Circuit help...it's in a clock.....

[orange_glow_fan]

See the picture. This little circuit was added to an older car clock (electric wind by solenoid) to eliminate most current drain across the wind contacts which tend to burn/pit  over time. It worked just fine for 4 days, then the solenoid went up in smoke and the mosfet shorted out ( I don't know which event happened first)

Firstly..a fuse or current limited power supply would have been a good idea.... hindsight is always 20/02...

They drive the solenoid very hard because it's only activated briefly every 2 minutes or so. .It's not designed for continuous operation of course, just a lot of pull to kick the wind actuator. In fact, if the contacts stick the coil gets HOT in a big hurry! The math says the current drain would be ~5 amps continuous, the coil winding looks like 30 gauge wire. When it was working I measured about 3.2 ohms across the coil.

My question is, do I need more protection for the transistor? Maybe a diode? There is a 'flywheel' diode across the coil.

It's possible the coil simply failed after 50 plus years, it's also possible the transistor failed and destroyed the coil. The transistor is rated at 19 amps so one wouldn't think it would fail...

Thanks for any input..

[gregebert]

If there was any kind of electrical surge/transient, it's possible the gate-source was damaged, though we'll never know for sure. MOSFETs can fail in the shorted (on) state, and that would also fry the coil. Automotive electrical environments are not well-behaved, and prone to transients when cranking the engine, or when the alternator kicks in/out.

The first thing I noticed is the 2 resistors really dont protect the gate-source from overstress (most-sensitive part of  MOSFET). You should put a zener diode between gate-and-source (anode to gate, cathode to source). I  think extra capacitance across gate-source will help because the MOSFET  has about 1500pF, which gives an RC time constant around 1.5usec; that should be enough for the zener diode to conduct, but if you want to play it safe, add 0.1uF across gate/source.

Next, what does the pulse look-like ? In order for the MOSFET to be off, the input signal needs to close to the VDD supply value. Is that 12 volts since this is a vehicle ?

Per the datasheet, the MOSFET will start turning on if Vgs is -4 volts. So, if your supply voltage is +12V, the input must be at least +8V to keep the relay off.

[orange_glow_fan]

The input is,purely mechanical, consisting of the original contacts. So I would assume the gate would live at or near the 12vdc supply  when the contacts are open.

Would a PNP transistor be a better idea?

The clock wasn't installed yet.

Thanks

:

[gregebert]

OK, if the input is mechanical, and basically shorts-to-GND for turnin, then you're OK because the resistors you have will act as a pullup, and shut off the MOSFET.

Next, we need to look very carefully at the circuit layout, because there are ways the free-wheeling diode could en-up not helping at all. If at all possible, mount the main free-wheeling diode directly at the solenoid/relay because it is the source of the kickback. Having a secondary free-wheeling diode near the MOSFET will help, in case the relay is mounted remotely.

Next, what does the relay control ? If it's contacts are switching power from the same source that drives the MOSFET (+12V, gnd, or both), you could have other unexpected glitches from shared inductance. Years ago, I fried an expensive transistor because it shared 6 inches of wire with another transistor that shorted-out.

Lastly, is it possible the mechanical gizmo that turns-on the MOSFET got stuck-on for an extended period of time ? For example, if a motor/gearbox drives it, it seems very possible the motor could stop with the contacts closed if power got pulled at the wrong moment of time.

[orange_glow_fan]

A quick explanation. The clock is basically auto wound by the solenoid. When the contacts close the solenoid is energized which pulls the arm/contact down quickly,  kicking the auto wind mechanism and moving the lower contact arm away from the upper, opening the circuit. As the clock runs, the lower contact moves upwards toward the upper arm/contact until they close and the cycle starts over.

As now modified, the contacts on the solenoid only ground the gate of the Mosfet. (plus the arm still supplies the mechanical kick to the wind asm. )

Clear as mud?

So yes, the contacts could have stuck, but as there is so little current involved in their operation it seems unlikely. ?  I guess the auto wind mech. could have jammed and the contacts wouldn't have been able to separate but there wasn't any evidence of that when the smoke cleared away.

The flywheel diode lives at the solenoid, however there is a 4 inch three conductor ribbon cable that exits the clock and goes to the small transistor board. Adding a diode there should be easy.

You also mentioned adding a zener diode to the gate. What voltage would you recommend?

[orange_glow_fan]

Here is a link showing the clocks operation though I'm not sure if it will work..

https://www.facebook.com/kerry.borgne/videos/3025090357515025/

[gregebert]

Oy vey, Rube Goldberg would be proud.....

