Hi,
Well, the "common mode signal" in this case was maybe 40 or 50V RMS, 50Hz with a healthy spray of harmonics, overlaid with spikes (when for example a light switch was operated).
Most SPDIF receivers easily lock at jitter levels exceeding 1UI, which for CD Sdif (which this was) is around 180nS, at the minimum.
So I guess nothing to worry about in the great scheme of things?
Ciao T
PS, yes, this scales well with level, so in this case 1V RMS should produce at least 3.6nS Jitter...
Well, the "common mode signal" in this case was maybe 40 or 50V RMS, 50Hz with a healthy spray of harmonics, overlaid with spikes (when for example a light switch was operated).
Most SPDIF receivers easily lock at jitter levels exceeding 1UI, which for CD Sdif (which this was) is around 180nS, at the minimum.
So I guess nothing to worry about in the great scheme of things?
Ciao T
PS, yes, this scales well with level, so in this case 1V RMS should produce at least 3.6nS Jitter...
In US all the modern-age receptacles (after 1960) have grounding pins. No way you can have that kind of crap in your audio supply.
Yes, this seems excessive & unusual levels of CM noise.
Abrax, hears the sonic signature of CM noise as sibilance & edginess.
I hear the sonic signature of jitter as edginess & sound stage but I've also heard some sibilance which was alleviated when the jitter was attended to with a battery supply on the clcoks.
At this point, I'm not sure how much, if any, was CM noise.
Abrax, hears the sonic signature of CM noise as sibilance & edginess.
I hear the sonic signature of jitter as edginess & sound stage but I've also heard some sibilance which was alleviated when the jitter was attended to with a battery supply on the clcoks.
At this point, I'm not sure how much, if any, was CM noise.
Hi,
1) I don't live in the US (thank Jah for that).
2) The earth was actually faulty in the building to cause that.
3) You can have easily have 50V or more with equipment that is "double insulated"*.
Ciao T
* you have no limit on the voltage in the safety standards, only on the current, so in theory (and practice) you can have 1/2 mains Voltage or more on the chassis in terms of voltage, though the short circuit current may only be a few 100uA (IIRC < 10mA are considered safe)
1) I don't live in the US (thank Jah for that).
2) The earth was actually faulty in the building to cause that.
3) You can have easily have 50V or more with equipment that is "double insulated"*.
Ciao T
* you have no limit on the voltage in the safety standards, only on the current, so in theory (and practice) you can have 1/2 mains Voltage or more on the chassis in terms of voltage, though the short circuit current may only be a few 100uA (IIRC < 10mA are considered safe)
But even in Germany you have the option of Schuko plugs, you just need to wire the house with three wires instead of two.
Hi,
You did not read my answer, right?
Or if you did, you did not understand it. Try again.
Ciao T
You did not read my answer, right?
Or if you did, you did not understand it. Try again.
Ciao T
Unless there is a faulty ground or faulty X/Y capacitor in one of the power supplies, the common mode currents should be small enough that, across a reasonably low coax shield (ground) impedance, they only generate a small voltage.
However,
- how small is small ?
- the common mode currents can contain not only mains but also various frequencies (picking up radio waves, switching power supplies, rectifiers, etc)
- "reasonably small ground impedance" is a usually a joke unless the thing is built like a real RF device... If it doesn't look like plumbing, then it ain't RF 😀
On a PCB with a solid ground plane and good RF connectors (BNC, SMA) properly connected to the ground plane, the common mode currents will go through the ground plane. If all connectors are on the same side, these currents won't go through the sensitive bits on the PCB. That is, if you got a ground plane.
However, consider a typical setup : a RF-like signal (SPDIF qualifies) is sent down a transmission line composed of unmatched PCB traces, unmatched junk-o-brand RCA connectors, some cable which may or not be 75 ohms, then some more connectors, perhaps a bit of flying wire or ribbon cable, some more unmatched traces, a bit of 75 ohm termination if you're lucky, and the SPDIF receiver's input impedance.
At every point where the transmission line isn't up to spec, parasitics mismatches between "ground" and "signal" will turn some part of the signal into common mode, and some part of common mode noise into signal.
However,
- how small is small ?
- the common mode currents can contain not only mains but also various frequencies (picking up radio waves, switching power supplies, rectifiers, etc)
- "reasonably small ground impedance" is a usually a joke unless the thing is built like a real RF device... If it doesn't look like plumbing, then it ain't RF 😀
On a PCB with a solid ground plane and good RF connectors (BNC, SMA) properly connected to the ground plane, the common mode currents will go through the ground plane. If all connectors are on the same side, these currents won't go through the sensitive bits on the PCB. That is, if you got a ground plane.
However, consider a typical setup : a RF-like signal (SPDIF qualifies) is sent down a transmission line composed of unmatched PCB traces, unmatched junk-o-brand RCA connectors, some cable which may or not be 75 ohms, then some more connectors, perhaps a bit of flying wire or ribbon cable, some more unmatched traces, a bit of 75 ohm termination if you're lucky, and the SPDIF receiver's input impedance.
