using for C1 nor EXACTLY what decoupling you have for U1
I'm using 0.1uF ceramic capacitors in both cases
You should also realize that all boost type circuits are duty cycle dependant and will not work in DC drive scenarios.
What do you exactly mean? If, for example, motor will operate under duty cycle ranges lower that 0.8 why it will not work?




Qg=12nC; Total gate charge
Iqbs=150uA; % Floating section quiescent current
Ilk=50uA; % Floating section leakage current
Ids=0A; %Desat current
Qls=5nC; % Charge required by the internal level shifters
% level shift charge required per cycle (typically 5 nC for 500 V/600 V MGDs and 20 nC for
%1200 V MGDs)
Qp=4nC; % Сharge absorbed by the level shifter. QP is approximately 4 nC at VR = 50 V and increases to 7 nC as the rail voltage increases to 500 V
Qcmos=30nC; % QCMOS between 5 and 30 nC, depending on MGD.

Vcc=15V;
Vf=0.75V;
Vgs_min=10V;
Vds=0.1V;
 
Ohh well I really don't want to go through all of that in fine detail, the app note is for IGBT's. The important stuffs is that Cboost is large enough to supply the switching charge and driver ICC for the LONGEST lower driver off time. The charging resistor must be small enough to ensure Vboost is charged in the SHORTEST lower driver on time. The diode must handle the peak charge current with the lowest practicable voltage drop. Decoupling around the chip must be able to supply the peak charge current of Cboost as well as the peak switching current of the lower mosfet, a single 0.1uF is unlikely to meet these criteria, as for the boost circuit it's a non-starter, why do you think your top mosfet is oscillating at turn-on.

As an example when driving IRFB4227's at 25Khz I use Cboost=470nF/cer D=BYV26C Rd=2R2 Boost supply decoupling 100nF//220uF poly, Vdd decoupling 100nF, Vcc decoupling 100nF (also shares boost supply 220uF) Gate resistors 10R. This switches 80A/180V moderately fast. Layout is super critical, all decouplers are ON the pins! The chip is as close as possible to the mosfets its driving, nice fat traces all round (minimize inductance).

Note poly is polymer electrolytic with extremely low ESR.
 
Last edited:
H-bridge converter for DC-motor drive.
"motor" but the schematic shows resistors. The scope pictures, is that for resistors or motor? It looks like motor.
There needs to be a capacitor from Q1-D to Q2-S, across the 24V at the transistors. Large cap!

It looks like the 24V has "ripple" on it because not capacitor.

Assume a inductive motor. There is "dead time" so for about 1uS no FET is on. This is OK. Bottom MOSFET turns off, the motor "fly-back" or kicks up into the 24V, then the voltage drops back to 0. It may oscillate a little then the top MOSFET turns on. (1uS I can not see on the scope what time is so just a guess maybe 500nS) See data sheet.
 

Thread Starter

GeorgiyKlimov

Joined Nov 24, 2020
4
"motor" but the schematic shows resistors. The scope pictures, is that for resistors or motor? It looks like motor.
There needs to be a capacitor from Q1-D to Q2-S, across the 24V at the transistors. Large cap!

It looks like the 24V has "ripple" on it because not capacitor.

Assume a inductive motor. There is "dead time" so for about 1uS no FET is on. This is OK. Bottom MOSFET turns off, the motor "fly-back" or kicks up into the 24V, then the voltage drops back to 0. It may oscillate a little then the top MOSFET turns on. (1uS I can not see on the scope what time is so just a guess maybe 500nS) See data sheet.
You are right. The problem was due to incorrect supply. There was no decoupling capacitors between Q1-D and Q2-S.
May I ask you for some application notes dedicated to choice of decoupling capacitors.
Thank you for your anser!!!
 

Thread Starter

GeorgiyKlimov

Joined Nov 24, 2020
4
Ohh well I really don't want to go through all of that in fine detail, the app note is for IGBT's. The important stuffs is that Cboost is large enough to supply the switching charge and driver ICC for the LONGEST lower driver off time. The charging resistor must be small enough to ensure Vboost is charged in the SHORTEST lower driver on time. The diode must handle the peak charge current with the lowest practicable voltage drop. Decoupling around the chip must be able to supply the peak charge current of Cboost as well as the peak switching current of the lower mosfet, a single 0.1uF is unlikely to meet these criteria, as for the boost circuit it's a non-starter, why do you think your top mosfet is oscillating at turn-on.

As an example when driving IRFB4227's at 25Khz I use Cboost=470nF/cer D=BYV26C Rd=2R2 Boost supply decoupling 100nF//220uF poly, Vdd decoupling 100nF, Vcc decoupling 100nF (also shares boost supply 220uF) Gate resistors 10R. This switches 80A/180V moderately fast. Layout is super critical, all decouplers are ON the pins! The chip is as close as possible to the mosfets its driving, nice fat traces all round (minimize inductance).

Note poly is polymer electrolytic with extremely low ESR.
Thank you so much for your ansers and support! I really appriciate that you spent your time to anser!
 
Top