The zener diode should be around 12V; it needs to less than the maximum-allowed Vgs for the MOSFET so that the MOSFET is protected. If the zener diode is less than 12V, it will consume a few mA when the solenoid is energized, but that's harmless.

Now, as for why the solenoid fried, I have one theory. If you've ever played around with relays and connected the coil in series with the normally-closed contacts, what happens ? You have a buzzer. The clock is similar in behavior; if the contacts close, the solenoid starts to activate and that causes the contacts to open slightly, there might not have been enough time to 'cock' the mechanism. So, the solenoid would turn on again, perhaps briefly, but still not enough to cock the mechanism. It's possible this could go on long enough that the solenoid is energized almost continuously.

You could use a 555 timer configured as a one-shot multivibrator so keep the solenoid energized for a long enough time to guarantee it gets cocked, and also guarantee the coil is not energized for too much time.

[redrok]

On Friday, January 10, 2020 at 2:26:26 PM UTC-6, orange_glow_fan wrote:

Hi orange_glow_fan;
Interesting winding mechanism!

Some thoughts:

1. Those contacts are quite "Beefy". They are meant to conduct relatively high currents.Often, they require a small spark, on opening, to keep the contact surface clean.
Adding a small capacitor across the contact provides this cleaning spark. (In this case the spark happens on closing instead of opening.)
I would suggest a 0.1uF cap.

2. This 0.1uF cap would also delay the MOSFET turning off as suggested by gregebert.
The RC time constant, for a 10K pullup resister, would be about 1mS. Seems a bit short:
Maybe by making the pullup 100K to get an RC of 10mS would be better.

3. The IRF9340 is OK with an on resistance of 200mOHM is a bit weak but should work fine since it's ON for only a short time.

4. It looks like the contacts need some adjustment. They look somewhat out of alignment.

5. Is that a little metal "Prussian Helmet Spike" I see in the lower contact?
This might have been the "Root Cause" of the failure.
For a clock that has run for 50 years the designers must have done something right.
Maybe just giving the contacts a little contact filing would do the trick.

redrok

[GastonP]

What are the specifications of the flywheel diode? A common mistake is to use a common rectifier in there, which is not useful at all for the intended purpose.

Gaston

[gregebert]

Neglecting switching characteristics for a moment, the only 2 parameters for a free-wheeling diode are (1) it must safely handle the peak inductor current, and (2) the reverse-breakdown voltage must be greater than the supply voltage. In operation, it acts as a regular diode to provide a path for the inductor current until it decays to zero, or a switching event happens.

Now, if we consider switching effects, the diode has a finite turn-on time, on the order of several nanoseconds. As the driver MOSFET turns-off, the current thru it decays (di/dt < 0) and the inductance of the solenoid produces a counter-voltage (= Ldi/dt). If the MOSFET turns of slowly enough, the  negative 'spike' produced by the solenoid might be negligibly small and you wont even need the diode. But that's a big 'IF' and requires detailed analysis of all operating conditions. 

If the MOSFET turns off very fast, it's possible the free-wheeling diode will not turn-on sufficiently fast to clamp the spike to a safe level; given the currents involved in this circuit, I doubt this would happen. But if this was an electric vehicle, engineers will be spending a lot of time optimizing the design tradeoffs and probing around with a scope. 

Lastly, there's reverse-recovery. In this design, the inductor current decays to zero long before the MOSFET turns on so you can ignore reverse-recovery. But in many circuits, such as motor-controllers and inverters, the freewheeling diode is still conducting current when the driver turns on again. Not only does the MOSFET need to supply the inductor's current, it also has to handle the reverse-recovery current (technically, charge or Qrr) of the free-wheeling diode; neglecting Qrr would result in an under-sized MOSFET which could get zapped.

[GastonP]

As we are talking of an event that took days to happen, on a solenoid coil that can take up to 5 amps and in an automotive environment I wouldn't discount that repetitive effects of seemingly negligible effects and perhaps marginal components and unexpected power surges could fry a MOSFET that drives a relay that uses an 1N4007 as freewheeling diode.

We also know nothing about the car itself and we cannot assume that we have the old watch in a new car. Actually I think it is reasonable to assume that the situation is quite the opposite: an old car with extreme power transients all over it at crank up time. One of this cranks, or a failing alternator or dynamo regulator plus a weak (high internal resistance) battery can send even hundreds of volts transients through the power line, zap the MOSFET and then burn the coil.

[orange_glow_fan]

Just to clear the air. The mosfet/clock failed without ever being in the car. It was test running on the service bench, safely connected to a well filtered 13.6 volt power supply. (unfortunately one that does NOT have current limiting..) So I think we can eliminate externally generated events. 

Kerry