Understanding Snubber Circuit Types for Capacitor

[billy fok]

i have did some research of snubber circuit. and i do know there is 3 types of snubber circuits:

• Capacitor (C)
• Resistor-Capacitor (RC) damping network
• Resistor-Capacitor-Diode (RCD) turn off snubber

can anyone explain snubber circuit for Capacitor? thanks

 

[billy fok]

when i do the research, it mainly tell me more about RC and RCD..

 

[Svein]

I am not quite sure, but I expect that in a capacitor snubber, they rely on the resistance in the coil to absorb the energy.

 

[Mike_In_Plano]

There's probably over 20 in all. There's all manner of R, C, D, and L. Then there are switched snubbers such as SCRs that switch to take the stress from other SCRs...

 

[Svein]

My personal favorite is this circuit. There is no need for a separate snubber - the PNP emitter follower ensures that no voltage spike can occur.

 

[meBigGuy]

That's crazy! (as in cool). How come I've never seen that? But, how is it better than a diode?

 

[Svein]

If you the coil is in the collector path of a transistor, the moment you turn the transistor off, it shows a high impedance to the coil. The coil reacts to the loss of current drive with a voltage spike that can reach twice the supply voltage value. That spike is the reason why we need a snubber circuit.

In my circuit, if you remove the current from the relay, the emitter presents a low impedance to the coil. If the voltage at the coil tries to get higher than a V_be above the supply voltage, the PNP turns on and damps the current.

The advantage? For one thing, you do not have to calculate and implement a snubber circuit. If the PNP is able to drive the coil, it is also able to absorb the kickback from the coil.

 

[donpacino]

to the original question, the capacitor in that case will be a decoupling capacitor. It will function very similarly to the RC scrubber.
look up decoupling capacitor.

 

[meBigGuy]

If you the coil is in the collector path of a transistor, the moment you turn the transistor off, it shows a high impedance to the coil. The coil reacts to the loss of current drive with a voltage spike that can reach twice the supply voltage value. That spike is the reason why we need a snubber circuit.

In my circuit, if you remove the current from the relay, the emitter presents a low impedance to the coil. If the voltage at the coil tries to get higher than a V_be above the supply voltage, the PNP turns on and damps the current.

The advantage? For one thing, you do not have to calculate and implement a snubber circuit. If the PNP is able to drive the coil, it is also able to absorb the kickback from the coil.

The voltage must be higher that Vbe + Vr and will spike higher because of PNP turn on time.

But, that aside, you missed my point. How is that better than a reverse diode across the coil to the supply. In fact, a schottky will make the spike even less, and turn on very fast.

 

[Baluncore]

Svein's circuit in post #5 must have an R value low enough to handle the base current of the PNP during flyback. Otherwise, the maximum collector voltages of both the PNP and NPN will be exceeded.One solution is to add a reverse diode in parallel with R.

I have considered circuits like that in the past but have always had a problem with controlling the PNP base voltage.
With R selected low enough to prevent over-voltage then the NPN must sink both Vs/R current and the PNP base current.

To turn off quickly requires a significant negative inductor voltage. A schottky diode with a series resistor can achieve that in a more controlled way than the emitter relay drive circuit of post #5.

 

[jim hardy]

hmmmm

Seems to me when NPN turns off and the edit: PNP comes out of saturation, PNP's Vce starts to increase,
the current through R_upper would become I_relay divided by Beta of PNP .So long as ratio of R_upper / R_relay is less than Beta of PNP ,
the peak voltage at bottom of relay will not exceed 2 Vsupply?

Use something like a 2N3906 where you can count on Beta >50 and make R_upper 5X R_relay ,
peak should be less than ~ 1.1 Vsupply.

Only trouble i see is relay's dropout time could depend on beta if that matters to the application.

Is my thinking straight?

 

[Baluncore]

Is my thinking straight?

Yes.
But beta is something I do not like to predict. Beta is usually specified as probably being in a wide ballpark.
On the other hand a schottky diode located at the relay coil, possibly with a series resistor, is much more predictable.
Today, a power MOSFET with low R_on would replace the complementary darlington pair which will always drop one V_be.

 

[Svein]

Seems to me when NPN turns off and the PNP comes out of saturation,
the current through Rupper would become Irelay divided by Beta of PNP .

Oops - an emitter follower does not go into saturation...

 

[jim hardy]

Oops - an emitter follower does not go into saturation...

Well, you're right.. Bad wording on my part.

okay,

"..when the PNP's Vce starts to increase,..."

i think is more correct ?

thanks - as we choose our words better we reason better...

 

[Svein]

On the other hand a schottky diode located at the relay coil, possibly with a series resistor, is much more predictable.

Yes - but Schottky diodes traditionally have a fairly low breakdown voltage. A quick check - well, they have improved. You can now get Schottky rectifiers with 100V_RRM. OK, I agree.

Today, a power MOSFET with low Ron would replace the complementary darlington pair which will always drop one Vbe.

For example http://www.onsemi.com/pub_link/Collateral/MCH3486-D.PDF.
There may be a cost issue, but otherwise I agree.

 

[Baluncore]

There may be a cost issue, but otherwise I agree.

If you buy only the specifications needed, then the price of a MOSFET is now less than that of the NPN + PNP.
The problem of cost is when buying a superceded legacy part number that is no longer in production, or less than 1000 units.

