The rise and fall differences, and dead time, are there whether your output waveform is near zero volts or isn't. It can affect the percentages of duty cycle, but the ratio of Off to ON changes most at large signal.
Unlike class AB, which switches only near zero volts, class D switches ALL the time and doesn't change operating modes. Class A doesn't either, but does depend on a device controlling current linearly (while class D depends on switching at correct times in correct proportions... FETS are pretty good at that).
Unlike class AB, which switches only near zero volts, class D switches ALL the time and doesn't change operating modes. Class A doesn't either, but does depend on a device controlling current linearly (while class D depends on switching at correct times in correct proportions... FETS are pretty good at that).
Yes, a class D output stage that allows for a total of more than 100% duty cycle... e.g. it allows negative dead-time without destructive cross conduction.
I have a paper somewhere - I'll see if I can find it.
\\\Jens
The amps I like to examine are those that achieve the most consistent success, commercially and critically. We can pontificate all we want but real world votes are what counts.
Some class D amps are now achieving just that, so perhaps we should pay attention.
OTOH, others (eg. most/all of the "Temple of Blameless" designs) achieve nothing more than pedestrian commercial/critical success, so you can put those into the 'yawwnnn' box.
Then there are the consistent faves - big class A amps either tube or SS. Many of which are low feedback. So maybe they are worth more than just a cursory dismissal ?
Of course one is always entitled to the view that the marketplace and everyone ($ paying clients, reviewers, dealers) are all delusional/wrong/fashion junkies but that would not be my view certainly.
Some class D amps are now achieving just that, so perhaps we should pay attention.
OTOH, others (eg. most/all of the "Temple of Blameless" designs) achieve nothing more than pedestrian commercial/critical success, so you can put those into the 'yawwnnn' box.
Then there are the consistent faves - big class A amps either tube or SS. Many of which are low feedback. So maybe they are worth more than just a cursory dismissal ?
Of course one is always entitled to the view that the marketplace and everyone ($ paying clients, reviewers, dealers) are all delusional/wrong/fashion junkies but that would not be my view certainly.
What would constitute consistant success, commercially would selling 100000 count or being in production for many years
Stuart
Stuart
I may be wrong but what I understand is distortions arising out of the use feed back can not be eliminated with another application of feedback because that distortion is ever changing. If general consensus on an ideal amplifier is required why not use most linear devices and design a circuit with commonly agreed threshold where one can hear distortion/s and be done with it. I guess people want uniqueness in their amplifier. Any device/material advances of modern times which are extremely linear ?
As someone has said, "Better to be thought of a ---- than to speak and remove all doubt." I have switched on my passive reader mode as I will carry on browsing the site. Lots of things needs to be learned.
Thanks in advance and best regards.
That was aimed at Waly as I couldn't understand why it would even matter. Sadly I never got a clear answer as to what I was missing as even the IMD argument didn't make sense.
I take a more simple view that if 3rd and 5th are better than 80dB down I have bigger problems in my system then happiness ensues and I can get on with the music.
Is it that easy though? I worry that a lot of this particular story has become part of the audiophile oral tradition. Happy to be proved wrong.
With the right system, everything makes a difference, I can even hear differences between different feedbackratios on an op-amp. 0-6-12 dB gain. (Sonny knows which op amp it is)
SY, trust me I would much rather that I could not hear any difference with the gain of the op amp. Now that I can raisers a whole bunch of questions that needs to be examined.
Interesting, somehow.
As the most knowledgeable people here many times stated that no difference can be heard in a controlled test between two well designed and sufficiently good measuring amps, what can be the reason in discussing global, nested and no feedback, class A, B, D, tubes, BJT, MosFET, circuit topology and whatever?
What is the reason to try to measure the last 0.0001% of distortion, which can't be heard anyway, in a controlled test?
As the most knowledgeable people here many times stated that no difference can be heard in a controlled test between two well designed and sufficiently good measuring amps, what can be the reason in discussing global, nested and no feedback, class A, B, D, tubes, BJT, MosFET, circuit topology and whatever?
What is the reason to try to measure the last 0.0001% of distortion, which can't be heard anyway, in a controlled test?
Bonsai "Not in the same box. Trust me."
--What a box have to do with PSRR , and Noise levels Yes Yes in the same box yes SMPS600 and NC400 in the same box.
sonnya "I did not write all, Hypex is in general very good, but still for monitoring purpose in studios one of my customers changed from hypex UCD400 to my design and ended up with less audible noise from the speakers"
--Your friend might have oscillation problem with his amp or is using a 106db at 2.8v hornloaded driver other wise there is nothing on the market that improves on UCD400 in the noise departments only some rail switcher like Benchmark Media AHB2 even this amp when is running at high rails is on UCD levels in the noise department, more info on your amp is it low gain or Class-H it would be nice.
sonnya "All CE test is most likely to be done in a shielded box to get them through the tests. I have measurement equipment here for the purpose"
--Again you can not shield agains EMI problem all the EMI comes from power fets switch at zero crossing and the lowpass LC filter becomes CL Highpass filter at Hundred Megahertz but you can minimize it up to benign levels, you use multiple layer wire wound ferrite core, ground plane at the whole board with no slices, all the cable connectors needs to be grouped together so you don't crate dipole and reduce parasitics and stabilize power stage after switch transition and you can put your Class D inside your FM or VHF receiver infact there are FM Integrated Class D receivers on the market without any shield whatsoever Pioneer, Sony and Panasonic has a couple models on sale.
sonnya "No it is not, again use an H-field probe. Also if you do an sweep up to 30MHz, you will clearly see the switch frequency and its harmonics."
