SMPS and grounding

[kboung]

On 11/2/2017 at 2:29 AM, Superdad said:

 

On 11/1/2017 at 4:59 PM, kboung said:

Regarding combination of iso-regen powered by LPS-1

 

(1) For "tick/dropout" issue with iso-regen, it was recommended to ground downstream of iso-regen

(2) For high impedance leakage with LPS-1, it was recommended to ground Meanwell output negative

 

In this case, both groundings required or (1) will actually cover for (2)?

 

Maybe Alex or John can clarify. Thanks.

 

Hi Kboung:

 

If one is using an UltraCap LPS-1 to power the ISO REGEN, then no, grounding [the zero-volt "neg./grd"] of an SMPS that is "energizing"/charging the LPS-1 is NOT the same thing as grounding the downstream side of the ISO REGEN (the DAC's USB input).

 

This is because the LPS-1 is an isolated/"floated" PS (notwithstanding the newly discovered high impedance common-mode leakage which can creep across the unit's 100pF of capacitance between power domains), and its input and output "grounds" are not common to each other.

By the way, most people do not end up needing to ground the output side of their ISO REGEN.

 

Hope the above clears things up for you.

 

--Alex C.

 

 

6 hours ago, Frojo said:

Hi Em2016

 

My system mirrors yours, IsoRegen/LPS-1 into Chord Hugo. I grounded the IsoRegens usb output and found it to be helpful, however, once i added a ground to the meanwells dc output sq took a backwards step. In my system, the best outcome was achieved by forgoing the IsoRegen ground and adopting the meanwell ground alone.

 

 

 

Hi Frojo,

 

I also experienced something similar. By grounding the meanwell output DC neg, I found that I no longer need to ground the downstream of ISO regen USB output.

 

I have ISO regen powered by LPS-1 to the Singxer F1. I get random "ticks" and "dropout" until I ground the ISO regen USB output as suggested by Alex. Now with Meanwell grounding only, the "ticks" are eliminated as well. It seems that only need either one of the groundings instead of both. Somehow the Silanna "built up charges" is able to drain across the "capacitive" barrier of the LPS-1 to the reach Meanwell ground ?!?

 

 

 

[jamesg11]

20 hours ago, Em2016 said:

Hi John @JohnSwenson

 

Once you shunt the high impedance leakage to ground, using any of the methods you've suggested - where does this leakage go?

 

Obviously it goes to ground but what does it do?

 

Does this shunted leakage 'circulate' via the ground path of other GROUNDED SMPS's that happen to be plugged into the SAME powerstrip?

 

Or is it hard to say exactly what this shunted leakage does and where it goes?

 

I know you're flat out on new products and I do know you've stated that leakage loops form between power supplies so no need to revisit that (unless that's where my question leads to?).

 

The background to my question kind of relates to my current recently re-jigged setup.

 

At my office desk I have a Hugo2. This DAC has 2 separate micro USB ports, one for 5Vdc charging and one for USB audio input.

 

For the DAC's USB audio input I have the ISO REGEN + LPS-1 with grounded Mean Well - but I need to ground the ISO REGEN's USB output, using an Grounding ESD Plug and Silicone wire.

 

For the DAC's 5Vdc input, it's powered by a grounded iFi 5V iPower.

 

Sure the leakage loop from the computer is blocked by the ISO REGEN but I guess when I look at my setup I keep thinking that there's a leakage loop between the Ground ESD Plug and the iFi 5V iPower, through the DAC's groundplane...

 

So is all the leakage that's been blocked upstream of the ISO REGEN now travelling between the Grounding ESD plug and DAC's 5Vdc grounded iFi 5V iPower, through the DAC's groundplane?

 

Much appreciated, Sean.

