As a relative newcomer, this is the area that I find myself most fascinated. As I (think I) learn more about how components may interact, I'm most curious how the 'chain' behaves up to the loudspeaker. I'm not ready to delve into loudspeaker measurements apart from perhaps a simple phase / impedance measurement. With that said, all loads for amplifiers are not purely resistive (as I've come to understand it). So, while standardized measurements with resistive loads are likely indicative of performance, there always seems to be some missing pieces / variables even when looking at how an amplifier performs.
My particular fascination is with the 'nice' harmonic distortions. Separate from the argument and discussion that no part of the chain should distort the signal audibly. I've come to understand that certain phases of certain harmonics relative to the fundamental may be appealing. Some may be less so. Many disagree with the concept, nevertheless the subjective impressions. Either way, I find it interesting. If I have a preamplifier with 'positive 2nd harmonic' dominant distortion and an amplifier with 'negative 2nd harmonic' dominant distortion, I think I've come to understand that some of those distortions would cancel. Similarly, if both had the same type of dominant distortion, then those could be additive. That's all separate, but related, to things that might come about with various noise introduced as components are connected in a typical home environment.
tl;dr - I'm interested in some 'real world' measurements too of combinations of gear, and I hope to be able do conduct some relatively soon.
Where's that leave us? Back to purely subjective reviews, with no constants in the system, all variables?
To me, a simple number of merit like THD probably isn't helpful. The spectra of the distortion products probably is more meaningful. That changes with level and frequency, of course.
Maybe a better figure of merit might be the results of a multitone test, with the tone levels weighted to simulate actual music. Pink weighting, I think. Do it at two power levels, like 1 and 10 watts. Or three. Whatever.
That's certainly easy enough to do now. There's at least two free software packages that will allow for that. Audio Precision systems will do that, too. Not for free, of course. I'm sure that there's more.
Still need system level testing, though.
There is no "nice" distortion really, since any distortion always produces intermodulation products, not just harmonics, and intermodulation products are not harmonically related to the input frequencies. Good type of signal to try for this might be a sweep of a bell tree producing ghost tones lower in frequency.
THD's main benefit is that it will robustly pick up non-linear behaviour in one simple figure - its a great smoke test for problems when troubleshooting for instance. And if its very low you know you don't have to worry a jot about which harmonics are in the spectrum, they are all extremely low. These days multi-tone tests are much easier to do than used to be the case, but there are too many options really, without a standard test you have difficulty making meaningful comparisons sometimes.
THD's main benefit is that it will robustly pick up non-linear behaviour in one simple figure - its a great smoke test for problems when troubleshooting for instance. And if its very low you know you don't have to worry a jot about which harmonics are in the spectrum, they are all extremely low. These days multi-tone tests are much easier to do than used to be the case, but there are too many options really, without a standard test you have difficulty making meaningful comparisons sometimes.
^ I added "sweep of a bell tree producing ghost tones" and "intermodulation products" to my working notes to research, learn, and hopefully try. A lot of new terms in one phrase for me.
"meaningful comparisons"
😄
Thank you!
"meaningful comparisons"
😄 Thank you!
Seems like a lot of different opinions about the measurement conditions. The simple answer to most of them is to standardise a set of measurements, one at 1W, another at 5W, another at 10W and another at full power. Same with frequency. I guess the THD vs Power Output plots would cover most of this. But I have my doubts that they are meaningful if rapid sweeps are done since there wouldn't be sufficient time for the amplifier to stabilize thermally at each power level. Then again, music is rapid too, so maybe it's OK.
Maybe DIY audio should have it's own THD measurement standard? Then we could compare our builds objectively!
Maybe DIY audio should have it's own THD measurement standard? Then we could compare our builds objectively!
^ I'd enjoy that. With that said, I've sat on standards boards / review committees ... oof. A doff of the cap to whomever manages to find consensus.
Whether it's misguided or not, I've used REW and ARTA along with the guidance of the forums here to begin looking at how I can roughly standardize even my own 'measurement package'. It's not free, but what can be done at (to me) a relatively low cost these days is remarkable. The downside is that it can be daunting for someone that does not understand (and doesn't really care to learn) the statistics and some of the functions behind the acquisition, manipulation, and display of the data. I'm delving a little bit into what's actually being measured and displayed (in loose terms), and that's fascinating, but I doubt I'll go deeper. A set package of measurements using something like those two packages and a basic DAC/ADC/Soundcard along with a few other tools would be wonderful.
At a higher cost, but ... still relatively inexpensive compared to the the AP packages commonly discussed are the Quantasylum units. I'm sure there are others.
When I worked on standards previously, a few aspects for consideration (among many) was the distribution / error of measurements within, between, and overall. Conducting that type of work is no small task.
It's one thing to say... run this test this way because everyone finally agreed upon a methodology using a certain set of tools. It's an entirely different (but related matter) to determine how reliable those results may be when conducted over time by multiple people using multiple sets of equipment. Then calibrations etc. etc. I've seen threads touting sub ppm distortions. I trust that it was measured, and generally I'd say that I trust the people doing the measurements, but ... could I measure that and achieve matching results?
That could all be simple with today's tools in this field, but I (per the norm) don't have the foggiest clue.
