sfogg Posted May 10, 2007 Share Posted May 10, 2007 Mark, "If you look at the signal alternating between On and Off, or 1 and 0, you get the impression it is representing 1's and 0's. But there no meaning in that interpretation." Of course there is meaning. One or off is a meaning. It is a boolean value... it is a single bit system, that is its entire meaning and the only thing that gets encoded into a single bit system. "If you looked at a series of pulses which had the following widths, 5uS, 7.5uS, 23.098uS, 16.876uS, how would you assign a "1" to any of those, or a "0" to any of the spaces between?" You would assign a 1 or 0 based on the modulation frequency. In that regard it is NO DIFFERENT at all of looking at a PCM signal on a scope. If you look at the data line feeding a PCM DAC on a scope it could look exactly like your PWM signal. How do you determine where one 0 ends and another begins? "Digital" means to operate with discrete values like 1 and 0 which carry the INFORMATION." Yes and the entire information it may carry is simply on/off. "Analog operates with a continuous spectrum of value such as 1.0001 volts, 1.0003 volts, 202.987 volts, 2,389 volts and so on." Exactly!!!! And a PWM signal DOES NOT HAVE THAT. It has one of two discrete values.... on or off as your own picture shows. Shawn Quote Link to comment Share on other sites More sharing options...
Deang Posted May 10, 2007 Share Posted May 10, 2007 Shawn -- when you get a chance: What do you think about dead time -- does it really exist? Quote Link to comment Share on other sites More sharing options...
maxg Posted May 10, 2007 Share Posted May 10, 2007 Thanks Guys - we now have 2 perfectly good explanations as to why they are, and are not digitial amps. Think I will tune back in when we have a winner. Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 "You are using "states" to describe amplitude. The word "state" means "it's information." The pulses are all the same amplitude, but that is not representing any "state" whatsoever." Of course it is representing a state... either on or off. Yes the width changes but that is still representing the state of on or off... just off or on for longer amounts of time. You can determine if it is a 1 or a 0 simply by overlaying the modulation frequency on top of the PWM signal. Each 'timeslice' of the modulation frequency is the cut point of if the value is one or zero and you tell that simply by looking at the amplitude.... it it high or is it low. "The PWM signal pulses represent infinite states of width." If it had infinite states of width it would therefor have infinite bandwidth. Which is not even remotely the case. The noise levels of PWM amplifiers increase dramatically has you go up in frequency because the system is not at all infinite. It (the width of the pulses) is limited by the modulation frequency. The modulation is what determines if the signal should be on or off and it can not make that decision any more often then at whatever frequency it runs at. Shawn Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 "Of course it has that! The width of the pulse is the analog representation of a different dimension, namely, amplitude relative to the modulating signal. " The signal per modulator cycle is either on or off. It has nothing more then that. If the modulator is thought of as a 'sample' the sample has a value of on or off. Nothing more. That you can line up the cycles and sort of see something that looks a little like an analog represenstationof amplitude doesn't make it analog. You can do that *exact* same thing if you look at a PCM signal on a scope too. I highly recommend you do it sometime. If you set your trigger on the L/R clock and look at the data bits of a PCM signal they look just like a VU meter. Doesn't mean that is an analog signal either. Shawn Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 "How many different values of width can exist between 3uS wide and 5uS wide?" What is your modulator frequency? Shawn Quote Link to comment Share on other sites More sharing options...
NOSValves Posted May 10, 2007 Share Posted May 10, 2007 Shawn -- when you get a chance: What do you think about dead time -- does it really exist? It absolutely exists...my heart just stopped while reading this thread Quote Link to comment Share on other sites More sharing options...
