Toslink, EMI, and jitter ?

[nagual]

Hi,
I´ve been using a Fiio D3 DAC, to connect the Toslink out of my motherboard to an Edifier M3400 speaker set. So, I´m far from beeing an audiophyle, what I really like to do is to tweak my rig, in order to "increase SQ". Well, since I´ve taken off a 120mm fan from the rear of my pc case, the audio from the speakers seem to be a bit more detailed, pure. First, I thaught it was some kind of placebo effect, but later, as I searched some papers about this matter, I found out that while the optical cable itself, doesn´t suffer from EMI, the Toslink connectors at its ends, where the eletrical signal is converted to optical, and vice-versa, may interact with the EMC, producing jitter. My question is: Must I proceed in my investigations regarding Toslink, EMI, jitter, and the possibility of producing audible noise, or it is all inside my unhealthy mind ?
Best regards !

[Tom F]

That's an interesting observation! A fan is likely to be quite an EMI-noisy device (big coils being switched quickly).

The audibility of jitter is a hotly debated subject which I don't think we should really get into here - but I suppose it is at least conceivable that there might be some susceptibility to EMI in the receiving end of the TOSLINK connection, as here a very small current is amplified quite substancially.

One way to test whether your changes are EMI related would be to put the fan back in and move your DAC around - see if you can replicate the same perceived increase in sound quality by putting the DAC a long way away from the case (source of EMI).

Neat discovery though, wherever it's come from!

Edit: My immediate reaction is that this -isn't- an EMI issue, but you never know...

[nagual]

In fact, what made me link the removal of a fan next the Toslink connector, inside the pc case, and sound quality, was a post in Audioasylum:

"EMI/RFI will not travel down the optical link. However, the timing of the optical pulses is affected by EMI/RFI inside the source box and when the pulses are reconverted to electrical signals at the DAC this EMI/RFI will be resurrected, albeit in an altered form.

Tony Lauck"

I can´t say Tony is right, but I´ve seen in optical cables advertising, praises to "metalic connectors, in order to lower EMI", or something like that. Always seemed to me, and I think for the Marketing too, optical connection means noiseless...
Now, I´m not so sure.

[Andrew LB]

When using a toslink optical output from a motherboard and an external DAC to convert the digital signal back to analog... you shouldn't pickup any noise such as EMI/RFI unless you are powering your DAC off the same electrical circuit as your PC. The most common noise caused by dirty power tends to be ground loops.

Try powereing your DAC and external amplifier/etc off a completely separate electrical circuit. Also try running your amplifier/receiver/etc using a cheater plug. (google it).

[Mufflore]

He looks to be commenting on RF/EMI from the fans affecting analogue circuits before being converted to optical.
Or perhaps the interference is transmitted to the audio circuits through internal PC power wires.
So the optical signal already has induced jitter by the time it leaves the PC.

There is no ground loop between his DAC and PC when using optical only.


Op, if you can reliably repeat the test, it may be worth pursuing.
Assuming it make s a big enough difference to your ears.

[TruMutton_200Hz]

The clock jitter in onboard S/PDIF output connectors of computer motherboards is already to cry for even without the EMI of your case fan making it worse still. If using cMP + cPlay is not an option to you, probably I would try a digital-to-digital converter such as the Audio-gd Digital Interface (the cheaper model of it will do, unless you really want that additional digital input for one reason or another). Either that, or invest in a better DAC (preferrably one that uses a proper galvanically isolated asynchronous USB input...) but then you'd be better off saving the money towards better speakers first IMHO.
There are more expensive solutions to be had than the one from Audio-gd (Musical Fidelity V-LINK, M2Tech hiFace 2,... just to name but a few). I am guessing these might be sort of overkill with the Fiio, though.

[ROBSCIX]

I agree...If you want to have a nice clean digital signal look for a USB to S/Pdif converter they would be much cleaner and offer lower jitter over an onboard offering. People really only start to worry about jitter at the higher price ranges as you probably have other places that can benefit from improvment prior to optimizing against jitter.

While it is true that optical is not susceptible to certain issues that can affect Coax, optical has it's own issues. I find that most of the people I know running high end external digital systems always seem to prefer coaxial.

