Richard, I basically agree with you that .01% (almost anything) is a useful limit to try to design below for any 'realistic' peak signal passing through the gain stage. For phono and preamps, this is an easy target, but with power amps, sometimes I have to let more distortion generate at high output levels (that are usually not reached). However, .01% at listening levels is mandatory, AND the distortion (usually harmonic is measured) must be essentially only 2, and 3'rd harmonic. This is not always easy, except with lots of extra negative feedback, which appears to create further problems, at least in my opinion. and so forth.
That article has bad premises. Fordism worked because there were already enough countries sufficiently scientifically and economically developed that their citizens in large numbers could buy primitive automotive products which were an advance on horses and steam for personal transportation and small freight loads. As economies developed further then automotive products became vastly more refined, affordable and ubiquitous.
The sale of artistic performance follows a similar economic historical process and thus the audio industry was borne and the products also have become vastly more refined, affordable and ubiquitous.. The refinements in audio have always been in the direction of greater realistic illusion and portability.
Improving the audio illusion is a more difficult problem than providing safe, reliable, comfortable transportation due to the horrifying opacity and complexity of human subjectivity.
You decide which tests... do multiple different distortion tests to find your own threshold limits for that/those particular test(s). What do you get for your result? But try to do what you find to be the most valuable and which you are the most sensitive. But, I'll take what ever you do as one persons threshold for that test.
THx-RNMarsh
Last edited:
Brad, John, Richard,
I am fully aware that THD and IM are interrealted. But exactly how I think is still not quite clear. I have seen and listend to amps with a very modest THD rating, but a very good IM rating, but also vice versa. Because their retaionship seems to be very variable with no guarantees in between, I mentioned them saparately, just to make sure the message has gone through.
I might also have mentioned load tolerance. I mean, if my amp produces say 0.02% at rated power into 8 Ohms, but 0.1% THD at the same voltage output into 4 Ohms, the message is clear - it not nearly as load tolerant as it should be, some increase of THD is to be expected, but if it doubles into 4 Ohms over 8 Ohms, I have a problem. Possibly resolved by using an extra pair of output devices, to make them share the load still better.
What the "objectivist" fedayeen here obviously cannot grasp is that measuring THD is not just a sales gimmick, but an excellent overall chekup of the whole amp. A good first overvirew, so to speak. Crossover distortion, if it's bad enough, eventually shows up as THD, typically in a sudden jump above some frequency, in most cases below 300 kHz. Like say having a THD of say 0.1% at 100 kHz, but like 5% over 300 kHz. It's a sure-fire way to actually see that something is wrong with your amp. Waiting for it to start oscillating is kinda stupid in my book, oscillation very probably had a prelude to it in sudden THD rise.
If THD rise is relatively linear and in some proportion to the increased test frequency, with no decay funnies, chances are it's going to work just fine, although some surprises are always possible.
As for souind qiality, ask yourself - what is the one BIG difference between tube and SS audio? How do the tube people get away with THD and IM factors of 5%, and still call it "the sound"?
A lot of tube audio has no GNFB loops, and consequently no waiting for the correction signal. If they do have a GNFB loop, it's usually very low, 3, 6, maybe 9, sometimes all of 20 dB of GNFB. In my book, at 5% THD and INm you are listening to some nicely packaged distortion, so I would not expect a dedicated tube man to ever become reconciled with relatively distortion-free sound. Heck, they WANT that distortion. So what do they do? They call it a gimmick because they can't do it. So, to me, they are of no consequence.
I am fully aware that THD and IM are interrealted. But exactly how I think is still not quite clear. I have seen and listend to amps with a very modest THD rating, but a very good IM rating, but also vice versa. Because their retaionship seems to be very variable with no guarantees in between, I mentioned them saparately, just to make sure the message has gone through.
