The SURE line of class D amps are economical amps deliver some fairly big specs. Class D means that they are switching amplifiers, which normally boast much improved efficiency over Class A and Class AB amplifiers. The silicon and efficiency also lends itself to straightforward miniaturization.
Parts-Express and others offers a range of these amps, ranging from a few watts up to several hundred watts. As we hit the final stretch in checking out the QA400, we wanted to make sure there were no issues with high power, class D amps and external attenuators.
The input limit on the QA400 is +6 dBV and +/-5V DC. Class D amplifiers usually operate from a single supply, and as a result they usually operate the speaker in a push-pull fashion. The average DC across the speaker is zero, but the outputs idle at roughly the supply voltage divided by 2. The SURE amp here was no exception. With a 25V supply, the outputs idled at 12.5V.
While the QA400 is readily suited for measuring line level signals, the high output levels of an amplifier as well as the high DC level presents a challenge. Additionally, the push-pull output of the amp means that achieving a reasonable distortion measurement requires you to measure both outputs differentially. A survey of 4 different approaches were considered:
1) Use scope probe in 10X settings. Combined with 100K input Z of the QA400, this would yield an attenuation of about 91X. Of course, too much attenuation will always cause problems with respect to noise measurements, and the high impedances combined with stray capacitances that normal aren’t an issue in audio circuits means we need to be particularly alert. This is a single-ended measurement.
2) A resistive divider (220 ohms each) from of the speaker outputs to ground, and then tap off the divider. This has an attenuation of 2X. the low output Z is agreeable with the QA100 input Z, and the loading relative to the 6.8 resistive speaker dummy load is negligible. This presents a ~6V DC to the QA400 input, which is a bit higher than the spec limit. This is a single-ended measurement.
3) A differential probe with a 20 MHz bandwidth and 10:1 divider. This is the only true differential measurement of the 3, and for an amp with a push pull output, this is ideal. And the 10X attenuation is right about what we want too. The questions here will be related to noise floor, as most commercial diff probes are more for measuring AC. These probes are quite expensive.
4) A low-cost NE5532-based differential amp with a gain of 0.05. This is extremely low cost, and extremely low noise. When powered from 9V batteries, it makes an ideal amp for measuring the outputs from push-pull amps as we’ll see below.
The specs on the amp are listed below:
For the tests here, the amp was run at 25V.
For the first test, our goal is to rule out why a scope probe in 10:1 mode can be misleading. First let’s look at a 1 KHz signal about 10 dB below clipping. Below we see the noise floor is very reasonable, but it looks like there is some peaking beyond 10 KHz. Also, the THD looks a bit high when you consider the amp specs.
Next, we look at frequency response of the amp. Notice below we see the frequency response at 10 KHz is +15 dB. This seems not plausible for an amp that is designed for audio. As we’ll see below, this quick and dirty measurement yields some very deceptive results.
Next, we look at the same tone with a 2:1 divider and the scope probe clipped onto the 2:1 divider but in 1:1 mode. With the same input level, we see the THD is dramatically reduced (although still not great), and we also see the bump at higher frequencies has flattened out. The green trace below indicates the measured signal with the scope probe grounded. This indicates that the noise overall measurement system is significantly lower than the SURE amp. And the difference between the yellow and green traces are purely from the amps contribution. A check was made with the input signal dropped 10 dB to confirm that the THD stayed the same, and it did. Had it dropped, that would have meant the QA400 was contributing to the THD due to it being close to overload.
A quick check of the frequency response confirms it is much flatter too. There’s about 1 dB of peaking and the response is down about 2.5 dB at 20 KHz. Quite reasonable for an audio amp. But can it be better? Can the distortion be improved by a more accurate measurement technique?
Next, we switch to the differential probe. This particular probe has a 20 MHz bandwidth, but it’s designed for measuring AC, so we need to be aware of it’s limitations for audio. The first measurement we make is with the diff probes shorted, and compare that to the QA400 baseline above. Below this shows the diff probe itself contributes a fair of noise to the process. Next, we mark the diff probe response and that will be shown in green for future measurements.
In looking at the distortion, it appears the differential probe has improved the measurement significantly. But the noise doesn’t come up, which means we are at the limit of the differential probe here. But this does underscore the importance of making a true differential measurement when facing a differential output stage of you want an accurate THD measurement.
And finally, a frequency response measurement shows even more improvement. Here, the response at 20K is down about 1.8dB and the peak at 10K is reduced to 0.5 dB. This is a decent improvement over the quick and dirty single ended measurement above.
Next, a small differential amp with a gain of 1/20 (0.05X) was constructed from an NE5532. The noise floor was measured (green) and the SURE amp output was measured differentially. Here we are showing the same good THD performance as achieved with the expensive diff probe, but notice too we have a substantially lower noise floor. The power out here is about 2W/
At power levels on the edge of clipping (-2 dBV is just below clipping on the SURE amp in “weak” gain), the power output is roughly 11W out and the distortion shows THD has climbed to 0.15%. The amp can be pushed harder, but the distortion rises quickly after this point.
Interestingly, the spec sheet on this part shows the THD should be sitting around 0.05% at 10W remain there up until 20W or so. While we were very close to that at the lower setting above, we’re not there at this particular power level. This assumes that the SURE engineers followed the ST guidelines for their design.
For measuring push-pull high power amp outputs, four different measurements were compared. The final solution showed accurate measurements can be made using a low-cost diff-amp constructed from an NE5532