The lowdown on loads for device testing
Once you have designed some kind of electronic device you have to test it. If it is a power supply, AC or DC driver or even a battery charger you need some form of load. In some of my designs the load is in fact used together with a supply to create a given input current and not to test for the output of the design. Depending on what is available and the cost you may buy a commercially available piece of test equipment or end up making your own. The load characteristics also play into the selection and sometimes you may need two or three approaches to completely test the unit. The proof of the pudding is in the eating, as they say, but often the targeted load is not available or cannot withstand the power dissipated when a bug arises during test, so you need to simulate.
Of course the heart of any load selection is the power dissipation. If you are designing a 4-20mA current loop 1/16W resistors may be all that you need. 0-10V at 10mA also doesn’t pose much of a problem, but a 1KW 24VDC power supply - that’s a different story.
Let’s start out by looking at an AC load. The simplest load for an AC output is a lamp. Figure 1 shows one of my creations that uses incandescent light bulbs (we used to call them “globes” back in Southern Africa). The load presented differs greatly from startup to steady state, but you can’t beat it for simplicity. Recently, replacement light bulbs have become more difficult to obtain although I am told the ones intended for farm use (higher reliability) are still available in rural areas.
Figure 1. Good old incandescent light bulbs (including some 150W and 200W ones) used as a load. It is possible to switch bulbs into and out of the circuit using the yellow contacts on the connectors. A great advantage of this system is that you really know when the load is active, and it is quite pleasant in winter!
There are a few commercially available AC electronic loads like the Model 63800 from Chroma or the 4600 series from NHR. I cannot vouch for these as whenever I have needed to generate an input current I have resorted to my AC Current Generator (“An AC Current Generator” Parts 1 through 4).
Logically, load resistors would appear after my discussion of the light bulbs, but since they can be used for AC or DC I will use the discussion as a metaphorical bridge. Since resistors are passive components they can be used for AC and DC loads. There are decade boxes that can handle some power like the IET HPRS series but they do have limits and we have needed quite a few over the years so we have built our own. Figure 2 shows our development using 4 banks of 3x3R 225W resistors paralleled, i.e. 4 banks of 1ohm at about 600W each. Each bank can be switched in or out using circuit breakers and we achieve fine tuning by adding a lower capability load sink (like a decade box) in parallel.
Figure 2. High power resistive load. Note the fans in the base for cooling and the fine tuning current sink in the orange on the right. The protective wire grille (matching the one at the back) has been removed for the photo. This is also an excellent source of heating in the winter.
One of the problems with this approach is the inability to control the load electronically. To do this you need to control the gate voltage on a MOSFET for a particular current. This uses the conventional op-amp driver with feedback as shown in Figure 3. Looks simple, but keep in mind it can dissipate a lot of heat, along with all the analog gremlins like oscillation, amplifier headroom to name but a few.
Figure 3. Typical configuration for an electronic load.
Figure 4 shows one of our models that we developed for in-house use.
Figure 4. A 240W electronic load. Of course it does not use a single MOSFET, but several connected in parallel in order dissipate the power. It also has a pair of fans in the base.
There are commercial electronic loads like the PEL-300 from GW Instek or the range from B+K. These can have additional features like constant current, constant voltage and constant power. I have even seen a Kickstarter project, but we need so many of these that it was viable to produce our own.
As I hinted at earlier, lab loads are not always the same as the real world. Often there is inductance and in the case of batteries there is a back EMF (see EDN’s article “Power supply simulates battery”). Batteries are generally problematic since you want to be able to test the output at different states of charge and using a real battery is impractical. Startup conditions can also produce different results, so do keep in mind that you should test in the real world or you may get a surprise.
This blog in fact was the result of an upcoming blog I am working on that describes a burn-in chamber for our range of switch mode power supplies (SMPS). Watch for it!