Calculating and storing the perfect signal is trivial. But having measurement instrument accurate enough to give good results, and algorithms good enough to interpret and make something meaningful of the results are harder. But if it was possible, then we would get a measurement that actually tells a lot about the transparency of the device (unlike the very basic measurements typically done today).
We are talking about an analogue signal, right? Something you can âcompare to the output of a DAC.â
So, if storing a âperfectâ analogue signal is trivial, why donât we all just do that and dispense with digital audio entirely?
I already explained that earlier, you donât store the analog signal, you store input to the math used to create it, and then just apply that math with the stored input. But unlike 44/16, you would do so with much higher precision and in such a way it can more easily be created lossless (i.e. not as samples but as control points for sine waves).
Or if you prefer, just use the 44/16 but unlike a DAC you calculate it accurately.
In other words, you store it as digital audio.
The device for doing so is called a Digital Analogue Converter (or DAC, for short).
What is commonly called âhigh-resâ digital audio.
Umh, so youâre proposing a new digital audio encoding scheme? One thatâs âbetterâ than the standard ones (high res PCM or DSD).
Thatâs great! I hope your new technology comes to market soon. Thanks for working on this!
What do you mean by âunlike a DACâ?
What you are describing is a DAC. Just one that (according to you) is way better than existing commercially-available DACs. I encourage you to bring that to market, too!
You will be rich, man, rich !
1 Like
Donât put words in my mouth, what I am proposing is a new way to measure sound, nothing more. If you donât understand it then read what I have said again and think about it (its not that complicated).
You âexplainedâ how to âstore input to the math used to create it, and then just apply that math with the stored inputâ and thereby create a âperfect signalâ which you can âcompare to the output of a DAC.â
Thatâs the holy grail of digital audio.
The only thing missing here are the actual details of how you succeed at doing that.
I already explained that you need the whole signal to do so, and it will likely be computational heavy. So its not something that can be done for a DAC, at least not as DACs are today.
I donât know if you have a hard time understanding what I am saying, or just trolling.
1 Like
Great. So I give you the whole signal, f(t). What do you do with that?
For simplicity, letâs assume the track is of finite length: f(t) =0 for t<0 and for t >T. Thatâs still an infinite amount of data (the value of f(t) â a real number â for every t between 0 and T).
What is your procedure for capturing that with a finite set of 64 bit float numbers?
(Hint: the miracle of the Nyquist-Shannon Sampling Theorem is that this is possible, given a certain set of assumptions about the signal f(t).)
Hey Magnus
You want to transform an analog signal to some reference points. Afterwards you want to compare the played music again to those reference points. Thatâs what I understood.
If I understand you correctly, you want to sample the analog result the comes after the DAC again and compare it with the original sample. Is that correct?
Why should I capture it? All I need is to compare result of f(t) with measured data on enough t to get a good enough subset. For example, I could compare on every millionth second. Each comparison will yield a difference between measurement and calculated signal, which in turn can be used to calculate an index for a âtransparency valueâ or something similar.
How are you going to store the analogue signal, f(t)? You already said you are going to store it in some digital form. What is that digital form?
So youâre going to sample f(t) (with a 1MHz sampling rate) and store the samples?
Please be specific here.
Sounds correct, but for completeness sake Iâll summarize here in more detail:
- Select a complex and well recorded music tune in 44/16 digital format
- Use the digital data in that tune to calculate a âperfectâ signal, which will be the sum of a number of sine wave peaks/valleys, and store control points for the sine waves, which will in effect create a function that can calculate the amplitude of the signal at a given time. Note that this can only be done accurately when you have the whole signal, from start to end. DACs do an approximation and usually a filter to get rid of high frequency garbage.
- Measure the analog output of a DAC on the same tune, store samples on a predetermined interval (for example every 1 microsecond), for each channel obviously.
- Compare measured samples with calculated results to see deviation. Difference between measurement and calculated value will be stored and analyzed to calculate a âtransparency indexâ
- Lots of possible further result analysis can be done, like transient response, crossover, etc. But the main goal is to get a value of a DACs transparency (i.e. how close to perfect it can generate a analog signal with reference voltage given a digital tune as input).
Note that this will measure the whole DAC and not only the reconstruction process/filter, including jitter from external source and clock-jitter, analog stage, handling of electronic noise and so on.
Normally, with a shelf with 4 legs you always have to assume that something is not 100% on the floor. Therefore I have planned a small compensation. However, the locksmith worked so precisely that it turned out to be unnecessary afterwards.
1 Like
This was an awesome video.
Thanks!
I guess answer be no. But im a person that want to buy quality gear, so im using Supra CAT8 cables and a Edgeswitch.
Fantastic design, and amazing workmanship. 
1 Like
Thanks Mike. As you know, Iâm the one with the room to drag you out of. And thatâs how I implemented my own ideas. 
I havenât read all the comments in this thread, but my understanding is as follows:
An ethernet cable transmits a bit-perfect, error-corrected, and buffered signal to a DAC. Bits at one end, the same bits at the other. One of the arguments here seems to be concerned with noise, and that somehow RF (or other) noise has some impact on the intelligibility or quality of this signal, so hereâs a visual analogy:
To my mind, this is analogous to the issue of noise and ethernet cables and, more specifically, its impact on the intelligibility of the signal they transmit, i.e. both images convey exactly the same amount of meaningful data. Yes, thereâs lots of noise in one of the images, but what impact does it have on your understanding? Thereâs no ambiguity in the second image, and as such thereâs no degradation in the meaning of the signal - the salient data is intact. My understanding of ethernet cables is that provided they transmit the necessary data, their job is done.
2 Likes
I donât think noise has any impact on the intelligibility or quality of the signal. If it did, there would simply be loss of signal. Under normal circumstances the DAC receives, processes and transmits that just fine. It also receives and transmits the electronic noise (from USB) at the same time. Some DACs handle incoming noise it better than others. Having tried the Ethernet isolation thing I donât buy noise in the network transport, at least not in my audio chain and none of the proponents use what Iâd term out of the ordinary hi-fi gear.
Yeah, and this is the point where my technical knowledge lets me down as I donât understand how this works. I was as a photographer for over ten years, and at no point during that time did I send an image to a client that didnât arrive in exactly the same form as it left me. Every single 16bit pixel ended up on their screen looking exactly like it did on mine. No change in brightness, or colour, or any other metric you might possibly use to measure their veracity, despite their journey down countless numbers of ethernet cables. The signal arrived at its source, as shiny and perfect as the moment it departed.