I might also have mentioned load tolerance. I mean, if my amp produces say 0.02% at rated power into 8 Ohms, but 0.1% THD at the same voltage output into 4 Ohms, the message is clear - it not nearly as load tolerant as it should be, some increase of THD is to be expected, but if it doubles into 4 Ohms over 8 Ohms, I have a problem. Possibly resolved by using an extra pair of output devices, to make them share the load still better.
What the "objectivist" fedayeen here obviously cannot grasp is that measuring THD is not just a sales gimmick, but an excellent overall chekup of the whole amp. A good first overvirew, so to speak. Crossover distortion, if it's bad enough, eventually shows up as THD, typically in a sudden jump above some frequency, in most cases below 300 kHz. Like say having a THD of say 0.1% at 100 kHz, but like 5% over 300 kHz. It's a sure-fire way to actually see that something is wrong with your amp. Waiting for it to start oscillating is kinda stupid in my book, oscillation very probably had a prelude to it in sudden THD rise.
If THD rise is relatively linear and in some proportion to the increased test frequency, with no decay funnies, chances are it's going to work just fine, although some surprises are always possible.
As for souind qiality, ask yourself - what is the one BIG difference between tube and SS audio? How do the tube people get away with THD and IM factors of 5%, and still call it "the sound"?
A lot of tube audio has no GNFB loops, and consequently no waiting for the correction signal. If they do have a GNFB loop, it's usually very low, 3, 6, maybe 9, sometimes all of 20 dB of GNFB. In my book, at 5% THD and INm you are listening to some nicely packaged distortion, so I would not expect a dedicated tube man to ever become reconciled with relatively distortion-free sound. Heck, they WANT that distortion. So what do they do? They call it a gimmick because they can't do it. So, to me, they are of no consequence.
Less than 01 is not a strict requirement...... unless you are talking about the whole process from mic preamp to the speaker..... the accumulated affect being above your detection threshold.
Notice I use the work detection. A speaker has highest distortion but what small added distortion level change can you detect? yes, what you can detect will to some degree depend on the speaker/headphone distortion. I use very low distortion speakers and headphones for music and to evaluate My detection limits/thresholds.
THx-RNMarsh
John, I'd like to remind you of something I know that you are aware of. Otala postulated that goo sound will result only in amps where all distrotion mechanisms contribute equally to te resultig sound. NFB will do its part, but if what it corrects was initially unequal, that correction will also be only partial. So far, so good, it's easy to understand that.
But, I aksed myself - what if I have harmonics which do not decay appreciably and have a relatively high amplitude from 2nd to 20th? How might that sound? So I developed an amp which behaved a lot like that, very small harmonic decay rate, PEAKING at around 0.05%. I got a very even, lnear sounding amp, which reminded me of both Otala and what Nelson Pass said here again (he had said it before) that he likes to leave a little distortion behind. Strange as it may soud, there is something in it - distortion levels will matter much less if they are even about linear on all accounts.
I was reminded of that by this discussion, so I dug up that particular project and will investigate it further in the coming days. This concept intrigues me and I'd really like to know.
Richard, I made that quite clear a few messages ago, but I'll repeat: for me, anything below 0.1% THD and 0.05% IM is essentially distortion free, assuming we are talking about not average, but PEAK distorion into the smallest designed for load impedance, whatever you as the designer specified as the smallest permissible load impedance, 4 Ohms or less.
Andrew (Bonsai) really hogged it with his 180W e amp, because he specified 3 Ohms with a -60 degree phase component as the least specified load impedance for his RMS rating. Since -60 dgrees phase means twice the current, this is i fact like specifying 1.5 Ohm loads, making his amp almost bullet proof under realistic work conditions. This cannot but please me by reminding me that there are still some damn serious people desiging audio for real world loads rather than lab resistors.
Yes. And... the preponderance of music of equal temperament makes most IM products not exactly harmonically related. Really only octaves are exact. But other sum and difference frequencies can be pretty close. Seventh is a bitch.
- Status
- Not open for further replies.