 

 

 

Apropos? -

https://www.computeraudiophile.com/forums/topic/35129-a-novel-way-to-massively-improve-the-sq-of-the-microrendu-ultrarendu/?page=18

 

[asdf1000]

47 minutes ago, jamesg11 said:

Apropos? -

https://www.computeraudiophile.com/forums/topic/35129-a-novel-way-to-massively-improve-the-sq-of-the-microrendu-ultrarendu/?page=18

 

 

Hi Jim, I had seen that. But that's about 'noise' going back into the mains line it seems, not leakage current specifically?

 

[jamesg11]

Well, js covers the high impedance leakage from the smps - hasn’t yet seen it on ac mains; but not fully explored clearly, so yes, your question awaits further investigation ...

 

Today I got round to getting rid of my sole smps - music files hard drive now on lps, using ghent splitter cable, shared with netgear switch. (it’s a 3 prong mains plug, so probably output grounded). Also, moved the macmini with the ethernet input further away from the AE, although still less than 1m.

Still liking the sound of the isoRegen downstream of the ultraRendu ...

[Cornan]

1 hour ago, Em2016 said:

 

Hi Jim, I had seen that. But that's about 'noise' going back into the mains line it seems, not leakage current specifically?

 

 

Well it does! This post by JS further down the thread do somewhat answer your question.

 

Specificly this part:

 

 

As someone else mentioned traditional ground loops are pretty rare in home audio systems. What is more likely is that you add another path for leakage current to flow through. In some circumstances this may actually help. In your example of a system with shielded Ethernet cable connected at both ends to the boxes, all leakage from the up stream devices power supply goes right on through. That would go right on through the renderer and into the DAC.  If the supply powering the renderer has a grounded output at least the high impedance leakage from the network will get shunted by the renderer before it gets to the DAC. At the DAC there are a number of possibilities. If the DAC or anything else down stream of it has a power supply you can get a leakage loop through its power supply even if it is not "grounded". Un-grounded leakage loops still happen and high impedance leakage likes to go through these.

[asdf1000]

10 minutes ago, Cornan said:

What is more likely is that you add another path for leakage current to flow through

 

Thanks Cornan, I guess this line in particular answers my question.

 

ANY path that forms a loop (with the DAC being part of a loop) can create a path for the leakage through the DAC.

 

Even if it's just grounding wire with a grounding plug I guess. Which could be a problem for those of us that need to ground the output of the ISO REGEN, if the DAC has other grounding. Or even if it doesn't - since leakage currents don't just travel via ground.

 

So even if you block all the leakage coming into the USB DAC input, then it can find it's way into the DAC via the grounding wire path and through the DAC's SMPS (if it uses SMPS, per my original example/question).

 

I am trying to piece together bits and pieces to draw these conclusions but hopefully @JohnSwenson can correct me if I'm wrong.

 

 

[Cornan]

Just now, Em2016 said:

 

Thanks Cornan, I guess this line in particular answers my question.

 

ANY path that forms a loop (with the DAC being part of a loop) can create a path for the leakage through the DAC.

 

Even if it's just grounding wire with a grounding plug I guess.

 

 

I have experienced it first hand in my own setup. I use to have all my floating SMPS DC- output grounded. It destroyed the sound, especially when I added external voltage regulators (LT3045 in series) in the DC cables. Now I only have DC- output grounding on the floating SMPS powering the Aqvox switch plus the Aqvox switch GND and t sounds much better. When I started to add the voltage regulator a it felt like I was playing a pin-ball game with leakage. Depending on where I put it the leakage seemed to bounce in different directions. Worst was when only the Aqvox switch had voltage regulator a and all SMPS DC- output grounded. Then the leakage probably sailed right through the ISO Regen into the Brooklyn DAC. Not nice at all!

[R1200CL]

@Cornan So basically follow John’s original guide, which says only one SMPS grounded, and preferably at the last chain (switch) in the audio paths seems to be the way to go then. 

 

Howeverer if you have several LPS-1 with the Meanwell, I suppose all those LPS-1 may need that ground shut. 

 

Maybe @Superdad can answer that ?

The question is, if you have more than one LPS-1 (not in series) in your system, and at same power strip, is it then sufficient to ground only one Meanwell SMPS, or all ?

I expect this is easy to measure. 

I hope the question is clear. Take two LPS-1, ground one Meanwell, not the other, and measure if leakage is affected equally on both LPS-1. (vs ground both).

 

And also if none of the Meanwell’s is grounded, but say you just ground your Netgear FS105/8, what effect does this have on leakage that can/will pass through the LPS-1 ?

 

Further if I understood @JohnSwenson findings in another post, ground the Meanwell internally vs grounding the same Meanwell “externally”, made a difference. Was my understanding correct, and if so, can any advice at the present be given on which method to me used on the Meanwell ?

 

[Cornan]

15 minutes ago, R1200CL said:

@Cornan So basically follow John’s original guide, which says only one SMPS grounded, and preferably at the last chain (switch) in the audio paths seems to be the way to go then. 

 

Howeverer if you have several LPS-1 with the Meanwell, I suppose all those LPS-1 may need that ground shut. 

 

I would say that it most likely is system dependant. Probably close to impossible for anyone to know unless they get the full picture of each perticular system. It needs to be tested/experimented  in each system. What you´ll need to know in my case is that I am using floating SMPSs that are connected to a balanced isolation transformer with floating center tap (balanced & floating) followed by a AC mains DC blocker trap filter. When I ground the DC- outputs of my floating SMPSs it does´nt sound good enough to ground it to the same star-earth wired powerstrip/PSD that the SMPS is connected to. In my case the optimal spot is connected to a Entreq Minimus grounding box, ie. on a separate 0v reference point. If you think about it the floating ITs safety ground reference is actually the floating ITs chassi. Lets say you connect a lot of DC- output grounding to a powerstrip/PSD that is connected to a floating IT. Where do the leakage end up? Most likely the most of it ends up in the floating ITs 0v reference point, but since you have a multiple ground wires connected to a powerstrip/PSD BEFORE it ends up in the floating IT you are likely providing alternative leakage paths into other devices in the chain before it even reaches the IT. As long as you do not limit the lekage paths the leakage will sail through anything on its way to the 0v reference point. Leakage does´nt care if the path is long and complicated. It just want to reach its destination. Personally I think it is wise to try it and evaluate it for each device. If it sounds good, keep it. If it sounds bad, remove it. But you´ll have to re-evaluate it everytime you add another DC- output grounding to the chain. You´ll never know where that leakage goes if you add another path for it to sail through. What you want to avoid at all cost is that the leakage goes into the USB chain into the DAC. You cannot stop the leakage. You can only limit its choises by narrowing down the possible leakage paths, hopefully ones chosen by yourself. 

 

For me the DC- output grounding adventure stopped with the floating SMPS powering the Aqvox switch-8. All other devices (ISO Regen, Brooklyn DAC & Aries Mini) with the DC- output grounding turned out worse. I do NOT think that everyone else will have to come to the same conclutions. It depends on their 0v reference and all possible leakage paths in their perticular system. Unfortunately it can be quite a few.

 

In my tests with LT3045 voltage regulators in the DC cable path it seems to complicate things as well. It´s like they do block the leakage to some degree. I had a Y-split cable powering my ISO Regen and Aqvox switch ffrom the same floating SMPS. When I added a voltage regulator to the Aqvox switch leg the SQ got much worse. When I use voltage regulators at both legs the SQ improved, but still did´nt sound right. When I powered the Aqvox on a separate floating SMPS and ISO Regen on a separate floating SMPS it amost stayed the same. When I finally removed the DC- output grounding to the floating SMPS powering the ISO Regen the SQ jumped sky high. That tells me both that the voltage regulators do block leakage to a certain degree and the ground wires is a leakage path into other devices.

[ambre]

Dear John Swenson,

 

Many thanks for the trick to ground my Netgear GS105(V5)  and also my Meanwell SMPS and Sbooster Ultra which both feeds the Uptone Regen(amber) to my Team UD 501. Upsampling  my music in DSD 128 by means of the superior HQ-player and to NAA Mac Mini and sometimes a Raspberry 3.

