I've read IRF's - Class D Tutorial.
When Half bridge and full bridge are compared, IRF name an advantage gained when using full bridge, which is:
Superior (No even order HD)
Can anyone elaborate on this? As in why even order THD are missing in this setup?
When Half bridge and full bridge are compared, IRF name an advantage gained when using full bridge, which is:
Superior (No even order HD)
Can anyone elaborate on this? As in why even order THD are missing in this setup?
Even order distortion products is a result of non-symmetric waveform distortion.
Think of a sine wave that is slightly flattened but more so on one half of the waveform than the other (the positive half for example).
I the waveform would be distorted/flattened symmetrical the result would be only odd order harmonics.
Now, if a circuit with nonsymmetrical distortion (even order) is bridged with a second identical circuit the new complex circuit will balanced so any nonsymmetrical distortion is cancelled out.
A little simplified but more or less so! 🙂
/Peter
Think of a sine wave that is slightly flattened but more so on one half of the waveform than the other (the positive half for example).
I the waveform would be distorted/flattened symmetrical the result would be only odd order harmonics.
Now, if a circuit with nonsymmetrical distortion (even order) is bridged with a second identical circuit the new complex circuit will balanced so any nonsymmetrical distortion is cancelled out.
A little simplified but more or less so! 🙂
/Peter
That all makese sense, but the general, communal wisdom about harmonic distortion is that it's the odd-order harmonics that sound worst.
It's even been postulated that much of what some people like about 'tube sound' is the 2nd order harmonic distortion.
I'm confused by why IRF would want to say "hey! We only have the bad distortion!"
I realize that it's probably more complex than that, but there is certainly an increase in the popularity of integrated amps that include a tube gain stage and a solid-state power output stage. So far i haven't seen one with a class-D power stage, but it wouldn't surprise me to learn that they exist.
It's even been postulated that much of what some people like about 'tube sound' is the 2nd order harmonic distortion.
I'm confused by why IRF would want to say "hey! We only have the bad distortion!"
I realize that it's probably more complex than that, but there is certainly an increase in the popularity of integrated amps that include a tube gain stage and a solid-state power output stage. So far i haven't seen one with a class-D power stage, but it wouldn't surprise me to learn that they exist.
Full bridge does indeed cancel out even order harmincs, whatever the class of operation.
The point is that class D distortion, which is mostly about dead time and power supply saging phenomena is inheritely odd order, so there is not so much to cancel.
The real case with bridged class D amplifiers is that switching mosfets are really bad when voltages over 150V are concerned.
On the other hand point of building a class D amplifier of below 100W while one can build a great class AB amp of similar wattage is questionable.
IMO class D is calling to be used in full bridge from the more general point if view.
Regards
Adam
The point is that class D distortion, which is mostly about dead time and power supply saging phenomena is inheritely odd order, so there is not so much to cancel.
The real case with bridged class D amplifiers is that switching mosfets are really bad when voltages over 150V are concerned.
On the other hand point of building a class D amplifier of below 100W while one can build a great class AB amp of similar wattage is questionable.
IMO class D is calling to be used in full bridge from the more general point if view.
Regards
Adam
Pick up a sound program like Audacity. Find a bunch of sound samples of acoustic instruments. And run it through the spectrum analyzer. You'll find some instruments produce primary even harmonics, others primarily odd - yet they all sound musical. In one case I know of a recorder (closed pipe, odd harmonics) was substituted for a flute (open pipe, even harmonics) and many were fooled thinking it was a flute. In some old music scores, flutes and recorders were used interchangeably.
A amps job is to just amplify, unfortunately nothing is perfect and there is some distortion introduced. But with a lot of these amps there is such a small difference you'd have to be super human to really hear a difference. When somebody starts pushing opinions that their amp is designed to produce even harmonics then walk away. It's not doing it's job of amplifying very well.
A amps job is to just amplify, unfortunately nothing is perfect and there is some distortion introduced. But with a lot of these amps there is such a small difference you'd have to be super human to really hear a difference. When somebody starts pushing opinions that their amp is designed to produce even harmonics then walk away. It's not doing it's job of amplifying very well.