Deang Posted May 10, 2007 Share Posted May 10, 2007 Shawn -- when you get a chance: What do you think about dead time -- does it really exist? It absolutely exists...my heart just stopped while reading this thread LMAO! Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 For those that don't know.... in a nutshell Pulse width modulation basically compares an input signal against some other signal and if the source is greater then the other signal it turns the output on otherwise it turns it off. (Or vice versa...simplified explanation) Mark's picture with the sawtooth wave overlaid on top of a sine wave is basically showing this. Working on that picture.... the modulator runs at a frequency and simply compares the triangle wave against the sine wave. At each comparison if the sine waves amplitude is greater then the triangle wave the output is on otherwise it is off. If for example in that picture the modulator made 100 comparisons you could simply overlay 100 evenly space divisions over that picture. Then at each and ever division look at the output from the modulator and determine if the value is on or off. There is your digital value. The Tripath difference is the modulator isn't at a fixed frequency and the comparison signal likely isn't either. Shawn Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 "How many different values of width can exist between 3uS wide and 5uS wide?....1MHz or 100KHz, take your pick, it doesn't matter to my question at all. " It totally matters, and why you don't see this is why you don't understand this. OK, so say the modulator is running at 100kHz that means it compares 100,000 times a second. A us is a microsecond... 1 millionth of a second. The different between 3us and 5us is .000002 of a second. The modulator is running at .00001. It won't be able to see anything at .000002 percision. So the answer to your question based on a 100kHz modulator is there are NO different values of width that can exist between 3us and 5us on a 100kHz modulating system. The modulator doesn't have the bandwidth to see to that frequency. Shawn Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 "I snagged this off Wikipedia...." Now read the section on how a modulator works and remember that the modulator runs at a specific frequency (exception Tripath)..... Shawn Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 Mark, "On a personal note, I hope you are approaching this in the spirit of a "backyard BBQ debate" and nothing more. " No worries! We are both simply debating the topic without getting personal. (Something a little too rare unfortunatly....) Shawn Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 Mark, "I did not ask how many values the modulator can recognize between strokes. I asked how many values EXIST." And the answer for a PWM system at that modulation frequency is the same... none. A digital system is limited by its sampling rate/modulator frequency. In the real world there are an infinite number of values. But in a pulse width modulation system that is totally and completely not the case. It is limited by its modulator frequency. "So, when you read that the duty cycle is modulated based on the average value of the input signal, did you interpret only 2 states for that duty cycle?" The duty cycle is the modulator frequency. The average value is basically derived by the comparison against the triangle (or whatever) wave. Is the signal wave greater or less then the comparison wave at that point? The answer is either yes or no. So yes, there is only 2 states for that duty cycle/modulator cycle... on or off... 1 or 0... whatever you want to call it. You know this to be true since at any one instant the output of a 'perfect' PWM system is either on or off. Where you get into the average is over time... say 1 is full power, 0 is full off... if you have a string of 1/0/1/0/1/0/1/0/1/0 the average is half power. You can take that string of 1 and 0 and save it. Then play back the same string of 1 and 0 on a totally different system that is at the same modulator frequency.... and you will have the same signal being reproduced. This is in essence what SACD does. Shawn Quote Link to comment Share on other sites More sharing options...
maxg Posted May 10, 2007 Share Posted May 10, 2007 Integrals and functions?!!! I dont log in here to see stuff that reminds me of the latter years at school - I log in here to tell everyone vinyl is better than CD!! My god this discussion is soaring - and still no clear winner. Figures. DD, I have built a somewhat similar amp based on Hypex boards. It did not perform very well but that is probably more due to my lack of building skill than anything else. I think I could have a much better stab at it a second time. It did work though - which amazed me to begin with. Mine was a 30 wpc model. Didnt look massively digital I must say - nothing obviously CPU or RAM like about it - but I met 2 people on their discussion board having much the same argument as to whether PWM is, or is not, digital. Quote Link to comment Share on other sites More sharing options...
