How does RF noise interfere with Digital music?
Posted by: MartinCA on 29 August 2011
Can someone explain to me how RF noise interferes with digital music?
The two contexts I have in mind would be a transporter feeding into an nDAC via a S/PDIF interface (assuming an optical connection), or a USB stick plugged into the nDAC. In previous threads, the possiblility of RF interference in these scenarios has been discussed, and I wanted to understand a bit more about what this means.
I understand how RF noise can be created (more or less) and I understand (more or less) how it impacts analogue music in and downstream from a DAC. But I'm not sure what is happening in a digital environment - does RF interference corrupt the music once it is unpackaged from its file format into a bit stream, or what?
Thanks to anyone who can explain this intelligibly!
Now you might say well surely these distortions are far higher than I can hear ? Fair comment. But a) the rf reflections can push an amplifier so it is non linear, probably a little unlikely with good designs. Or b) the intermods caused by the amplifiers and related components not having enough bandwidth to handle the RF. Intermods, are kind of the inverse of harmonics, they are the sums and differences of the rf reacting with the pass band audio. These are audible distortions, and usually extremely low level. But on a revealing system they can add hardness, smear, rob micro dynamics etc.
You mention digital audio. Well this distortion can occur within the analogue low pass ant aliasing or re construction filter and out put stages.
Finally a DAC or equivalent should be designed to with stand it's own RF sources (ie it's clocks) however rf from other sources can infect the circuitry. By it's nature rf can jump across conductors etc ( it's obvious if you think about it...) and so can be quite hard to cure, and prevention is a lot preferable.
Finally it's worth pointing out on cheap in accurate electronics, the noise floor is quite high and so the RFI distortion although it might be present it can't be heard. On quality electronics such as Naim the signal to noise ratio is a lot higher, ie the system noise floor a lot lower and so the noise from RFI can become more noticeable.
Simon
Hi Simon
Thanks for this. But am I right in thinking that most of the impact of what you describe is in the analogue stages? In these stages, I can understand that RF noise will corrupt the electronic wave form, and in interesting ways that would not be immediately apparent - like those you are are describing. Moreover, looking at the Naim DAC white paper, it is evident that a lot goes on in the DAC circuitry that could be sensitive to RF noise. I'm sure that the nDAC is not totally immune to RF interference within its own circuitry, but I trust Naim to get it as right as they reasonably can for the price.
I thinking about further upstream of the DAC chips, and back to some of the threads we were both part of some weeks ago ..... there were the two scenarios that left me puzzled ....
1) the S/PDIF interface out of the UnitiServe (and similar transports), where the discussion suggested that this was subject to RF interference. I didn't properly understand how RF interference might occur.
As you know, computer files (including music) sent across a WAN are sent in packets with error-checking so that if a packet arrives corrupted the network will keep trying to resend until it arrives correctly - and much the same happens when computer data is read off a data store. So this should, whilst not immune to RFI, cope with RFI, and the music file should arrive at the start of the S/PDIF process intact and correct. Then the S/PIF interface unpacks the data and transmits it as a bitstream with the clock. I understand about jitter across S/PDIF, but that is not what I am trying to get to grips with. I am guessing that RF interference could cause corruption of the bit stream within the SPDIF interface?? The use of an optical cable might isolate the DAC itself from RFI from the S/PDIF, but it won't stop problems within the S/PDIF?? Or is it that the circuitry that processes the conversion of the files into S/PDIF pulses is sensitive to RFI?
2) the hypothesis that RFI causes slight degradation of SQ off a memory stick into the DAC, which was something you suggested. Would this be the RFI generated within the memory stick interfering with the DAC processing? Or would it be similar to what is happening in a S/PDIF interface?
Hope the question makes sense!
Thanks
Martin
But for now let's deal with analogue and SPDIF.
First of all forget about spdif being a digital signal. It is primarily an analogue signal that is driving an amplifier input via a galvanic isolator. The isolator removes earth loops, but any HF noise on top of the spdif shaped signal will couple through to the spdif input amplifier. This in itself can couple into power lines in the device etc. Potentially this noise if not decoupled can propagate to the analogue stages.
Now it's fair to say the levels we are talking about are probably hovering around the noise floor. But some of us pick up on this.
Where does this noise come from in the first place? Switch mode Power supplies, and noisy electronics are often culprits.
With regard to memory sticks, USB is purposefully a loose standard. The serial line effectively sees the USB connector and any lead as a transmission line with reflections. If reflection and impedance mismatches occur the voltage can vary quite a lot. Again the electronics reading this can cause a signature on the power lines to varying degrees based on this variation.
Everything can affect everything else. The art is to get these affects below the noise floor so we don't notice them, and that in my book is the art of good engineering. However when dealing with lose standards such as USB this might not be possible, in other words a case of trial an error.
It is when these artefacts poke just above the noise floor from time to time then that's when some of us pick up on it.
Simon
Simon
You say that S/PDIF is primarily analogue - but it is still a means of transmitting digital sound, essentially a stream of 0s and 1s, and I think it is analogue only in the sense that the information is transmitted using a biphase mark code which is a sort of analogue for the digital 0s and 1s. As I understand it, a transition (from high to low value or vice versa) counts as a '1' and staying in the same state counts as a '0'.
I just looked up biphase mark code. What's interesting is how this data is merged with the clock information (i.e. the transitions are the sum of the data and the predictable clock transitions- they are not separate), which seems to me to suggest that in order to read the data part 100% accurately, the clock data must also have been accurate - and that suggests in turn that poorly clocked output would not only suffer from jitter (which the nDac should be able to handle, up to a point) but also from degradation of the ability to retrieve the bit stream accurately.
If this is correct - this would help explain why we hear differences in digital sources in the nDac despite the de-jittering capabiliities of its buffering approach.
Martin
From a signals perspective, the SPDIF digital encoding is in the time domain (the digital domain) and the analogue waveform is in the frequency domain ( the analogue audio domain). It's time domain noise that is the jitter and frequency domain noise that includes RFI.
Simon
Well.... you've get digital data that is transmitted by pulses, like over ethernet, or it can be transmitted using a sine wave representing the bit stream, like over wireless comms. Both are digital in terms of the data they are carrying.
I don't know whether it matters for S/PDIF which way it is carried - I had thought that an optical cable would have to use light pulses, wouldn't it? So I thought the S/PDIF interface is more about the protocol for using transitions to carry both the data and the clock. And then it would be waveform over BNC and pulse over Toslink? Please note - I have no expertise in this, unlike you - I am just trying to make sense of it!
Anyway, the sine wave that the digital data represents isn't being directly affected by any RF interference. As I understand it, the question is whether the amplitude of the interference results in '1s' being flipped to '0s' or vice versa. If the amplitude of the noise is not that great, it can easily be disregarded. I think this is the same as what you are saying?
So - two things from this...
(1) Back to the optical connection (between the Unitiserve and the DAC) - this should neither create nor transmit RF interference, yet BNC connections (which have the potential to do so) sound better - so the implication is that the worst of the RF interference in that setup is happening in the S/PDIF interface on the transporter side and in the digital world, rather than in the analogue circuitry of the DAC.
(2) I am not sure that I understand your last paragraph. Doesn't the S/PDIF transmit the two PCM channels that make up the digital recording? It is still digital data at this stage - basically the collection of sample amplitudes that the DAC translates into an analogue signal. My understanding was that this digital data is combined with the clock to produce the encoded stream that SPDIF transmits. I know Wikipedia is not the last word on this but here's is a link that describes this....
http://en.wikipedia.org/wiki/Biphase_mark_code
Not simple, trying to understand this stuff is it? I have a colleague at work who occasionally queries me about the wonders of digital music, and it doesn't take long before he gives up muttering darkly about how he's always known that vinyl is better.
Can someone explain to me how RF noise interferes with digital music?
The two contexts I have in mind would be a transporter feeding into an nDAC via a S/PDIF interface (assuming an optical connection), or a USB stick plugged into the nDAC. In previous threads, the possiblility of RF interference in these scenarios has been discussed, and I wanted to understand a bit more about what this means.
I understand how RF noise can be created (more or less) and I understand (more or less) how it impacts analogue music in and downstream from a DAC. But I'm not sure what is happening in a digital environment - does RF interference corrupt the music once it is unpackaged from its file format into a bit stream, or what?
Thanks to anyone who can explain this intelligibly!
RF interferes with the timing of the clocking speed of the digital to analogue clock and creating a hard and unnatural sound sound to your music. This can be seen clearly when RFI is removed or partially removed.
You are right optical coupling goes a long way to remedy this. But some noise will modulate the emitting diode. This noise modulation will be transferred to the receiver at the physical optical receiver component, ie the device that is turning the light back into an electrical pulse. If the received noise (the analogue noise extracted from the light) that is say transferred to electrical noise on the receiver power lines is above the noise floor it may be heard, but I believe this should be less likely than a direct electrical connection. However there are often other unrelated compromises associated with consumer optical connections that mask any benefit.
I hope that helps???, but don't worry these issues are no way specific to digital audio :-)
Simon
Simon
Thanks - that makes sense, and yes - it helps.
Where I was coming from was trying to understand where why one source for a DAC is better than another (especially if playing the same file) - i.e. how SQ is lost between the digital stored music and the DAC (recognising the compromises inherent in both digital stored music and in a DAC).
I wouldn't say that I have a crystal clear understanding of this now! But it is becoming less hazy.
Did you look at the url I attached in the last post?
This is apparently the encoding S/PDIF uses and what I think it is saying is that clock and data are combined. I can't reconcile this with the statements that the music and clock are in different domains (I mean, conceptually they are, but physically it would seem they are not). The article suggests that you get clock 'ticks' represented by a bit stream (representing peaks and troughs of the clock waveform). Then you get the music data on very other beat which is combined with the clock data. Because the clock data is predictable it can be subtracted from the combined stream at the receiver, and the music data recovered. But then if the clock has been modulated by RF interference, as Steve suggests, it would imply that some of the music data would not be correctly recovered.
Thanks for the URL. It is a good summary of the physical encoding method used by SPDIF.
To answer your query If the level of the RF is significant compared to the SPDIF then you are correct it will disturb the pulses by clearly modulating them. However RF that causes interference that we typically refer to here is vey low level and so will be say a less than 1% of the full pulse voltage, so although they modulate the pulses, the shape doesn't materially change. Digital circuits need the shape of the waveform, as they measure a signal at a discrete time or level, signals we hear are more about frequencies than shapes... But your analysis is on the right paths, as you see how rf contamination of the spdif can affect the jitter of the signal by modulating the pulses even by a small amount.
Simon
Electromagenetic Interference (EMI ) can affect other electronic and lectrical apparatus in many ways
Here are just 500 ways from the funny to the fatal.http://www.compliance-club.com/pdf/banana%20skins.pdf
Interesting read with radio frequency some audio frequency examples
Site may need registration
Keith Armstrong's (technical) articles on the site are excellent if you want to read in depth. There's a section with device & circuits layers and one with the systems layer. Shame they are not all in PDF, though the systems layer one is
Simon
Interesting link! Thanks.