 

Much better engaging sound in bass, voices and cymbals. An absolute must. So easy to do yourself.

 

 

Thank you very much

[Frojo]

6 hours ago, R1200CL said:

 

Further if I understood @JohnSwenson findings in another post, ground the Meanwell internally vs grounding the same Meanwell “externally”, made a difference. Was my understanding correct, and if so, can any advice at the present be given on which method to me used on the Meanwell ?

 

 

I modified a meanwell to internalise the 'grounding' wire, it did not operate as well as my external ground wired meanwell. I concluded (without any technical knowledge) that the internal wire was picking up other artefacts which detracted from the grounding effect. 

[MikeyFresh]

Last night I grounded two iPower units used in my bedroom/desktop system.

 

Dramatic improvement, this technique has improved the sound everywhere I've tried it so far, in all 3 of my systems.

[One and a half]

A quick measurement today revealed that if the host PC is grounded, the frame of an iMicroUSB3.0 is also grounded and the USB shell is as well input and output. 

This of course doesn't mean that the 0V is grounded, and it isn't unless 'forcefully' connected at the power source, but at least the shield is connected to ground/earth.

[Cornan]

14 minutes ago, One and a half said:

A quick measurement today revealed that if the host PC is grounded, the frame of an iMicroUSB3.0 is also grounded and the USB shell is as well input and output. 

This of course doesn't mean that the 0V is grounded, and it isn't unless 'forcefully' connected at the power source, but at least the shield is connected to ground/earth.

 

This means that you use shielded USB cables right?

In my case I'm using JSSG (floating shields with drain wire connected to both shield ends) starquad on both USB cables and DC cables. The AC mains cables is unshielded starquads.

Different setups probably gives different end results.

[sandyk]

13 minutes ago, Cornan said:

This means that you use shielded USB cables right?

 

 

I have found that not all longer USB cables even have a shield, possibly to reduce capacitance so that they will even work at the longer length ?

 

Incidentally, USB connected Mouse and keyboard cables can also radiate quite a bit of RF/EMI that may affect DTV reception where an indoor aerial is not too far from the computer.

I have some clamp on 5mm inside diameter ferrite cores on order to try and alleviate this problem.

[Cornan]

22 minutes ago, sandyk said:

 

 

I have found that not all longer USB cables even have a shield, possibly to reduce capacitance so that they will even work at the longer length ?

 

Incidentally, USB connected Mouse and keyboard cables can also radiate quite a bit of RF/EMI that may affect DTV reception where an indoor aerial is not too far from the computer.

I have some clamp on 5mm inside diameter ferrite cores on order to try and alleviate this problem.

 

Like with ethernet cables USB cables clearly sounds best with a floating shield a la JSSG IMO. The problem occurs if the shield is connected to the plugs. 

I prefer using my cables without ferrite cores since they could dampen the sound. I choose to use JSSG, grounding, starquads and cable distance instead.

[sandyk]

Just now, Cornan said:

Like with ethernet cables USB cables clearly sounds best with a floating shield a la JSSG IMO. The problem occurs if the shield is connected to the plugs. 

 

 That would depend on the computer itself, with a possible benefit, as in my case, if the shield is connected at the PC end ONLY  with a Desktop computer which is earthed, as mine is.

I would only use ferrites on items such as Mouse and Keyboard leads.

[barrows]

12 minutes ago, Cornan said:

 

Like with ethernet cables USB cables clearly sounds best with a floating shield a la JSSG IMO. The problem occurs if the shield is connected to the plugs. 

I prefer using my cables without ferrite cores since they could dampen the sound. I choose to use JSSG, grounding, starquads and cable distance instead.

As with most things, this will be system dependent.  Where the USB shell is terminated, and where the USB ground is terminated will matter, as I would expect so will whether or not USB in the receiver is isolated.

[Cornan]

15 minutes ago, barrows said:

As with most things, this will be system dependent.  Where the USB shell is terminated, and where the USB ground is terminated will matter, as I would expect so will whether or not USB in the receiver is isolated.

