From my perspective. No.
Once it is determined that certain criteria matter above others, it becomes a subjective evaluation. IMO, All attempts should be made to portray each facet equally. I think that the observer / user should determine how the information should be analyzed for their use case. I also think "quality" should not be a part of the definition / criteria.
However, some may say that "scaling" is OK, and I would agree. As an example, I would consider a power metric scaled logarithmically as "fair"... since in-general that's (to me) more meaningful than linear. To me, it may actually be an easier comparison. 10W vs. 20W is much more meaningful than 100W vs. 120W (to me). Scaled logarithmically ... I think power (and then by definition power to "cost" could be more interesting.
--------
Overall
I think that cost should be left out, but... I also think that there could be more discussion around how the amplifier's power 'rating' should be determined if it is to be included as any part of a metric. To me... as a layperson... I think it's currently one of the most useless metrics in all of audio (at least in the USA). I think it could be very useful.
When the final result of a test can be used as continuous data (numbers); I tend toward scaling the result (if needed) vs. bucketing results based on assumed needs and/or arbitrary break points (however accurate those assumptions may be). A graphical depiction of a 1.1 vs. a 1.0 is totally fine (if the results are statistically different) Why does it need to be a 1 or 2? Are certain results rounded / truncated?
That removes all the noise in the discussion around what's "better". It also removes (with the exception of operator time and potential equipment costs) the discussions around which tests are "more valuable" to include than others. It also removes discussion around where to set the "break points".
-----------------------
Separately - Unless a direct correlation / causation has been shown between two tests, perhaps include them as options. IMO, minimum criteria could be set for some sort of "submission". As an example, if it's determined that one of the distortion tests is perhaps the most indicative of linearity, then that test could be set as the minimum requirement in that category for the submission of a result for comparison. The other tests would / could be run, because they may reveal additional information. However, IMO, testing should not be a "just measure everything" approach. That can be costly / time consuming, and would likely yield fewer participants.
I advocate for a defined series of tests with defined parameters for data presentation with no subjectivity involved. Other than what tests may be included as the "minimum", it removes all argument re: validity. Use it, ignore it... whatever.
Perhaps... sometime down the road, if a database is compiled, people can have a more thoughtful discussion around what amps may be "better" => higher "quality" overall. Also, then, maybe it could be somewhat fun to see if any of the data correlate with properly published subjective evaluation data vs. anecdotes. The discussions around listening tests of any kind (however fun they may be to read) are completely irrelevant.
As always, it's simply an opinion. I'm offering mine from the layperson's perspective.
It's worse than that. One could — in principle — go full reverse Volkswagen 😆 (or is that just full VW?). A processor that does the power monitoring, ramping the bias up or down based on temperature, etc, checks for "stupid signals" and lets the amplifier run super-hot for THD tests, and back to normal for normal usage.
One rare example ;
https://www.diyaudio.com/community/threads/anybody-have-scan-of-an-483b-motorola.99967/
Black Devil is the name of one simple low power AB class amp where regulated PSU directly regulate soft cliping of output signal , (#13) .https://www.diyaudio.com/community/threads/anybody-have-scan-of-an-483b-motorola.99967/
For consideration re:
I'm trying to think of other durable consumer goods that have clearly objective performance properties along with "safety" and or "reliability" along with cost factors / aesthetics etc. The first thing the came to mind (triggered by the VW response) is cars.
Perhaps there could be a "performance" data section along with a "features" and cost section. Think consumer reports / car rags etc.
#9 is a little trickier (for me) as a layperson. It's not a performance metric. Providing "points" or a rating for the inclusion of parts that may not be needed in certain designs seems counter-intuitive. It also (almost by definition) would increase the cost / Watt. Thus, it's a conflating metric. Is it better to have an amplifier that does not need these "safety features" than one that was perhaps poorly designed and uses a couple of bolt-on boards as a 'Band-Aid'? Also, the discussion of what type brings value (works properly) is really tricky. It almost gets into an engineering evaluation of design vs. an objective measurement.
#11 and any discussion re: environmental impact (energy usage / materials sourcing etc. etc.) ... oooooooof. I strongly encourage just leaving it out. Talk about an endless discussion ...
