Digital is the representation of a quantity with a number. Like 'the signal level is 0x4ff'. In analog a quantity is represented by a voltage or a current. Hence the use of the word 'analog'. You can have both analog and digital sampled systems, of course.
I agree that it is in itself not very important whether class-D is analog or digital, but Mr. Grasso has the view that because class-D is digital, it cannot output small signals, there is a limit to that and therefor it is an inferior technique . This is incorrect; it can easily output 0.000000012nV as long as the noise is lower.
So that wrong view led to this discussion; I thought that if he realised that class-D is not digital he might release his wrong idea that class-D cannot output extremely small values.
But I can understand it looks strange if you fall into the discussion without the proir statements.
Anyway, I think I have explained things as well as I am able to. It's now up to you to save the ship ;-). Have a nice weekend!
Jan
I agree that it is in itself not very important whether class-D is analog or digital, but Mr. Grasso has the view that because class-D is digital, it cannot output small signals, there is a limit to that and therefor it is an inferior technique . This is incorrect; it can easily output 0.000000012nV as long as the noise is lower.
So that wrong view led to this discussion; I thought that if he realised that class-D is not digital he might release his wrong idea that class-D cannot output extremely small values.
But I can understand it looks strange if you fall into the discussion without the proir statements.
Anyway, I think I have explained things as well as I am able to. It's now up to you to save the ship ;-). Have a nice weekend!
Jan
Last edited:
no, its easier to do in the analogue domain.
the simple example is with a triangle wave and a sine wave. (the triangle is much faster than the sine) Overlay them, At every crossing between the two, you switch the pulse direction, creating the pwm signal. That signal is analogue, it is not limited to discrete numbers. To amplify the signal reverse the process with higher rails and filter.
That is a simplified description of a type of pwm conversion. Bruno has much nicer white papers about the more complex real world at the Hypex site.
BTW, the description I gave is pretty much a description of the method Ampex used in the ATR100 tape recorder for the reel servo system, that is another place to read a systems design description, the manual has a well done description and is readily available on line. That was designed in the late 70"s.
PWM has been used for many things audio since the 70"s, AM broadcast transmitters began using this technology because of its efficiency. Imagine the cost difference between AB and PWM technology at 10kW.
Cheers
Alan
the simple example is with a triangle wave and a sine wave. (the triangle is much faster than the sine) Overlay them, At every crossing between the two, you switch the pulse direction, creating the pwm signal. That signal is analogue, it is not limited to discrete numbers. To amplify the signal reverse the process with higher rails and filter.
That is a simplified description of a type of pwm conversion. Bruno has much nicer white papers about the more complex real world at the Hypex site.
BTW, the description I gave is pretty much a description of the method Ampex used in the ATR100 tape recorder for the reel servo system, that is another place to read a systems design description, the manual has a well done description and is readily available on line. That was designed in the late 70"s.
PWM has been used for many things audio since the 70"s, AM broadcast transmitters began using this technology because of its efficiency. Imagine the cost difference between AB and PWM technology at 10kW.
Cheers
Alan
Last edited:
Just in case you haven't seen this. Explains the short comings and ways around them in class D technically.
http://www.infineon.com/dgdl/an-1071.pdf?fileId=5546d462533600a40153559538eb0ff1
http://www.infineon.com/dgdl/an-1071.pdf?fileId=5546d462533600a40153559538eb0ff1
ClassAB is also level-discrete at clipping (which is equivalent to classD at shortest on-time).
Danke, Voltwide!
Every power amplifier to be sold as HiFi should have a data-sheet mentioning
DIN(A)-weighted (4KHz pronounced) noise floor voltage (or noise floor voltage over x Ohms, if the amplifier has output resistance) and
THD (not THD+N) at 1KHz and 0, 1, 2, 3, ... Bel above that noise floor
and ascertaining, that a signal of full-scale input voltage and any frequency up to 100KHz will not cause aliasing. Above that frequency, allowed input voltage shall fall linearily with frequency.
and mentioning, if its output can feed another amplifier of the same type without causing aliasing.
Every power amplifier to be sold as HiFi should have a data-sheet mentioning
DIN(A)-weighted (4KHz pronounced) noise floor voltage (or noise floor voltage over x Ohms, if the amplifier has output resistance) and
THD (not THD+N) at 1KHz and 0, 1, 2, 3, ... Bel above that noise floor
and ascertaining, that a signal of full-scale input voltage and any frequency up to 100KHz will not cause aliasing. Above that frequency, allowed input voltage shall fall linearily with frequency.
and mentioning, if its output can feed another amplifier of the same type without causing aliasing.
No, that is rite neither. One cannot detect least bit with THD measurement -- one needs to measure linearity, say if half input level makes half output level. With a 4KHz signal and all noise filtered out, Class-D amplifiers will at some point put out less than half level, say start to supress signal.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.