cupoftea
Advanced Member level 6
Hi,
Just received another free email from
...i always read their articles with interest.
However, i have never found an EMC tutorial site, or EMC book, which tells the "real dirty truth" of EMC for SMPS.
For example, its a fact that you can worsen common mode emissions for an SMPS by overdoing the Diff Mode filtration.
EMC is full of such wonderments, but no website speaks of such phenomena.....or do you know of one that does?
Another "real dirty truth" is that floating metal, or a floating metal enclosure, around or near an SMPS can cause it to stop working. (By floating i
mean there are no direct connections to circuit or capacitive connections.)
Where do you find these dirty truths?
This article is as close as it gets to giving away dirty truths....
..in there, they confess that virtually no hard switching offline SMPS can ever truly pass Radiated Emissions EMC testing (to EU/A/NZ/UK/USA standards) unless it is caged in a metal enclosure.
Another filthy truth is that if you have a large number of paralleled SMPS's, then you set yourself up to get massive ground loop problems for EMC. You can mitigate this by using low value common mode chokes at the input and output of each SMPS thats paralleled.
Another filthy truth is that with an offline SMPS, you can actually often pass EMC by literally turning the common mode choke through 90 degrees.
Yet another dirty truth is that often you can pass EMC for offline SMPS by adding in a small common mode choke wound with a few turns of insulated wire (not ECW)...the insulation keeps the metal strands apart such that interwinding capacitance is reduced.
Another dirty truth is that if you have an offline SMPS screwed to an earthed heatsink, (which often also acts as its enclosure) then you must connect (either directly or capacitively) the circuit's GND (or other quiet node), to the earthed heatsink.
Another dirty truth is that often connecting a small common mode choke wound with insulated wire (as discussed above) right by the mains input connector often passes EMC.
And another dirty truth is that connecting Y caps L-E and N-E upstream of the common mode choke of an offline SMPS can often make common mode EMC worse (even though it generally improves common mode EMC immunity).
Another dirty truth concerns offline SMPS for which the secondary is connected to earth....if so, then far better EMC is achieved if you make a shielded earth connection from the earth wire in the mains input cable, to the secondary ground (or if not shielded then torroid wrap it). If the secondary is then also earthed from another connection there (which it often is), then you should add a ferrite bead into that earth connection (or wrap it round a torroid a few times) to avoid a big low Z earth loop. Also, adding a y cap from pri to sec of isolation transformer helps here.....and also increasing the Y cap from "downstream sides [L&N] of common mode choke to earth".
In relation to this last point, is that no EMC book states how the EMC is usually made far worse by connecting the secondary of an offline SMPS to earth ground via an external earth connection (ie, not via the mains input cable earth). -It obviously results in an enormous & horrendous earth loop.
Another dirty truth is that the "L-E" Y capacitor in many offline SMPS's is often dead. (Also the y cap across the transformer).
This is because common mode transients are the most common of the large transients and they eventually pop open
the "L-E" y cap. The "N-E" y cap isn't so badly affected as Neutral is usually connected to earth at most installations anyway. Even if there
is a MOV "L-N" it has "let-through" voltage. This is one reason why when you connect a y cap from an offline SMPS
to an earthed metal enclosure...if you only can fit in one Y capacitor....then put it "N-E" instead of "L-E".
Another DT is that MOVs are usually connected L-N, even though common mode transients are far more common and damaging. (N is connected to earth at most installations)
Another DT is that connecting a large 450V electrolytic via a diode bridge to the incoming L/N is generally the best form of mains transient protection since it will soak the surge up and not die like a MOV which unfortunately crowbars the fuse. There is only one piece of literature on the web that confess's this dirty truth, and its in a Synqor datasheet.
Another DT is that streetlights often don't have their neutral connection connected to earth anywhere near
the actual streetlight. This means that common mode transients can wreak havoc. Often the streetlight's
local L and N are taken transiently up to thousands of volts above earth. (ie above the voltage of the actual earth
wire that's in the mains cable coming to the streetlight)
So basically for a streetlight it would often be best to connect a power resistor in this earth wire say 10 Ohms.
Then you connect a large electrolytic from L to earth via diode bridge (downstream of the resistor)
...and ditto Neutral to earth.
This gives the best transient protection for the streetlight.
Add a TVS in parallel to the lytic, and a high value discharge resistor.
The dirty truths of Streetlight transient protection still aren't understood, and this is why today, many miles/kilometres of the UK motorways have no streetlights...even the M25 "ring of hell".
A streetlight on a motorway should be non isolated...its PSU should be in the base near the ground where its easily accessible for failure replacement.....any LED heatsink should simply be connected to neutral. You dont want any earth on that streetlight head MCB LED heatsink , since it may well end up thousands of volts apart from L and N due to lightning strike. The earth to the streetlight should be earthed at local to the streetlight, but very often isnt.
