Blog

Sep 30, 2023

Learn And Build A High Side Switch

As electronics engineer I have a mental collection of circuits that I’ve

As electronics engineer I have a mental collection of circuits that I’ve gathered over the years, much like a mechanic collects specialized tools as they work. All engineers do this and the tools in their tool boxes usually represent their project history and breadth.

A useful circuit to have in designer's toolbox is the "high side switch". Like it sounds, this is a circuit that switches the "high side" or positive voltage to a load.

We usually tend to switch things to ground as seen by outputs such as an Open Collector output, the reason being that ground usually is a known entity and is usually low impedance and is at a known voltage. But there are advantages to using a high-side switch in your circuits.

Switching the high side deals with more unknowns than low side; the input voltage, the required output voltage, and the impedance of the source voltage are pretty much all variable. Most often we also need to present a low impedance output meaning that the resistance of the high side switch itself doesn't form a voltage divider with the load where a significant voltage is dropped across the switch.

We could make a high side switch with a relay for example, and there are times when this is still done. Typically the properties of current usage, current capability, coil voltage, cost and size are at odds with each other.

If we use a standard transistor it's a given that we are going to have to live with a voltage drop of some sort. On one end this means that we can't have a 5 volt output from a 5 volt source as we typically lose 0.3v in the process. At high currents the power dissipation also quickly gets out of hand.

Using a Field Effect Transistor (FET) we can make use of some of its best qualities to make a switch. To narrow down on which FET we would use we can start by saying we want a part that normally is turned off and has to be turned on by applying a control voltage, meaning we want an Enhancement Mode FET. Next we decide whether we want to control the device by using a voltage greater than the voltage we are switching (if available) or less than the voltage. For example if we want to turn on 5 volts do we want to do that using 8 volts or more or 4 volts or less? In the example here we want to turn on the high side switch without an additional voltage, in fact grounding a signal is somewhat attractive. That leaves a P-Channel Enhancement FET as our choice.

The traits of any part can be wide and varied so we start by looking for a few important parameters. In switching applications, as opposed to something like a linear audio amplifier application, a low Drain to Source On Resistance is important. This parameter known as Resistance Drain to Source ON or RDS(ON) and a good usable part typically is measured in milliohms. Using ohms law a quick shortcut tells us that at one Amp of current, the voltage drop of milliohms will be millivolts.

Next we want to make sure that we can turn the part on with the voltage we have available. This equates to the Voltage Gate to Source Threshold VGS(thresh) specification. A VGS(thresh) of -1 v means that if we want to switch 3.3 volts we need to pull the gate at least 1 volt below 3.3 v. Using a transistor or open collector device typically can pull a signal within 0.3-0.5v of ground, plenty of room in this case to switch 2.5V using a part with a VGS(thresh) of a volt or so.

Looking at the specifications for several devices shown in the table, we see lots of trade-offs happening. If we select smaller TO-92 packages we get larger, unusable in our case, RDS(ON) values of an ohm or greater. If we go too small of an RDS(ON) the price quadruples. Other parts have too large of VGS(thresh) but the reality is it wasn't too hard to find parts that were usable for the project shown here.

For those interested in peeking under the covers, the reason a larger case like a TO-220 has lower ON resistance is because the case holds a larger chip die. A larger chip die has a larger surface area which offers less resistance. The fact we are using a P Channel device means we need more surface area also, as P-Channel device are generally less efficient than N-Channel devices as they use "holes" for their carrier instead of electrons. The simplest statement is that hole mobility is less than electron mobility.

The switch circuit consists of two basic components, not counting add-on components that we would look at for a production worthy design such as reverse protection diode for the FET. With that said the FET shown does have a reverse avalanche diode built in for both overvoltage and reverse protection but it is the engineer's job to determine if additional protection is needed.

The resistor shown is a bias resistor and keeps the gate at a known value with no other input present, in this case it keeps the FET in the turned off or non-conductive state. In short the resistor keeps the FET in a VGS state of 0v when at least -1v is needed to turn it on.

To demonstrate the effects of low RDS(ON) I have a circuit shown with a 5 Ohm load. Remembering Ohms Law and that E=IR, 5 Ohms load on 5 Volts yields a load current of 1 Amp. Using the same equation the RDS(ON) is easy to demonstrate using 1 Amp of current: A voltage drop of .057 volts at 1 Amp means that the resistance of the FET in this case is .057 Ohms! To put that in perspective for us, the power being dissipated by the device is P=I2R, or .057 watts or 57 milliwatts. We don't really even have to do our thermal calculations to know that no heat sink is needed and that the device will work reliably used this way. To put it in perspective, the power dissipation of the load resistor is 5 watts, 87 times that of the switch itself.

Since we only lose .057 volts in the process of turning on the load this means we can use a 5V supply as a source and still turn on a 5V load such as one of the many 5V controller or microcomputer boards available today…. this is almost as good as a mechanical switch.

With the addition of a simple transistor and a resistor on its base to limit current, we can invert the signal needed to turn on the FET. This means we can create a push-on situation by having the load side voltage connect to the inverting transistor; once high the transistor will keep the FET turned on until something turns off the transistor.

My thoughts in showing a push-on circuit stems from thoughts about battery powered assemblies or applications where maybe a timer is used to turn off after a some amount of time. A controller based load can even turn off its own power supply, though this is a little more complicated where the load needs time to shut down cleanly such as a Linux based system, Raspbery PI for example.

Hopefully this is an example of a simple circuit which you may find useful at some point, at the very least we have explored the properties of a high side switch and a little bit about selecting a component by its specifications.