So I have a project where everything runs at 12V except the brains of the operation, a Raspberry Pi, which needs 5V. This is a situation where I knew I didn’t want to reach for a 7805 linear regulator. Sure, I used one for prototyping because I didn’t want to wait for parts to arrive, but a linear regulator is really inefficient and the project I’m building is to run 24/7. So I set out to find a cheap DC-DC converter and settled on the VXO7805-1000, which is meant to be a drop-in replacement for the LM7805 pictured on the left. Of course, as soon as it arrived I wanted to know just how much more efficient it was, and how long it would take to make up the difference in price, so I set up a test.
I started the test with the same regulated 12V supply I expected to use in the final product, plugged into my homebrew inline current and voltage monitor. From there the power went to the protoboard where I set up each regulator with a 10uF capacitor on the input side and a 22uF capacitor on the output side. Those values were recommended by the manufacturer of the DC-DC converter. From there the power goes through an ammeter and then to (swoon) one of those giant rheostats I’m always so excited about. Let’s talk about it.
This is by far the coolest of the Ohmite rheostats I’ve picked up at hamfests. You can see that the tap in this picture is resting in a spot where there are no windings. That’s how they designed this rheostat to have a “switch off” position. There’s a bump so you can feel it “click on” when it reaches the windings. Then you can see they used three different gauges of wire in the windings. This is how they made it roughly logarithmic. The whole thing is 75 ohms and perfect for an adjustable DC dummy load application like this.
The procedure for the test was to run each regulator at a number of output currents and measure the input currents. An additional meter helped measure the output voltage so we could calculate the power output. Subtracting from the input power, we found the power lost inside the regulator. Working from the cost of electricity in our area and the cost of each regulator, I could then find the break-even point for the more efficient DC-DC regulator.
The data for the 1000mA test should have an asterisk. Somewhere around 700mA the heat in the linear regulator becomes too much and it shuts down. I added a heat sink so I could continue the test, but that changes the cost of the linear regulator significantly, making the DC-DC regulator an even better deal. My application will draw about 250mA. It should take less than 1.5 years to have the energy savings of the DC-DC regulator more than pay for its higher cost. Not bad! I wonder, though, if I should be comparing the regulators at the same power output rather than current output. The output voltages are similar, so this analysis should be good as a ball-park estimate regardless.