Load is inductive, but the wires and power FET's are not, may be that is the problem. Input capacitance of this FET's is more than 4nF.
Could it be possible, that the FET driver opens both FET's for 2us and they just sink current to the ground ?
P.S. May be it's a prob's fault ? Calibration or something ?
That could be case when H mosfet is opening , due to Miller effect of L mosfet it opens as well - thus lower part of bridge will short H transistor Source to ground. It takes time to charge L transistor's dg and gs capacitance , then Vgs comes to 0 due to mosfet driver's output impedance.
You can check that by measuring voltage on L transistor gate .
I think it is good to decrease mosfet driver output impedance.
I use IRF type ready driver for the MOS. Series gate resistors only 10 ohm and gate voltages too clean.
I opened the probe and i saw wrongs.
My current probe use small resistor for current sensing purpose. and this resistor is wirewraped. One small transistor take a bias from the current sense resistor.
Inductance effect of this resistor mislead to me.
How do you measure current by one transistor(if possible present schematics here), and why do you use transistor if scope can directly be connected to resistor (current sensor)?
It also can case for late L transistor OFF switching, but you said that driver impedance is low enough. Anyway what is transition parameters of your driver and max freq it can handle?
Electronic fuse circuit follow the outputs of current measurment unit. Therefore current measurment circuts include transistor. (Current probe has own supply)
I use two independent driver signals and I use IR2106.
IR2106 has H and L side driver and inputs not common.
Put another current probe serially with load, and provide scope outputs. Then repeat same for the Lo transistor drain.
Then we can locate where this currrent is originated from .
If this is an actual transient and not due to the inductance in your probe, then your initial assumption was probably correct. This current spike is likely due to the reverse recovery current of your lower diode summed with the input current which is high because of the poor recovery characteristics of the irfp460 body diode (860nS @ 25C).
To reduce the effect of the lower diode either use a SIC diode or an APFC specific diode, or slow down your top switch turn on, or a combination of the both. Recommended transition times are 3 times the recovery (trr) of your diode.
To deal with the body diodes, block them by placing Schottkys in series with your drain. See this:
Is your inductor saturating ?I don’t really see how you could be generating a significant counter emf to ramp down the current in your inductor? I could be missing something though .Because if that were the case your top FET and bottom diode would fail.
Is your inductor saturating ?I don’t really see how you could be generating a significant counter emf to ramp down the current in your inductor? I could be missing something though .Because if that were the case your top FET and bottom diode would fail.
Given schematics only is simplified model. You thing this is half of H bridge. Load is coils of stepper motor. Electronics current regulator regulate the current of coils.
Yes problem is reverse recovey time of internal diodes of MOS. I found.
Your choice is good but i dont want to use two fast diode per mos. My driver has two H bridge. (8 MOS)
16 Fast diodes expensive also this configuration need more pcb and cooler area.
Can you suggest to me good MOS without 16 extra fast diodes.
Max:
Vdd 150V
Idd 15A
PWM frequency 20Khz
Load is highly inductive !!!
What about IRF740 ?
VDSS 400V
Rds(on) 0.55Ohm
max ID 10A
trr Reverse Recovery Time ––– 360 ns, IF = 10A
Qrr Reverse RecoveryCharge ––– 2.9 µC
di/dt = 100A/µs
you can use the selection guide of international rectifier to choose the best transistor according to your load requirements: **broken link removed**
There are several Mosfets that have better diode recovery characteristics you just have to wade through the manufacturers data sheets. Unfortunately body diode recovery characteristics aren't listed in parametric search.