--If you measure cheap Class D with toroid inductor offcourse.
sonnya "No i am not talking speaker wires here my friend, but class D has very high dv/dt values, and to save money many designs is only a 2 layer pcb. I know that IRF reference designs are two layer, but they have massive ground copper layers. But still you have the EMI which could be lowered a lot by using 4 layer and use the layers as shields."
--You can not shield EMI from a bad design once you connect the speaker wire it will come out from the wire.
john curl "Class D can, and usually has lots of distortion that is roughly equivalent to xover. "
--Not true, you can listen to zero feedback amps out there like the Equibit/TI Amps Panasonic XR-55, Onkyo has some also on the same concept Lyngdorf/Tact Millennium they don't have feedback because they use digital PCM to PWM conversion into a couple of FET's switching at 400kHz into a LC filter. It does not matter if its digital modulated or analog infact analog modulater zero feedback Class D has more potential then these amps and people complementing the sound compared to the previous cheap Class AB amp, Non feedback Class D distortion is not horrid like Linear crossover distortion but it is not enhancing anything its not a valve amp but also it is not benigh so its best to get rid of it.
Hiten "I may be wrong but what I understand is distortions arising out of the use feed back can not be eliminated with another application of feedback because that distortion is ever changing."
--Yes you are wrong in a well designed stage negative feedback reduces distortion it does not create new one, in a non linear like single transistor it creates more distortion until it stabilize then it start to reduce distorion Nelson Pass talk about this somewhere i think you can read in Bob Cordell's book in more detail.
--What a box have to do with PSRR , and Noise levels Yes Yes in the same box yes SMPS600 and NC400 in the same box.
sonnya "I did not write all, Hypex is in general very good, but still for monitoring purpose in studios one of my customers changed from hypex UCD400 to my design and ended up with less audible noise from the speakers"
--Your friend might have oscillation problem with his amp or is using a 106db at 2.8v hornloaded driver other wise there is nothing on the market that improves on UCD400 in the noise departments only some rail switcher like Benchmark Media AHB2 even this amp when is running at high rails is on UCD levels in the noise department, more info on your amp is it low gain or Class-H it would be nice.
sonnya "All CE test is most likely to be done in a shielded box to get them through the tests. I have measurement equipment here for the purpose"
--Again you can not shield agains EMI problem all the EMI comes from power fets switch at zero crossing and the lowpass LC filter becomes CL Highpass filter at Hundred Megahertz but you can minimize it up to benign levels, you use multiple layer wire wound ferrite core, ground plane at the whole board with no slices, all the cable connectors needs to be grouped together so you don't crate dipole and reduce parasitics and stabilize power stage after switch transition and you can put your Class D inside your FM or VHF receiver infact there are FM Integrated Class D receivers on the market without any shield whatsoever Pioneer, Sony and Panasonic has a couple models on sale.
sonnya "No it is not, again use an H-field probe. Also if you do an sweep up to 30MHz, you will clearly see the switch frequency and its harmonics."
--If you measure cheap Class D with toroid inductor offcourse.
sonnya "No i am not talking speaker wires here my friend, but class D has very high dv/dt values, and to save money many designs is only a 2 layer pcb. I know that IRF reference designs are two layer, but they have massive ground copper layers. But still you have the EMI which could be lowered a lot by using 4 layer and use the layers as shields."
--You can not shield EMI from a bad design once you connect the speaker wire it will come out from the wire.
john curl "Class D can, and usually has lots of distortion that is roughly equivalent to xover. "
--Not true, you can listen to zero feedback amps out there like the Equibit/TI Amps Panasonic XR-55, Onkyo has some also on the same concept Lyngdorf/Tact Millennium they don't have feedback because they use digital PCM to PWM conversion into a couple of FET's switching at 400kHz into a LC filter. It does not matter if its digital modulated or analog infact analog modulater zero feedback Class D has more potential then these amps and people complementing the sound compared to the previous cheap Class AB amp, Non feedback Class D distortion is not horrid like Linear crossover distortion but it is not enhancing anything its not a valve amp but also it is not benigh so its best to get rid of it.
Hiten "I may be wrong but what I understand is distortions arising out of the use feed back can not be eliminated with another application of feedback because that distortion is ever changing."
--Yes you are wrong in a well designed stage negative feedback reduces distortion it does not create new one, in a non linear like single transistor it creates more distortion until it stabilize then it start to reduce distorion Nelson Pass talk about this somewhere i think you can read in Bob Cordell's book in more detail.
Last edited:
The 'dead zone' in class D amps is VERY difficult to control. I have measured class D amps at modest output levels, have you? I have even offered my results to class D designers to their embarrassment.