Why Use Isolated Power SuppliesWith most hobby electronics, all your system parts tend to operate from a common ground. This makes things easy to understand, as you just measure what a device is 'doing' with respect to ground, and you can make things happen in your design.
For example, you can decide that 5V above ground is logic high, and if your micro sees that level on a pin ou can turn on a light. Or if it's an analogue value of 3.2 volts you might think your room is warm enough and decide to turn off your heater.
Sometimes you might here this referred to as an 'absolute value' above ground.
But sometimes, you don't want to know what the difference is between two points that are different to ground.
Low Side Measurement.To make my point clear, lets start with (still) ground referenced example that people may already be familiar with.
Constant Current Load
When I needed to test a Power Supply design, I knocked up this constant current load. Yes I forgot the gate resistor so it's also a nifty little oscillator when limiting, but let's just ignore that error for now...
Constant Current Load
Rshunt lets your measure the current flowing through the MOSFET, where the voltage across Rshunt is equal to the current flowing through Rshunt times the value of Rshunt, and it just happens that we are referencing that voltage with respect to ground.
This is known as 'Low Side' monitoring.
This works for this application, but if you wanted to use this low side monitoring in a power supply, your supply will fluctuate above ground - it will rise as your draw more current, and this can be bad.
I'll Take the High Road
To get around this problem, you need a 'high side' shunt. This puts the sense resistor before the voltage output and thus the voltage output no longer fluctuates around ground (but you do lose 'headroom' before the output, that's the trade off).
Here a differential amplifier measures the voltage across the shunt resistor, and uses this to modulate the MOSFET. And again, all the voltages are still referenced to a common ground.
High Side Shunt Example
Now, the above example works over a limited voltage range. Why? Well, if you are limiting a 12V rail, you can run the Opamp off the same 12V rail, but you need a opamp that can swing it's output right to the positive rail. The LM324 in the above example cannot do this so you need a a slightly higher supply for the Opamp (i.e. 15V) but all is still good.
However, what if you want to limit a 50V supply? If you check the LM324 datasheet the maximum supply rail is 30V and the input voltages cannot exceed the supply voltages.
You might try to find an opamp that can operate at these high voltages (very rare and *very*) expensive or you may try another approach.
If you take your multimeter, you can easily measure the voltage across the shunt resistor. This is because it's operated off it's battery, and none of the circuit is referenced to your system ground.
If you then consider the case where you are measuring the efficiency of a system (i.e. a boot converter) you need to measure voltage and current in, and voltage and current out, and both of these current measurements require high side current shunts. In fact Dave Jones explains this very well at the EEVBlog here.
As Dave explains, you need to isolate each channel. Each reason basically boils down to the fact that if your measurement system has a common ground, you will short out your system under test. THe following pictures illustrate the problem here.
If you are using 4 battery operated meters (isolated supplies) you can easily measure the input and output powers. If you think about it the current meters above are really voltage meters measuring the voltage across the shunts, so your system is really 4 isolated voltage meters.
Now, if you have a 4 channel data logger, the temptation might be to replace the 4 meters above with the single device.
You might then be surprised to realise that you now have introduced short circuits in your measurement system - which can be eliminated with using isolated supplies per channel.
These isolated channel data loggers are out there, but they are expensive. You might have guessed by now I want to perform the above measurements, don't want to pay a fortune for equipment to do so, and have decided to roll my own meter.
In the teaser above you can see I've started to load my prototype. I'll follow up with design and build details soon, but first there's domestic work to be finished....
What's this Sunshine stuff all about anyway?