• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Mains Leakage Currents

Joined
Apr 3, 2018
Messages
55
Likes
11
#1
I was not familiar with the term Mains leakage currents. Looking it looks just like AC input artifacts and the multiple harmonics of the power line frequency.
When I saw tests run on switching supplies these are very evident. Yet the measured noise levels are sometimes unbelievable low level. The 60, 120, 180 Hz noise bands should push the noise levels way beyond the micro volts levels reported.
What am I missing here?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
16,770
Likes
12,970
Location
Seattle Area
#2
The mains leakage is due to emissions control Y capacitors. The values are small so the impedance is pretty high at low frequencies. Hence the reason the measured levels are seemingly low. Is this what you are asking?
 

solderdude

Addicted to Fun and Learning
Joined
Jul 21, 2018
Messages
761
Likes
1,176
Location
The Neverlands
#3
I was not familiar with the term Mains leakage currents. Looking it looks just like AC input artifacts and the multiple harmonics of the power line frequency.
When I saw tests run on switching supplies these are very evident. Yet the measured noise levels are sometimes unbelievable low level. The 60, 120, 180 Hz noise bands should push the noise levels way beyond the micro volts levels reported.
What am I missing here?
The leakage currents can be caused by capacitive coupling between transformer windings (primary and secondary).
They can be caused by capacitors in mains filters.
It also depends whether or not the cable is connected to safety earth. Things may go wrong when a filtered device intended to be connected to PE mains sockets is used in a mains socket without PE.
Currents can also be coming from SMPS with a small capacitance between the secondary DC side and the mains side and depends on L and N and/or PE being connected.
The currents can run from a device through a human body (hum comes or lowers/disappears) or via a second device connected via audio cables or digital cables.
It also depends on make and model. Not all devices are created equal !

The problem in analog audio is usually limited to mains frequencies, its harmonics (can go up to quite a few and call this 'hum') and usually is more irritating with higher harmonics.
The mains is swamped with HF switching noises. They may be small in amplitude but could be enough to cause some problems in some equipment under certain circumstances.
Noise that passes through virtually unattenuated is HF/RF noise generated by numerous devices as the frequencies are high they pass almost unattenuated through small capacitances.
This could be more of a problem for certain digital devices than for analog ones.
PCB design, wiring and wires and number of connections between boxes can also give more or less 'potential problems'.

There are simply a lot of factors in play.

The good thing is... When (properly designed) equipment is connected properly with the right type of cables there should not be any audible or other problems.
 
Last edited:
Joined
Apr 3, 2018
Messages
55
Likes
11
#4
Maybe you have answered my question. It really is why a devise such as the iFi iPower has large leakage “currents” when measured. But the measured noise level is micro volts.
I do not understand how the mains leakage does not increase the level of measured noise on the dc output.
I purchased a Gophert to drive a Jensen 4-pole capacitor filter. I think it was oscillating. Gave me a headache after an hour.
It was good to see how much current the dac is consuming. Never saw over 1.25 amps. So a 1.8 amp supply like an iPower 12 volt should work fine. Same as with my Jay’s LPS.
 

solderdude

Addicted to Fun and Learning
Joined
Jul 21, 2018
Messages
761
Likes
1,176
Location
The Neverlands
#5
Noise on the mains can be both common mode and differential mode.

In differential mode the voltage/current is measured between Live and Neutral and for (secondary) low voltages between + and - (5V, 12V or whatever voltage)

All differential mode garbage on the mains will also be present at the output of a mains transformer but lowered with the same factor.
Most mains transformers start to drop off somewhere between 2kHz and 30kHz (depends on make/type/model etc).
When the AC voltage is rectified and perhaps regulated all these differential mode voltages will be gone.

For countering differential mode garbage RC/LC filters help as well as regulators (up to a certain frequency). A LOT of differential noise is removed by rectifiers and buffer capacitors. Bigger capacitors (values) can remove lower frequencies. Smaller (de-coupling) capacitors (often in parallel) can remove HF noise. Depending on their value these are most effective in a certain frequency range. This is why sometimes you see 'compound' decoupling capacitors. 2 or 3 small caps in parallel say 10uF ceramic + 100nF + 1nF for instance.

