Square waves or 1's and 0's?

When my pc is sending signal to my avr via ethernet cable, is it sending 1's and 0's or is it sending square waves? When my transport is sending signal to coax input on my processor, is it sending square waves or 1's and 0's?

Hi Lynne,

The most accurate answer is probably "neither."

A square wave, in the context of electrical signals, is a voltage that alternates periodically between a higher voltage level and a lower voltage level, spending an equal amount of time in each of the two states. An ideal square wave has infinitely fast transitions between the two states, and each voltage level is perfectly precise and constant, i.e., no noise (random fluctuation of the voltage levels) is present. Neither of those conditions is possible in the real world, so what are referred to as "square waves" are approximations of ideal square waves.

A square wave cannot normally be used to convey information, because its pattern of alternating between the two voltage states remains the same all the time. It can be used in many applications as a "clock signal," however, which controls the timing of whatever operations are performed by the circuit that is involved. Square waves can also be useful as test signals, to evaluate circuit or component performance.

1's and 0's are just numbers. The decimal (base 10) numbering system that humans like to use utilizes numbers whose individual digits can range from 0 to 9. The 1's and 0's you refer to are based on the binary (base 2) numbering system, where the only allowable digits are 1 and 0. Either system can represent all possible numbers; it just takes more digits to do it in the binary system. Computers and other digital devices are designed based on the binary number system because their practical implementation is facilitated by the fact that only two states have to be distinguished from each other.

A series of 1's and 0's can be used to convey information. An example of "information" is the amplitude (volume) of a music signal at a given instant of time. Since those 1's and 0's are numbers, though, they in turn have to be represented by something else, such as a voltage level, before they can be sent or communicated or processed by a physical circuit. In some applications, a 1 may be represented by a higher voltage, and a 0 by a lower voltage, or vice versa.

In the two cases you mentioned, though, those approaches aren't used, in part because clock and data are combined into a single signal, in such a manner that the receiving circuit can separate the two. S/PDIF encodes the 1 and 0 data, together with the clock and additional necessary information as described in the writeup, into something called Biphase Mark or Differential Manchester Code. Ethernet, since it is a networking standard that is designed to provide communications between multiple devices at arbitrary and intermittent times, and in its modern forms at very high speeds, is complex and is described further in this Wikipedia writeup and at the links it provides. Different codings, all of them combining clock, data, and other necessary information, are used for each of the commonly used link speeds (10, 100, or 1000 mbps).

Hope that clarifies more than it confuses :-)

Best regards,
-- Al
first you need to understand a meaning of analogue signal, than modulation, than sampling, than binary algebra.

otherwise any technical content will seem to be confusing.
Paul McGowan ran a huge post on these issues with DACs. Go to www.psaudio.com and read "Paul's Posts". It is a ton of reading...
Thanks, everybody. I've been reading on the net. Someone says the signal is 1's and 0's and the next person (critic) says square waves. You open the door so that I can understand the concepts. What started it is that I have an entry level (RCA)MIT digital coax and a signalcable digital coax. I knew that the MIT didn't cut it so I tried the signalcable and the sound opened up. Now I'm reading overwhelming positive reviews of Oyaide DR-510 and wondering if it would be a worthwhile upgrade from the signalcable at $140 or around there. Don't know if I will try it but I feel smarter already.
It's 1's and 0's. A cd is basically a file of 1's and 0's. DAC is a computer that converts digital (1's and 0's) to an analog signal ... D to A converter. Data must be loaded into memory before the CPU can process it whether streamed from a computer or transport.

Ideally it should work like any program. When it starts, loads into memory before running. This eliminates jitter, complex synchronization between DAC and tranport ... Can you imagine running your browser or MS Office off a CD drive?

Even HD video is streaming with more data demand than audio. Netflicks almost bankrupted before streaming to compete with the cable companies.

04-03-13: Knghifi
It's 1's and 0's.
Kng, while 1's and 0's are certainly being COMMUNICATED between the two components, the references to the signal, and to the possibility that it might be a square wave, would appear to indicate that what is being asked about is what is being "sent" in a physical/electrical sense.