Just received another free email from
EMI Basics and Board Level Shielding Design | Interference Technology
Board level shielding has become a requirement for most PCB designers to reduce EMI as circuitry grows more complex. Learn the basics here.
interferencetechnology.com
...i always read their articles with interest.
However, i have never found an EMC tutorial site, or EMC book, which tells the "real dirty truth" of EMC for SMPS.
For example, its a fact that you can worsen common mode emissions for an SMPS by overdoing the Diff Mode filtration.
EMC is full of such wonderments, but no website speaks of such phenomena.....or do you know of one that does?
Another "real dirty truth" is that floating metal, or a floating metal enclosure, around or near an SMPS can cause it to stop working. (By floating i
mean there are no direct connections to circuit or capacitive connections.)
Where do you find these dirty truths?
This article is as close as it gets to giving away dirty truths....
..in there, they confess that virtually no hard switching offline SMPS can ever truly pass Radiated Emissions EMC testing (to EU/A/NZ/UK/USA standards) unless it is caged in a metal enclosure.
Another filthy truth is that if you have a large number of paralleled SMPS's, then you set yourself up to get massive ground loop problems for EMC. You can mitigate this by using low value common mode chokes at the input and output of each SMPS thats paralleled.
Another filthy truth is that with an offline SMPS, you can actually often pass EMC by literally turning the common mode choke through 90 degrees.
Yet another dirty truth is that often you can pass EMC for offline SMPS by adding in a small common mode choke wound with a few turns of insulated wire (not ECW)...the insulation keeps the metal strands apart such that interwinding capacitance is reduced.
Another dirty truth is that if you have an offline SMPS screwed to an earthed heatsink, (which often also acts as its enclosure) then you must connect (either directly or capacitively) the circuit's GND (or other quiet node), to the earthed heatsink.
Another dirty truth is that often connecting a small common mode choke wound with insulated wire (as discussed above) right by the mains input connector often passes EMC.
And another dirty truth is that connecting Y caps L-E and N-E upstream of the common mode choke of an offline SMPS can often make common mode EMC worse (even though it generally improves common mode EMC immunity).
Another dirty truth concerns offline SMPS for which the secondary is connected to earth....if so, then far better EMC is achieved if you make a shielded earth connection from the earth wire in the mains input cable, to the secondary ground (or if not shielded then torroid wrap it). If the secondary is then also earthed from another connection there (which it often is), then you should add a ferrite bead into that earth connection (or wrap it round a torroid a few times) to avoid a big low Z earth loop. Also, adding a y cap from pri to sec of isolation transformer helps here.....and also increasing the Y cap from "downstream sides [L&N] of common mode choke to earth".
In relation to this last point, is that no EMC book states how the EMC is usually made far worse by connecting the secondary of an offline SMPS to earth ground via an external earth connection (ie, not via the mains input cable earth). -It obviously results in an enormous & horrendous earth loop.
Another dirty truth is that the "L-E" Y capacitor in many offline SMPS's is often dead. (Also the y cap across the transformer).
This is because common mode transients are the most common of the large transients and they eventually pop open
the "L-E" y cap. The "N-E" y cap isn't so badly affected as Neutral is usually connected to earth at most installations anyway. Even if there
is a MOV "L-N" it has "let-through" voltage. This is one reason why when you connect a y cap from an offline SMPS
to an earthed metal enclosure...if you only can fit in one Y capacitor....then put it "N-E" instead of "L-E".
Another DT is that MOVs are usually connected L-N, even though common mode transients are far more common and damaging. (N is connected to earth at most installations)
Another DT is that connecting a large 450V electrolytic via a diode bridge to the incoming L/N is generally the best form of mains transient protection since it will soak the surge up and not die like a MOV which unfortunately crowbars the fuse. There is only one piece of literature on the web that confess's this dirty truth, and its in a Synqor datasheet.
Another DT is that streetlights often don't have their neutral connection connected to earth anywhere near
the actual streetlight. This means that common mode transients can wreak havoc. Often the streetlight's
local L and N are taken transiently up to thousands of volts above earth. (ie above the voltage of the actual earth
wire that's in the mains cable coming to the streetlight)
So basically for a streetlight it would often be best to connect a power resistor in this earth wire say 10 Ohms.
Then you connect a large electrolytic from L to earth via diode bridge (downstream of the resistor)
...and ditto Neutral to earth.
This gives the best transient protection for the streetlight.
Add a TVS in parallel to the lytic, and a high value discharge resistor.
The dirty truths of Streetlight transient protection still aren't understood, and this is why today, many miles/kilometres of the UK motorways have no streetlights...even the M25 "ring of hell".
A streetlight on a motorway should be non isolated...its PSU should be in the base near the ground where its easily accessible for failure replacement.....any LED heatsink should simply be connected to neutral. You dont want any earth on that streetlight head MCB LED heatsink , since it may well end up thousands of volts apart from L and N due to lightning strike. The earth to the streetlight should be earthed at local to the streetlight, but very often isnt.
Last edited:
