Wiggly Electron

Hi all
Here is the update on how to make your own inductor to filter out the noise coming from the inverter.
First off, you will need a core to wind your wire around. A toroidal core will be the easiest to work with and will also radiate the least amount of EMI. The bigger the core, generally speaking, the better as you will be able to get more windings on it with a thicker wire. Here is the biggest one I could find with a good permeability:
https://www.mantech.co.za/ProductInfo.aspx?Item=82M1485
 
The enamel wire diameter will determine the amount of amps it will be able to handle. I would recommend using 1-1.6mm diameter wire. You can find a good variety from Communica.
https://www.communica.co.za/products/eccw-1-25
Just note that thicker wires will be more difficult to work with, but will obviously be able to handle more current. Here is a website to give you an idea of the current carrying capability of different wire thicknesses:
https://www.powerstream.com/Wire_Size.htm
 
How many turns do you need? Well you would want to wind the inductor to get you an inductance of at least 1mH, but more than 5mH would probably be overkill. You can calculate the exact number of windings you'll need with a nifty program called Coil64:
https://coil32.net/download-coil64-for-windows.html
For the above-mentioned core and wire, you will need about 50turns to get 1mH and 75turns for 2mH. (The inductance goes up with the square of the number of windings and the resistance of the wire goes up linearly with number of windings).
 
As mentioned before, you would place this filter "In-Line" by breaking either live or neutral and join the break through the inductor. 
Also, be sure to use a surge protector on the load side of the filter as inductors might cause voltage spikes when the current draw through it changes.
 
I still need to test my toroidal filter to see how many amps it can handle, but for now I can definitely say that it works in filtering out the noise to a level that is completely inaudible. The previous filter I made with the E-core handles 8 amps without getting too hot. I would guess that the smaller toroidal core will do about 3 amps, as I had to use a thinner 1.25mm wire instead of the 2mm wire.
Hope it helps!
 

Hi everyone
I have found a solution to the problem. Before you get too excited, I need to say that it isn't an easy nor cheap fix. It also only gets rid of the 15kHz whine on your appliances (output side of the inverter), while the noise from the inverter itself remains unchanged. That being said, it was still worth it for me. Our microwave was driving me insane!
Anyway without further ado, to fix the issue you need to place a big filter inductor on the live line between your inverter and your appliances. The inductor must have a ferrite core (preferably the MnZn type as they have a higher core saturation point). Iron cores wont work as they will filter out the 50Hz power as well and probably cook itself within minutes. I got my E-cores from Mantech Electronics (about R350 for two half cores), although they seldom have stock of these large cores. With these cores you will need between 50-100 turns (resulting in 1-5mH) and the wire size is determined by the amount of current your load is going to draw. I used 10m of 3.14mm^2 enamel wire (sourced from Communica @ R210) which got me about 50 turns and 1.25mH. At 1.25mH it should attenuate about 14dB of the 15kHz noise. After I installed it, I couldn't hear the whine at all. So it definitely works.
Also, another note to those who haven't wound an inductor/transformer yet, it took me a whole day to wind that thing. A solid core wire with a 2mm diameter is quite difficult to work with...
Hope this might help someone.
Regards

Hi everyone
I have found a solution to the problem. Before you get too excited, I need to say that it isn't an easy nor cheap fix. It also only gets rid of the 15kHz whine on your appliances (output side of the inverter), while the noise from the inverter itself remains unchanged. That being said, it was still worth it for me. Our microwave was driving me insane!
Anyway without further ado, to fix the issue you need to place a big filter inductor on the live line between your inverter and your appliances. The inductor must have a ferrite core (preferably the MnZn type as they have a higher core saturation point). Iron cores wont work as they will filter out the 50Hz power as well and probably cook itself within minutes. I got my E-cores from Mantech Electronics (about R350 for two half cores), although they seldom have stock of these large cores. With these cores you will need between 50-100 turns (resulting in 1-5mH) and the wire size is determined by the amount of current your load is going to draw. I used 10m of 3.14mm^2 enamel wire (sourced from Communica @ R210) which got me about 50 turns and 1.25mH. At 1.25mH it should attenuate about 14dB of the 15kHz noise. After I installed it, I couldn't hear the whine at all. So it definitely works.
Also, another note to those who haven't wound an inductor/transformer yet, it took me a whole day to wind that thing. A solid core wire with a 2mm diameter is quite difficult to work with...
Hope this might help someone.
Regards

That's great news! A lot easier than to make a filter!

Hi all
Here is the update on how to make your own inductor to filter out the noise coming from the inverter.
First off, you will need a core to wind your wire around. A toroidal core will be the easiest to work with and will also radiate the least amount of EMI. The bigger the core, generally speaking, the better as you will be able to get more windings on it with a thicker wire. Here is the biggest one I could find with a good permeability:
https://www.mantech.co.za/ProductInfo.aspx?Item=82M1485
 
The enamel wire diameter will determine the amount of amps it will be able to handle. I would recommend using 1-1.6mm diameter wire. You can find a good variety from Communica.
https://www.communica.co.za/products/eccw-1-25
Just note that thicker wires will be more difficult to work with, but will obviously be able to handle more current. Here is a website to give you an idea of the current carrying capability of different wire thicknesses:
https://www.powerstream.com/Wire_Size.htm
 
How many turns do you need? Well you would want to wind the inductor to get you an inductance of at least 1mH, but more than 5mH would probably be overkill. You can calculate the exact number of windings you'll need with a nifty program called Coil64:
https://coil32.net/download-coil64-for-windows.html
For the above-mentioned core and wire, you will need about 50turns to get 1mH and 75turns for 2mH. (The inductance goes up with the square of the number of windings and the resistance of the wire goes up linearly with number of windings).
 
As mentioned before, you would place this filter "In-Line" by breaking either live or neutral and join the break through the inductor. 
Also, be sure to use a surge protector on the load side of the filter as inductors might cause voltage spikes when the current draw through it changes.
 
I still need to test my toroidal filter to see how many amps it can handle, but for now I can definitely say that it works in filtering out the noise to a level that is completely inaudible. The previous filter I made with the E-core handles 8 amps without getting too hot. I would guess that the smaller toroidal core will do about 3 amps, as I had to use a thinner 1.25mm wire instead of the 2mm wire.
Hope it helps!
 

Hi all
Here is the update on how to make your own inductor to filter out the noise coming from the inverter.
First off, you will need a core to wind your wire around. A toroidal core will be the easiest to work with and will also radiate the least amount of EMI. The bigger the core, generally speaking, the better as you will be able to get more windings on it with a thicker wire. Here is the biggest one I could find with a good permeability:
https://www.mantech.co.za/ProductInfo.aspx?Item=82M1485
 
The enamel wire diameter will determine the amount of amps it will be able to handle. I would recommend using 1-1.6mm diameter wire. You can find a good variety from Communica.
https://www.communica.co.za/products/eccw-1-25
Just note that thicker wires will be more difficult to work with, but will obviously be able to handle more current. Here is a website to give you an idea of the current carrying capability of different wire thicknesses:
https://www.powerstream.com/Wire_Size.htm
 
How many turns do you need? Well you would want to wind the inductor to get you an inductance of at least 1mH, but more than 5mH would probably be overkill. You can calculate the exact number of windings you'll need with a nifty program called Coil64:
https://coil32.net/download-coil64-for-windows.html
For the above-mentioned core and wire, you will need about 50turns to get 1mH and 75turns for 2mH. (The inductance goes up with the square of the number of windings and the resistance of the wire goes up linearly with number of windings).
 
As mentioned before, you would place this filter "In-Line" by breaking either live or neutral and join the break through the inductor. 
Also, be sure to use a surge protector on the load side of the filter as inductors might cause voltage spikes when the current draw through it changes.
 
I still need to test my toroidal filter to see how many amps it can handle, but for now I can definitely say that it works in filtering out the noise to a level that is completely inaudible. The previous filter I made with the E-core handles 8 amps without getting too hot. I would guess that the smaller toroidal core will do about 3 amps, as I had to use a thinner 1.25mm wire instead of the 2mm wire.
Hope it helps!
 

Hi all
I finally received some toroidal cores. As soon as I get the enamel wireI that I ordered, I'll wind a new filter and test it. 
Should be a few days to a week before another update. 

Hi all
I finally received some toroidal cores. As soon as I get the enamel wireI that I ordered, I'll wind a new filter and test it. 
Should be a few days to a week before another update. 

