This is an informational post for folks doing power amp work.
Over the past three-plus decades I've spent doing audio engineering, I've accumulated a small collection of dummy loads. For the uninitiated, a dummy load is a high-power resistor used to present a near perfect 8 or 4 or whatever ohm load to a power amp for testing. Something I learned the hard way just now is that they're not all the same.
While I was doing performance verification of an experimental power amp design this week, I just wasn't seeing the THD performance the theory was predicting.1 The only thing I could think of that might possibly be causing the performance shortfall was the grounding scheme I used. In theory it was perfectly sane, but it was possible I might have missed something. So, I decided to try some others. No matter what I did though, I couldn't get the amp to do better than 0.04% THD at 10kHz at 5 watts. That's not completely wretched for 10kHz, but I knew the circuit was much better than that.
I spent most of yesterday on other things and eventually came back to this in the evening with yet another grounding alternative to try. At that time, I had grown tired enough of the slightly inconvenient way the I had to attach the dummy load I was using to the output terminals that I decided to switch to one that made the frequent re-connections a little easier.
And there it was: 0.00052% THD at 10kHz at 5 watts. That's not a typo. The THD was now almost 106dB down from the fundamental and mostly 2nd harmonic instead of 3rd harmonic observed previously. This was the expected performance and a full two orders of magnitude better than the previous measurements. To verify it was the dummy load that was responsible and not the changes to the grounding layout, I re-measured the other channel, which still used the original grounding scheme. It showed nearly identical, much-improved results.
In retrospect, I feel sheepish for not validating the suitability of the first load for the intended application. While the device is marketed as a 200W non-inductive dummy load, the supplier doesn't make any voltage linearity or thermal modulation claims.2 And to be fair, for the common use case of making sure output stages don't melt or blow up under load, they suffice. But as devices for making precision measurements, they clearly don't.
With this post, I really wanted to provide you a working example of how things that aren't a big deal in more mundane settings become a big deal when it comes to precision performance and measurement rather than warn you about a specific dummy load. While you're almost always best off questioning whether things aren't working as expected because of something you did, sometimes, just sometimes, it is your tools. Don't be afraid to question everything.
And in case you're wondering what does work well, the load I switched to is the classic Dale NH-250 I typically default to. If you decide you want one of these or a smaller sibling, make sure to get the non-inductive NH version rather than the standard RH version.
