But the signals that are held in the flip flops and latches, for example, in the SRAM of the buffer of the node should be in a modulated (with BPSK etc.) form, right? If so, the transmission delay actually (and mostly) results from sampling the input voltages and thus, reading from the memory (buffer in this case), I suppose.

+ Do you state that the transmission time limitations of different PHY protocols result from their modulation requirements? I suspect this because the ethernet with a more complicated modulation phase should have had a longer transmission time than the legacy RS-232 then. Therefore, even if the contribution to the transmission time coming from the memory read operations on the buffer is clear, I still cannot understand the PHY protocol side of it.

Additionally, this came to my mind: If these "digital signals" are really analog signals modulated with digital information, what do the good old modems do / did? Do we say that nodes have built-in modems that already modulates and demodulates analog signals?

"But the signals that are held in the flip flops and latches ..." - What you describe ("BPSK") is nothing like how TTL logic are implemented. The 4th paragraph in my answer describes how TTL logic works. Or read a textbook. You need to stop guessing or wildly connecting unrelated ideas.

I did not say I was talking about TTL, do you see a statement or do you even see I say the word TTL? I am talking in general and there is an etc. word there to imply this. The emphasis is clearly on modulation. Maybe you should first read the answer. Plus I am not wildly connecting unrelated ideas. Can you give an example what is unrelated to the concept please?

Then when you write "flip flops and latches" specify what kind of logic you mean if not the convential TTL. I can't read your mind.

flip flops are like elementary particles and they are used everywhere. It does not mean that only TTL uses them just because it is called "Transistor to Transistor Logic", am I wrong? Anyway, stored (analog) data is in a modulated form, right?

@ConventionalProgrammer A flip-flop stores a single (digital) bit, not analog data, and certainly not a waveform.

@Bergi What does it change? Say registers if you want a byte-read. A bit is also a signal. Flip flops hold signals behind it and copies (samples) it to the output at the edge of a clock. sawdust explains about non-existance of digital signals above, so what's your point?

@ConventionalProgrammer I just agree with sawdust that you're totally wrong about flip-flops storing signals in modulated form. They store one bit of information, and are able to produce a new analog signal on read that depends on the concrete physical implementation and which of the two binary states was stored. Not sure what you mean by "a bit is also a signal".

@Bergi Firstly, sawdust never stated I am wrong about flip-flops storing signals in modulated form, quote if you will. Secondly, a bit is indeed a signal, what else would it be? You say that flip flops store a single bit of information, which changes nothing. Bits are modulated into analog signals through different modulation schemes. Do you know about the inputs and outputs of a flip flop? Quote from D. Harris: "... a D flip-flop copies D to Q on the rising edge of the clock, and remembers its state at all other times".

@Bergi Input and outputs are nothing but voltage oscillations wrt a reference point and as the time passes the voltage sampled at a point takes the form of a signal, e.g., you see on an oscilloscope. Additionally, it is not the flip flop that modulates the signal, which implies the bit that is to be stored must be modulated already. In many undergraduate textbooks these signals are given as square (logic) waves. Check out David Harris' "Digital design and Computer Architecture" for example.

They stated in their first comment here. A bit is an abstract unit of information. "A D flip-flop copies D to Q on the rising edge of the clock, and remembers its state at all other times." - that means it remembers the logical state of the input, from the last time when the clock was not rising. It does not mean that it somehow remembers the signal shape. It will again output a simple logical value.

@ConventionalProgrammer I'm not trying to prove anything, I'm trying to explain, or at least to discover your misunderstanding. But it's still not clear what you are talking about concretely, especially if you don't want to assume TTL.

@Bergi I talked about flip flops because they are built into the SRAMs of the router buffers. The discussion is not specific to TTL, the discussion is rather about the tranmission time as the header goes. I stated that the tranmission time can also result from the read operations on the buffer SRAM, and it indeed does because the read operations require sampling, which brings a delay itself. Think of the first bit that "leaves" the node. The transmission time starts from then on. If the sampling rate is considerably low, the time required to read the bytes from the buffer will increase.

@Bergi In fact, in an extreme scenario in which the sampling rate goes to zero, the time required to read from the memory, for instance, the time required to fetch the "second bit" from the memory (these bits will be consecutive in the virtual memory, they do not need to be side by side on the hardware level though) and put it onto the wire will go to infinity and the transmission time will go to infinity as well. Of course, this putting bits onto the wire is an abstraction that sweeps the real analog case under the rug.

@Bergi The thing is, I did not misunderstand anything, this is what I am trying to say. sawdust thought that I was talking in the context of TTL and stated that BPSK has nothing to do with TTL. If you read carefully, he replied: " Then when you write "flip flops and latches" specify what kind of logic you mean if not the convential TTL. I can't read your mind." , which amounts to mean that I am actually the one who is misunderstood.

@ConventionalProgrammer : "I stated that the tranmission time can also result from the read operations on the buffer SRAM, and it indeed does because ..." _ This is wrong and absurd. Nobody builds comm equipment with such limitations. "I did not misunderstand anything" - I'm pretty sure I am not the only one to disagree with that statement.,

@ConventionalProgrammer Good old modems modulate the simple high-voltage/low-voltage digital signal into something more complicated. For example, one ancient type of modem emits a 1270 Hz sine wave when it gets a low voltage from the computer, and a 1070 Hz sine wave when it gets a high voltage from the computer.

unrelated to modems: imagine you are transmitting a digital signal by turning a flashlight on and off (someone else is looking at the flashlight to receive the signal). What prevents you from sending a trillion bits in half a nanosecond? What prevents you from turning the flashlight on and off a trillion times in half a nanosecond?

@sawdust “I am sure nobody bıilds comm equipment with such limitations”, can you give a proof document etc.? It seems to me you are the one who makes unrelated claims. The fact that you are not the only one to disagree with that statement changes nothing about the concept.

@sawdust + Even if nobody builds comm equipment with such limitations does not mean at all that they’ve built it at some point in time. The read operations are indeed ‘possible* reasons for transmission time.