# Mains filter design

#### RGood1

Joined Jan 25, 2016
6
Hi

I am designing a filter to go between a product and the mains to help with conducted emissions failures. All the material seems to point towards using RLC networks but could a typical RC network work instead? I have designed one to be a 9th order with 9 100 pF caps and 9 36 Ohm resistors.

Thanks

#### Sensacell

Joined Jun 19, 2012
3,180
The power loss of the R is not going to be negligible.

#### Papabravo

Joined Feb 24, 2006
20,122
I could more readily believe an LC filter, but fabricating suitable inductors is always a challenge.

#### RGood1

Joined Jan 25, 2016
6
My concern is the power ratings of the resistors using the classic high-pass filter design. When you do the calculation

P = V^2 / R

P = 110^110 / 36

P = 336 W per resistor is this correct?

#### Papabravo

Joined Feb 24, 2006
20,122
My concern is the power ratings of the resistors using the classic high-pass filter design. When you do the calculation

P = V^2 / R

P = 110^110 / 36

P = 336 W per resistor is this correct?
Only if each resistor had 110 volts across it. The TS/OP did not specify the nature of the filter, only that he was interested in conducted emissions. Without a schematic it is hard to say what we are dealing with.

You might find the following paper of interest:
http://www.egr.msu.edu/em/research/goali/notes/module11_conducted.pdf

#### RGood1

Joined Jan 25, 2016
6
Its a filter to go between a power supply and the mains. The power supply is injecting a series of frequencies back onto the mains and is failing EMC conducted emissions.

I looked into filters and found a high-pass filter which is essentially a capacitor in series with a resistor that goes down to ground like this:

http://hyperphysics.phy-astr.gsu.edu/hbase/electric/filcap.html

#### RGood1

Joined Jan 25, 2016
6
This has already been tried and has not been able to reduce the frequencies enough hence the need for a designed filter.

#### RGood1

Joined Jan 25, 2016
6
I am not trying to filter the 50/60 Hz mains, I am trying to filter the rubbish coming from the product.

#### AnalogKid

Joined Aug 1, 2013
10,299
There are a LOT of power line filters to choose from, ranging from a few ounces of PC board components to TEMPEST-grade that weigh a few pounds. None of them use resistors for anything other than bleeding of leftover charge in the filter capacitors (some applications require this), but those resistors are in the megohm range. Read a few catalogs and you'll get a feel for how the pros do it. Mains filter design frequently starts with the power line frequency and leakage current require limits for the country you're in. Then pile on the inductance (and stages) to get the characteristics you want. The best way to contain EMI is to build the AC power input connector into the body of the power line filter's hermetic can, and mount the filter directly on a metal surface of an all-metal enclosure. What country are you in?

P = V^2 / R
P = 110^110 / 36
P = 336 W per resistor is this correct?
Only if the resistors are directly across the power line.

ak

#### Papabravo

Joined Feb 24, 2006
20,122
I am not trying to filter the 50/60 Hz mains, I am trying to filter the rubbish coming from the product.
You have a basic misunderstanding. A high pass filter, like the one you pointed to , will pass the garbage in the range of 450 KHz. to 30 MHz. that is the subject of conducted emissions testing. You might be better off improving the design of the non-compliant product rather than putting a band-aid on after the fact.

You might find the following paper of interest:
http://www.egr.msu.edu/em/research/goali/notes/module11_conducted.pdf

The best advice I can give you is to find a consultant with expertise in this field. You will end up spending less time and money if you do this now rather than six months from now when all the patchwork solutions have failed to achieve your goal.

Last edited:

#### Papabravo

Joined Feb 24, 2006
20,122
There are a LOT of power line filters to choose from, ranging from a few ounces of PC board components to TEMPEST-grade that weigh a few pounds. None of them use resistors for anything other than bleeding of leftover charge in the filter capacitors (some applications require this), but those resistors are in the megohm range. Read a few catalogs and you'll get a feel for how the pros do it. Mains filter design frequently starts with the power line frequency and leakage current require limits for the country you're in. Then pile on the inductance (and stages) to get the characteristics you want. The best way to contain EMI is to build the AC power input connector into the body of the power line filter's hermetic can, and mount the filter directly on a metal surface of an all-metal enclosure. What country are you in?

Only if the resistors are directly across the power line.

ak

#### AnalogKid

Joined Aug 1, 2013
10,299
I want a raise.

#### RGood1

Joined Jan 25, 2016
6
The product is not ours and I can't change it I have to implement a retro-fit filter.

Thanks for the help guys, the reason I pointed to a high pass is that the frequencies are in the 100 kHz - 1 MHz range so I designed a high pass filter to pass everything beyond 1 MHz but to suppress the frequencies below it.

#### Papabravo

Joined Feb 24, 2006
20,122
The product is not ours and I can't change it I have to implement a retro-fit filter.

Thanks for the help guys, the reason I pointed to a high pass is that the frequencies are in the 100 kHz - 1 MHz range so I designed a high pass filter to pass everything beyond 1 MHz but to suppress the frequencies below it.
So get them to fix their retrograde product; and since the conducted emissions test (US) goes from 450 kHz. to 30 MHz., you figure that will be sufficient. I hate to tell you this, but you're trying to push on a rope. Some systems are so bad that no retro-fit can fix it; and you need to understand when that situation occurs.

What would be the harm in buying one as an experiment and as a control for your own implementation should you decide to go that way?

Last edited:

#### ronv

Joined Nov 12, 2008
3,770
The product is not ours and I can't change it I have to implement a retro-fit filter.

Thanks for the help guys, the reason I pointed to a high pass is that the frequencies are in the 100 kHz - 1 MHz range so I designed a high pass filter to pass everything beyond 1 MHz but to suppress the frequencies below it.
Usually low pass filters are used with the corner frequency at about 400 HZ.
What frequency is it out at, and by how much?

#### AnalogKid

Joined Aug 1, 2013
10,299
ronv asked the real question. If this is a retrofit to an existing product, that effort must be caused by the product needing a filter. This means that someone thinks the product is out of compliance at some frequencies.
What is the compliance standard you are trying to meet? FCC-A, FCC-B, MIL-STD-461, etc.
What are the frequencies at which the product fails?
What is the magnitude of the fail?

ak

#### crutschow

Joined Mar 14, 2008
32,031

#### Papabravo

Joined Feb 24, 2006
20,122