RCD snubber design for switching converter

[powertec]

Snubber circuits are commonly used in switching converters like flyback etc.
There are a variety of types used, one of these is the RCD turn off snubber. This snubber controls the rate of voltage rise at turn off and thus reduces the switch losses. For more detailed analysis and simulation example see the link below.
There is also an online design calculator that helps design and validate the RCD snubber circuit.


http://switching-power.blogspot.com/2013/09/turn-off-rcd-snubber.html

 

[legg]

Snubber circuits are commonly used in switching converters like flyback etc.
There are a variety of types used, one of these is the RCD turn off snubber. This snubber controls the rate of voltage rise at turn off and thus reduces the switch losses. For more detailed analysis and simulation example see the link below.
There is also an online design calculator that helps design and validate the RCD snubber circuit.


http://switching-power.blogspot.com/2013/09/turn-off-rcd-snubber.html

This is a current snubber. Calling it an RCD snubber risks confusion
with voltage limiting circuits that may also use the same component
types.

As with any circuit, it's effectiveness depends upon the ability of
the components and layout to react within a useful time frame,
particularly in low voltage or high current applications. Models may
be limited in duplicating both the switching performance of the
devices used, and the physical effects of the layout.

The use of shunts to measure switch current may represent a problem,
in that they are more easily situated in the low voltage circuitry,
where drive currents will both interfere and be interfered with. They
are also difficult to couple into with real measuring probes and may
have their own bandwidth issues.

Mind you, in simulations, such 'measurements' can also be too perfect.
The operating word is 'ideal'.

In fact, the switch current must begin to fall, even if only modestly,
for the active node voltage to rise, even if only to charge stray
capacitance. In mosfets, also, the effects of Crss can easily dominate
the current fall time - considering Ciss alone may be a mistake
regardless of the effectiveness of the current snubber in controlling
dV/dT.

I guess other types of switches are irrelevant.....

What's NGSpice?

RL

 

[powertec]

A LRCD snubber is a combination snubber, being a combination of a turn-off (rate of voltage rise control) and current (rate of current fall) snubber.
This way you achieve both switch turn-off and turn-on loss reduction.

 

[powertec]

This is a current snubber. Calling it an RCD snubber risks confusion

with voltage limiting circuits that may also use the same component

types.



As with any circuit, it's effectiveness depends upon the ability of

the components and layout to react within a useful time frame,

particularly in low voltage or high current applications. Models may

be limited in duplicating both the switching performance of the

devices used, and the physical effects of the layout.



The use of shunts to measure switch current may represent a problem,

in that they are more easily situated in the low voltage circuitry,

where drive currents will both interfere and be interfered with. They

are also difficult to couple into with real measuring probes and may

have their own bandwidth issues.



Mind you, in simulations, such 'measurements' can also be too perfect.

The operating word is 'ideal'.



In fact, the switch current must begin to fall, even if only modestly,

for the active node voltage to rise, even if only to charge stray

capacitance. In mosfets, also, the effects of Crss can easily dominate

the current fall time - considering Ciss alone may be a mistake

regardless of the effectiveness of the current snubber in controlling

dV/dT.



I guess other types of switches are irrelevant.....



What's NGSpice?



RL

Thanks for the very detailed and exhaustive comment, glad to find someone willing to discuss about the topic. Let me answer to your remarks.
- this is a RCD, rate of voltage rise control snubber, it is composed of a resistor, capacitor and a diode. A current snubber contains an inductor which is a current limiting device, a capcitor is a voltage limiting device onthe other hand.
- yes, using shunts to mesure switch currents would add parasitic inductance that would interfere with the curent rise/fall times. The most important thing is a good layout of the switch section in order to minimize strain inductances. A possibility would be to use a low resistance/inductance shunt (metal film) and a current high bandwidth sense amplifier. Simulation is much easier here...
- Well, I agree...simulating real world is almost impossible. What I would do here is modelling the trace inductances using a calculator and insert the value in the circuit model.
- You are correct about that the current must first fall a little, Coss is the key here. It is like an embedded voltage snubber, in fact a skilfull designer would take Coss in consideration when determining the final snubber Cs.
- Regarding the current fall time, Ciss= Cgs + Cgd = Cg, Crss = Cgd. In my case Ciss already incorporates the sum of Cgs and Cgd. Cg is the total capacitance seen from the gate, not only gate-source.
Thanks for your opinion, please let me know if you find any other mistakes.

AK

 

[powertec]

Thanks for the very detailed and exhaustive comment, glad to find someonewilling to discuss about the topic. Let me answer to your remarks.

- this is a RCD, rate of voltage rise control snubber, it is composed of a resistor, capacitor and a diode. A current snubber contains an inductor which is a current limiting device, a capcitor is a voltage limiting device on the other hand.

- yes, using shunts to mesure switch currents would add parasitic inductance that would interfere with the curent rise/fall times. The most important thing is a good layout of the switch section in order to minimize strain inductances. A possibility would be to use a low resistance/inductance shunt (metal film) and a current high bandwidth sense amplifier. Simulation is much easier here...

- Well, I agree...simulating real world is almost impossible. What I would do here is modelling the trace inductances using a calculator and insert the value in the circuit model.

- You are correct about that the current must first fall a little, Coss is the key here. It is like an embedded voltage snubber, in fact a skilfull designer would take Coss in consideration when determining the final snubber Cs.

- Regarding the current fall time, Ciss= Cgs + Cgd = Cg, Crss = Cgd.. In my case Ciss already incorporates the sum of Cgs and Cgd. Cg is the total capacitance seen from the gate, not only gate-source.

Thanks for your opinion, please let me know if you find any other mistakes.



AK

NGSpice is a spice circuit simulator, http://ngspice.sourceforge.net/presentation.html

 

[legg]

The use of inductors in snubber circuitry is almost always an attempt
to perform the function of the resistor - in a 'loss-free' manner. The
current snubbing function is still performed by a capacitive element
in the circuit. The inductor forms a passive discharge path, sometimes
usefully resonant to allow extra formation of capacitor 'precharge'
for the next dynamic node change.


Crss is charged by output voltage changes, dominating the charge
transfer into the gate node during the switching interval, though
physically smaller than the static Ciss. This is one reason why Qg@Vds
is noted on spec sheets for switching characteristics.
Not really practical to use newsgroups in this manner, due to
increasing irregularity of propagation and use. For example - previous
half dozen posts from this machine are not presented on my own news
server.

 

[powertec]

The use of inductors in snubber circuitry is almost always an attempt

to perform the function of the resistor - in a 'loss-free' manner. The

current snubbing function is still performed by a capacitive element

in the circuit. The inductor forms a passive discharge path, sometimes

usefully resonant to allow extra formation of capacitor 'precharge'

for the next dynamic node change.

This what you are talking about is the so-called lossless or resonant snubber, what I am talking about is lossy snubbers. A lossy current snubber has an inductor as a current limiting device (in series with the switch). It isnot a very common type of snubber though, but it exixts.

Crss is charged by output voltage changes, dominating the charge

transfer into the gate node during the switching interval, though

physically smaller than the static Ciss. This is one reason why Qg@Vds

is noted on spec sheets for switching characteristics.

Agree, Crss definitely impacts the switching time. In my equations Ciss =Cgs + Crss. Ciss is NOT equal to Cgs (gate-source).