High voltage high-side MOSFET contactor switch

[VK3DRB]

Hello.

I have the task of designing a device which will switch a contactor which has a high inductive load and a capacitance to ground as high as 1000uF. The rail voltage is 110V, but can vary from about 70V to 156V or even much higher, and though the contactor is designed for 110V nominal.

I would consider a N-channel MOSFET low-side, except that it is not allowed. Only high-side switching is OK. So can I use a P-channel MOSFET? Yes, but they are too expensive. The IXYS MOSFETS IXTP10P50P have a VDS of -500V but cost $5 each in qty. So I am considering using an N-channel MOSFET to switch high-side. They are cheap, low RDS, plentiful and support a high VDS. I would prefer to use a charge pump circuit to switch the high side N-channel MOSFET on at several volts at above the high voltage rail whatever that be. I have a total of 1 inch by 1 inch real estate (double side) to play with, cost has to be low and the circuit fairly simple.

I am considering a soft start approach (maybe a series resistor and and a capacitor across the gate to the source) to limit the inrush current from the load capacitance at switch-on, or maybe a series inductor to the load, or a current limiter to the FET  (MOVs are not allowed!).

Has anyone ever done anything like this? If so, any suggestion on maybe a single charge pump circuit or IC to drive one N-channel MOSFET high-side? I don't see the reason (or added cost) to use a half bridge output with two N-channel MOSFETS. Or, if I use a P-channel FET, does anyone know of a lower cost device than the IXYS MOSFET? There is a Fairchild device, but the rail voltage is too low.

An SSR to do the whole thing that switches DC is too pricey and probably too big I think.

This circuit has to run in a train for 30 years without fail and work from a -40 deg C to +85 deg C environment! Any suggestions would be welcome. A challenging project, and I have to model the whole thing. I'll try using LTspice, rather than Altium simulator.

cheers!

[T3sl4co1l]

So, the MOSFET is switching the contactor?  Or replacing the contactor?  What's high inductance and what's high capacitance?  The two are contradictory, unless you mean the load is both, in order.

Draw a circuit including your source, load, and the switch you're describing.

Tim

[Marco]

Do you have a khz range clock available? If so any bog standard cheap bootstrap gate driver is easy to change into a charge pump gate driver AFAICS (without the clock too obviously, but it takes more components).

PS. you don't want a gate-source capacitor to slow down turn on di/dt. Better to just increase Rg, with a parallel diode for fast/reliable turn off.

[ajb]

Since you don't care about switching speed, a photovoltaic gate driver could be an option.  The output is inherently fully isolated, and it's basically a single-part solution.  I don't know how offhand how they compare to other options in your application territory, cost-wise.  There's also the Hypertex HT0440, which does the same thing but without the photo part (it's a dual part, but the outputs can be connected in series for higher Vgs).

[VK3DRB]

So, the MOSFET is switching the contactor?  Or replacing the contactor?  What's high inductance and what's high capacitance?  The two are contradictory, unless you mean the load is both, in order.

Draw a circuit including your source, load, and the switch you're describing.

Tim

Good comment! What makes this difficult is I have to date only been told the load is complicated and can vary. They are some sort of active ABB contactor which is not just a simple LCR load, but some sort of complicated load with active parts in it apprently. I am trying to get part numbers because understanding the load is important.

[VK3DRB]

Do you have a khz range clock available? If so any bog standard cheap bootstrap gate driver is easy to change into a charge pump gate driver AFAICS (without the clock too obviously, but it takes more components).

PS. you don't want a gate-source capacitor to slow down turn on di/dt. Better to just increase Rg, with a parallel diode for fast/reliable turn off.

Hi Marco. Thanks. Yes, I did see that spec by IR. Notice the typo in that document under Design Considerations on page 2:

"The 100k 1 kW resistor should be sized according the maximum supply current at the high side of the I R2125..."

Sized accordingly? 1 kW resistor would be bigger than Texas.

I guess not many people have read the document, including IR's own document reviewers.

[VK3DRB]

Since you don't care about switching speed, a photovoltaic gate driver could be an option.  The output is inherently fully isolated, and it's basically a single-part solution.  I don't know how offhand how they compare to other options in your application territory, cost-wise.  There's also the Hypertex HT0440, which does the same thing but without the photo part (it's a dual part, but the outputs can be connected in series for higher Vgs).

Thanks, I never saw the HT part which has an on-board charge pump oscillator. It might be a simple enough solution. Will investigate this further.

[VK3DRB]

Actually the HT0740 is a little brother of the HT0440 chip that it is even cheaper (about 84 cents in qty 100), but the trade off is single channel. Minimum +/-400V isolation, which is fine because there are three 58V bidirectional TVS in series clamping the rail (I would have used one or more unidirectional TVS, but that part of the circuit is outside my control).

For the MOSFET the Infineon IPD50R380CE seems OK - low Rds, high Id, high Vgs and 40 cents each from Arrow. I'll consider a diode across the gate resistor for faster turn-off, but an extra cap between the gate and source is still warranted because it is OK if this thing takes a few milliseconds to turn on, but faster turn-off is best (spec is <50ms from command to opening the contactor), whilst acting as a soft start. Continuous load ground return current monitoring could be used to detect a load short (op-amp and uC, so the threshold is programmable).