As you will realize, numbers cannot be sent through wires, in that sense. So when "someone says the signal is 1's and 0's and the next person (critic) says square waves" (quoting from Lynne's second post), they are both wrong.

-- Al
Square wave is a generic term for a non-continuously changing voltage signal. It changes in discrete steps. It does not necessarily mean a repetitive signal or actually square. IF it were repetitive, then it is a digital oscillation, also known as a clock. Because even clocks have non-50% duty cycles, even clocks are not actually "square".

1's and 0's are defined as: the high state of the signal is "1" and the low state of the signal is "0". the high and low can be defined as any voltage depending on the logic family and physical interface.

All digital voltage signals actually contain analog components since they do not switch in zero time from 0-1 or 1-0, and because drivers and transmission-lines are not perfect, there is also resulting overshoot, ringing etc..

Steve N.
Empirical Audio
Thanks, Steve.

I would add to your comment, though, the clarification that the 1's and 0's that are referred to in your definition are NOT the same thing as the 1's and 0's which constitute the audio data that may be communicated via S/PDIF or ethernet, which are the focus of this thread.

Also, although your definition of a square wave is a reasonable one, it is a looser definition than many others would apply to the term, their more narrow definition also being reasonable IMO. See, for instance, the first paragraph of this Wikipedia writeup, in which a square wave is defined as being periodic, and as having equal durations in its two states.

-- Al
Its not sending an exact square wave, but it sends the signal as an analog waveform.

So anyone who says 'square wave' is about 1000x more accurate than the person who says 1s & 0s.
Al and Agisthos are right. I was not asking the content of the message (signal). I was asking for a physical description of the signal. I thought I was asking in effect for someone to dispel a myth. Now I'm not sure of that because there seems to be a gray area of confusion depending on what exactly is the question.

I also read that a digital coax cable between transport and DAC should be 5 ft long. This certainly qualifies as myth.

Hi Lynne,

Perhaps surprisingly, the 5 ft/1.5 meter length suggestion is not a myth. See Steve's paper here, which makes sense to me, and is also supported by experimental results that have been reported by some A'gon members I consider to be credible. What length will be optimal in a given system is dependent on a number of hardware-specific variables, however, which are generally unspecified, and IMO that recommendation should be viewed as a length that is LIKELY to be optimal in MOST cases, but is not guaranteed to be. There have been at least a few reports I have seen here from members who have compared different lengths, and have found shorter lengths, such as 1 meter, to be preferable in their systems. Also, if a very short length is practicable, such as 8 inches or less, IMO that stands a very good chance of being an optimal choice.

Some further comments on the definitions Steve provided in his post above:

As I indicated in my previous post, what he defined as "1's and 0's" has nothing to do with the 1's and 0's which comprise the data content of the signal. A less ambiguous way of referring to what he defined as "1's and 0's" would be to refer to them as "logic levels," or more specifically as "logic 1" and "logic 0" levels, respectively.

If we go by the assumptions I stated in my initial post, that "1's and 0's" refers to data, and that square waves are periodic and symmetrical, then the parties you quoted, who were disagreeing with each other, were both wrong.

If we go by Steve's definitions, then those parties, who were disagreeing with each other, were both right!

I would agree that a plausible case could be made on the basis of either set of definitions. However, IMO it would be a safe bet that the presumably non-technical person who was arguing that what is being sent are 1's and 0's was referring to data, and not to logic levels.

Best regards,
-- Al
Hi Al,

Thanks for articulating the original question. Steve's paper is very helpful.

I can do close to 8" but will have to measure to make sure. It will be interesting to experiment and very helpful to know that length is almost always a factor.

Best Regards,

digital signal not only has ones and zeroes, but rather characterised by sampling frequency and sampled amplitude as mentioned before voltage level.
an arbitrary point of signal could be captured knowing the particular voltage and the value of the signal 0(low) or 1(high). The low could be let's say 0V and high could be +50mV etc...

0-s and 1-s are only rererence values of digital signal, but not the actual physical ones for those who thinks that 0s and 1s are 'traveling' accross the digital cable.
electrones are the only ones capable of doin' that
Almarg - with S/PDIF the transport level is still 1's and 0's, but the protocol of the interface requires encoding the data to limit the number of consecutive 1's or zeroes, allowing a clock to be recovered as well as the data.

