SMPS and grounding

[JohnSwenson]

Over the last month I have been performing extensive tests on leakage currents coming from SMPS power supplies. One of the most important results from this investigations is the makeup of this leakage current.

 

Previously my understanding was that leakage current was mainly a low frequency phenoninum, 60Hz, and its harmonics (60, 120, 180, 240 etc). My early investigations seemed to show that this was true with some high frequency components from the switching operation, but that this was fairly small in comparison to the whole. This turned out to be not true. It turns out that SMPS leakage is very hard to measure, it consists of some VERY high impedance components, on the order of 300 Mega ohms and some much lower impedance components.

 

None of the standard electronics test equipment will properly measure this, there impedance is simply too low and drastically changes the signal while trying to measure it. My only option was to build my own ultra high impedance differential probe (around 10 Giga ohms) so I can measure leakage current directly (rather than its affect on other equipment).

 

Because of this ultra high impedance it turns out leakage current can go all kinds of places you don't think about. It turns out to be extremely difficult to block using normal techniques, the blocking device has to have over a giga ohm to significantly attenuate it, this is VERY tough to achieve.

 

It turns out the best way to deal with this high impedance part is to shunt it around the audio equipment, there is a very easy way to do this, ground the negative output of the SMPS. It seems that even SMPS that include a ground pin don't actually use it for anything. This sounds too simple, but it actually works. PLEASE do not under any circumstances  attempt to modify an SMPS to do this, this is DEADLY, DO NOT attempt!!!

 

Fortunately there is a real easy - inexpensive way to do this, it doesn't take any complex knowledge, I'm calling this the power supply grounding adapter, here are a couple pictures:

 

 

This consists of three items and some wire. The yellow item is a three pin AC power plug (shown is the US model), the black parts are male and female DC barrel connectors with screw connectors, no soldering necessary! Amazon has a whole bunch of places selling these for CCTV cameras. The spec is 5.5mm outside and 2.1mm inside. Most you will see will be this spec. The + of the male is wired to the + of the female and - on the male goes to - of the female. Look at the close up picture, in order for the wire to go straight across one of the connectors has to be upside down. This is extremely important to get right. You can use many different types of wire for this, I used solid core 14AWG wire with the insulation stripped off.

 

A green wire goes from the ground pin of the AC plug to the - pin of one of the connectors, strip a little off and just stick it in with the wire going between the two connectors, use a screwdriver to tighten the screws, you are done.

 

The green wire does NOT have to be a heavy duty high power wire. The SMPS are already designed to be double insulated so the AC cannot show up on the DC output, grounding it does not pose any electrical threat. I just used some nice silicon rubber wire I bought on Amazon, but really any green wire will do.

 

You plug the output of the SMPS into the "jack" side, and the "plug" side goes into where you would have plugged the SMPS. The AC plug goes into the SAME power strip or duplex outlet where you plugged in the SMPS. This is very important to properly shunt the high impedance part of the leakage current. If it goes into a different circuit you can wind up making it worse.

 

You should do this on every SMPS that is somehow electrically connected into your sound system. Some items may have different connectors such and Ethernet switches etc. There are a bunch of different adapters available that can convert the 5.5/2.1 to just about anything out there, you may need to use a pair of these.

 

You may ask "how effective IS this?"

 

Here are some graphs, the first is the leakage current of an SMPS, the second is with this adapter plugged in:

 

 

 

That is a significant reduction in leakage for a $10 shunt adapter.

 

If you want to get it all you will have to use something like the LPS-1 which will get rid of it all.

 

Have fun,

 

John S.

 

[rickca]

52 minutes ago, JohnSwenson said:

Fortunately there is a real easy - inexpensive way to do this, it doesn't take any complex knowledge, I'm calling this the power supply grounding adapter

This is awesome, John!  It's going to be interesting to hear the impact of using this grounding adapter on things like routers/switches vs using a linear power supply.  Could I even do this with my powered studio monitors?

