Varistor starts to conduct at higher voltage than Vf of the diode, so it won't prevent the breakdown of the diode.
S1 diode has maximum average forward current If 1A.
Just use a diode with high enough If, about 3A should be ok.

 

Varistor starts to conduct at higher voltage than Vf of the diode, so it won't prevent the breakdown of the diode.
S1 diode has maximum average forward current If 1A.
Just use a diode with high enough If, about 3A should be ok.

The situation is , I have already bought varistors.

 

How was the pump energized in the original application, are you sure it is a solenoid/piston pump? as most auto petrol pumps are DC motor driven.
Based on the Bosch model.
Max.

 

So, should I remove the diodes and only use varistor?

No, I need convincing that a Varistor is useful as a snubber.

My recommendation: Get the proper diode & use just one.

 

Here is the circuit:
View attachment 78719

The 60N03 MOSFET is only rated 30V, any diodes with higher reverse voltage than that should survive. An 18V MOV (if you can still get them that low) across the MOSFET should give peace of mind.

The diodes failing is a bit puzzling, but then it is a lot of inductance to clamp - fast-soft recovery diodes are probably the best choice, Shottky-barrier types would probably be cutting it fine, reverse voltage ratings start at only 20V.

 

How was the pump energized in the original application, are you sure it is a solenoid/piston pump? as most auto petrol pumps are DC motor driven.
Based on the Bosch model.
Max.

It is 100% solenoid/piston pump. If you want to see the inside of the pump, I will upload a photo.

 

It is 100% solenoid/piston pump. If you want to see the inside of the pump, I will upload a photo.

Thats OK, I figured it might be, just confirming it.
Max.

 

The 60N03 MOSFET is only rated 30V, any diodes with higher reverse voltage than that should survive. An 18V MOV (if you can still get them that low) across the MOSFET should give peace of mind.

The diodes failing is a bit puzzling, but then it is a lot of inductance to clamp - fast-soft recovery diodes are probably the best choice, Shottky-barrier types would probably be cutting it fine, reverse voltage ratings start at only 20V.

I have seen circuits of other similar pumps having just 241K or 271K varistor. In fact, I have also tried with 241K and 271K, they both work fine.
The actual current drawn by those pumps is 2.5 - 2.6A.

 

I have seen circuits of other similar pumps having just 241K or 271K varistor. In fact, I have also tried with 241K and 271K, they both work fine.
The actual current drawn by those pumps is 2.5 - 2.6A.

All the MOVs I've seen with numbers on were marked in the familiar format of other component types, 271 would be 27 followed by a single zero.

I'm guessing a 270V MOV wouldn't save a 60N03 from much!

 

I'm guessing a 270V MOV wouldn't save a 60N03 from much!

You are right but, the fact is they are working.

 

You are right but, the fact is they are working.

It won't stop them from working - but it won't stop a transient from stopping them working either.

 

You are right but, the fact is they are working.

How do you know?

 

How do you know?

I meant to say the pumps are working with those varistors and their respective mosfets are okay.

 

It won't stop them from working - but it won't stop a transient from stopping them working either.

Got it. But how long the mosfet can bear the transient voltage? Should it get bad immediately or it can take some time?

 

When the diodes fail, how do they fail, open, or shorted? How do you know that the diodes have failed?

I reran the simulation to show that the diode should not be failing because of their peak inverse voltage being exceeded. I plot the reverse differential voltage across the diode V(cathode,anode). Note that the diode is not subjected to a reverse voltage exceeding the 14V power supply.

The forward drop while the diode is conducting is ~1.2V. If you look at the previous simulation back in post #14, the voltage at the drain of the NFET is never exceeds 15.2V (14V+Vf).

If the diode D1 was not there, when M1 turns off, the drain voltage would shoot up to the breakdown voltage of the parasitic drain-to-source diode inside the NFET.



This brings us back to the diodes blowing because their repetitive forward current pulses are exceeding the thermal capabilities of the package. The varistor will do absolutely nothing...

 

When the diodes fail, how do they fail, open, or shorted? How do you know that the diodes have failed?

They get broken, can be seen on the PCB.

 

They get broken, can be seen on the PCB.

If the back emf diodes physically crack in two - how do the 30V MOSFETS survive?

 

If the back emf diodes physically crack in two - how do the 30V MOSFETS survive?

It is understood when the diodes get damaged, the mosfet also get damaged and stop functioning.

 

So, to quote myself:

...This brings us back to the diodes blowing because their repetitive forward current pulses are exceeding the thermal capabilities of the package. The varistor will do absolutely nothing...

The diodes are not up to the task of dissipating the heat that is produced by repetitive 7A peak pulses. The original designer of the board knew there was a problem. The designer tried to fix that problem by paralleling two diode. Bad idea, because without equalizing resistors, one diode always takes more current than the other. Eventually, the first diode blows, causing the second to blow shortly thereafter. When it goes, that instantly blows the NFET.

The correct solution is to replace the two small diodes with one bigger one; one that is up to taking the repetitive 7A pulses...

 

So, to quote myself:



The diodes are not up to the task of dissipating the heat that is produced by repetitive 7A peak pulses. The original designer of the board knew there was a problem. The designer tried to fix that problem by paralleling two diode. Bad idea, because without equalizing resistors, one diode always takes more current than the other. Eventually, the first diode blows, causing the second to blow shortly thereafter. When it goes, that instantly blows the NFET.

The correct solution is to replace the two small diodes with one bigger one; one that is up to taking the repetitive 7A pulses...

Hmm, I am now just adding a 3A diode. Removing the 2 x 1A diodes will be time consuming.