Loudspeaker THD figures are in the 0.1% to 1% range, at least at moderate playback levels because they can rise up to 10% at high levels. They produce both odd and even harmonics at all frequencies and particularly in the lower end, whose harmonics are quite audible because ear sensitivity to midrange is much higher than to the bass fundamental, which is also much stronger than the original midrange components.
Solid state amplifier THD figures (including class D) are well below 0.1% during most of the time, at least an order of magnitude lower than loudspeakers. Harmonics are mostly produced in the higher end, where they are harder to perceive because ear sensitivity is much more uniform... So there is not much to worry about.
The problem is that most amplifier freaks think about speakers as ideal black boxes (same for room acoustics, etc...) You can even see people developing sophisticated ultra low distortion amplifiers and using them with boom box speakers...
Solid state amplifier THD figures (including class D) are well below 0.1% during most of the time, at least an order of magnitude lower than loudspeakers. Harmonics are mostly produced in the higher end, where they are harder to perceive because ear sensitivity is much more uniform... So there is not much to worry about.
The problem is that most amplifier freaks think about speakers as ideal black boxes (same for room acoustics, etc...) You can even see people developing sophisticated ultra low distortion amplifiers and using them with boom box speakers...
Reducing 2nd ordeer HD also means lower IMD and that is essential in HiFi gear.
Lower sensitivity to higher frequencies does not mean that we can allow high levels of distortion up there since IM products fall down in the midrange where we are most sensitive.
AFAIK class D amps typically have some nasties in the upper range making them non-transparent in blind tests.
As for looking at amplifiers as ideal black boxes, of course, what would we do otherwise? Use them as effect-processors? ;-)
/Peter
Lower sensitivity to higher frequencies does not mean that we can allow high levels of distortion up there since IM products fall down in the midrange where we are most sensitive.
AFAIK class D amps typically have some nasties in the upper range making them non-transparent in blind tests.
As for looking at amplifiers as ideal black boxes, of course, what would we do otherwise? Use them as effect-processors? ;-)
/Peter
for class d there are two mechanisms for distortion, the first is swiching related and down to dead time, ringing etc, and the second is down to frequency of the sample/self oscillation to the amplified frequency.
the first distortion is analogous to crossover distortion in class ab amps;
the second is the number of sample steps you have for the feedback to correct the amplifier error; sampling at 400KHz you have 4000 samples for a 100Hz signal, 400 for 1KHz and 40 for 10KHz- however the distortion at higher frequencies appears to the ear lower, as the 2nd harmonic of 10KHz is 20KHz this above most peoples hearing.
this is why class d amplifiers have lower distortion at lower frequencies and it tends to rise with frequency; but perceived distorion is less
the first distortion is analogous to crossover distortion in class ab amps;
the second is the number of sample steps you have for the feedback to correct the amplifier error; sampling at 400KHz you have 4000 samples for a 100Hz signal, 400 for 1KHz and 40 for 10KHz- however the distortion at higher frequencies appears to the ear lower, as the 2nd harmonic of 10KHz is 20KHz this above most peoples hearing.
this is why class d amplifiers have lower distortion at lower frequencies and it tends to rise with frequency; but perceived distorion is less
Pan said:
How can you justify this statement? mechanisms for THD and IMD are different, especially in class d
The linearizer effect of a feedback depends on the open-loop gain at the desired frequency, the frequencies of the harmonics. The sample rate influences the stability, because of the spectral overlap.
The linear-phase one-cycle delay of calculation always causes the same phase margin degradation on the z-plane. The question is that how can the process dynamics fit in the range of -Fs/2...Fs/2, and it depends on the sampling caused spectral overlap. If you can assure the Nyquist-criteria, the fed-back system will be stable. Of course if you choose higher sample-rate, the process's poles will be placed closer to z=1, so the calculation delay caused phase-margin degradation will be lower at the cutoff frequency.
Previously I read about some ******** of the must be preferred sample-rate of 15...20 times higher than the fastest pole of the process. I think this assumption is made for the cases when the overlap caused exact amplitude and phase characteristics never been inspected and the original characteristics are used. Despite of these theories only the Nyquist-theorem must be applied for the proper overlapped spectra and the exact fed-back system can be calculated.
