I've got this little problem with my formula student project that i need to solve. Basically i'm building a paddle shift system for the car gearbox, and to select gears we have a linear actator that we would like to use. The actuator draws 35A at 12V for a very short period of time, no longer than 1 second at any time. But obviously some kind of switching system is necessary.

Initially i planned on using a relay switch, driven directly from a 555 chip. But i have since decided to use logic gates instead and its output current is not sufficent for the relay, which requires 120mA, therefore an additional switching stage will be needed.

I honestly don't know much about electronics, being a mechanical engineering student. What do you think will be the best way to switch it? Drive the actuator directly through a MOSFET? or use a smaller MOSFET to drive the relay. The datasheets give so much information but i am not sure what i need to look at. I am worried about heat dissipation if using the MOSFET to drive the actuator directly, but on the other hand it seems very redundant to use a MOSFET to switch on the relay.....

Any thoghts?

 

How short is that period of time?

By driving the actuator with a MOSFET directly you will require a quite large heat sink. However, since you have the relay, you can use a BJT or a MOS to drove the relay.

 

most of the time it should be less than 500ms, would that be long enough to warrant the use of a heatsink?

 

most of the time it should be less than 500ms, would that be long enough to warrant the use of a heatsink?

You need to use a low Rds(on) MOSFET, like the IRL2203, but you will still need a heat sink.

If you use the IRL2203 which has a Rds(on) of 0.007 ohm and you pass 35 Amps through the MOS it will dissipate 8.6W.
Without using a heat sink, if the MOS dissipates 2.41W its temperature will rise to 150 degrees above the ambient temperature and it will burn.
If it dissipates 8.6W for 0.5 seconds then its temperature will rise much more and it will get destroyed. Thus you need to use a heat sink.
Also, note that the cooler the MOS runs the more its life.

 

For the current levels you are working with, MOSFETs would work great.

The only remaining issue to mention is safety. Use a good double sided PC Board with thick tracks on Source and Drain, use 2-4 identical MOSFETs in parallel to reduce the total resistance, lower size of individual heatsinks, and extend the life of the MOSFETs. In addition, keep the driver board physically separate from the controller board, or at least the power rails. Overall heatsink area may be larger than with a single MOSFET, but a few smaller ones are easier to find and install than one very large one. Old CPU/chipset heatsinks are cheap to free and work well in pulsed applications.

Calculations suggest a minimum size as 8 gauge stranded wire for a 1 second burst at 45A, with a wire length of 5 feet. This is double the time your design anticipates, for good reasons.

Wire burning is a nasty thing, I've seen some pretty badly scarred hands on people who tried using 14ga wire to power amplifiers, electric motors, and other high draw items. When it starts smoking, first reflex is to grab it and pull it out of the quick release terminal. Often, enough time has elapsed to the point where the wire has since transformed into a hot wire cheese slicer.

 

Thank you very much for the info everyone, I'll probrably go for the IRL2203 and find a suitable heat sink for it.

The only remaining issue to mention is safety. Use a good double sided PC Board with thick tracks on Source and Drain, use 2-4 identical MOSFETs in parallel to reduce the total resistance, lower size of individual heatsinks, and extend the life of the MOSFETs. In addition, keep the driver board physically separate from the controller board, or at least the power rails. Overall heatsink area may be larger than with a single MOSFET, but a few smaller ones are easier to find and install than one very large one. Old CPU/chipset heatsinks are cheap to free and work well in pulsed applications.

You've mentioned using a PCB to mount the driver board stage with the MOSFET, How are PC Board be made? would something like a stripboard work as well?

 

Yes, a vero board will work too, if the circuit is made properly.

 

But wait...you have to take the maximum rate of shifts -how many times per minute the driver can shift gears. Compare this to the transient thermal characteristics of the MOSFET and you can decide how much heatsinking you really need.

BTW, using a MOSFET is a much better choice than a mechanical relay, particularly with repsect to operational lifetime. Make it work.

 

If you use veroboard, keep the length of tracks between input and output as short as possible, and use two tracks each for sources and drains, with a wire tinned over them to make up for the lost copper area with all the holes in the board. Using a single, unmodified track will result in several slow-blow fuses - note width of track around holes.

--ETA: I don't mean to sound paranoid, scare you, etc. I've just seen too many projects where "it burns" was the mode of failure, rather than something less smelly.