[Cornan]

2 hours ago, JohnSwenson said:

Over the last month I have been performing extensive tests on leakage currents coming from SMPS power supplies. One of the most important results from this investigations is the makeup of this leakage current.

 

Previously my understanding was that leakage current was mainly a low frequency phenoninum, 60Hz, and its harmonics (60, 120, 180, 240 etc). My early investigations seemed to show that this was true with some high frequency components from the switching operation, but that this was fairly small in comparison to the whole. This turned out to be not true. It turns out that SMPS leakage is very hard to measure, it consists of some VERY high impedance components, on the order of 300 Mega ohms and some much lower impedance components.

 

None of the standard electronics test equipment will properly measure this, there impedance is simply too low and drastically changes the signal while trying to measure it. My only option was to build my own ultra high impedance differential probe (around 10 Giga ohms) so I can measure leakage current directly (rather than its affect on other equipment).

 

Because of this ultra high impedance it turns out leakage current can go all kinds of places you don't think about. It turns out to be extremely difficult to block using normal techniques, the blocking device has to have over a giga ohm to significantly attenuate it, this is VERY tough to achieve.

 

It turns out the best way to deal with this high impedance part is to shunt it around the audio equipment, there is a very easy way to do this, ground the negative output of the SMPS. It seems that even SMPS that include a ground pin don't actually use it for anything. This sounds too simple, but it actually works. PLEASE do not under any circumstances  attempt to modify an SMPS to do this, this is DEADLY, to NOT attempt!!!

 

Fortunately there is a real easy - inexpensive way to do this, it doesn't take any complex knowledge, I'm calling this the power supply grounding adapter, here are a couple pictures:

 

 

This consists of three items and some wire. The yellow item is a three pin AC power plug (shown is the US model), the black parts are male and female DC barrel connectors with screw connectors, no soldering necessary! Amazon has a whole bunch of places selling these for CCTV cameras. The spec is 5.5mm outside and 2.1mm inside. Most you will see will be this spec. The + of the male is wired to the + of the female and - on the male goes to - of the female. Look at the close up picture, in order for the wire to go straight across one of the connectors has to be upside down. This is extremely important to get right. You can use many different types of wire for this, I used solid core 14AWG wire with the insulation stripped off.

 

A green wire goes from the ground pin of the AC plug to the - pin of one of the connectors, strip a little off and just stick it in with the wire going between the two connectors, use a screwdriver to tighten the screws, you are done.

 

The green wire does NOT have to be a heavy duty high power wire. The SMPS are already designed to be double insulated so the AC cannot show up on the DC output, grounding it does not pose any electrical threat. I just used some nice silicon rubber wire I bought on Amazon, but really any green wire will do.

 

You plug the output of the SMPS into the "jack" side, and the "plug" side goes into where you would have plugged the SMPS. The AC plug goes into the SAME power strip or duplex outlet where you plugged in the SMPS. This is very important to properly shunt the high impedance part of the leakage current. If it goes into a different circuit you can wind up making it worse.

 

You should do this on every SMPS that is somehow electrically connected into your sound system. Some items may have different connectors such and Ethernet switches etc. There are a bunch of different adapters available that can convert the 5.5/2.1 to just about anything out there, you may need to use a pair of these.

 

You may ask "how effective IS this?"

 

Here are some graphs, the first is the leakage current of an SMPS, the second is with this adapter plugged in:

 

 

 

That is a significant reduction in leakage for a $10 shunt adapter.

 

If you want to get it all you will have to use something like the LPS-1 which will get rid of it all.

 

Have fun,

 

John S.

 

 

Great write-up John! ?

How about when you are using a floating SMPS, where the safety ground is not connected to the negative output or chassis? I pretty much use these everywhere with truly great results. Especially effective powering network devices.

[JohnSwenson]

1 hour ago, Cornan said:

 

Great write-up John! ?

How about when you are using a floating SMPS, where the safety ground is not connected to the negative output or chassis? I pretty much use these everywhere with truly great results. Especially effective powering network devices.