Common mode voltages/currents are measured/present between the (PE/ground/safety earth) and either voltage rail.
It cannot be measured across the mains wires nor across the output terminals.
When there is a common mode voltage the voltage is thus 0V between L and N for instance but is present between L(and N) and (PE/ground/earth)

Common mode currents can be lowered with ferrites. These work optimally when you loop the wire about 3 to 4 times through a ferrite ring.
Another effective part is a common mode choke which has 2 windings in opposite directions where L and N (or + and -) run through.
Could be combined with capacitors, ferrites or multiple chokes + caps and form filters that have a certain attenuation in a specific freq. band.


When you measure the output voltage of a DC power supply you might see a very clean DC with little to no noise on it.
When you measure the + (or -) to (PE/ground/safety earth) you will probably see a LOT more noise.
The circuit it feeds does not really care about these common mode noises as the total power supply varies along with the common mode noise.

This can become an issue when the device it feeds is connected to another device (say a PC to a DAC to an amp) and all of these devices have different 'leakage' to mains / ground.
In this case the different currents also can (depends on several factors) flow between the ground/shield/common of the interlink cables.
Currents introduce voltages and these can be amplified.
 
Last edited:
Joined
Apr 3, 2018
Messages
55
Likes
11
#7
That is a very good article. I need to work my way through it a third time.
Build a new amp a month ago. It is “floating” power supply design. Uses an input transformer to convert balanced to single ended. Had to ground the input jacks to chassis to eliminate a very low level hum. Using 106 dB efficient speakers, real hum magnets.
That tied the analog output ground from dac to chassis ground on amp. Very complicated grounding with amp, optical Thunderbolt connect, and various power supplies tried with Dac.
Have tried switching (iPower) and LPS (Jay’s) with great results. Tried a big Golphert with filter that had some artifacts.
I need to let my ears decide whether a switcher or LPS is best sonically. The clarity was better with switcher. LPS seems fuller. No Clear Winner.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
16,770
Likes
12,970
Location
Seattle Area
#8
Maybe you have answered my question. It really is why a devise such as the iFi iPower has large leakage “currents” when measured. But the measured noise level is micro volts.
That's because in their testing, they use an isolation transformer and float the power supply! In other words, they work hard to remove mains leakage from the equation.
 
Joined
Apr 3, 2018
Messages
55
Likes
11
#9
This helps explain the SOTA test results with iFi iPower supplies and multiple reports of leakage currents much higher than the “gimme” switching supplies provided with equipment.
I have an ancient “Noise Trapper+ “ in the attic. I think it is a 250 watt isolation transformer with some crude filtering on several outputs.
No sign of mains disturbance with just iPower now. Will drag the Noise Trapper down and see if it makes audible any difference.
 

Superdad

Active Member
Joined
Aug 6, 2017
Messages
100
Likes
48
#10
That's because in their testing, they use an isolation transformer and float the power supply! In other words, they work hard to remove mains leakage from the equation.
That is incorrect. Plugging an SMPS into an isolation transformer does absolutely nothing to eliminate the transfer of AC leakage to the DC -ve (zero-volt "ground") of its output.
John Swenson ran a whole series of direct leakage current measurements (using a differential probe and a proper impedance loading circuit) on a dozen different SMPS units. All of his tests were with the units plugged into a 2.5kVA Topaz Ultraisolation transformer (with 0.005pF interwinding capacitance).

Here are direct measurements of the iFi iPower 9V/2.0A unit at three different bandwidths (1kHz, 50kHz, and 1MHz):
lkdetect_iPwr_1khz_30k.gif

lkdetect_iPwr_50khz_30k.gif

lkdetect_iPwr_1Mhz_30k.gif


For comparison, here is the AC leakage from a Mean Well GST40A07 (7.5V/5.34A) run with the same setup (1kHz bandwidth):
MW40_1khz_0929.gif


As we have explained numerous times elsewhere, the pernicious high-source-impedance form of leakage (what all the above are measuring, and which yes, Amir's early test results with our original UltraCap LPS-1 into single-ended DACs prompted us to research this lesser-known form) is easy to get rid of--simply by shunting the DC -ve ("ground") of the SMPS output to AC mains ground. So here is the same Mean Well SMPS with that done:
MW40_1khz_internalgnd_0929.gif


And with the same test set and the same "ground-shunted" Mean Well used as the AC>DC charging supply, here is the 1kHz leakage from the output of our original UltraCap LPS-1:
MW40_1khz_internalgnd_lps1_0929.gif


