I find it odd that this so common with Class-D chip-amps, given that the onset of clipping and the rise in distortion is more abrupt compared to linear ones. I guess it's after decades of looking at specs, but I'd always be far more impressed at first glance with, say, "30w @ 0.03% THD" than "40w @ 10% THD"".
Your video confirms my doubt on why companies do the 10% trick. Thanks John. It may be useful to listen to the clipping amplifier playing music or test tone. For that one can add an attenuator resistor of, say, 100 ohm in series with the speaker and connect them in parallel with the dummy load.
I am often asked what 10% distortion sounds like? The answer of course depends on the type of distortion. A good way to hear clipping distortion would be to play a test tone though a loudspeaker at, say 1khz. Then slowly turn up the gain until. you hit clipping. Audible distortion can be heard by most people at around 1-2%. 10% will sound really hard and harsh. On the other hand distortion from a tape recorder could well be 2-5% in normal use and not really be audible due to more harmonic distortion rather than clipping. This is why some people like the sound of a valve amplifier that tend to round the waveform at maximum power unlike a solid state amplifier that just chops the top and bottom off the waveform. Running an amplifier into clipping can damage the tweeters in your loudspeaker as it's like putting square waves into then, lots of high frequency harmonics. You can always get lots of loudspeakers in for repair after parties with the HF unit burned out, I love it!!
They'll also complain that some 0 - 100 000 Hz +/- 0.01 dB, 0.0000001% THD @ 100W will sound too dry for their taste. People want more musical stuff, not strictly hi-fi.
Yes it depends on how the distortion is created. 1% from clipping will make your ears bleed from sounding so bad. 5% from a slightly mis-shapen sine wave is virtually inaudible. A very experienced audio engineer once commented to me that if you can't see the distortion on an oscilloscope, you can't hear it, regardless of what a spectrum analyzer or THD meter shows. A "clean looking" sine wave will sound clean - the slightest amount of clipping is plainly audible. John mentioned car audio. Most of those devices are measued at "hard clipping" - so over driven that a sine wave at the input becomes a square wave at the output.
I've blown many of the cone type tweeters over the years, I couldn't understand how a 30w amp could destroy a tweeter in a 100w speaker system. The Tandy (Radioshack) salesman said it's amplifier clip
John, you can calculate THD on your scope as it measures true RMS. Just use pure sine input, take note of the Vrms value measured by the scope, then using Vpp value measured by scope calculate pure sine Vrms and divide with measured Vrms. You had around the 10% mark 12.3Vrms with ca. 31Vpp (eyeballing as it is covered by menu @7:56) which is ~11Vrms for pure sine. Divide and you get ca. 12% THD. To get 10% distortion, adjust input level until you see the scope read ~12Vrms. Actually, that'll be THD+N as not only harmonic content will be calculated into Vrms value the scope computes. However, hopefully at 10% that won't matter if it's THD or THD+N or something else would be very very wrong.
That's interesting. I never heard that before. It seems like it would go awry if the distortion caused a high crest factor (large pk to pk). I guess it is only for clipped signals? I missed the large 5th harmonic in the 10% measurement (notice how I ignored it). I'm so used to seeing the reference signal there, that I overlooked the 5th as the reference causing me to turn up the signal more and get 12%.
The THD calculation is similar procedure to analog domain THD analysis like on HP334A except now we do the notch filtering manually using ideal sine as etalon. Main requirement is that the measured Vrms must be a true RMS measurement. Naturally, there are limitations to using it this way.
Sound like marketing 101! Most people eyes go straight to those wattage outputs. Uneducated people most likely will over look the 10% distortion. Hey buy my chips!!!
10 % THD rating is more useful than might be obvious. Bass might get to this level and not sound too bad. Above 10% is a disaster. 10 % will be momentary if acceptable. Some believe they hear 0.01%. I doubt it. They might hear a concealed distortion found by a loudspeaker more than a resistor. TDA2050 was a nice device. 10 % at 1 kHz wouldn't be good .1950s tube amps were said to be good if 1%. This was made 0.1% if class B types or whatever. This is as true today as then. The concealed distortions usually include stability or slewing problems that mostly were not common with tube amps. TDA 2050 is class AB. That is a small proportion is class A. 1/10 watts perhaps. 10% THD is an engineers rating. My first amplifier a Teleton GA 202 by Mitsubishi said 20 watts 1% THD. At 15 watts 0.06%. I almost didn't buy it due to that. They never said at 15 watts. A Siemens tube cinema amp of 100 watts gave 0.13% 1 watt and 1.4% 10 watts 10% 100 watts. It was a good compromise. I am certain the 100 watts would have been used. It was a serious piece of engineering. They weren't stupid.
Thank you for answering my comment John. I am old school like you but with a 50 year gap went from valves straight to D-Class. For me making a D-Class is all but impossible. The ST TDA7498E PDF shows figures up to 30% THD and 3 Ohms, the graphs are at 4 Ohms - 8 Ohms is nowhere to be seen. They state 160 W but looking at the graph nasty things are starting around the 80 W mark. I have built a linear and a LT1083CP (pair) regulator power supplies and I have buck converters. What are the pitfalls of using a buck converter to drive a D-Class ? The research I have done has just left me very confused. I have just bought a scope, wish me luck.