maxg Posted May 10, 2007 Share Posted May 10, 2007 Mark, Haven't you just described FM? How does PWM differ from that. It seems apparent to me that there are 2 systems to choose from - DSD type digital ala SACD and FM analogue like radio. This discussion now boils down to which system the amps in question are using - unless there is a third one I dont know about (other than a variation on FM). Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 "It's not a greater or less, yes or no comparison. The whole point of the PWM scheme is to represent amplitude as time. ( X volts = Y time)." Yes, it is a yes/no comparison. That is all it does. The amplitude as time comes in simply because of the modulator frequency and that that comparison happens many many times per second. Time is not encoded per sample, it is a byproduct of the fact that the sampling happens very very often and is not a one shot deal. "Turn the signal on it's side, so to speak. The "average value" difference between the triangle wave and the signal wave is in fact the amplitude of the signal at that moment. This average value in the X axis during a clock cycle, is now translated into the y-axis or duty cycle, or "on" time." Think this through.... if the modulator encoded the difference between the triangle wave and the signal wave into y-axis 'duty cycle' what exactly happens on the next modulator cycle when it has a totally different duty cycle to encode yet the preceeding values duty cycle is not complete? The system is simpler then you think it is. The width of the square wave is nothing more then a collection of 1 or 0 over time. How fine of a division that square wave can be is determined by the modulator frequency. "This varying duty cycle is now "carrying the information" about the signal. That information is not 1 or 0, it is some value of time that is varying as the amplitude varies. That's the information content." Nope, it is a 1 or a 0, nothing more. Lets attack this from a different angle.... Draw any width square wave you want on a piece of graph paper. The modulator 'frequency' is the width of one square on the graph paper. Convert the square wave to ones and zeros. As long as your square wave is within the 'frequency response' of the modulator 'freqency' you can digitally encode that square wave to 1 or 0s. Repeat this time with a collection of different square waves on the paper. Again as long as your square waves are within the 'frequency response' of the modulator 'freqency' you can digitally encode that square wave to 1 and 0s. If you make the square wave have a higher 'frequency' then the modulator is capable of you can not properly encode it. This is why digital systems are bandwidth limited to avoid aliasing. Shawn Quote Link to comment Share on other sites More sharing options...
DrWho Posted May 10, 2007 Share Posted May 10, 2007 Ya know...at the University of Illinois at Urbana-Champaign, digital circuits are explicitly described as any circuit that uses "switches." Turning a transistor on and off is a switch. The definition for digital has absolutely nothing to do with bits, or words, or even monkeys in spaceships. And the reason digital circuits are defined in this matter is because the major design criteria for all digital circuits is the same. Since UIUC is top 3 in electrical engineering, I think it safe to consider this a reliable definition of digital. But if that's not enough, I've got a bunch of textbooks that say the same too...written by the people that pioneered the wonderful world of the digital domain. Quote Link to comment Share on other sites More sharing options...
DrWho Posted May 10, 2007 Share Posted May 10, 2007 Square waves have nothing to do with the definition of digital Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 "Think of your calculus a moment here." But that is not what it is doing. Build a PWM circuit and see... http://www.cpemma.co.uk/pwm.html "But the argument is that the 1 or 0 state has no information." It does though. It is on or off. A boolean value holds information. Did you leave the soldering iron on last night when you went to bed... yes or no.... 1 or 0.... boolean value... it has information in it. When that is then looked at over time (based on the modulator frequency) there is where you get far more complex information. Again, this is a 1 bit digital system. This is how they work. You can argue that a single on or off holds no information but anyone here can go listen to a SACD (which is also a 1 bit system) and hear for themselves that when you string a bunch of them together they convey information within them. Shawn Quote Link to comment Share on other sites More sharing options...
sfogg Posted May 10, 2007 Share Posted May 10, 2007 "Square waves have nothing to do with the definition of digital" I think what Mark is trying to say is a transistor that switches on/off by definition makes a square wave. The obvious response is when you play back a square wave in an amplifier if the amp is not at full power the output devices are not simply on/off or off but are acting in a linear manor to the input signal. In fact they are acting in a linear manor even at full power too.... just one that looks like what a switch would do at that operating point. Shawn Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.