By the way, the new generation UltraCap LPS-1.2, with uniquely paralleled LT3045 regs and wider 5/7/9/12V output selections, comes with an UpTone-branded 36W SMPS charger that is already internally "ground--shunted," so high-source-impedance leakage never enters our piece. Mean Well units are not designed that way and they failed to understand/comply our request, even for 1,000 units, so I found another reputable manufacturer whose 3-wire units are already designed that way.
(SMPS manufacturers only understand the traditional low-impedance "touch current" leakage--which has ALWAYS been 100% blocked by our UltraCap units, but the few tens of picofarads of total capacitance of the transistors we use for bank-alternating--instead of mechanically noisy large relays--is the path for the high-source-impedance form to pass through. As shown above, prevention of it ever entering is an easy thing to do.) And we offer our new internally ground-shunted chargers for $15 to original LPS-1 owners. Not that anyone is able to hear that difference, but it is technically better.
 

Superdad

Active Member
Joined
Aug 6, 2017
Messages
100
Likes
48
#11
This helps explain the SOTA test results with iFi iPower supplies and multiple reports of leakage currents much higher than the “gimme” switching supplies provided with equipment.
I think you are conflating two different things. iFi adds about 6 extra parts to their $4 wall-wart (I've spend a lot of time pricing such things so I assure you that even $4 is generous) to greatly reduce the DC output noise of the iPower SMPS. Standard SMPS units usually have about 80mV RMS noise, and while iFi gets that way down, do note that their noise measurement claims are only in the 20-20kHz bandwidth--fairly irrelevant to typical digital components which respond best to power supplies that have ultra-low-noise and high PSRR across an extremely wide bandwidth. For example, the Linear Technology LT3045 regs offer 0.8µV RMS noise from 10Hz-100kHz and 76dB PSRR at 1MHz.

But as explained above, the iFi iPower units do have the typical very high leakage of an SMPS (actually higher than most we tested). And while the posted graphs don't show it, John Swenson's realtime measures of the iPower units lead him to reward that their leakage pattern was the "most bizarre and asymmetric" he has seen. Perhaps that is a byproduct of the circuit elements they add to get the standard DC output noise down.


I have an ancient “Noise Trapper+ “ in the attic. I think it is a 250 watt isolation transformer with some crude filtering on several outputs.
No sign of mains disturbance with just iPower now. Will drag the Noise Trapper down and see if it makes audible any difference.
As explained, an isolation transformer will do NOTHING to reduce AC>DC leakage from an SMPS (or linear PS for that matter--those have leakage too) into your system. Remember, these are AC currents that travel over ALL the DC cable connections in your system. The "ground" connections of analog cables, digital cables, USB cables--even some Ethernet cables--are the pathway for this stuff. Don't believe me? Look back at many of Aimir's measurements and you will see it in the low frequency ranges of his graphs (though there are plenty of very high-frequency components too).
 

solderdude

Addicted to Fun and Learning
Joined
Jul 21, 2018
Messages
761
Likes
1,176
Location
The Neverlands
#12
I might add that with a lot of SMPS some HF 'noise' (generated by the high frequency oscillator in the SMPS that provides the high switching frequency) 'rides' along across interlinks, power lines, transformers and what not into circuits.
HF/RF (common mode) noises can thus pass through small capacitances (windings of transformers in fairly close proximity over a relative large area) like a hot knife through butter and can be AM modulated with the mains frequency. The power devices in the SMPS (that generate the actual power) are mostly directly fed from the rectified mains. That rectified mains voltage has a substantial 'ripple' on it as small capacitances are used as reservoir caps.
In such cases, where lots of transistors etc., are present in amplifiers/opamps, that HF noise is travelling via interlinks (the screens/return paths) can be 'rectified' (as in AM radio detector) and the hum products (the modulation of the HF/RF noise) can become audible again.
This way mains frequencies can still be 'conducted' via RF through small capacitances.

It is the task of the designers to keep the RF/HF signals below 'detectable' values so 'AM detection' can not happen.
Some designers cut corners (mostly due to profit margins).

With decent gear, connected 'correctly' there should not be any noise or hum audible.

If there is.. happy hunting. Sometimes it can be a real b***h to find out what the culprit(s) is/are.
sometimes running a strategically placed extra ground wire can help, sometimes filtering is needed or other cables (or the way they run).

Proper filtering near the noise source(s) and decent (screened) interlink cables as well as well layed-out PCBs and internal wiring can of the used equipment can go a long way.
Cheap Chinese (or other) gear often does not comply, nor can some very expensive boutique audio products.
 