Previously I read about some ******** of the must be preferred sample-rate of 15...20 times higher than the fastest pole of the process. I think this assumption is made for the cases when the overlap caused exact amplitude and phase characteristics never been inspected and the original characteristics are used. Despite of these theories only the Nyquist-theorem must be applied for the proper overlapped spectra and the exact fed-back system can be calculated.
raintalk said:
Hardly anybody selling or buying amps is willing to say "This amp sounds good because of the harmonic distortion".
However. hollow-state amps are literally at an all-time high of popularity. There are more brands and models available today - if you include both instrument amps and listener amps - than ever before.
Very few people who have any idea what they're talking about ever try to make the claim that tube amplification produces less distortion than solid-state amps of equal quality of design and manufacture. I've only ever run into one kook on a forum who insisted that tubes are the most linear, distortion-free amplification device ever invented -- that it's just that modern designers have forgotten everything that their grandparents knew about amplifier design.
Some amount of serious research has been put into why some people prefer "tube sound" - and this is from the "if it sounds good, it is good" perspective, where the measuring device is the rubber yardstick of the human ear rather than any kind of serious metrology.
Some people just like it - and i have to admit that from time to time I'm one of them. Just last night i spent about an hour with an all-tube futterman headphone amp acting as preamp for an old Kenwood KM-105 driving vintage Sony uni-electret headphones. It was bliss.
Most of this research came down to this: When a tube goes into overload, this results in diode current, which produces 2nd order harmonics, which sound warm and chorousey.
When a transistor goes into overload, it often results in odd order harmonics - especially 7th order harmonics - and these sound rather harsh.
The issue at hand is not accuracy or purity - it's euphonics.
You can, however, make a transistor amp that sounds very "tube-y", and you can make a tube amp that lacks what most people would identify as "tube sound".
One vendor even set up a booth at a trade show with a blind A/B test of an excellent tube amp vs. an excellent solid state amp, and challenged people to guess which was which. Pretty much nobody could coherently tell the difference.
So, to sum up, you are correct that hardly anybody is going to try and market an amp on the strengths of it's 2nd order harmonic. Other than guitar amps. And one solid-state preamp that Paia once offered as a kit.
But people often love amps for reasons other than the strict definition of their performance as an accurate reproducer of sound. And people do make purchasing decisions based on euphonics.
Agreed, but... I think there's more going on than just that. Even when a tube amp is designed for very low distortion and excellent recovery from overload, there is still the question of why its sound is "more-ish." What I mean is that, yes, when you design it right, it will sound only subtly different than a good solid state amp, and it will not be distinguishable in a short-term double-blind test, but in my experience (and that of many others), a good tube amp is just easier to listen to for a longer time.
Maybe it's purely psychological (I can't discount that), but I would certainly like to see some long-term controlled listening tests to see if there's something there beyond the null results seen to date.
I exclude amps that are deliberately designed to sound different (high output Z, high second order distortion); like you, I can have a nice one-night stand with those amps, but can't tolerate them day-in and day-out.
Anthony C Smith said:
If a nonlinearity is present in a stage you will have harmonic distortion on a singel static signal like a sine wave, feed a two-tone signal and you will see modulation between the two, IMD.
/Peter
ericj said:
Hi Eric - do you have any modern references for this research?
And what class amp are you talking about? This forum is class-d, which other than some old patent I have never seen implemented with valves. And really - does class-d overload?
(I don't mean clip - yes they clip)
I'm all for euphonics - but it can always captured at the source of the recording, or added in through signal processing. Again, it's not the amplifiers job to muck with this. Just my $.02 😀
I'm not referring to any specific amp - just expressing bewilderment over the touted feature.
It was my understanding that even-order harmonics are rarely any kind of major issue in reasonable solid-state designs.
Having none at all is just fine, of course.
The larger question would be why we're using metrics designed to measure the performance of a different technology.
It was my understanding that even-order harmonics are rarely any kind of major issue in reasonable solid-state designs.
Having none at all is just fine, of course.
The larger question would be why we're using metrics designed to measure the performance of a different technology.
- Status
- Not open for further replies.