Snubber Recommendation

[Noway2]

I am working on a circuit that will use a mosfet to supply 12Vdc or 24Vdc to a water / fuel solenoid on a diesel engine. The choice of solenoid is outside of my control and I have no idea what its design parameters are.

I plan to drive the mosfet with a high side driver so that I can use an N-Channel fet. The only real operational spec I have is that I have been told the circuit should be designed for 20 Amps DC.

I am thinking that a wise move would be to put a form of snubber across the mosfet and / or possibly a diode at the source terminal (drain is at power supply) to clamp the reverse voltage that will appear on Vgs.

I understand that exact values are in short supply in this application, but this is what I have to work with.

Does anyone have any recommendations or rules of thumb for a values or a topology to use?

 

[VEBill]

A suitable diode to control the inductor current. That's reasonably obvious.

Make sure you design the circuit for alternator 'load dumps'.

Think about the Arizona mid-summer sun and worst case of everything. Temperature and heat might be major challenges! That's what would make me worry. The other two are 'just' voltage (easy).

Stand-by for other advice.

 

[Noway2]

I agree that a diode is a virtual no brainer. I was attempting to run some LTSpice simulations of this, to get a feel for what I am up against. It was showing me that with a large solenoid (inductance) that I had a moderate current flowing through the diode for a period of time which equated to a bit of power dissipation. I think this is addressable by putting a resistor in series with the diode. This will mean that the solenoid will turn off slower, but that isn't a problem in this application.

Again, a lot of the circuit is outside of my control so there is a limited amount I can do but I would like to make a reasonable attempt at protecting my equipment.

 

[itsmoked]

The voltage will rise on the cut-off if you put a resistor in. Make sure it remains below the FET's smoke portal.

How about a mondo MOV that can take the entire solenoid's I²t at 24VDC say a 30V one?

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[OperaHouse]

putting a resistor in series with the diode. This will mean that the solenoid will turn off slower

Just to clear a point, any snubber or diode will slow the opening of a solenoid. Adding a resistor to a diode will speed it up slightly, but increase the spike voltage. A look at any diode spec will give a peak inrush current. Use this to select a diode. Save space and select a FET with a built in zener.

 

[Noway2]

OperaHouse: You have a very good point about the diode inrush current. Even a measly 1n4001 can withstand currents well in excess of its continuous forward rating for short periods of time. I would just have to be sure to pick one sufficiently large that it will last in the application, ie provide sufficient derating.

Smoked, you too have an excellent suggestion. The maximum current would be 20A per the given specification. Once the circuit opens, this would decay exponentially. However, I don't know of any way to determine the L and R of the load circuit to get an idea of the time factor for calculating an i2t rating. Please correct me if I am missing something here. When you said pick a 'mondo' mov were you suggesting just picking a really large one for this reason?

My thinking is to use a diode to provide a path for the inductive kick in addition to a good sized 30V MOV across the drain - source terminals. I am thinking of something like a MUR460 (they are used elsewhere in the system) which has a forward surge current capacity of 110A which is about 5x what I would throw at it. The MOV, on the other hand, would act as a belt and suspenders clamping the source terminal voltage (the drain is at a power supply rail) to a level that would prevent device damage in extreme circumstances.

Does this sound reasonable to everyone?

 

[catserveng]

I am not a controls designer, but from the standpoint of someone who has a lot of years in industrial engines, please consider your design to use an interposing relay for the solenoid circuit. We have piles of ECMs and controllers that are useless because the fuel and/or starter solenoid output drivers are smoked. A relay driver circuit and a cheap 20 or 30 amp rated relay really is a better real world solution in my opinion.

You would be surprised to see the number of things that get done to those circuits when engines are down, and based on field experience, the styles using interposing relays for larger power circuits always have better service life than those that try to use embedded high power drivers, even with good protection designs.

Hope that helps.

 

[Noway2]

A relay was actually the first choice, but it was jettisoned because of other problems. Specifically in this application, we need to have a manual control switch that operates in such a way that the automatic control circuit can't prevent the manual control from functioning. We are sidestepping the issue in that the solenoid drive circuit is independent of either the manual or automatic control circuit.

The problem I have with a relay is two fold in that it either requires the system power supply (violates that condition of automatic circuitry interfering with manual operation) or has to be selected according to whether the system is a 12V or 24V application which is undesirable. While selecting a relay according to the voltage is simple in principle it creates a whole second 'system' that needs to be maintained and documented which we have been avoiding since the inception of this project.

I had thought about putting a simple linear regulator on the board for the relays, but the T90s that we are currently using can pull a large amount of current (upto ~2.5A) which makes the dissipation a real problem, especially in a 24V application.

This leaves the idea of wiring the manual controls to the terminal side of the relays. Unfortunately, this means that we need 20A switches and have to wire them in parallel with the relay terminals (board mounted) which creates a issues elsewhere in the system. I actually discovered that the existing design that I am attempting to replace violates this item.

Hence, I came up with the idea of using a transistor as a power switch instead. It can be driven by low power, low voltage control signals that can be diode isolated and can operate over the entire supply voltage range of the 12V and 24V systems.

Your point, though, is well taken and is a concern I have.

 

[VEBill]

"...we need to have a manual control switch that operates in such a way that the automatic control circuit can't prevent the manual control from functioning."

Conceptually, 'full-authority' manual control can be implemented using a SPDT switch with a center-off position. The three positions would be labled AUTO, OFF (center), and ON. The OFF and ON positions are 'hard' (simple wiring only and the automatic circuit has zero control).

The actual implementation varies with the details of your application. But the concept of using a center-off switch is fairly obvious.

 

[Noway2]

"The actual implementation varies with the details of your application. But the concept of using a center-off switch is fairly obvious."

Agreed. As I have learned on many occasion, the devil is in the details. I have spent several days considering how to make a relay work cleanly and I am not coming up with anything.

This is one of those occasions where I learn of a "requirement" after the design is 90% complete that had that "requirement" been better understood up front, different architectural decisions may have been made. While the situation which causes this may not be my fault (lack of specifications), I still have to live with the consequences.

 

[VEBill]

The three position manual switch concept applies to ANY automatic system (relay or whatever) wherever you need to have positive manual control. Once you switch away from the AUTO position, you have full manual control no matter if the automatic system is stuck on or stuck off.