Inverter filters

[john fletcher 1Participant @johnfletcher1]

Some Inverters include a filter hoping to reduce or eliminate main borne interference created by the inverter. Has any one made their own filter, and is there a simple way/method of testing to find out if an inverter is creating interference. Also would a filter from an older type washing machine be of use. Amateur Radio enthusiasts make their own filter so what is special about those in Inverters, such that  Model Engineers are unable to make their own ?  Ted

[John HaineParticipant @johnhaine32865]

Why an “older type” wasking machine?  Modern machines often use induction drives with 3ph VFDs and mains filters for them should be a standard spare part I’d have thought.

You could make your own but messing around with high voltage caps etc not my idea of fun, less hassle to buy and probably work out not much more expensive.

[John Haine]

On John Haine Said:

Modern machines often use induction drives with 3ph VFDs…

Would there be any potential difficulty with the power rating of the devices? A person might have a 3kW VFD. Not many washing machine motors would demand that power.

The plated power draw of a w/m might not be a good guide as a lot of that would be eaten by the water heater.

[John HaineParticipant @johnhaine32865]

Not many amateur lathes with 3 hp motors.

[SillyOldDufferModerator @sillyoldduffer]

The filter design is complicated because power inverters switch square waves at radio frequency, and square waves are full of harmonics and voltage spikes.  The filter cleans up in several stages, each stage removing part of the muck.  The stages are carefully matched to cooperate with each other.  The design is difficult.

The components aren’t ordinary:  high-speed low ESR capacitors, and inductors wound on ferrite cores selected to absorb part bands of unwanted frequencies.  The inductors are probably wound on toroid cores rather than solenoids. All mains rated or better.   The circuit is balanced, so double the number of components.

The components are internally shielded inside a screened box and often arranged at right angles to minimise pick up.  Circuit something like this:

So, 8 feed-through capacitors, 14 filter capacitors and 6 inductors.

The filter I recovered from an old washing is single stage, unbalanced and narrow frequency.  I mean to try it on my lathe one day.  I expect it will help a little, not much.

Home made is possible.  Trouble is, the interference spectrum would have to be identified and fed into the component value calculations.  I have a book and the maths is intimidating.   There might be an online calculator, but then suitable components have to be found and assembled in a well-made box.  A loose lid ruins performance.

Could take a by-and-large approach, building a filter with guesstimated components, and accepting it will underperform.  Not difficult to filter out 6dB, problem being a professional unit does 60dB or better.  And the scale is logarithmic.  Suitable filters must be available second-hand, I’ve never seen one!

Some Radio Amateurs do make filters in this class; not me, though I have made a few simpler bandpass filters.  For example, transmitting on 7MHz, suppressing the second harmonic on 14MHz and the third on 21MHz is relatively easy because the three single signals are spaced 7MHz apart.  Inverters are much messier.

Dave

 

 

 

 

 

[Julie AnnParticipant @julieann]

The two types of filter mentioned by the OP are different animals.

Most RF filters used my radio amateurs are what I would call classic filters. That is they have a standard characteristic, lowpass, highpass, bandpass or bandstop and use one of the standard filter arrangements, ie, Butterworth, Chebyshev, Elliptic and others. Most RF filters are passive and use LC components, as opposed to active filters using opamps and RC components.

In order to understand these filters a grasp of poles and zeros on the complex s-plane, ie, Laplace transform theory, helps visualise what is going on. In practice the pole and zero positions for all the common filter types are tabulated along with unit frequency component values for common circuit configurations. For frequencies other than unity a simple scaling gives real component values.

Of course plug ‘n’ play computer programs are available to do all the calculations needed above which will lead to a design but not to understanding.

The filtering needed for a VFD input is a little different. First it is important to understand that a VFD is a component, it needs to be assembled with other components to make a working system. For that reason, and cost, input filtering is not normally included. The input filter needs to perform several functions:

• Protect against nasties on the incoming mains, mostly fast spikes up to several kV
• Limit RF being conducted back down the mains
• Reduce harmonic current distortion due to non-sinusuiodal input current draw from the mains

The filters are normally designed using general techniques, the noise spectrum for a specific VFD is not needed, or indeed known, as it will depend on the complete system integration.

Protection against spikes is simple, usually metal oxide varistors across the various lines. High frequency filtering is fairly basic, low value capacitors across the lines and possibly in line lossy chokes. Most of the high frequency noise generated by a VFD is on the output, not the input.

Reducing harmonic distortion is the key function of the filter. Essentially it is a low pass filter looking out towards the mains. But the filter needs to deal with both common-mode and differential-mode distortion. The most common arrangement is a common-mode choke on both lines (live and neutral) and a capacitor across the lines and other capacitors from both lines to earth.

There are some special considerations. The common-mode choke needs to carry the total input current without saturating the core. The capacitors need to be special types for safety reasons. The capacitor across the lines needs to be X-rated, these will fail short and hence blow an external protection circuit like a fuse. The capacitors from line to earth need to be Y-rated. These will fail open so that earth is not connected to a line.

To summarise it is of course possible to design ones own input filter but since off the shelf filters, in metal shielding cases, are readily available it seems a rather pointless exercise.

Julie

 

 

Julie

[Robert Atkinson 2Participant @robertatkinson2]

I came late to this one. Julie has pretty much said it all.
Generally in small quantities the components will cost more than a complete filter.
From experience most small VFD’s and similar unit will need a two stage filter. Going for a higher current rating than you need is not good as apart from costing more they tend to have lower value inductors and thus lower perormance.

Another issue is leakage current. The capacitors to ground (often called Y capacitors because they should be Y safety class) pass current to ground. This unbalances the live to neutral current and if large enough will trip safety devices like RCDs and RCBO’s. The current is cumulative and a VFD filiter will sometime tip the system over the limit. It also means that if the earth is disconnected e.g. unbolting a filter from a chassis can make the body of the filter live.

Used filters are available.

Robert.

[Author]

Posts