Information on Jitter and other subjects surrounding digital audio are always recommended reading as the info can help when making future buying decisions.
External digital audio can get expensive quick especially when you want everything to be of the same high quality level. Such as you buy a new digital source and realize your DAC needs replacement. You replace the DAC and figure out your speakers are lacking..etc Lot of fun!

Good Luck.

[Tom F]

Do you have a coax output available? It'd be interesting if you can replicate the experience using that...

Not sure I buy the explanation of the fan affecting the transmission / reception of the pulses - the field strength would have to be insanely high to actually be able to actuate the optical transmitter to a point where it was emitting light above the noise floor of the fibre.

It could be something else - maybe upsetting the stability of a PLL somewhere - in which case a digital-digital interface might indeed help by re-clocking the signal.

An interesting design note from Fairchild on the subject: http://www.fairchildsemi.com/ms/MS/MS-566.pdf

Consider where else these components are used though - things like fibre optic networking / communications are way more demanding than digital audio over fibre optic, where you're much more sensitive to jitter and lost counts - but it still works perfectly. Of course, these products have been designed much better than an afterthought bodged onto a motherboard without a proper power supply or stable clock or anything - but still. You can send Gb's a second with these - a few Mb's / sec for audio and the transceivers are practically standing still

Of course, signal integrity is a career in itself - without proper measurements it's virtually impossible to guess what's going on. Still, it's interesting to investigate nevertheless =)

[ROBSCIX]

I can see a fan introducing noise into a circuit but if that was the case it would be present in both coax or optical. However as Tom pointed out, it might be something else. Either way I would look for a better source for your digital.

[TruMutton_200Hz]

For a clock to be able to output a clean clock signal it not only has to be of astounding quality, but it also has to be fed a clean power signal. There is only so much the internal PSU of a computer can do, and the computer hardware (especially the motors of cooling fans) are feeding noise back into this PSU due to the way the hardware draws current.

So it's not just a matter of what components are used to drive the transmitter / proper shielding it against EMI and RFI. The cleanliness of the power supply's output that's entering the transmitter is at least equally important. However, that's still not the end of the story. If you use a USB-to-S/PDIF converter, you'll still introduce a certain amount of transport jitter which could have been completely avoided.
To be able to understand this, you need to understand how S/PDIF works. First off, it's a one-way communication. Meaning, data re-transmission requests cannot be sent by the receiver each time when it has received corrupt data. Secondly, the information required to partially clean up any transport jitter (i.e., jitter caused by both the transmitter and the digital cable) has to be derived from the signal received. There is no separate clock data being sent along with the signal itself.

While one can argue a high quality S/PDIF chain will reduce jitter beyond that which the human hearing system can discern, the facts I described above remain. Asynchronous USB was designed not to have any of these problems. In asynchronous USB, the receiver feeds the data into a memory buffer (RAM). As soon as the buffer is full, the receiver changes its status from "ready" to "non-ready". Erroneous data is detected by the receiver, and re-requested via two-way communication. The master clock completely ignores the transmitter, it just grabs the data it needs, immediately when it needs, from the buffer. All the receiver has to do is to just make sure the buffer is never empty when the master clock wants more data (i.e., no buffer-underruns should ever occur).

The transport mechanism that's between the buffer and the input pins of the DAC chip itself has jitter also, and every DAC chip has jitter also. No matter still, completely eliminating all jitter caused by the transmitter-and-digital-cable combination generally means half the battle is won, especially if the DAC unit's asynchronous USB input is galvanically isolated properly from the electric noise that's coming from the computer motherboard's USB port (and especially if the DAC unit's asynchonous USB input supports 24-bit 192 kHz). A good example of a DAC unit that uses a downright wonderful implementation of asynchronous USB is my Eastern Electric MiniMax DAC Plus. After having removed its vacuum tube from it to get rid of some unwanted distortions, the extremely low jitter output of my DAC unit via USB now contributes to a three-dimensionality in the soundstage that's so convincingly real it surpasses all of my expectations by a mile or two. My mind is permanently blown. Be warned, though... listening to music can easily become more addicting than customizing the liquid cooling in a computer!