It turns out the floating ones are the worst, the leakage current has nowhere to go except what it is powering. None of the SMPS I have connect the DC output to safety ground so all these tests were done with floating SMPS (I actually have over a hundred tests), grounding really does dramatically reduce leakage from these devices into what they power.

 

John S.

[JohnSwenson]

3 hours ago, rickca said:

This is awesome, John!  It's going to be interesting to hear the impact of using this grounding adapter on things like routers/switches vs using a linear power supply.  Could I even do this with my powered studio monitors?

Every test I have done shows that grounding the DC output from a SMPS significantly reduces leakage current.

 

I have been doing other tests on networks and leakage, I will be making a post on that in a few days (probably in the thread on the Sonore forum).

 

John S.

[Cornan]

30 minutes ago, JohnSwenson said:

It turns out the floating ones are the worst, the leakage current has nowhere to go except what it is powering. None of the SMPS I have connect the DC output to safety ground so all these tests were done with floating SMPS (I actually have over a hundred tests), grounding really does dramatically reduce leakage from these devices into what they power.

 

John S.

 

Very interesting. What I found interesting is that in my setup where I use a floating SMPS to power the ISO Regen, Aqvox switch and Aries Mini the ISO Regen did´nt improve much. In fact it descreased the SQ, unless I had the ISO Regens galvanic isolation off. When I tried a Regen Amber pre the ISO Regen the difference was not subtile at all, and it actually improved SQ. I concluded that this had something to do about the Regen Ambers voltage regulator, so I have ordered a bunch of LT3045 voltage regulators (2 regulators per device) to use pre ISO Regen and Aqvox switch. 

For notes. When I had the ISO Regen with ISO off and 5v injection it sounded great as well. When I tried to use the Regen Amber as a voltage regulator for the Aqvox switch it decrease SQ, which I concluded that the leakage went right throught the ISO Regen into the Brooklyn DAC. This told me that there indeed was a internal leakage loop going on (since everything is connected to a IT with grounded secondary and the SMPS is floating).

 

However, I have a star-earthed (and starquad wired) power distributor on order that will have a separate ground pin. What you are telling me now is that I can connect a ground wire this ground pin directly to the negative DC output (screwable banana sockets) of the floating SMPS to reduce this problem?

 

Why do you think Regen Ambers voltage regulator improves SQ pre the ISO Regen in my setup with floating SMPSs?

 

 

 

[Cornan]

@JohnSwenson

Still I wonder? The floating SMPSs have improved SQ vastly in every single spot that I have used them. They are perticularly effective on ethernet connected devices such as routers, network switches, streamers etc. Plugged into a isolation transformer with floating secondary they have even improved upon unregulated/regulated Li-ion battery packs (charger disconnected from ac mains). Still you are telling me that floating SMPS is worse than a normal SMPS? I find that extremely hard to beleive (not to say impossible to beleive in), unless there are other explanations why floating SMPSs have that great impact on SQ that is much more prominent than the leakage?

[sandyk]

 

Cornan

Silicon Chip magazine published an article on this in July 2006, complete with measurements from different DVD players, showing the HF rubbish on the AC residual from the SMPS, which could measure as high as half the A.C. Mains voltage.. They used a 3 pin mains plug with only the earth wire connected, and a capacitor to the equipment's metal case.

 Unfortunately, a HDD crash saw me lose my copy of the original article. However several members were sent copies of this and another connected article. several years ago

Attached is a link to a partial copy of the "Tingle Terminator" article.

Regards

Alex

Attached is a link to a copy of one of the pages that I somehow managed to retain.

https://imageshack.com/i/po722bNDj

 

http://archive.siliconchip.com.au/cms/A_107106/article.html

[Cornan]

30 minutes ago, sandyk said:

 

Cornan

Silicon Chip magazine published an article on this in July 2006, complete with measurements from different DVD players, showing the HF rubbish on the AC residual from the SMPS, which could measure as high as half the A.C. Mains voltage.. They used a 3 pin mains plug with only the earth wire connected, and a capacitor to the equipment's metal case.

 Unfortunately, a HDD crash saw me lose my copy of the original article. However several members were sent copies of this and another connected article. several years ago

Attached is a link to a partial copy of the "Tingle Terminator" article.

