Mains EMI filter and surge protector high leakage current

[wascodagama]

Dear All,

To protect my precious audio equipment, I've made an EMI filter/surge protector combination.
However, when I measure the RMS current to the pcb with my cheap multimeter, it draws a constant 25,5 mA of current at 240V = 6Watts without load!
None of the components get hot, even after an hour of being plugged in without load.
When I remove the Y2 capacitors to ground, the leakage current drops to 21,4 mA, which is still a whopping 5Watts.


What is going on here? Why is the measured current 'consumption' so high? Is the cheap multimeter at fault? Is my design bad? Is my power very dirty?
Since standards for leakage current speak of a maximum 0,4mA permitted limit for commercial use appliances, I'm not happy with the hundredfold reading on my multimeter...

Please refer to the attached picture for PCB and schematics.
Thanks for your help!



For EMI filtering I use

- 2 Y2 capacitors to ground (47nF - rather big?)
- a common mode choke (9 turns)
- a differential mode choke on line and neutre (9 turns)
The chokes are ferrite coils with self-wound regular 1,5mm copper conductor, stripped and re-insulated with transparant crimp insulation - for the looks
-3x 100nF x2 capacitors across line-neutre.

For surge protection I use both at the input side as the output side 2 MOV's rated at 275vac secured by a series Thermal fuse.

[pardo-bsso]

What is going on here? Why is the measured current 'consumption' so high? Is the cheap multimeter at fault? Is my design bad? Is my power very dirty?

What is the impedance of all those capacitors at your line frequency (50Hz / 60Hz)?
What is the result of 240V / Z ?

[Gyro]

[Edit: Sorry, I didn't notice that it was your first post - Welcome to the forum  ]

Firstly, please don't use a cheap multimeter to measure mains current. Unless it is properly designed and equipped with HRC fuses, one accidental probe (ie. Live-Neutral on the current setting) could cause your meter to explode and cause you burns or worse.

The current you are measuring is that of a capacitive load (assuming that the VDRs aren't conducting at all). The Power factor is a long way from unity (the current draw waveform is out of phase with the voltage waveform) so your calculation for power dissipation will be a long way off. It's also the reason why you are not finding anything warming up. Try https://en.wikipedia.org/wiki/Power_factor as a starting point.

[wascodagama]

What is going on here? Why is the measured current 'consumption' so high? Is the cheap multimeter at fault? Is my design bad? Is my power very dirty?

What is the impedance of all those capacitors at your line frequency (50Hz / 60Hz)?
What is the result of 240V / Z ?

Hi there!

Indeed... 3x100nF + the 2 Y-caps in series give a total of 396nF between line and neutre. At 50hz that's about 8038Ohm, resulting in a current of 30mA at 240Vac... Which already comes close to the measured value...Thank you!
@ Gyro: Thank you for your remark. Would this discrepancy between calculated (30mA) and measured (25,5mA) explain the difference between apparent (measured) and real power (calculated)?

What would be the acceptable value for leakage current and filtering caps for a power conditioner as I'm trying to make?
The triple 100nF + Y-caps was the only example I found on the net of specific values for this circuit...

Just thinking forward...What kind of values should I consider instead of these?
Reversing the calculation with a target value for 'leakge budget' of for example 3 mA at 50Hz gives me around 30nF equivalent at 50Hz.
With the same topology in the PCB, I could go for 3x 10nF and smaller values (say 4,7n ?) for the Y capacitors to earth.

The smaller the caps, the higher the peak attenuation at thigher frequencies until parasitic inductance compensation...
If cost / floorspace is not an issue, I assume a combination of different capacitor values (eg. 10nF & 1nF) would result in better attenuation across a wider frequency range?

Thanks again for your help!

[Yansi]

it draws a constant 25,5 mA of current at 240V = 6Watts without load!

25,5 mA of current at 240V = 6VA, not watt!

[wascodagama]

Dear Gyro & Yansi,

In the meantime I've refreshed my memory on real/apparent power, but I need some help to understand the impact on my situation.

I've read that in a capacitive circuit (at no load, I guess the inductors in the circuit don't act, as there is nearly no current), the current is 90°out of phase with the voltage, hence no real power (Watts) is consumed.

If my power meter (supply company) only measures real power (kWh), I'm in the clear ?

[Gyro]

Yes that's right, your supply current meter will be reading real power. Your filter will be dissipating some power through losses (you don't get anything for free), but nowhere near your VA calculation, otherwise you would be detecting the heat (real power) dissipation. 5W would be quite noticable.

As you say, yes with a capacitive load the current leads the voltage by 90° (in simple terms, the capacitor current is highest when the rate of change of voltage is highest - that's as the voltage waveform passes through zero). The inductors are too small in value to 'participate' in an LC network at 50Hz, so the circuit is effectively capacitive at mains frequency.

[T3sl4co1l]

FYI, those series chokes aren't doing anything -- they probably saturate at less than 100mA load.  You need powdered iron or air core chokes in that position.

