FFT Sizes with the QA401 & QA450

The QA450 is designed for really fast burst testing of amplifiers. Because of its size, the QA450 can handle an average power of 5-10 watts, and a peak power of 200W or so. The QA450 is NOT the tool you want if your aim is a load that can soak up lots of power while you tune in real time.

The manual goes into a lot of detail on the power levels and time durations for testing. A lot of engineers like to use really long FFTs for their measurements. But beyond a certain point, there’s not much to be gained from longer FFTs in terms of measurement results.

Let’s take a look at a distortion versus input level plot using the Amp GainAndDistortionVersusAmplitude plugin. For testing the TPA3255 with a QA450 we run with the following plugin options (these tests were conducted on a device similar to the QA450 in many respects, but identical to the QA450 thermally):


Let’s first run this with a 2K FFT into a 4 ohm load:

This was run into the left channel only, so that we can see the temperature comparison with the right channel. At 4 dBV input level, with 26.74 dB of gain, our output is 30.74 dB = 34.4 Vrms = 296W into 4 ohms. Note that after running this test, the temperature of the left channel only climbed to 34C (a 5 degree rise). Note our sample rate is 192K, and so a 2K burst is just 11 mS (2048/192K) in duration:


If we run it again with an 8K FFT, we get the plot below. The gain measurement is unchanged, but we do see some improvement in THD–the red trace (8K FFT) is shows about 5 dB better than the blue trace (2K) at very low levels. But at the higher power levels, the difference is just a few dB.

The temperature of the load has increased, too, with this run:


Running again with a 32K FFT we see some more improvement for THD at the lower levels (green trace). But at the high power levels, the difference is still just a few dB.

But look at the temperature rise: The QA450 is very close to the 60C cutoff.


In short, make sure you understand the minimum FFT size you can get away with when making high power measurements. When the power levels are high, you usually don’t need large FFTs. And in many cases, you might be surprised how low you can go. And look at the speed you get in the end: 55 test points–1 dB apart–done in about 6 seconds.

1 Like