Alternatively (slightly more costly and complicated), I could use the dual channel HT0440, and use the second output to drive a second FET to a short out a current limiting resistor at, say, 10ms after turn-on. At the end of the 10ms just before shorting out the resistor with the FET, I detect the load current. If it is too high, a fault condition occurs and the first FET is switched off. So we have a soft start (not using VGS) with short circuit protection and detection, irrespective of the load connected. The problem is, without a current limiting circuit, the FET could blow with a dead short before the monitoring has time to switch the FET off, so maybe a FET with a greater RDS is needed to limit the current until detection.

In the end the whole circuit could be under $2 or $3. And previously, just the P-channel FET alone was $5. An improvement - if it works!

 

[MagicSmoker]

So, the MOSFET is switching the contactor?  Or replacing the contactor?  What's high inductance and what's high capacitance?  The two are contradictory, unless you mean the load is both, in order.

Draw a circuit including your source, load, and the switch you're describing.

Tim

Good comment! What makes this difficult is I have to date only been told the load is complicated and can vary. They are some sort of active ABB contactor which is not just a simple LCR load, but some sort of complicated load with active parts in it apprently. I am trying to get part numbers because understanding the load is important.

You didn't answer what Tim was asking: does the MOSFET switch the contactor or the load itself?

If the MOSFET is switching the contactor then it really shouldn't matter whether it is in the positive or negative line to the coil (ie - high side or low side, respectively, for an N-type switch). The prohibition on the contactor" being in the low side suggests that the MOSFET is the contactor here, and if that is the case then you need to be wary of spending too long in transitioning from on to off or vice versa - the MOSFET will be operating in the linear region during the transition and peak power dissipation limits could be exceeded, even when switching relatively modest  currents.

But again, you need to clarify whether the MOSFET switches the contactor, or *is* the contactor...

[VK3DRB]

So, the MOSFET is switching the contactor?  Or replacing the contactor?  What's high inductance and what's high capacitance?  The two are contradictory, unless you mean the load is both, in order.

Draw a circuit including your source, load, and the switch you're describing.

Tim

Good comment! What makes this difficult is I have to date only been told the load is complicated and can vary. They are some sort of active ABB contactor which is not just a simple LCR load, but some sort of complicated load with active parts in it apprently. I am trying to get part numbers because understanding the load is important.

You didn't answer what Tim was asking: does the MOSFET switch the contactor or the load itself?

If the MOSFET is switching the contactor then it really shouldn't matter whether it is in the positive or negative line to the coil (ie - high side or low side, respectively, for an N-type switch). The prohibition on the contactor" being in the low side suggests that the MOSFET is the contactor here, and if that is the case then you need to be wary of spending too long in transitioning from on to off or vice versa - the MOSFET will be operating in the linear region during the transition and peak power dissipation limits could be exceeded, even when switching relatively modest  currents.

But again, you need to clarify whether the MOSFET switches the contactor, or *is* the contactor...

Sorry, the MOSFET is switching the contactor coil, not the contactor's load itself. I did get some specs on the various contactors. They are not as nasty as the customers spec says and I see no evidence of the 1000uF. ANd it does not sound right. (The customer is someone who took over from a guy who left their company, so I will need to discuss with him his spec.) About the transition stage, the time constant  needs to be quite short if its controlled by the gate voltage. As you said, tread carefully.

[David Hess]

I agree with ajb; use a photovoltaic gate driver if at all possible unless the circuit already shares a common ground with the supply powering the contactor and maybe even then.  My next choices would be an AC driven pulse transformer which is actually better than a photovoltaic gate driver in all respects except cost and simplicity and then a capacitively coupled charge pump.

If the supply for the contactor shares a common ground with the circuit, then a direct level shift with a high voltage NPN or n-channel MOSFET driving the PNP or p-channel MOSFET is feasible but I like using isolated drive schemes to limit the extent of catastrophic failure.  If I did do it this way, I would use constant current drive to the high voltage switch to limit the potential short circuit current.  Then add a series resistor so only about 1/2 of the typical voltage is across the constant current source.  This makes it feasible to handle high voltage surges without damage.

[VK3DRB]

Thanks Dave.

It does have a common ground (8 of these circuits) and the junction not on the board. Hence there is no hope of low side of anything, which made things a bit more messy from the outset.

I am trying an isolated level shifter driving a N-channel. Turn on/turn off is much faster than the application requires (diode in parallel with a gate series resistor). I am using a high side current sense resistor, and op-amps/flipflop to measure current (into a PIC32), and turn off the FET an overcurrent condition. If the FET can get damaged with an immediate huge current with a dead short on the output (I'll model it with the soft start), I'll look at adding add a single common current limiter upstream to the supply voltage using a P-channel FET and a PNP feedback tranny, to protect all 8 N-channel FETS - not too bad because the common P-channel will only need to be 250V rated VDS.

This is not like , and there are regulations to consider  :police:. Cheers, and thanks for everyone's suggestions. I owe you all a :beer:! (An excuse to add these new emoticons.)