So again the same proposal I've made earlier with different types of complex load. Yes, it will be more valid than a resistive load because of it being more akin to what one will connect the amp to in real world scenario. No one tries to say we need testing for all scenario just that we can do tests better and closer to reality if done with complex load.
"Gives one the capability to calculate the response into any complex load.."
-Yeah, the same way that any review could just consist of a parts list or BOM and then it's up to anyone to personally calculate the values they are interested in?
Doing more consumes more time, yes. That's not the point though. The point is to bring testing closer to reality which it will be with complex load testing.
What types of complex loads would you have in mind? What types of speakers do we want to model? How low should the minimum impedance be? How high? Since they come in so many flavors its hard to say what a worst case should be. If we really wanted to do "worst case", that would likely be something like something esoteric that presents either a near dead short or becomes virtually purely reactive at ultrasonic frequencies just above the amplifiers range. The results of that test would at best be it going into protect. Some may tolerate it, but it would severely stress the amplifier.
Secondly, in a well designed amplifier, if the load is able to induce genuinely audible distortion one of three things is likely to happen in short order:
- The protection circuit kicks in, and it illuminates an LED or shows some error message on its TFT display.
- It blows all of its fuses.
- There is a *pffffffftttt*, *POP*, or *BANG!* sound accompanied by smoke and unusual smells.
Or possibly some combination, or even
all three at the same time. This is because there is likely going to at least be some form of instability, or even outright full-power oscillations taking place in the amplifier. We can make test loads that will cripple virtually any amplifier. As I said, things that are purely reactive or nearly a dead short at very high frequencies where the amplifiers gain and phase margin are low are good candidates. However, since most of us here don't have the money for such oddities, we will be using normal speakers. For a well designed amplifier that measures well here, we can already get a good idea of how it will perform from the THD+N and IMD tests. Yes, at higher frequencies in some designs like linears it can induce increased distortion, but this is seldom an issue for two reasons. The first is that its at higher frequencies where there is not as much spectral content, and the second is that, provided its not IMD, many if not all of the harmonics will fall outside of the audio spectrum. If it is IMD, Amir's multi-tone tests at varying loads will give a good indication if we need to worry. If its literally just 32 tones with some noise at the bottom than forget it. Not an issue. If, however, at around -40 dB and below the main tones the test plots are just clogged with shiet then you are likely to have serious issues with even mild loads. And lastly, the resistive loads will already induce distortion. If the range of loads nicely bookends your speaker's impedance minima, then you already have a good indicator of how much you can expect in practice.
So, that really leaves the easiest to hear and most common fault we often see in amplifiers: Channel imbalance and FR issues. For the load to impact the FR the amplifier has to exhibit substantial output resistance or resistance variations. There's no need to use anything reactive. Just load the output down to the minimum rated impedance and see how the FR changes. Channel imbalance is even easier. That will show up on the dashboard at the start of the review as a gain variation between the two channels.
We could certainly do studies with reactive loads and how they impact amplifiers, but for the most part these have already been documented very well in the literature. Tougher is stability as its not always easy to predict, but its also unlikely to be an issue for the majority of speakers that the amplifier could be used with.
Again, all of these issues
can crop up. I ordered a TDA1554A amplifier kit in high school a few years after the IC was introduced to the market. This was the schematic for the amplifier:
And the actual amplifier, complete with its delightfully vintage parts I bought from Radioshack along with my Sony SS-MB150H's that I bought way back in the early 00's:
While it brings back nice memories, even the most casual observer with some amplifier design experience should see what the problems are, not to mention me being greedy and running it just below its absolute maximum voltage rating of 18V for max power (Edit: It could have been a fake IC as well as it should have at least worked ok-ish with its minimalist circuit). The part that was not no nice was it had pretty much no bass and treble. Every EQ had to be the Joker's smiley face, and it wasn't the speakers. Later on I wondered why and hooked an oscilloscope (edit: Fluke scope-meter, to be precise, since we know that grounding a BTL amp's output is a no-no. Still, I wonder how... But it died so it will be a mystery) up. I expected to see audio waveforms. What I didn't expect was what I actually saw. It wasn't oscillations, it was FULL CARRIER DSB AM being generated from the audio, and it was from the audio as the modulation. My only guess is that somehow the carrier from the nearby AM commercial broadcast station was getting into the amplifier and somehow it was modulating it with the audio.
However, absolutely NONE of these effects would escape testing with professional analyzers. Rather, if Amir was asked to hypothetically measure it he wouldn't be able to, because every plot from his AP Analyzer would look like it was drawn by a retard with a box of crayons. If an amp has deficits enough to be audible with speaker loads, it will show up in testing. The only grey area was already pointed out, and that is with headphone amps since their designs can vary so much. Maybe there enhanced testing would be a good idea. But, I see no need to lose any sleep over coming up with some unusual test methodology for speaker amplifiers to incorporate reactive loads as part of the test procedure.
Again, anything is possible as you say, but its unlikely enough to even be remotely audible that it really doesn't justify doing it as part of normal testing, and it has already been investigated pretty much ad-nauseum already in the course of conventional amplifier design.