 

[Windadct]

Back to the OP... In general I do not like to call a Diode in the Coil example a snubber - it is a free wheeling diode. Snubber - implies more of a suppression role - as when there is a V spike and the capacitor prevents excessive overvoltage, in many cases due to L * dI/dT at some level- ( ideally a FWD will prevent the V spike to start with)

As for why not always use a diode - there are many switching circuits where the inductance being switched off is not so "accessible" Consider l line rectifier or inverter. The snubbers often applied at the switch terminals and across the DC bus ( i.e. not on the load side) are not to counter the effect of the LOAD ( there is typically a Free Wheeling Diode for this) -- the snubbers used are to suppress the V spike generated in parasitic inductances. - often on the DC bus.

In the example below -- the snubber (RC) is not included to account for the V spike of the load (usually that is too much energy for this cap (in this case 0.1uF and 100ohm R) - but for all of the inductance between the Diodes and the load. When the diodes commutate some are turning off and basically every conductor in the circuit has some minute inductance - the higher the current, and the faster the switching the more this becomes apparent.

As for the resistor - in general you need the energy to go somewhere - otherwise you and up with a LC circuit ( agreed with very small R "always") - but without the R this tends to generate excessive ringing - or an oscillation that can be disruptive to the circuit or emitting unwanted EMI.

Note - in the circuit above, if the load is a large inductor(or long leads to the load) - you may still want to add a FWD in parallel to the RC shown, this tends to need to be a fast diode, and/or sized to handle a large percentage of the load current - adding cost. And the diode requires some time to switch on - the Capacitor is much faster.

 

[Baluncore]

In the example below -- the snubber (RC) is not included to account for the V spike of the load (usually that is too much energy for this cap (in this case 0.1uF and 100ohm R) - but for all of the inductance between the Diodes and the load.

You show a single RC snubber on the DC output of a three phase bridge rectifier. Surely that is a job for the load. Any one phase will not take over conduction until the previous phase stops conducting. The output impedance is therefore always that of the load, it will self snub. At any point in time, only one line phase will not be in conduction.

I would have expected to see a delta of three RC snubbers on the input side, to prevent spikes on the line side of the bridge, where the line inductance of the non-conducting phase will generate transient switching spikes and noise back up the line when the diodes switch off. The snubber resistors are actually impedance matching diodes, AC coupled by the capacitors, to prevent transients on the line side, when each phase in turn ceases to conduct.

 

[Windadct]

This is a basic topo for a subsystem - sold as such, and we can not predict all of the conditions the end user will apply(line or load side). So this is in place to provide some robustness to the rectifier diodes- it can be made much more robust- but then it depends more and more on the total application. In higher current assemblies we apply the RC snubber to each individual diode ( or thyristor). From a system level design standpoint we do advise to employ the AC side - but this seemed to be the simplest application for an RC case (in response to the OP) - and not a FWD.

 

[Svein]

My personal favorite is this circuit. There is no need for a separate snubber - the PNP emitter follower ensures that no voltage spike can occur.

Now I see why so many found fault with this circuit - I forgot the diode in parallel with the collector resistor (supplying base current to the PNP when the base tries to go above the supply voltage).

 

[jim hardy]

I forgot the diode in parallel with the collector resistor (supplying base current to the PNP when the base tries to go above the supply voltage).

I don't see why one is needed. R only needs to carry Irelay/Beta without letting volts at bottom of relay exceed Vce of the transistors..
I guess a diode would allow higher R and would remove consideration of Beta , though.

 

[Mike_In_Plano]

The greater part of the schematic is the part unseen. If it wasn't for leakage inductance and parasitic capacitance, I don't know if we would have ever invented snubbers.

 

[Svein]

I don't see why one is needed. R only needs to carry Irelay/Beta without letting volts at bottom of relay exceed Vce of the transistors..
I guess a diode would allow higher R and would remove consideration of Beta , though.

The basic reason for the resistor in the collector is to tackle the problem of leakage current. This does not require a low value resistor, 10k - 33k is usually OK. As long as the circuit is in linear mode, this value is sufficient, But when the NPN cuts off, all base current to the PNP has to be supplied through that resistor. If the PNP has a β of 50, it needs a base current of 1 mA to handle the relay current. 1mA and 10k makes 10V, which is unacceptable. Therefore - a diode.

 

[billy fok]

does anyone know how to calculate the value of Rs and Cs?
when i search on the internet, the equation is so confusing

 

[zoki85]

does anyone know how to calculate the value of Rs and Cs?
when i search on the internet, the equation is so confusing

Transients one may expect are main design guideline.

 

[billy fok]

meaning?

 

[zoki85]

meaning?

That various circuits can generate various transients depending on their topologies and modes of operation. And they use various components (some require more care than others in that respect, some don't at all)

 

[jim hardy]

1mA and 10k makes 10V, which is unacceptable. Therefore - a diode.

I'd have said instead

"which if unacceptable requires a diode." 12V supply + 10V more from inductor is probably okay with 40 volt transistors.A clever little circuit that one, using the inductor to keep PNP in conduction until coil has discharged.

With diode present, i wonder what would be effect of beta on relay dropout time ?

 

[Baluncore]

With diode present, i wonder what would be effect of beta on relay dropout time ?

Beta variation would have no significant effect. Drop-out time with the diode would be longer because there is less reverse voltage to decelerate the inductor current. di/dt = V/L

Another way to quickly turn off a relay is with a flyback diode to the unregulated supply. That way the reverse inductor voltage is significantly higher but controlled, and part of the energy is recycled.

 

[Averagesupernova]

No one has mentioned the fact that if supply voltage suddenly disappears on that PNP relay driver the only place for the flyback current is completely through the base-emitter-junction of the PNP. It will kill the transistor for sure.

Edit: Or not? Trying to wrap my head around it yet.
Edit: Think I got it. Power supply simply opened would cause damage. A shorted supply would not. I will stick with the simple diode.