You'd likely have two general categories of people: those that don't care one iota; and those that care so much that they'll argue incessantly about how detailed you need to be to really understand environmental impact. A simple power usage may be a meaningful metric... but ... still...
And any discussion re: environmental impact etc.
I'm trying to think of other durable consumer goods that have clearly objective performance properties along with "safety" and or "reliability" along with cost factors / aesthetics etc. The first thing the came to mind (triggered by the VW response) is cars.
Perhaps there could be a "performance" data section along with a "features" and cost section. Think consumer reports / car rags etc.
#9 is a little trickier (for me) as a layperson. It's not a performance metric. Providing "points" or a rating for the inclusion of parts that may not be needed in certain designs seems counter-intuitive. It also (almost by definition) would increase the cost / Watt. Thus, it's a conflating metric. Is it better to have an amplifier that does not need these "safety features" than one that was perhaps poorly designed and uses a couple of bolt-on boards as a 'Band-Aid'? Also, the discussion of what type brings value (works properly) is really tricky. It almost gets into an engineering evaluation of design vs. an objective measurement.
#11 and any discussion re: environmental impact (energy usage / materials sourcing etc. etc.) ... oooooooof. I strongly encourage just leaving it out. Talk about an endless discussion ...
You'd likely have two general categories of people: those that don't care one iota; and those that care so much that they'll argue incessantly about how detailed you need to be to really understand environmental impact. A simple power usage may be a meaningful metric... but ... still...
I do think the assessment categories and criteria have to be kept reasonably simple. You could end up with many measurements that would only be possible to do with a lot of time and specialist gear.
I would envisage the measurements to be able to be done with a good quality sound card, a scope, 2, 4 and 8 ohm load resistors, an as yet to be defined speaker dummy load, and a selection of capacitors between 1nF and 2.2 uF.
It is in the nature of a standard like this that not everyone will be happy and not all amplifiers will be covered. If however most cases are covered fairly, it is a good step in the right direction IMV.
We should be very clear about what each measurement criteria is attempting to do, what major flaws it will uncover and how it can improve overall amplifier design quality in the long run.
I would envisage the measurements to be able to be done with a good quality sound card, a scope, 2, 4 and 8 ohm load resistors, an as yet to be defined speaker dummy load, and a selection of capacitors between 1nF and 2.2 uF.
It is in the nature of a standard like this that not everyone will be happy and not all amplifiers will be covered. If however most cases are covered fairly, it is a good step in the right direction IMV.
We should be very clear about what each measurement criteria is attempting to do, what major flaws it will uncover and how it can improve overall amplifier design quality in the long run.
Last edited:
Environmental impact is a good reason not to design for much higher powers than you normally use; lower power means less silicon area, smaller heatsinks and for class-(A)B, smaller product of the supply voltage and the quiescent current. So 5 points up to 30 W per channel and then 1 point subtracted for every 10 W more 😉 ?
I think you will find that a Class B amp of the same power output as your Class A amp will have nicer sounding overload if both used 'conventional' power supplies. I've tried this in Jurassic times and done some listening tests.
You get the audible benefits just by using laptop PSUs for a small Class A amp. It doesn't have to be 'noiseless'. Just smooth enough not to make an overloaded signal sound more yucky. It's the sawtooth @ 100Hz that makes overload with conventional supplies yucky.
There are different issues with SMPS power supplies, namely limited output current, strictly limited. Speaker with violent complex impedance dips and EPDR dips eats a lot of current. Current limiting (and shutdown) in such cases is much more audible and unpleasant than usual ripple clipping of linear PSU. Been there.
It’s recovery from clipping that’s important. When an amp clips it is running open loop, which is not a normal situation.
Excellent... so overload characteristics may conceivably go into 'safety'.
All the discussion seems to be around how it sounds when / if an amp is overloaded. Potentially, but not with certainty, a better understanding of the power (voltage and current) delivery capabilities of an amplifier would help prevent people from doing such things. I can't say with 100% certainty, but even as a layperson, I don't think I've ever unintentionally taken an amplifier to clipping during normal listening.
I'm not in any way saying that clipping/overload behavior isn't important, particularly if it can drive an amplifier into oscillation and take out the amp or other gear. However, if we're discussion the audibility of clipping... It seems like maybe... make it audible and nasty. Then people may not drive their amp to clipping. Is that foolish? Assbackward?
