What? Is this really a question?
Because of high speed switching principles. The return currents follow the path of least resistance, or in this case, least inductance. Travelling across the PCB, possibly through sockets/card interfaces, down a USB cable, through the ground plane of the DACs PCB, down the audio cables shield back into the PC, then back to the PCs PSU is NOT the path of least inductance.
There are reasons why sensitive, high impedance circuitry can coexist on boards with high rise time, high current switching signals because with the right design principles you can keep the various return currents self contained.
Local decoupling is usually specified to keep the majority of the high rise time return currents local to their specific device, minimising radiated and conducted EMI. Where in sensitive applications additional filtering is often provided to keep things completely under control.
Okay dude.
Edit - The point is that sensitive and high current switching electronics coexist within the PC environment. If the latter were allowed to wreak havoc with the former then the PCs would not work. It's not like the design engineers at ASUS etc go out of their way to botch the layout and BOM to make things worse for themselves. Yes certain 'extras' often included on motherboards, such as on board audio, can suffer, but this is mainly because they are simply added onto the board wherever there is space left as a result the audio performance suffers. It doesn't have to be that way though.
Last edited:
You'd better believe it.
Sure, but you've not yet so far identified what's the source of the high speed switching. Wanna have another try at it ? Hint - its nothing residing on the motherboard in this instance.
Yep. You've got a switching transistor on there typically swinging over 350V (UK mains) on its collector with fast edges. What's the lowest impedance path for the currents through the parasitics to return to earth?
Here's an article that might help - http://www.eetimes.com/author.asp?doc_id=1279630
Here's an article that might help - http://www.eetimes.com/author.asp?doc_id=1279630
Last edited:
You do realise that the majority of PC SMPS do not conduct or radiate much in the way of high frequency grunge right? And that if you scope their outputs, or the input for that matter, any rubbish that's present is usually present as a result of back flow from the PC itself.
Actually they swing considerably more than that with the circuit parasitics internal to the PSU. You get ringing present between the drain and the source such that you need to specify 550-850V parts to ensure reliability. Still the currents flowing around the switching nodes within the SMPS are tightly controlled via layout. The return currents are all internal.
On another note lots of PC SMPS use LLC converters which do not use hard switching in their main power switches and use sinusoidal switching patterns instead.
You could sign up to EETimes and be the first to leave a comment on Mr Kollman's (very helpful) article that he's tilting at windmills, all parasitic currents are already resolved internally 😀
No, obviously not all parasitic currents are going to be controlled, in fact none ever really will, there is always some bleed through, but it's a matter of degree. The point being that they can be reduced to insignificant amounts if correct design procedures are followed.
Now this isn't to say that every PC environment is going to be suitable. There are certainly going to be knock off products from China that skimp heavily and will have less than satisfactory performance. But if you buy from reputable manufacturers and don't spend as little as possible one each and every part for your PC then you are not going to have any problems. The same goes for the sound card itself. If you buy a cheap knocked down product don't expect guaranteed performance across platforms.
I mean do you read power supply reviews for PCs? I certainly follow Kollman's power tips over at TI. Supply ripple and noise are one of many aspects that are generally covered with the good in depth reviews. Most good PC power supplies are actually very quiet with minimal noise and ripple.
I don't build my own PCs any longer so given up reading PSU reviews. When I did read them I never noticed any talking about common-mode noise so I can't see the relevance. The output (i.e. normal mode) noise and ripple has no bearing on the CM noise as far as I can see, they're orthogonal things.
Sorry i didnt.
I compared directly side-to-side with the NAD M51, the smaller Weiss DAC2 and the internal DAC from the DEQX HDP-3.
Only thing i can say is: that Weiss DAC1 mk3 is one hell of a beast. I understand why the pro sound guys love it that much.
Common mode noise on the input to the PSU will be controlled as regulations cover line contamination. The output side of the PSU should be fairly clean too as it's completely isolated from the primary switches and the PFC controller/switch too. With the output currents of the transformer following the sinusoidal input current there will be far less of an issue than if the converter was hard switched.
Like I said, most of the noise that gets thrown around comes from the PC itself, not the PSU. All of the stuff, besides parasitics, will be wanting to return to the PSU and not down the USB cable, which is also differential.
Good system design, in general, solves most of the problems that audiophiles like to spout off about regarding PCs.
What fun is that?!
It wouldn't surprise me in the least that older systems using ancient motherboard-based codecs with poorer layout/PSU's and OS's that glitched the audio stream on interrupts are where these biases originated from.
But things have marched on and, some folks haven't. What was exceptional before might well be mainstream now.
And all this handwringing about hash would show up in a second on an two-tone IM test. Actually, just a single tone should work, no? I mean these spurious signals are what we'd expect to IM...
Exactly. If you've dealt with mixed signal hardware where emi and cross contamination do occur then it shows up plain as day.
That was with my phono stage in the loop. So that must be the thing cancelling all the distortion and noise. I'm an awesome designer.😀
Just thought I'd pop this up.
Even Berkeley now with their Reference Dac have gone back to using Multibit conversion.
"Alpha DAC Reference Series, use multi-bit D/A converters because they provide better performance than 1-bit converters - even DAC’s who advertise “native” DSD compatibility. "
Cheers George
Even Berkeley now with their Reference Dac have gone back to using Multibit conversion.
"Alpha DAC Reference Series, use multi-bit D/A converters because they provide better performance than 1-bit converters - even DAC’s who advertise “native” DSD compatibility. "
Cheers George
As far as I was aware, BAD haven't used 1-bit DACs in their products. The consensus seems to be that they used AD1955 in their previous DAC which is a multibit S-D.
Maybe, their early one sounded smooth but lifeless to me and others when compared to my Cary 303/200 PCM1704 AD844 I/V unit, and the reviewer here who's place we used for comparing in Australia said it was Delta Sigma, I said, "that's why" back to him.
Maybe it was just bad implementation of Multibit.
Cheers George
Last edited:
'Bad implementation' aka 'multibit without enough bits'. AD1955 is delta-sigma, I think using 5bit DAC.