At every point where the transmission line isn't up to spec, parasitics mismatches between "ground" and "signal" will turn some part of the signal into common mode, and some part of common mode noise into signal.
First point to note is that ground is rarely ground at RF - inductance rules. So even if the case is grounded securely enough for safety purposes (50Hz) its not going to look much like ground at 5MHz. Very few CM chokes on the mains do much above 5MHz - they become capacitive normally at lower freqs than this. I'm not by any means certain that frequencies above 5MHz are the problem but ISTM that opamps have no CMRR at such frequencies.
When we have both a CD player and a DAC powered from the mains, the SPDIF cable between the two completes the loop at RF. There's a further loop formed of the DAC and pre/power amp via the grounds of the audio cables. Ground loops are known to cause problems at LF, why would we assume they're completely benign at RF?
But this is going off the topic of jitter somewhat ;p
When we have both a CD player and a DAC powered from the mains, the SPDIF cable between the two completes the loop at RF. There's a further loop formed of the DAC and pre/power amp via the grounds of the audio cables. Ground loops are known to cause problems at LF, why would we assume they're completely benign at RF?
But this is going off the topic of jitter somewhat ;p
Well, it's not really off the point as the O/p asked is jitter a non-issue or have we given up - I see the RF & CM noise as still on topic & very relevant to the Q - correct me if I'm wrong.
I think it's a very fruitful line of discussion. I presume, these issues, if they are of importance in audio, will show their sonic signature in the audio band?
Going back to an example I gave of a very low jitter device showing improvement when it's USB PS was substituted by an external PS, leads one to the premise/conclusion that jitter is not the only important measure in digital audio transmission.
I think it's a very fruitful line of discussion. I presume, these issues, if they are of importance in audio, will show their sonic signature in the audio band?
Going back to an example I gave of a very low jitter device showing improvement when it's USB PS was substituted by an external PS, leads one to the premise/conclusion that jitter is not the only important measure in digital audio transmission.
Noise correlates directly with jitter. In a logic signal with a finite rise time changing the amplitude with noise moves the threshold with respect to time.
Noise in the Digital signal cable interface will cause jitter even on a dedicated clock line. Common mode noise is one the worst offenders. Thats why smart guys use pulse transformers and differential receiverers or optics. This is very basic stuff and well documented. Noise on the SPDIF interface is a sonic disaster. That's why many people use Toslink even with its bandwith issues.
Noise in the Digital signal cable interface will cause jitter even on a dedicated clock line. Common mode noise is one the worst offenders. Thats why smart guys use pulse transformers and differential receiverers or optics. This is very basic stuff and well documented. Noise on the SPDIF interface is a sonic disaster. That's why many people use Toslink even with its bandwith issues.
I'll just mention again that in my references to CM noise I'm not particularly interested in it on the signal - I agree it translates to jitter there. Rather what concerns me is its presence in the system as a whole. I point this out because it appears stormsonic made the same mistake in one of his posts, in saying that CMRR in a video opamp will deal with the problem. It won't - the CM noise is common mode (by definition) so is present on both signal AND ground.
Seems to me he's saying they use Toslink because they consider CM noise a more pressing issue than jitter. Kinda like me 😀
Decoupling the ground at the Dac with a 100 ohms or so and using the other side of the resistor as the coax ground point and taking the signal differentially will get rid of much of the common mode noise. There are differential video op amps made just for the purpose. One of these is the AD830. Works great with improvements in imaging and naturalness. Built many of them as a commercial product.
I know 🙂 but my understanding from many, many sources is that Toslink implementations are seriously jitter laden & sound bad so it would seem CM noise reduction is NOT enough to trump jitter increase?
An useful experiment might be to have a DAC, like Thorsten's with the local re-clocking & compare SPDIF via coaxial with SPDIF via Toslink. This might give an idea of the possible contribution of CM noise?
Ah so you're saying do not connect the shield of the SPDIF cable to the local DAC ground directly, but only through a 100R resistor? I agree that's a good start so long as the receiver can cope with the CM signal. I will check out the AD830 its not a part I'm familiar with - thanks.
CM noise a more pressing issue than jitter
NO..... CM is a major source of jitter. There is no common mode with toslink
and well done Toslink can have less jitter than many coax SPDIF links.
NO..... CM is a major source of jitter. There is no common mode with toslink
and well done Toslink can have less jitter than many coax SPDIF links.
But Thorsten's one of the more competent designers out there, so he'd minimize the audible effect of switching between the two. Rather we need a DAC where the designer was unaware of CM noise issues...
Yes, but do you know of a DAC where there the only difference between Toslink & Coaxial SPDIF is going to be the CM noise i.e the jitter is controlled & common?
But if I understand your AD830 suggestion correctly, your 100R resistor in the ground lead actually ADDS to the CM noise at the receiver yet you're saying it gives a better sound. So one of us is misunderstanding the other 🙂
Yeah, that's what peufeu was suggesting a few posts back. I agree - easier to fix the jitter probs with Toslink than fix CM noise with cable.
- Status
- Not open for further replies.