Superdad

Active Member
Joined
Aug 6, 2017
Messages
100
Likes
48
#13
I might add that with a lot of SMPS some HF 'noise' (generated by the high frequency oscillator in the SMPS that provides the high switching frequency) 'rides' along across interlinks, power lines, transformers and what not into circuits.
...
Another spot-on Solderdude post! So refreshing to read what you write--often within a sea of others misunderstanding this stuff in a big way.
Thanks,
--Alex C.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
16,770
Likes
12,970
Location
Seattle Area
#14
That is incorrect. Plugging an SMPS into an isolation transformer does absolutely nothing to eliminate the transfer of AC leakage to the DC -ve (zero-volt "ground") of its output.
Mains leakage occurs due to differential in ground between equipment. Floating the grounds on each gear will absolutely change the amount of leakage you get. Here is ifi themselves in their CA forum response:

1544312318324.png


1544312404197.png
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
16,770
Likes
12,970
Location
Seattle Area
#15
Another spot-on Solderdude post! So refreshing to read what you write--often within a sea of others misunderstanding this stuff in a big way.
Thanks,
--Alex C.
That is a strange proclamation Alex seeing how both you and John were quite mistaken about this topic altogether as reflected in the design of his/your product. It took a lot of effort on my part to conclusively demonstrate that misunderstanding last year. Now you are standing tall saying others are confused with that sarcastic remark???
 

Superdad

Active Member
Joined
Aug 6, 2017
Messages
100
Likes
48
#16
Mains leakage occurs due to differential in ground between equipment. Floating the grounds on each gear will absolutely change the amount of leakage you get. Here is ifi themselves in their CA forum response:

View attachment 18581

View attachment 18582
Well they are talking about setting up for properly MEASURING, with steps that if you incorporated into YOUR bench setup would greatly improve the validity of your own tests. :cool:

Yet the sort of leakage current interaction they are talking about falls more into the category of “leakage loops,” whereas the measurements I presented above were of direct leakage currents from SMPS units (with the DUT plugged into an isolation transformer and a differential probe and resistor used to make sure that ALL we were measuring was the leakage from the supply and not an interaction with the test gear.)
 

jtwrace

Active Member
Joined
May 31, 2017
Messages
237
Likes
71
Location
Orlando, FL
#17
Well they are talking about setting up for properly MEASURING, with steps that if you incorporated into YOUR bench setup would greatly improve the validity of your own tests. :cool:
Says the guy who has NEVER published any real objective data for any of the millions of dollars of product you have sold. Come on man! Have you even ever thanked Amir and this site for the flaw in your product that he found?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
16,770
Likes
12,970
Location
Seattle Area
#18
Well they are talking about setting up for properly MEASURING, with steps that if you incorporated into YOUR bench setup would greatly improve the validity of your own tests. :cool:
Boy, you are back to your old self. Regardless, OP is asking about power supply measurements and why mains leakage may not appear in it. That is exactly what I explained. That they use unusual setup to avoid it which in reality does not exist in actual systems that have no isolation transformers.
 

Superdad

Active Member
Joined
Aug 6, 2017
Messages
100
Likes
48
#19
Now you are standing tall saying others are confused with that sarcastic remark???
No sarcasm was meant. Remarkably—based on mis-statements you frequently make concerning leakage in various product reviews—it is only you who still seems to be confused. :D

As for the high-source-impedance leakage through our original LPS-1 due to the ungrounded Mean Well charging brick, not only has that been litigated to death, but I have repeatedly acknowledged that your indirect measurement of its effect lead us to put a couple of weeks of effort into ferreting out this broadly overlooked form of leakage, resulting in the simplest possible solution to it (see our graphs above showing the efficaciousness of “ground-shunting” the zero-volt VE-, and the end result out of the LPS-1).
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
16,770
Likes
12,970
Location
Seattle Area
#20
No sarcasm was meant. Remarkably—based on mis-statements you frequently make concerning leakage in various product reviews—it is only you who still seems to be confused. :D
Yeh right. As I noted, you are acting like a student that got an F in math just a year ago, and now pretends to be a professor teaching the topic.

You and John designed and sold a product to audiophiles as battery replacement where in reality it was pumping mains leakage into audio gear. After weeks of back and forth and countless tests, you conceded this basic issue. Even beyond that, unless you had the world's most broken audio product, the thing did nothing useful.

Did you ever buy an audio analyzer so that you can test what your products do to improve the fidelity of the system? Or are still trying to confuse people with micro-measurements which may be totally at fault as I showed in the past?
 

Similar threads

Top Bottom