Steve N.
Empirical Audio
My question was oversimplified and came out of ignorance. Your anwsers will give me all the reading I want tonight because I have to re-read. You guys are over my head but it's fun trying to understand it. You are a great source, better than anything else I've found on the net.

I was tired of reading the argument that the quality of cable doesn't matter because the signal is simply ones and zeros. I have two digital coax cables and it clearly does matter.

Thank you.

that was a true statement Lynne simply because digital cable carries simple shape signal compared to an analogue music.
OK. An analogue wave form is continuous in terms of time and voltage and frequency. A digital wave form is a square wave, as it were, which means it is repetitively maybe not on-off but high-low, higher-not-so-high, low-lower in terms of time and voltage as it represents the encoded information. And the binary system is the only one that can work for this Pulse Code Modulation because the language is ones and zeros. The vehicle for this language is the square wave because it is not continuous but repetitive.

That's my homework.
Lynne, if you haven't already, take a look at the figures shown in the Wikipedia writeup I linked to earlier for Biphase Mark/Differential Manchester Encoding, Biphase Mark (shown in the second figure) being the encoding method used for S/PDIF. The paragraph above the figures helps to clarify them.

Think of all the waveforms shown in the figures as being graphs that depict voltage along their vertical axis, and time along their horizontal axis.

As you'll see, 1 and 0 data information is conveyed by virtue of whether one "transition" or two "transitions" occur within each "clock period" (defined below). A "transition" being defined as a CHANGE from either the higher voltage ("logic 1") state to the lower voltage ("logic 0") state, or vice versa.

The higher voltage (logic 1) state is the upper of the two possible voltage levels of each signal waveform that is shown, and the lower voltage (logic 0) state is the lower of those two levels.

A "clock period" is defined as the amount of time either between one positive-going (logic 0 to logic 1) transition of the clock waveform and the next positive-going transition of that waveform, or, equivalently, between one negative-going (logic 1 to logic 0) transition of the clock waveform and the next negative-going transition of that waveform.

That encoding method allows both clock and data to be conveyed in a single signal, as Steve and I indicated earlier.

-- Al
Jitter on a digital streaming interface is another thing entirely. This is the time variation of the switching transitions, not just 1's and 0's. The digital feed to a D/A is sensitive to this because precise timing matters. It is also important that the A/D used when recording has low jitter. These two add to make more frequency modulation distortion at the D/A.

See more info here:


Steve N.
Empirical Audio
Al & Steve,

Yes, Al, I read the Wiki piece before but was somewhat unclear until you have now further explained it. Visual aids--graphs, charts, schematics are difficult. I need to learn first the meaning before they make sense. I was unclear on the voltage being a constant value and on the transition period as it relates to the binary data.

I was struck with the beauty of the S/PDIF system where the clock and data are one signal until I read Steve's explanation of jitter and how the BMC signal is vulnerable to it.

I'm very happy to have learned this basic concept of the digital signal. Many thanks for hanging in.

Al, on an unrelated topic discussed in a former thread which I should discuss in a follow up to that thread but since the website made changes nothing works on my pc the way it did and I'm not sure you would find it on that thread, and in regard to your lack of enthusiasm for autoformers in SS, I did demo the autoformers with the hk990 integrated driving the AR9's. the 990 is rated at 150w/8ohms, 300w/4ohms, and the 9's are rated at 4ohms. So using the 4ohm tap gave the set-up 150w/8ohms versus 300w/4ohms without autoformers. There was no apparent difference except that sound quality was a little better without the autoformers apparently because of running the signal through another device with more connections. It was a horse a piece.

But with an amp that is limited into 4ohms, the autoformers did enhance sound quality when the amp was driving 4 ohm speakers turned into 8 ohm speakers. Presumably converting 8 ohm speakers into 16 ohm speakers would cause more loss than gain. If all this makes any sense.

Hi Lynne,

Here is a link to the other thread you are referring to. Everything in your post above sounds reasonable to me.

-- Al
There's no waveforms in the Universe that are not continues. That also applies to square pulse.
OK. It probably needs to be qualified. I got it from
en.wikipedia.org/wiki/Analog_signal. Please explain.