 

I totally agree! It is both system dependent and could vary based on sound preferences as well. All I know is that if you want to go down the unshielded path you'll need to go all in or not at all. Everything else will not show you the full picture.

[JohnSwenson]

I have been looking into low frequency emissions from cables, both Ethernet and USB. The high frequency DATA from these cables is fairly well attenuated due to the twisted pair construction because the data is differential. Leakage is different because it is common mode (the same on both data wires). Because of this the twisted pair doesn't attenuate anything. UTP (un-shielded) radiates leakage like crazy.

 

A traditional shielded (STP) ethernet cable with both ends connected to ground doesn't help because the shield is ALSO part of the antenna (leakage current going through the shield). A traditional USB cable will also have the shield acting as a very good antenna for leakage. (maybe this is why some people like un-shielded USB cables?) A USB data pair may NOT be perfectly differential so the twisted pair may have some attenuation of radiated leakage. So a shieldless cable may be a benefit if there is leakage on the cable.

 

For both of these it seems like the JSSG technique (Shield not connected at either end except for a wire connecting both ends of the shield) would be of benefit. This would seem to provide attenuation of leakage radiation WITHOUT the shield carrying the leakage current and being part of the antenna.

 

Of course this is immaterial IF there is no leakage current going through the cable in the first place. So IF you can keep leakage out of the cables, a lot of these cable weirdnesses go out the window.

 

It would be interesting to see if getting rid of leakage in these cables means less cable sensitivities. 

 

John S.

[jamesg11]

Righto then -

macmini has hd drive connected to it by usb, both lpsu powered ...

router is ethernet into macmini but, even with no netgear switch in there, low freq emissions not an issue on ethernet ...?

Best to cover bases by getting a jssg usb cable.

 

[sandyk]

James

 Did you sort out your screw terminal type plug problem ?

 I noticed in a more recent post that a larger photo of the screw terminal plug showed + and - markings.

These markings were not visible with the small Jaycar photos.

 

Alex

[One and a half]

For those of you have the Groundhog, a similar principle applies to the 0V connection and the frame of a microUSB 3.0. When the switch is down, the resistance from the frame to 0V is 56 Ohms, and the other way is 28 Ohms, not quite 0 Ohms. Also IIRC, the regen had a smaller resistance of 3 Ohm from a 0V, caused a colour change in versions. Seems that there's a deliberate impedance quite different to AC power technology where the neutral and ground is hard bonded with 0 ohms on the secondary of transformers. If the noise needs to travel a certain path, this is one to achieve with a higher than 0 but less than 60 Ohms...

 

[One and a half]

9 hours ago, Cornan said:

 

This means that you use shielded USB cables right?

In my case I'm using JSSG (floating shields with drain wire connected to both shield ends) starquad on both USB cables and DC cables. The AC mains cables is unshielded starquads.

Different setups probably gives different end results.

Yes, I always use shielded cables everywhere, AC, signal, the lot. The shields channel the noise where it's under control, rather than out of control. Ideally, but will never happen in domestic audio equipment, the shield should be bolted to the chassis of the source and the receiver as in the photo. This would work for XLR, forget it for RCAs. Then install a 6mm or 10AWG +  bonding earth wire between the source and receiver. Like you do in countless badly noisy industrial environments for the last 20-30 years for digital signals.

 

 

[jamesg11]

3 hours ago, sandyk said:

James

 Did you sort out your screw terminal type plug problem ?

 I noticed in a more recent post that a larger photo of the screw terminal plug showed + and - markings.

These markings were not visible with the small Jaycar photos.

 

Alex

Yes, stumbled on the correct plug in jaycar, smps in hand ...

So, now no smps to bother about. Just an lpsu guy here.

 

Not that there aren’t things to bother about - a tweaker friend, who I tuned into the js grounding idea & who ordered some of the plugs etc for such, was told by an electrical engineer that the tweak was ‘illegal’ in Oz.