Regards

Alex

Attached is a link to a copy of one of the pages that I somehow managed to retain.

https://imageshack.com/i/po722bNDj

 

http://archive.siliconchip.com.au/cms/A_107106/article.html

 

Thanks for the link Alex! However, I am not sure that this is totally tranferable to a floating SMPS that have both the negative DC output and chassi totally isolated from the safety ground while (quote) "they used a 3 pin mains plug with only the earth wire connected, and a capacitor to the equipment's metal case".

[sandyk]

Just now, Cornan said:

 

Thanks for the link Alex! However, I am not sure that this is totally tranferable to a floating SMPS that have both the negative DC output and chassis totally isolated from the safety ground while (quote) "they used a 3 pin mains plug with only the earth wire connected, and a capacitor to the equipment's metal case".

 

 

If you were able to read the original articles and see the measurements, you would see that it is.

Most CD/DVD/BR players use SMPS, and are double insulated with no connection to mains earth.(floating)

[Cornan]

10 minutes ago, sandyk said:

 

 

If you were able to read the original articles and see the measurements, you would see that it is.

Most CD/DVD/BR players use SMPS, and are double insulated with no connection to mains earth.(floating)

 

I am going to ground the DC terminal strip in my B setup tonight. This DC terminal have a GND screw that is connected to the DC negative side of the terminal and it is powered by a single Gophert csp-3205II floating PSU. All devices 5v devices (Chromecast Audio/TP-Link UE-300/OTG power split adapter/TP-Link DSG-105 and Oehlbach DAC 40 edition) are connected to this DC terminal and the active speakers are connected to a Ultra IT on a separate ac mains line (but ending up in the same ac wall outlet). I will ground this DC terminal strip to the safety ground receptacle of the Hugo powerstrip where the Gophert is plugged in to see how that turns out.

 

I have actually tried this before with great results, but since I also had a grounding box connected to the same safety ground receptacle at the time I probably made the wrong conclutions that it was the groudning box that made that improvement and not the safety ground connection to the DC terminals GND screw.

 

Anyway I´ll be back with my impressions asap!

 

 

[sandyk]

The schematic at the link shows the mandatory culprits responsible for most of the problems.

 

http://imageshack.com/a/img923/6873/7NUrjX.jpg

 

 

[octaviars]

4 hours ago, JohnSwenson said:

I have been doing other tests on networks and leakage, I will be making a post on that in a few days (probably in the thread on the Sonore forum).

 

Really looking forward to what you have found regarding networks and leakage? Do you by any chance also involved fiber optics in the test to see if that will solve all problems or will there just be another problem with the FMC at the end? I am about to go down a route of new network equipment and endpoint so the input that you share here will be usefull for me.

 

Another question have you done any measurments on linear PSU regarding the leakage when the DC minus is connected to earth?

[Cornan]

10 minutes ago, sandyk said:

The schematic at the link shows the mandatory culprits responsible for most of the problems.

 

http://imageshack.com/a/img923/6873/7NUrjX.jpg

 

 

 

Thanks Alex! Here is how the Gophert csp-3205II looks like on the inside (borrowed picture):

 

 

I am no good with schematics. Are we talking about the same thing?

 

 

 

[sandyk]

Cornan

 I just located a backup copy of the 2 original articles .

I will PM you the DL link. If J.S. is also interested , perhaps he can PM me for the link to the articles.

Alex

 

 

[sandyk]

15 minutes ago, Cornan said:

 

Thanks Alex! Here is how the Gophert csp-3205II looks like on the inside (borrowed picture):

 

 

I am no good with schematics. Are we talking about the same thing?

 

 

 

 In the schematic, you will see 2 capacitors highlighted in yellow.

They connect the mains side of the SMPS to the secondary side.

IIRC, they are mainly there to meet RF/EMI regulations.

 

Please check your PMs for a DL link to a copy of the 2 articles mentioned.

Do not repost the link, as it is copyright material.

