Trying to solve ringing issues in switching power supply

[charliehorse55]

I am building a switching power supply to drive large currents into an inductor for a short period of time (<10ms). The circuit is working well at low current, but I see large ringing on the scope and I believe it will destroy something when the current set point is increased from 20A to 200.



Some notes on the circuit diagram:

• The main capacitor bank are electrolytics 4x5800uF in parallel
• The Mosfet on the schematic is actually an IGBT, there wasn't a symbol for it in the software
• Both the 100nF caps are film type, mounted about 1cm from the igbt
• The 430nH inductance in series with the main cap is the measured parasitic inductance of the bus

I have added both a decoupling capacitor and a switch snubbing circuit to try and suppress ringing, but there is still a fair amount of ringing on the emitter node. It only seems to appear once there is a current built up in the inductor - the first pulse has very little, but when the IGBT is turned on for the second time there is a substantial ring:



Yellow is main cap bank voltage
Green is emitter voltage
Purple is current sense resistor voltage

What can be done to reduce the ringing? How can I analyze the circuit to find a solution?

[camivoss]

Looks like inductive spiking. What frequency are you switching at?

If you are switching above ~1kHz, you probably need a fast switching flyback Schottky diode. The 1N5404 might not be cutting it.

If you aren't familiar with inductive spiking:

[MrAl]

I am building a switching power supply to drive large currents into an inductor for a short period of time (<10ms). The circuit is working well at low current, but I see large ringing on the scope and I believe it will destroy something when the current set point is increased from 20A to 200.



Some notes on the circuit diagram:

• The main capacitor bank are electrolytics 4x5800uF in parallel
• The Mosfet on the schematic is actually an IGBT, there wasn't a symbol for it in the software
• Both the 100nF caps are film type, mounted about 1cm from the igbt
• The 430nH inductance in series with the main cap is the measured parasitic inductance of the bus

I have added both a decoupling capacitor and a switch snubbing circuit to try and suppress ringing, but there is still a fair amount of ringing on the emitter node. It only seems to appear once there is a current built up in the inductor - the first pulse has very little, but when the IGBT is turned on for the second time there is a substantial ring:



Yellow is main cap bank voltage
Green is emitter voltage
Purple is current sense resistor voltage

What can be done to reduce the ringing? How can I analyze the circuit to find a solution?

Hello there,

First, you need to show a more complete schematic.  Drive circuit and how it is connected to the mosfet and even perhaps lead lengths.
Second, you dont use rectifier diodes in switching circuits you have to use diodes with fast recovery.
Third, i have to question what the 1N54xx diode is used for.  It looks like you are forcing the mosfet intrinsic diode to do all the work.  If that is what you want ok, but if not you would need a high speed diode in parallel with the drain and source.
Fourth, make sure you show a ground and maybe also how the input caps get their power.
Fifth, 200 amp switching circuits require more attention to detail than 2 amp switching circuits.

Sorry to lay all this on you all of a sudden, but i assume you want to get this up and running nearly perfect.

[xavier60]

As well as the unsuitable diodes, the ringing frequency appears to match the resonant frequency of the 430nH parasitic inductance and 100nF , 765Khz.

[T3sl4co1l]

How is drive connected?  What is ground in this circuit?

Resonances can usually be traced to a combination of known capacitances and/or inductances; xavier's note for example.

You may find it's better to use a TVS, than a dV/dt snubber.  If nothing else, the dV/dt snubber needs to have a capacitor much larger than the nearby capacitors, so that the peak voltage is limited proportionally.

1N4007 is wholly unsuitable here.  It turns on much too slow, and isn't rated for enough peak current, anyway.  1N5404 probably the same.

I've demonstrated a small current limiting switch of broadly similar design, which is capable of drawing up to 110A peak, turning off in 3 microseconds (while offering excellent lifetime from battery power).  To handle inductive loads, it has an RCD snubber (470nF, 1 ohm and, B560 schottky diode I think?) in parallel with a TVS (SMDJ48A, with the transistor being rated for 80V and the actual peak drain voltage being 65V or so).

Regarding stray inductances, we can calculate limits for the transistor and snubber.  Say the transistor is capable of turning off in about 200ns (typical for an IGBT, though if it's a newer type, it may well be significantly faster than this!).  That means your goal of 200A peak, in 0.2us, gives a peak current slew rate of 2000A/us (or probably a bit more than this, in the middle of the falling edge, since it's a curve, not a trapezoid).  If the transistor is rated 600V and the supply is 300V, we need to limit peak induced voltages to less than 300V (300 + 300 = 600V limit).

Inductance is defined as the ratio between voltage and rate of current, i.e., L = V / (dI/dt).  Or equivalently, the ratio of integrated voltage, i.e. flux (V dt), to change in current (dI).  300V / (2000A/us) = 0.15uH.

This translates to a physical length, given some known geometry.  For stray wiring, the factor is around 1/4th.  The ratio comes from the permeability of free space, mu_0 = 1.257 uH/m.  1/4 is 0.3uH/m, so we might guess a maximum distance between transistor and snubber (preferably a 300V TVS in this example) of 0.5m.

If the transistor is faster and the current is larger, or the voltage limits are more restrictive, all of these measures are reduced proportionally.  For a 30V overshoot, we need less than 5cm; for a MOSFET switching in 20ns, we need less than 5mm -- and that includes component lead length and body length!

Tim