The current compensated choke is fine.

Tim

[floobydust]

For audio equipment (mains) filtering, I don't place Y-caps right at the mains input.
They just shunt all powerline RF and garbage to your ground. Since ground is many metres of #14AWG wire and has a fair impedance, it just makes your audio gear's ground noisy.

I prefer the medical-grade line EMI filters, which have no Y-caps.

[wascodagama]

FYI, those series chokes aren't doing anything -- they probably saturate at less than 100mA load.  You need powdered iron or air core chokes in that position.

The current compensated choke is fine.

Tim

Hi Tim,
Thank you kindly for your comment. I understand that when saturated, the coil will no longer perform the intended filtering at higher frequencies. Should I worry -beyond the functional impairment- about saturation, especially with regards to safety (overheating)?

Unfortunately, my local supplier doesn't have any of your proposed solutions, which got me thinking... How can I solve this?
Can I reduce the number of windings to prevent saturation without losing filter functionality?
One thought that crossed my mind was to cut an air gap -or- cut out the top half in the series choke cores to reduce the chance of saturation.
Another solution would be to remove the core completely and keep the coil as is (9turns)

What would you reckon to be the effects of the proposed solutions?
I intend to test the current design as is with increasing current and assessing the coil temperatures.
Thanks again for your useful remark.

[T3sl4co1l]

Saturation is a magnetic thing, not a thermal thing.  (Well, it's affected by temperature -- but it's not a power loss, not at these frequencies.)

It's a reduction in the inductance at high current, that's all.

A gap would work, but ferrite is a ceramic.  You have three problems:
1. What do you cut it with?  You need a carbide abrasive.  (A cutoff wheel would work.)
2. It's very brittle, so you might chip or break it in the process.  If nothing else, it's more likely to break from handling, or other stress.
3. How much gap to create?  (This can be calculated, at least!)

Usually a different core shape is used, like an 'E' or rod or bobbin type.  These can be gapped (by adding an insulating shim), or are natural air gap shapes (a rod or bobbin is half air gap already!).

Powdered iron cores and ferrite shapes are commonly found in various equipment, or are affordable from various suppliers.

The thing is, you probably don't have any means to assess the RF performance of this filter?

If that's the case, then, the added series inductance doesn't matter -- for all you'd know, it could be a link of wire, right?  Which begs two further questions:
1. Do you really need honesty in the construction of a device which does not serve an honest purpose?
2. What ultimately is the purpose of this device?  If it is to perform its stated duty (RF filtering), then you must have some means to test it, right?  If you don't know whether it will work or not, and you can't determine whether it's working or not, why bother with it at all?

This unfortunately collides with a value at the heart of audiophile culture: that double-blind testing, or indeed much testing at all, is not only useless but counterproductive.  Understand that, while we can offer technical solutions here, within that culture, they're just a bunch of hot air...  Point being, #2 may well have a positive answer (why bother? because: ...), just a non-technical reason.

Tim

[floobydust]

Having a mains EMI filter is usually better than none, psychoacoustics play a large role in the audio community.
A common belief is that the filtering should be for incoming (mains) EMI- not outgoing EMI from a SMPS input which EE's are usually concerned with.
It's all ruined if you have something with a SMPS, like TV or DVD player plugged in on the same "filtered" mains. That TV or DVD smps now pollutes everything, so what was the point?

OP is really copying commercial EMI filters, similar to these Tripp-Lite power bars I use.
Inside I always see solenoid-style inductors as they don't saturate at 15A or so.
But I have no idea what OP's core material is, so his magnetics may look good but be saturating at low currents.

[wascodagama]

Dear Tim & floobydust,

Indeed, this is nothing more than a copy of commercial mains EMI filter. The intent is to make only 1 - no commercial goals here.
My main goal was to add overvoltage/lightning protection for my equipment until I saw many commercial examples with the filter section.
I decided to add this element to the design while I was at it.
Indeed the goal would be to have only my amplifier with toroidal transformer behind - no SMPS equipment.


Without (unfortunately) any means of testing the filter section I was left with the option to look at as many as possible designs and guesstimate what would be required.
Added to that the limited product catalog of the local supplier led me to copy the series filters of the design as floobydust showed.
I wish I could take it a step further and test its actual working. Unfortunately without testing equipment, I'm stuck at this point relying on seasoned forum participants that can a sanity check by looking at the design.

Why bother making this? - With the overvoltage/surge suppression working safely with the thermal fuses,I'm already happy. The filtering is a nice add.
Added to that, there is always the fun of the project and learning.


To wrap up: Inspired by Tim's comment - if you can't measure it, why bother? -l'll leave the design as-is except I've replaced the large (47nF) Y-caps by 4,7nF ones.

[floobydust]

The core material is the big unknown, if it is saturating then the filter will not work well. If you know the core make/model, we can check.

You would notice the filter's effect right away on high-gain low-level gear, like mic preamps, turntables etc.