In addition...
I don't want to steer it too far away from the OP's intention, but I have a few goals. Again, I'm a newcomer to electronics in general and to DIY. So, my goals will likely differ from a manufacturer (assuming they have good intentions) that wants to substantiate marketing claims and/or run R&D / QA / Compliance checks.
One of my first goals is to verify that I've actually reached an intended outcome with a number of the amplifiers I built. Sure, they all 'make music', and I did a few of the basic voltage checks with each build to check operating points. However, do they actually work as intended? I don't know. Maybe the one I prefer the least doesn't "work properly". Others seem to adore it; I don't. I don't think it's terribly relevant to post comments (even though I've done it) about the impressions of the "sound" of a DIY amplifier without first making sure that "it works" and posting accompanying measurements.
Then, I don't think I differ from too many when I wonder why I tend to prefer some amplifiers over others (in my system, in my room, with my ears, etc. etc.). Is it a real 'performance issue' (see first goal)? Is it some sort of interaction with other gear upstream and/or the speakers? Was it a well-executed build, but it's just not the right amplifier for my system and/or my tastes?
Lastly... the ever-elusive... and potentially controversial... do I particularly like an amplifier with a certain distortion profile... or similarly do certain distortions 'turn-me-off'?
So much to learn...
Whether it's misguided or not, I've used REW and ARTA along with the guidance of the forums here to begin looking at how I can roughly standardize even my own 'measurement package'. It's not free, but what can be done at (to me) a relatively low cost these days is remarkable. The downside is that it can be daunting for someone that does not understand (and doesn't really care to learn) the statistics and some of the functions behind the acquisition, manipulation, and display of the data. I'm delving a little bit into what's actually being measured and displayed (in loose terms), and that's fascinating, but I doubt I'll go deeper. A set package of measurements using something like those two packages and a basic DAC/ADC/Soundcard along with a few other tools would be wonderful.
At a higher cost, but ... still relatively inexpensive compared to the the AP packages commonly discussed are the Quantasylum units. I'm sure there are others.
When I worked on standards previously, a few aspects for consideration (among many) was the distribution / error of measurements within, between, and overall. Conducting that type of work is no small task.
It's one thing to say... run this test this way because everyone finally agreed upon a methodology using a certain set of tools. It's an entirely different (but related matter) to determine how reliable those results may be when conducted over time by multiple people using multiple sets of equipment. Then calibrations etc. etc. I've seen threads touting sub ppm distortions. I trust that it was measured, and generally I'd say that I trust the people doing the measurements, but ... could I measure that and achieve matching results?
That could all be simple with today's tools in this field, but I (per the norm) don't have the foggiest clue.
In addition...
I don't want to steer it too far away from the OP's intention, but I have a few goals. Again, I'm a newcomer to electronics in general and to DIY. So, my goals will likely differ from a manufacturer (assuming they have good intentions) that wants to substantiate marketing claims and/or run R&D / QA / Compliance checks.
One of my first goals is to verify that I've actually reached an intended outcome with a number of the amplifiers I built. Sure, they all 'make music', and I did a few of the basic voltage checks with each build to check operating points. However, do they actually work as intended? I don't know. Maybe the one I prefer the least doesn't "work properly". Others seem to adore it; I don't. I don't think it's terribly relevant to post comments (even though I've done it) about the impressions of the "sound" of a DIY amplifier without first making sure that "it works" and posting accompanying measurements.
Then, I don't think I differ from too many when I wonder why I tend to prefer some amplifiers over others (in my system, in my room, with my ears, etc. etc.). Is it a real 'performance issue' (see first goal)? Is it some sort of interaction with other gear upstream and/or the speakers? Was it a well-executed build, but it's just not the right amplifier for my system and/or my tastes?
Lastly... the ever-elusive... and potentially controversial... do I particularly like an amplifier with a certain distortion profile... or similarly do certain distortions 'turn-me-off'?
So much to learn...
This only applies if the simulation models are accurate and additional parasitics are placed in the appropriate locations on the simulation schematic to immulate the behaviour of the pcb. Inductance, capacitance etc.
We added all that to our simulation. This increased the accuracy of the simulation predictions.
That would be nice. I could then use any 'standard' that would favor my amp. 😎
Jan
The Wolverine was tested at 80wrms we felt this level represented an appropriate amount of power for home listening.
Plot attached by @fireanimal
Plot attached by @fireanimal
Attachments
Doesn't mean you need to use it all 😉
You'd be surprised how much power you need to cover the full dynamic range of some songs. There's actually a handy calculator attached to the first post of the Wolverine build thread which you may like to check out.
Pano once devised a simple test to see how much power people actually use. You have to read the first few posts to find out what the numbers mean.
https://www.diyaudio.com/community/...h-voltage-power-do-your-speakers-need.204857/
https://www.diyaudio.com/community/...h-voltage-power-do-your-speakers-need.204857/
The power people use would probably have good correlation to the age of the listener. Other factors would be living situation. e.g. single or with family or how close your neighbours are.
It could even be gender dependent: monotonic increase with age for women and non-binary people as their hearing degrades with age, big peak in puberty, then a local minimum and then a gradual increase with age for men. Anyway, Pano's test did not include things like that.