MOSFET to drive load, cables

Hi there,

I have a circuit dilemma that I was hoping someone might be able to help with. I'm trying to drive a load with MOSFET, but the load is located far away from the circuit, and needs to be driven with a cable (in this case it's a solenoid and a lamp, but maybe something else tomorrow). The most common way to drive things with a MOSFET that I've seen to to connect the load up to V+ and have the FET provide a path to ground, like this:

Now my dillema is that I'm connecting it over a cable, more like this:

If I do it like this, one of the cable pins are always going to be at +24V relative to ground, so that if there's a short near the load (or something grounded touches the pins while plugging it in, etc.), current will flow, even if the device is "switched off" - that is the logic line is low.

I'm wondering if there's an easy way to configure the circuit so that the cable pins are [0V, GND] when off, and [24V, GND] when on (rather than [24V, ?] when off and [24V, GND] when on)?

Any suggestions are greatly appreciated!!


You probably a high-side driver of some kind. International Rectified sells some under the Intelligent Power Switch line, such as the IPS6011 or IPS6041 (both available from their website). They also have some nice over-voltage/over-current/over-temp protection on the die along with the power switch.

That's really the lowest-effort way to solve your problem -- if they don't have enough current drive capacity, you have some other options. You can use a P-channel MOSFET, or you can use of IRF's high-side gate drivers with an N-channel MOSFET.

Yeah, I had seen these high-side switches, but they're relatively pricey. I suspect a p-channel mosfet is the way to go, though i'm having trouble wrapping my head around what that circuit looks like - seems like it'd need another transistor in it to do the equivalent. Anyone know of a simple circuit for that?


You could use an open-collector high voltage buffer to drive the P-channel gate. But the IPS6041s aren't all that expensive -- direct from IRF, they're US$2.50/ea in a TO-220-5 package for singles.

That's great; thanks for the voltage buffer tip - I'll give that a go, if it works out to be less than the switch. I suspect it still will; at $.80 for the FET x 64 of them, the price difference is enough for me.

Thanks again.


I haven't tested it yet, but it should work:

Be careful with the resistor values... if you pull the gate down too far below the rail you'll blow the MOSFET. Make sure that the difference between the gate and the high rail does not exceed the Vgsmax listed on the datasheet. For instance, the circuit as drawn is going to impose something like -23V on the gate, and that's not good for most MOSFETs. +/-20V is a pretty common limit for Vgs on power parts.

One more thing -- the simple stuff we're talking about will only work if you can accept the MOSFET switching relatively slowly. If you're just turning something on and off once every few seconds, there won't be a problem. If you're trying to do PWM on it, there might be. Calculate how long it takes to charge/discharge Qg and make sure that you're ok with the MOSFET remaining in the linear range for most of that time.

More great suggestions. OK, so I understand about not maxing out the gate-source voltage (which is indeed about 20V for the p-channel IRF9530 that I've got). Resistance values seem like a delicate balance here.

Now for the slow switching time, I think I don't quite see why it would be slow. So the total gate charge for the part is 38nC - that seems like a small amount of charge. If it got 1mA, that'd take 38 microseconds to charge, which seems pretty good. (I need it to be about 1 millisecond or less). But I'm not quite sure how to plug that into the circuit here. Any suggestions?


I don't quite see why it would be slow.

If it's any consolation I don't see why it would be slow either. You should have plenty of umph to drive the FET.

Slow is relative, and I forgot to adjust for the domain. I've been doing a bunch of high speed PWM stuff lately which has somewhat different requirements. 38 us is probably more than fast enough. :slight_smile: