So in a particular case PWM can be generated by digital circuits, from digital signals. We are talking such circuits. But you said they are still analog. Don't you see any contradiction here?
Do I mean correctly that you are denying this:
https://en.wikipedia.org/wiki/Digital_signal_(signal_processing)#/media/File:Digital.signal.svg
Voltage is also an analog quantity. Does this mean any voltage is analog even when it represents digital values? Then what you allow for digital? Is there any digital electronic system in the world? I already asked, and you didn't answer: "can you show an electronic digital system that doesn't use voltage (or current) to represent the signal?"
Do you want to tell that PWM being analog or digital is related to Classes?!?
And the answer is: not arbitrary. Some PulseWidthModulators are purely digital circuits, with finite input and output value possibilities.
Can a Class be digital? If not, then is there any meaning in saying a Class is analogue? If yes, then which Class is digital?
This what? This forum? Yes, it is called diyAudio. This topic? Really???? But let it be anything: so what? Is anything else not allowed to know by members?
J1850 is a data transfer protocoll using digital PWM. What you denied. Google it. I already linked a picture telling everything about it that is important here.
To grasp this one has to sort out the concepts of carrier and coding and presence of any coded information. The envelope of a signal carrying information coded in digital form is, yes, still analogue. From a signal envelope perspective, the DDR{0/1/2}_CAS Column Address Strobe on an Intel i7 processor chip is analogue but the information passed is digitally coded. The question here is if there is any digital coded information passed in a PWM based Class-D amplifier.
The answer is: No.
//
The answer is: No.
//
Last edited:
This is wrong. A rate is an integer number, and those reside in the digital, say discontinuous domain: The difference between neighbouring numbers cannot be infinitesimally small.DF96 said:
Pafi, symmetry does not matter.
A step or needle impulse is not symmetric, yet audio systems must be able to reproduce them. Any signal is a chain of impulses, some positive, some negative. A Class-D amplifier contains two bucket voltage droppers laid in parallel, one for positive and one for negative impulses.
Now crossover distortion is the disability of any leg to supply small signals in controlled fashion. It does not matter, that power actors of Class-D play a square wave all the time, since if it is symmetric (1:1 pulse width), no current flows. If it becomes asymmetric, then one leg must supply current. But below switching range, it cannot (as snappily as within switching range), so Class-D becomes Class-B.
Are you confusing "switched" with "clocked"?
Class D isn't restricted to switching only at particular clock intervals, it can switch at any time as required to generate the analog waveform. There isn't a necessary limit to the switch time resolution. There will be some practical limits to how soon it can switch back after a switching event, though, due to need to prevent shoot-though (both output polarities turned on simultaneously).
If an amplifier is to work as a Class-D say an internally switching one at infinitesimally low signal level, it must be able to switch infinitesimally fast.
If a class A amplifier is to work at an infinitesimally low signal level, it can't have any noise. It's just dynamic range. Nothing works at infintesimally low signal level.
Guys, I don't have time to wade through 100+ posts so I hope I have not duplicated what others have said.
Amplifier classes are named in order of their development.
Class A was first
Class B was next
Class C was third
Class D was fourth.
Class E was fifth
Class F was sixth
Class G was seventh
Class H was eighth.
Now classes C, E and F are for Radio Frequency power amps, that is for TV and Radio transmitters and have no place in audio.
Class A is really inefficient, but theoretically capable of the lowest distortion.
B is more efficient and A/B is lower distortion than B and much more efficient than Class A.
So what about Class D? Forget Digital. In my humble opinion, Class D is an analogue amplifier.
To be more precise it is a power codec. It takes a low voltage audio signal and using Pulse Width Modulation, encodes the analogue signal. This allows fast switching devices (FETs, Bipolar Transistors or Vacuum Tubes/Thermionic Valves) to magnify the signal and a simple filter arrangement to decode the amplified signal back to baseband audio.
It is not digital, it is more like a radio transmitter and receiver in the same circuit. Remember AM, FM and PWM are all types of modulation to allow signals to be transferred efficiently. So whether you use Frequency Modulation. Amplitude Modulation or Pulse Width Modulation, the modulated signal is analogous to the original electrical signal at the input.
The higher the Modulation frequency, the smaller the components (inductors).
Sent from my iPhone using Tapatalk
Amplifier classes are named in order of their development.
Class A was first
Class B was next
Class C was third
Class D was fourth.
Class E was fifth
Class F was sixth
Class G was seventh
Class H was eighth.
Now classes C, E and F are for Radio Frequency power amps, that is for TV and Radio transmitters and have no place in audio.
Class A is really inefficient, but theoretically capable of the lowest distortion.
B is more efficient and A/B is lower distortion than B and much more efficient than Class A.
So what about Class D? Forget Digital. In my humble opinion, Class D is an analogue amplifier.
To be more precise it is a power codec. It takes a low voltage audio signal and using Pulse Width Modulation, encodes the analogue signal. This allows fast switching devices (FETs, Bipolar Transistors or Vacuum Tubes/Thermionic Valves) to magnify the signal and a simple filter arrangement to decode the amplified signal back to baseband audio.
It is not digital, it is more like a radio transmitter and receiver in the same circuit. Remember AM, FM and PWM are all types of modulation to allow signals to be transferred efficiently. So whether you use Frequency Modulation. Amplitude Modulation or Pulse Width Modulation, the modulated signal is analogous to the original electrical signal at the input.
The higher the Modulation frequency, the smaller the components (inductors).
Sent from my iPhone using Tapatalk
Last edited:
Yes, as you said D is only the next letter in alfabet. Nobody here thinks it was stand for digital. ClassD is not digital. Nor analog. It is not even an amplifier. This class contain many different circuits, for example power suppies also. Some ClassD circuits are analog, some are digital, or mixed signal. Only 1 thing is common: switching mode operation. This is all about ClassD.
But some people thinks ClassD is a synonime for some of the amplifiers he heard about, and they try to apply their associative thoughts about these particular amps to a whole Class.
But some people thinks ClassD is a synonime for some of the amplifiers he heard about, and they try to apply their associative thoughts about these particular amps to a whole Class.
What is class A, or maybe to put I different is a class A circuit one where there's always excess current an no switching of output devices or is Class A restricted to circuits drawing constant power..??
Regarding class D then I belive that the major drawback is the needed output filter, where induction in the coil limits the instant current capability. I don't believe there's any absence of resolution.. I have some expience with Pascal, Ice Power and Hypex modules. Some are great some are not so great. But compared to an ultimate linear amp they seem to suffer somewhat
Regarding class D then I belive that the major drawback is the needed output filter, where induction in the coil limits the instant current capability. I don't believe there's any absence of resolution.. I have some expience with Pascal, Ice Power and Hypex modules. Some are great some are not so great. But compared to an ultimate linear amp they seem to suffer somewhat
Have you heard about RATIOnal numbers? No, this is not an unintentional coincidence. They are all a rate of 2 integer numbers.
What?!? Your previous deduction is based on operating point is set the most asymmetrical way possible (at the limit). If this premise is deleted your whole deduction collapses. You cant say it doesnt matter.
Symmetry of the system, obviously not the signal.
Well, in the small minority it contains almost 2 buck (no, not bucket. Are you using google translator?) converters (crown BCA topology), but even in this case both buck is used to cover the whole input signal range. Output voltage generated by each half bridges is the same. Only current is what splitted.
And this disability is not present around idle state. Try or simulate if you don't believe!
I guess you mean output current. That is irrelevant. Power actors are attached to the inductor, not the output! The current of inductor is matter.
Both legs supply current at idle stage. I already mentioned both cross-conduction and inductor ripple.
Yes but the basic principle of a class D is entirely analogue. You simply modulate the audio signal so that instead of the voltage representing the information the pulse width does. No data is lost in the respect that nothing is directly quantised. The pulse width directly represents the analogue input signal and can be recovered if you remove (filter) out the switching aspect. Yes class D is bandwidth limited by its switching frequency (among other things) such that performance metrics will start losing linearity when the input frequency gets too high, but this isn't simply a limitation of class D, the others are limited similarly but for different reasons.
You say that class D is used in power supplies, yes PWM controllers are used in combination with MOSFET drivers and MOSFETs to regulate an output voltage or current, where the pulse width is modulated in relation to the demands on the output. You'll find these devices under the analogue section of most semi conductor manufacturers. Other amplifier classes are quite similar to linear regulators too so there's not much difference.
One point worth mentioning is that a linear regulator, or switching regulator using PWM as it's drive signal, are called regulators, or power supplies. They are not called amplifiers.
There are power supplies out there that use PWM as their drive signal but that use micro controllers to generate it and that use A/D converters to control all aspects of the power supply. Is this power supply now magically digital? No, it's entirely analogue in its intrinsic operation it's just that a digital system is used to manage it.
A class D amplifier with an I2S input is no different. The bit that's performing the amplification (or providing the current drive) is done entirely within the analogue domain. Somewhere inside the amplifier the I2S signal is processed and converted into a PWM signal. You could theoretically filter this PWM signal and end up with the analogue waveform, but instead it's sent through a power stage.
I think we can all agree that whatever your digital audio system, if you send the data through a simple low pass filter it's never going to look like the analogue signal it that it was supposed to be storing. Likewise all the data will be lost.
As has been mentioned, PWM can be used to represent digital data, it's encoded within the pulse width. You can apply your simple low pass analogue filter and generate a voltage instead if you wish. The data will not be lost. Granted if you input the voltage directly to your micro controller or whatever it'll cease to operate properly, but that's just a minor problem of the micro controller not understanding the new language, or so to speak. Input the voltage into an A/D converter and alter the code to interpret the data differently and everything will carry on working.
So I suppose in answer to the original question what is the resolution of a class D amplifier? Infinite within the performance boundaries of the amplifier itself. Just like any other audio amplifier on the planet.
Last edited:
Not true ! It can switch quite sluggish (within reason) but still be able to vary the pulse-width in infinitesimally small steps. This accounts for the classic analog PWM amplifier. If it is one of the "power DAC" kind the pulse-width is varied in discrete steps, like the 254 possible steps of the TI ones.
Regards
Charles
Last edited:
When a continuously variable quantity (e.g. voltage, pulse width, pulse rate) is used to represent another continuously variable quantity (e.g. air pressure, voltage) then you have an analogue system. When a continuously variable quantity is used to represent numbers, or parts of numbers, in a discrete sense then you have a digital system.Pafi said:
When pressed on a technical level, I notice that you revert to word games. If all possibe amplifier classes were analogue then it would still be true and meaningful to say so; something does not become untrue or meaningless simply because it is necessarily true or always true, although it may be trivially true. The claim under discussion is that Class D is digital; the truth or falsity of this statement does not rest on whether (for example) Class A is analogue or digital.
Where did I deny that PWM, or any other analogue technique, can be used to transfer digital information? My point is not that Class D is PWM and therefore must be analogue; it is that Class D is using PWM (an analogue technique) to transfer analogue information and therefore is analogue. Class D involves no numbers, so cannot be digital.
I am not so sure. Some people do think like that, but I doubt if anyone here is going to put his hand up and admit it.
At least it doesn't give out numbers but voltage and current. Therefore all amplifiers are analog. The output power is not generated numerically nor logically but physically. As I said before: With some tolerance the ones with a digital modulator might be called digital amplifiers for marketing purposes. But I wouldn't use the term in a technical publication.
Ah and yes, there is still the OPs question: What is the resolution of a class-d amp ?
For the PWM ones with analog modulator the resolution is theroretically infinitesimaly fine like a linear amplifier in theory is. In practice it is of course plagued by physical restrictions like noise and distortion.
For digital modulators it is theroretically restricted but the exact resolution does of course depend on the implementation.
Regards
Charles
Forgive me, if i try to gain air dominance here, but the topic is " Class d amplifier resolution". I have intuitively known the answer to that for more than year, (re-?)invented the concept of Schaltumfang (switching range) in order to quantify it, and am now trying to prove it.
With a full-scale signal, each of both power actors is switched on for half of the time. With a smaller signal, on-times of power actors must be shorter, because a smaller signal contains fewer power. The smaller signal is, the shorter on-time is. But transistors are not infinitesimally fast. At some point of brevity, they start to ignore their inputs and start to conduct only partially, do not switch anymore. This point is lower limit of switching range.
With a full-scale signal, each of both power actors is switched on for half of the time. With a smaller signal, on-times of power actors must be shorter, because a smaller signal contains fewer power. The smaller signal is, the shorter on-time is. But transistors are not infinitesimally fast. At some point of brevity, they start to ignore their inputs and start to conduct only partially, do not switch anymore. This point is lower limit of switching range.
Isn't it almost full time with a full scale signal? If each 'power actor' (assume an N and P half of the output stage) is on half the time, 50% duty cycle, Vout is zero I think.
Seen from each device, close to 100% DC is max output voltage, close to 0% DC is zero output signal. No?
Jan
This would only happen with a badly implemented BD switching scheme. For the normally used AD switching scheme (both power devices switched on for 50% of the time at idle) this is definitely not true. You may try to use your intuitition as much as you like, but your findings are not valid.
Regards
Charles
Yes, right.
They are. Without input signal, power actors are not switched on; their internal resistance is still hi, no power is transferred. What matters is transferred power. Which is to be switched, if the amplifier under analysis is to be of Class-D. But switching is not infinitesimally fast, and so every Class-D amplifier has a lower power limit, below which it is not Class-D anymore.phase_accurate said:
That is the point that you get completely wrong. They are being switched constantly at their idle frequency (or carrier frequency for non-selfoscillating ones). At zero input voltage a symmetrical rectangular is given out that does have an average voltage of 0 V (after low pass filtering).
What you are talking about is a possible but very unusual switching scheme. And guess why it might be like that ! 😉
Regards
Charles
Last edited:
- Status
- Not open for further replies.