For an audio power amplifier, I find significant RF also comes in on the loudspeaker wiring (forms dipole antenna). You can try add come CM clamp-on ferrites there.

This filter need a PE-grounded metal enclosure because the MOV's burn up when they fail. The thermal fuse can take a long time to trip, or it can just add to the arc as it sits next to the MOV's. I just plan for total meltdown.

[wascodagama]

Hi Floobydust,

I've tracked the limited data from the manufacturer of the chokes:

Series choke (2x)  http://www.amidoncorp.com/ft-82-77/
Common mode choke   http://www.amidoncorp.com/ft-114-43/

Especially the series chokes seem to have a large AL Value (which I deduct to be microHenry per 100 turns from the common mode choke page)

With regards to safety: I've soldered the series thermal fuses in between parallel MOV's with a small rag with ice water wrapped around them to prevent premature tripping, but they are much less vulnerable than I thought... Also note the crimp around the 2 parallel MOV's with the fuse in the middle which should squeeze them even tighter together when overheating - hope it will trip in time when the day comes.

With regards to the case: it is an aluminium case that is earthed. The metal case I absolutely required for a DYI situation incorporating MOV's even with thermal fuses... especially after reading up on MOV's reputation.
I would like to add a plastic cover however to have a visual inspection of the PCB - again both for looks as easy check.
I'm thinking PMMA - not sure if it exists in clear in UL94-V0 or EU E class. It stays open for a while until I find what I need.

I've attached a picture of it.

[CopperCone]

I just removed a GFCI outlet in my lab, it could not handle 2x 8904a, 2x wavetek filter (where I installed a EMI module myself) running at the same time.

Previously traveling wave tube amplifiers gave it trouble as well. I wish you could adjust them to allow for a higher leakage current with a potentiometer or something. Guess that would violate safety directives, but I wish I could have something in between a residential GFIC and no GFIC at all. I guess I need to look for something industrial, but I doubt it will fit in a home dwelling outlet.

[T3sl4co1l]

Isolation transformer.

[wascodagama]

Hi Coppercone,

Not sure what the electrical installation regulations in the US are, but in Europe - in Belgium specifically - for domestic installations regulations require earth leakage current detectors in your home central circuit box.
These are cascaded from 300mA earth leakage on the mains input of your installation down to lower values to protect specific circuits e.g. bathroom, where 30mA is required for protection.
The smallest value you can buy is 10mA to my knowledge.
I know that for industrial applications, there are different types of these: faster and slower acting earth leakage current detectors to prevent tripping with heavy reactive loads and special types that can detect DC faults on top of AC.

As Tim indicates, an isolation transformer will make the earth leakage detector blind for this type of fault: the earth leakage detector checks the delta between neutre and phase return current.
With an isolation transformer, I assume the line and neutre current will always be equal, unless there is a leak in the primary winding?

[T3sl4co1l]

Right.  It's a somewhat heavy-handed approach, but it has the advantage that it's dumb -- a lump of iron and copper -- and it's more or less the approved method to use inside or with equipment.

Do remember to connect the equipment ground to earth!  Secondary neutral can also be bonded to earth (in which case the offending ground leakage is shunted on the secondary side, without increasing the hazard much).

Another way to think of it is, you're actually isolating the mains, rather than isolating the device.  That is, the mains side (transformer primary) has no galvanic or leakage path to ground, so it won't trip the breaker that way.  Meanwhile, the secondary can do whatever, and consequently, should be grounded and all that.

Tim

[Gyro]

Hi Floobydust,

I've tracked the limited data from the manufacturer of the chokes:

Series choke (2x)  http://www.amidoncorp.com/ft-82-77/
Common mode choke   http://www.amidoncorp.com/ft-114-43/

Especially the series chokes seem to have a large AL Value (which I deduct to be microHenry per 100 turns from the common mode choke page)

With regards to safety: I've soldered the series thermal fuses in between parallel MOV's with a small rag with ice water wrapped around them to prevent premature tripping, but they are much less vulnerable than I thought... Also note the crimp around the 2 parallel MOV's with the fuse in the middle which should squeeze them even tighter together when overheating - hope it will trip in time when the day comes.

With regards to the case: it is an aluminium case that is earthed. The metal case I absolutely required for a DYI situation incorporating MOV's even with thermal fuses... especially after reading up on MOV's reputation.
I would like to add a plastic cover however to have a visual inspection of the PCB - again both for looks as easy check.
I'm thinking PMMA - not sure if it exists in clear in UL94-V0 or EU E class. It stays open for a while until I find what I need.

I've attached a picture of it.

Just looking again at your implementation. I very much doubt that those little PCB terminal blocks are suitable for the 15A rating of your fuse.

I would swap the fuse for something like 5A or 6.3A. Go for a ceramic too, something with a higher interrupt rating - those glass ones are only capable of interrupting 35A at mains voltage so you would get a big bang and evaporated copper on a short. A decent ceramic fuse of the same dimensions will interrupt 1.5kA.