What are calculations and criteria involved in an RC snubber used in back-to-back mosfets for switching AC?

I know that without it, the amount of noise becomes horrible, not to mention the immediate damage to the fets themselves. On the right, there's a sim (using realistic data from a split-capacitor 120VAC motor that I'm currently using in one of my machines) of the circuit without the snubber. And on the left, there's the same sim, using a snubber with R=51 ohms, and C=4.7 nF

​

Funny enough, before I was able to more or less understand the noise issue involved in this application, what I did was just place a 180V bi-directional TVS diode across the mosfets, to inhibit the noise the best I could. But the machine was still producing a considerable amount of noise, and it wasn't until I looked at this simulation that I realized that what was happening was that the noisy waveform was only being clipped at ±180V by the TVS, but that the noise itself was never going to go away.

It wasn't until I added the RC (in parallel with the TVS, I saw no reason to remove that level of protection, just in case) that the noise subdued significantly. And LTspice has made it a lot easier for me to understand why.

So I reasoned, "gee, if I use a much larger capacitor for the snubber then lots more noise will be inhibited, right?" ... and yes, I was right ... but the large cap allowed for so much current to flow through the arrangement that the motor started spinning without the fets being active!

So here's my question, what's the math involved in calculating the best values for both R and C, depending on PWM frequency, voltage source, and motor size?

 

Googling 'snubber calculation' gives many promising hits. I expect the maths is buried in there somewhere.

 

If you google "snubber design mosfets" several application notes available.

Regards, Dana.

 

So here's my question, what's the math involved in calculating the best values for both R and C, depending on PWM frequency, voltage source, and motor size?

Math will not help.
Schematic is completely wrong.
Spikes are from energy, stored in motor's inductance.
You are trying to dissipate in FETs and snubber almost all energy, destined for motor.
AC currents in motor windings are not 60Hz, but mostly PWM frequency.
As always, you should use "flyback diode", when feeding inductive load by PWM voltage.

Without bipolar "flyback diode":


With bipolar "flyback diode":

 

Here also.
Max.

 

Math will not help.
Schematic is completely wrong.
Spikes are from energy, stored in motor's inductance.
You are trying to dissipate in FETs and snubber almost all energy, destined for motor.
AC currents in motor windings are not 60Hz, but mostly PWM frequency.
As always, you should use "flyback diode", when feeding inductive load by PWM voltage.

Without bipolar "flyback diode":
View attachment 156755

With bipolar "flyback diode":
View attachment 156756
View attachment 156757

Thank you, Danko. You've shown me that design before, and it's been on my mind ever since. I have a few questions about it, though. The secondary mosfet arrangement that you're calling a "flyback diode" (I'd call it something like a "dynamic inductive kickback suppressor", but then again, English is neither one of us' mother language ... ) shorts the motor to ground right after the mosfet pair in charge of generating PWM switches off. My question would be, for how long should that mosfet pair short the motor to ground? For as long as the PWM pair is switched off? ... or just for a brief moment while the kickback spike dissipates?

The reason I ask is because doing that would effectively brake the motor and slow it down, and maybe efficiency would be compromised.

 

Googling 'snubber calculation' gives many promising hits. I expect the maths is buried in there somewhere.

Thanks Alec ... of course I tried to use google before creating this thread, but believe it or not, it seems that I was using the wrong keywords for my search. I used "rc snubber for a pwm dc motor" ... and the results were quite confusing ... your suggestion yielded far more useful results.

These are the pages that seem the most promising to me, at this moment:

http://paulorenato.com/index.php/electronics-diy/197-rc-snubber-calculator-spreadsheet
http://electronicsbeliever.com/snubber-circuit-design-analysis/​

 

My question would be, for how long should that mosfet pair short the motor to ground? For as long as the PWM pair is switched off? ... or just for a brief moment while the kickback spike dissipates?

Motor should be shorted to ground all the time, when M1,M2 pair is off.

The reason I ask is because doing that would effectively brake the motor and slow it down, and maybe efficiency would be compromised.

When motor is shorted, working current still go in windings with same working direction and generates right torque with maximal efficiency.

 

I think you may require more for dynamic braking of a Universal motor, there have been quite a few patents over the years aimed at doing this.
e.g.
For example, a series-connected universal motor operating from an AC. source can be made to generate DC. by disconnecting the motor from the source and by shorting an inverted field winding across the armature winding. If the motor is sufficiently large to meet a critical efficiency requirement and if the circuit is not interrupted during the portions of the current cycle when the magnetic flux passes through zero, the conditions for generation are satisfied and the inertial energy stored in the motor rotor and the load are rapidly dissipated by ohmic losses of the generated electric power. The armature brakes rapidly to a halt.
Mac.

 

The secondary mosfet arrangement that you're calling a "flyback diode" (I'd call it something like a "dynamic inductive kickback suppressor", but then again, English is neither one of us' mother language ... )

R1,C3 and R2,C4 are "inductive kickback suppressors", but M3,M4 pair provides continuity of working current and forms 60Hz part of windings current, exactly as flyback diode.

 

What are calculations and criteria involved in an RC snubber used in back-to-back mosfets for switching AC?

Working with much less power than you, I do not recall doing any maths but following a procedure suggested in several application notes (buried right now on my defunct PC). The last adjustment was made probing with the scope. Worked every time.

Ashamed of recalling so little of it. Sorry!

 

@cmartinez:
Circuit below contains 2 flyback diodes, one (BD3) for positive half-wave of AC voltage, other (BD4) for negative half-wave.
Diodes BD3 and BD4 are body diodes of MOSFETs M3 and M4 agreeably.
Flyback diodes are switched by voltage V3 in counterphase with AC voltage.
This circuit have not spikes at all, by definition.

 

@cmartinez:
Circuit below contains 2 flyback diodes, one (BD3) for positive half-wave of AC voltage, other (BD4) for negative half-wave.
Diodes BD3 and BD4 are body diodes of MOSFETs M3 and M4 agreeably.
Flyback diodes are switched by voltage V3 in counterphase with AC voltage.
This circuit have not spikes at all, by definition.
View attachment 156951
View attachment 156952

That is very impressive, Danko. Thanks!

So, what you're basically doing, is shorting all inductive spikes by switching mosfets M3 and M4 in an opposite (or as you say, "counterphase") direction to the current being applied to the motor.

Question, what do you think would be the best way to generate V3? Would a DC-DC converter be necessary, such as this (I'm working with a 12V power supply), or could zener diodes be arranged in such way as to indirectly use V1 as the source?

 

An user here once recommended me this e-book: Snubber Circuits For Power Electronics

Don't know if you have checked it, but I find it useful.

 

An user here once recommended me this e-book: Snubber Circuits For Power Electronics

Don't know if you have checked it, but I find it useful.

Thanks for the suggestion, Xavier. I had not seen that document before, but I'm sure going to give it a look.

 

Question, what do you think would be the best way to generate V3? Would a DC-DC converter be necessary, such as this (I'm working with a 12V power supply), or could zener diodes be arranged in such way as to indirectly use V1 as the source?

I need some time for brainwork. Maybe completely redesign this circuit will best way for simplify it. I will try.

 

I will try.

So will I ... ... many thanks!

 

Danko, how's this?:

​

Only problem is that R1 and R2 are dissipating 12.58 watts of power , according to the sim .

Still, this is my first attempt at a self-sync flyback usinc SCRs.

 

Danko, how's this?:

See:
Equivalent circuit and same effectiveness, unfortunately.

"Like" is because of activity.

 

See:
Equivalent circuit and same effectiveness, unfortunately.
View attachment 157226
"Like" is because of activity.

....



​

Back to the drawing board ... ...