All the discussion seems to be around how it sounds when / if an amp is overloaded. Potentially, but not with certainty, a better understanding of the power (voltage and current) delivery capabilities of an amplifier would help prevent people from doing such things. I can't say with 100% certainty, but even as a layperson, I don't think I've ever unintentionally taken an amplifier to clipping during normal listening.
I'm not in any way saying that clipping/overload behavior isn't important, particularly if it can drive an amplifier into oscillation and take out the amp or other gear. However, if we're discussion the audibility of clipping... It seems like maybe... make it audible and nasty. Then people may not drive their amp to clipping. Is that foolish? Assbackward?
I have not read all the 9 pages, but one of the first things to come to my mind is testing and measuring power amps with real, preferably large woofer loudspeaker systems as as load, thus not a simulating network. Purely out of curiosity, I for one would be interested to see if Blameless amps- that is i.m.o. what we are basicaly talking about- in reality remain unaffected by large woofers "talking back" (back emf) to the amp, especially when driven harder than just a few Watts. Large systems such as the AR9, KEF105, B&W 80x and the likes thereof would be very nice as test bed. To stay sane & healthy, that speaker should be placed in isolation somewhere, in order not to deafen the measuring technician/hobbyist.
^ You'll now rule out 99%+ of anyone considering these tests if that would be a required criteria.
Is there a way to simulate it somewhat realistically with any kind of (readily available) testing gear / loads? Again... novice here...
Is there a way to simulate it somewhat realistically with any kind of (readily available) testing gear / loads? Again... novice here...
If you make R-L-R//L//C load
it acts exactly like the speaker, with "energy exchange" as well. No difference. There is no difference whether the energy is stored in mass/compliance or capacitor/inductor. Same "back EMF".
So back drive your amplifier under test from another amplifier via a series resistor of around 4 Ohms. I doubt that you will find anything unusual.
And one needs an SMPS with THREE TIMES the current capability that is demanded for full sine wave operation - for at least 50 milliseconds, repetitively to ensure that this never happens. With nominal speakers impedance. You want 4 or 2 ohms? Double it and double it again. It needs to supply the PEAK current, not average, without misbehaving. That can be one very big very EXPENSIVE power supply. Lighter than a 60 Hz transformer yes, but not cheaper. And totally forget about the idea of sizing it down just to handle the average draw at 1/8 power with pink noise. You can get away with laptop bricks for small amplifiers because they are oversized for the applications. 19 volts with 8 ohms load is under 3 amps.
Conducted 120 Hz ripple during clipping isnt very audible when it’s only a couple of volts out of a 60 or 80 volt swing. Those same two volts are a lot when your amp will only give 20V and you’re not half deaf from how loud it is during the event. It can be eliminated entirely by limiting the output swing to never exceed the bottom of the ripple trough. Easy on big amplifiers, not so easy (and expensive, relatively) on little ones. Dont even get me started on cheap subwoofers with 3300 uF caps in the supply. By far the worst offenders I’ve heard. Cheap price point “pro” amplifies pull the same $#1+.
I don't want to get too far off topic, but this is very interesting (to me) and perhaps particularly relevant to testing for failure modes / worst cases. Again, forgive the layperson's terms, but... safety features.
As a novice, I've been fiddling with using SMPS for a few amplifiers where more traditional linear supplies were typically used. In one particular application, I'm using two SMPS as a bipolar supply.
Is it a realistic / conceivable failure mode for one side of the waveform to "hit" one of the supplies quickly enough to where it would put one supply into protection mode / shut it down while leaving the other working properly? In some amps, that may lead to taking out a decent part of the circuit.
Edited to add - and ... perhaps taking speakers with it also...
With thanks in advance for any insights either of you could offer (or anyone for that matter).
Even if you had two completely separate supplies, one for - the other for +, you need both up for the amplifier’s front end to work properly. If you shut one side down from overcurrent the VAS isn’t going to like it very much. It could cause all manner of misbehavior - up to and including latching the speaker to the rail. Better have that DC protect relay…..
Output DC protection board with independent small linear PSU is a cure. Or circuit design with clamp diodes that does not run to one rail if the other one fails. As a designer, you have it in your hands. As a consumer, not so much.