[Cornan]

3 minutes ago, sandyk said:

 In the schematic, you will see 2 capacitors highlighted in yellow.

They connect the mains side of the SMPS to the secondary side.

IIRC, they are mainly there to meet RF/EMI regulations.

 

Please check your PMs for a DL link to a copy of the 2 articles mentioned.

Do not repost the link, as it is copyright material.

 

Thanks Alex! I will check it out asap and ofcourse not repost the link or the paper.

 

Micael

[Cornan]

Here is my first impression grounding the DC output of my Gophert csp-3205II! ?

 

A small step towards better sq, but not at all a night and day improvement.

[Cornan]

5 hours ago, sandyk said:

 In the schematic, you will see 2 capacitors highlighted in yellow.

They connect the mains side of the SMPS to the secondary side.

IIRC, they are mainly there to meet RF/EMI regulations.

 

Please check your PMs for a DL link to a copy of the 2 articles mentioned.

Do not repost the link, as it is copyright material.

 

Thanks for the info Alex! I have quickly read it through. Even if the "Tingle Terminator" is a fairly similar idea it is connected to the case of a CD player (which is grounded) and almost used as a shield for the AC mains power cable and a drain for shield currents. Since the Gopherts safety ground is not passed through to the case of the Gophert I am not sure it will work. Does this stop me trying? Ofcourse not! ? I will try this as well asap. Do you know if the capacitor is absolutely neccessary? Can I just run an additional ground wire from the chassi to the safety ground pin, maybe twisted with the DC output ground wire?

[octaviars]

1 hour ago, Cornan said:

Since the Gopherts safety ground is not passed through to the case of the Gophert I am not sure it will work.

 

Are you really shure about that? If the case is not grounded then it should have a dubble isolated symbol. 

 

If you look at the picture of a Ghopert you see a ground connection screw next to the incoming ground wire from the IEC inlet. Have you measured between the case and the ground pin in the IEC to see if there is any connection? 

 

 

[Cornan]

12 minutes ago, octaviars said:

 

Are you really shure about that? If the case is not grounded then it should have a dubble isolated symbol. 

 

If you look at the picture of a Ghopert you see a ground connection screw next to the incoming ground wire from the IEC inlet. Have you measured between the case and the ground pin in the IEC to see if there is any connection? 

 

 

 

Yes, I am 100% sure. I have both got it confirmed from the manufactorer and confirmed it myself using a hand-held battery powered multimeter between safety ground pin as DC output and chassi.

[octaviars]

9 minutes ago, Cornan said:

 

Yes, I am 100% sure. I have both got it confirmed from the manufactorer and confirmed it myself using a hand-held battery powered multimeter between safety ground pin as DC output and chassi.

 

Ok good to know that. Do you know what the chassi is connected to on the PCB? 

 

[Cornan]

6 minutes ago, octaviars said:

 

Ok good to know that. Do you know what the chassi is connected to on the PCB? 

 

 

It is the IEC fuse (soldered) as far as I can see.

[octaviars]

1 minute ago, Cornan said:

It is the IEC fuse (soldered) as far as I can see.

 

No I mean the screw next to the green and yellow wire that is soldered to the PCB? If you look at other photos of Ghopherts there is a metal flange going up from the PCB which is attatched to the screw on the chassi. To me this did look like the chassi was grounded but if you have measured it I belive you so hence my question what the purpouse of connecting the chassi to the PCB if it is not for safety grounding?

[Cornan]

12 minutes ago, octaviars said:

 

No I mean the screw next to the green and yellow wire that is soldered to the PCB? If you look at other photos of Ghopherts there is a metal flange going up from the PCB which is attatched to the screw on the chassi. To me this did look like the chassi was grounded but if you have measured it I belive you so hence my question what the purpouse of connecting the chassi to the PCB if it is not for safety grounding?

 

The green/yellow cable is a ground wire ofcourse. It would be really dangerous without the IEC inlet grounded. However, it is not passed by to the chassi or to the DC negative output. It is isolated. You can clearly see the isolation border on the picture I posted.