Really good demonstration. Great channel. Have you got any demonstrations of how to set up the scope for this kind of measurement and how to get the pilot signal in there?
Thanks for watching! Each scope has its own requirements for setting up the FFT mode and getting good measurements. I created the pilot signal in free software called Audacity by making a tone and adding the pilot tone at 1% level.
There are two chips Id like you look into 1. Is the LA4440 IC. This is a tiny IC yet I have a car stereo 4 channel. With a 6 band Graphics equalizer. Hooked up to a 3 way speaker 10 inch woofer, 8 inch midrange and a tweeter and it plays so amazing and loud on a 12 volt supply that I cant wrap my head around how its so good. 2. I tried building a Phono pre amp for my Hitach HT350. And have found that nothing sounds better than the ICs used in old Akai Amps. I cant get any Op amp to work as good as these old ICs in Akai amps. Why is this.
The 10% distortion figure was probably reasonable for voice communications products like loud hailers and CB radios. To use it for anything more than basic communications gear usually indicates they are fudging on power numbers. The question is how low the distortion can be in normal use. Looking at the graphs, it looks like the rise in THD+N at the lower levels is due to noise. 0.1% THD+N is okay. You might want to be a little more complete in explaining that push-pull circuits cancel out even order harmonics, leaving only the odd order ones.
Dunno if you've shown it before (i.e. can't find it) but I'd be interested in a demo of what that distortion sounds like if you could find a way of equalising the volume at different power outputs.. Guess some applications you don't worry so much about distortion - I was in a penny arcade a few weeks ago and one of the machines periodically played a shouting pirate voice at a high volume 'Arrr you win' etc through a car-type speaker. audio quality was crappy as you'd expect - Guess there was a single chip amp suffering away in there somewhere, might have been a retrofit feature cos the speaker looked like it had been bolted on as an afterthought.
It is called a oscilloscope, but you can build an indicator circuit that lights a (red) LED when the circuit begins to clip. Many amplifiers have this feature.
Please re-read the comment. Ideally you would use an oscilloscope, but a cheap circuit can be designed to light at the voltage that will cause clipping. Many commercial amplifiers have this feature. Another way is using a VU meter. If it is proper calibrated, the zero db point is the point of clipping.
Wow, I told you they exist, and with Google and less than 10 seconds I found a RU-vid video that shows a simple circuit that anyone with any electronics aptitude can build. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-L9sVY_wdDKw.html
Your videos are interesting but very theoritical. I have seen many of them but I have not understood yet how I can reduce the distortion in an amplifier that I have (class d). If you can explain this with a simple diagram or a real demonstration, this would be great.
Sir. I dont understand what all about the electronics. It just that i like listening music. And someone recommended this chip tda2050. My question is over all is it a bad chip? Jezz i just order the chinese board already before i find this video.
John: please explain the use of the 4.5kHz pilot tone with the 1kHz test tone during testing. I have been Googling an hour and haven't figured it out. This is very interesting. (edit...) Nevermind: I found your video on this exact question. Chee
I would not even look at or take any interest in am amp rated at 10% distortion if I was shopping. I am amazed that the chip company would even consider someone run it into clipping. By everything I learned some time ago, clipping destroys speakers,.. at least clipping on a high end amp into high end speakers, even if the wattage into the speakers is not above the maximum rating of the speakers. Also, who would want to listen to music at 10% distortion? It would sound absolutely terrible. If I recall correctly, when a super audiophile friend of mine was demonstrating distortion vs sound quality, I think I could tell the difference below 1%, just detectable to me there. But in reality, based on his seemingly infinite knowledge and experience, best to get an amp with .1% or lower distortion, which does not even make it more expensive. Most amps are far below that amount anyway. Maybe the super cheap stuff has more distortion, that I would not know because never had any interest in cheap stuff.
Clipping cannot destroy a speaker per se, the increased RMS levels resulting from clipping are what do it. Those same RMS levels from an unclipped source would do the same damage.
One thing I remind myself of is that "normal" people REALLY don't care about distortion. They just want volume, and they'll crank it up to that 10% distortion level and like it.
I did a service call on a system that complained that his system had lost volume. I get there and his system included four JBL L-100 Century speakers. His listening area was closer to the left, front speaker. He turned it on and played it quite loud. Of the four speakers, only one was operating properly. The other three had only the 12 inch woofers working, and the mid range and tweeters dead. He was running his source through a tape recorder preamp and then into his main amplifier inputs because he got more "volume". The signal going to his right front and rear speakers was in hard clipping. (I brought my oscilloscope). I explained to him that this is what fried his mid range and tweeter units in the three speakers. I advised him to NOT run the signal through what was actually TWO preamplifiers and also get his speakers repaired or replaced. The thing is that he was listening to horribly distorted music and did not care because it was "louder".
In order to reach 10% THD on earbuds, even the cheap 15bucks ones they should be blasting a stupidly insane amount of SPL through it, probably enough to cause eardamage.