Hi @DiSiNTiGR8
 
Most of these off-the-shelf filters are not very likely to work unfortunately. I have wasted a couple of bucks to try these out too. Almost all of them are designed to filter out frequencies in the CRISPR defined conducted emissions band, which starts from about 150kHz and ends at 30MHz.
Some do seem to have good filtering properties at 15kHz on their datasheets, but one should be careful for the "50Ω/50Ω symmetrical" condition the tests were run at. This condition refers to the characteristic impedance (transmission line theory) of the test system and is usually only applicable to RF circuits and not mains power. The test condition for an inverter would be closer to something like 0.1Ω/100Ω. Under this test condition most of these filters perform very poorly or can even cause resonances making the problem worse as in you case.
If you are looking for an off-the-shelf solution, you should get a filter with an inductance in the 1-5mH region (the higher the better but past 5mH it would be overkill). Just be careful, some datasheets also list the Common-Mode inductance, which wont be helpful in our case. A differential inductance (aka a normal series inductor as everyone knows it) is needed.
I am yet to find an off-the-shelf solution that has a large enough inductance and current rating. Usually you get to pick either or, but not both.
Regarding a DIY solution, I am looking for a supplier of a large toroidal core, which will make the whole winding process very easy. This will make it a viable solution for the general hobbyist. Will make a post as soon as I find something.
 
Cheers for now
Wiggly

Hi @DiSiNTiGR8
 
Most of these off-the-shelf filters are not very likely to work unfortunately. I have wasted a couple of bucks to try these out too. Almost all of them are designed to filter out frequencies in the CRISPR defined conducted emissions band, which starts from about 150kHz and ends at 30MHz.
Some do seem to have good filtering properties at 15kHz on their datasheets, but one should be careful for the "50Ω/50Ω symmetrical" condition the tests were run at. This condition refers to the characteristic impedance (transmission line theory) of the test system and is usually only applicable to RF circuits and not mains power. The test condition for an inverter would be closer to something like 0.1Ω/100Ω. Under this test condition most of these filters perform very poorly or can even cause resonances making the problem worse as in you case.
If you are looking for an off-the-shelf solution, you should get a filter with an inductance in the 1-5mH region (the higher the better but past 5mH it would be overkill). Just be careful, some datasheets also list the Common-Mode inductance, which wont be helpful in our case. A differential inductance (aka a normal series inductor as everyone knows it) is needed.
I am yet to find an off-the-shelf solution that has a large enough inductance and current rating. Usually you get to pick either or, but not both.
Regarding a DIY solution, I am looking for a supplier of a large toroidal core, which will make the whole winding process very easy. This will make it a viable solution for the general hobbyist. Will make a post as soon as I find something.
 
Cheers for now
Wiggly

Hi there
When looking for a filter in this application it's all about inductance. The bigger the better. If you look at the specs of this ferrite sleeve it is specced to be 225Ω @ 100MHz. So if we use the formula Z=2πfL we can solve the inductance to be 358nH. That is very, very small when trying to filter 15kHz. At this frequency it will only offer about one thousandth of a decibel of attenuation (0.001dB). Or in simpler terms it will only filter 0.02% of the noise we are struggling with.
These sleeve ferrites are designed to filter out much higher frequencies in the AM and FM radio bands. The design issue with these inverters are very unfortunate. I believe most inverters have similar noise emissions, although at slightly higher frequencies that are out of the hearing range for humans. I have measured an Axpert 5kW  VMIII and it had similar noise levels but only at 19.5kHz which is basically inaudible. The problem with the Dye/Sunsynk inverters are that their switching frequency is just a bit too low at 15kHz where we can hear it.
I am still on the lookout for more off-the-shelf solutions, if I stumble across anything I will be sure to post it here.
Regards

Thank you Coulomb for your insightful reply!
 
I only bought my inverter a month or so ago, so it *should* be quite new...
The earth wiring seems to be all good after I double checked it with my multimeter.
 
When I power everything down, I get >100MΩ (yes mega, not milli) between earth and neutral and also between earth and live on the output of the inverter. Does this mean that something in my inverter is broken?
My machine does not have setting 38. I have checked the manual, which states that the dry contact is operated depending on the battery voltages defined in Program settings 12 & 13. Probably useful when using lead acid batteries, but useless in my scenario.
 
Is there a firmware that I would be able to flash onto my inverter to get to setting 38?
 
Otherwise I guess I'll have to build an op-amp circuit to measure the voltage between earth and output neutral. I can then use this op-amp circuit to energise the N to E bonding relay. If there is a large enough voltage between E and N, the relay will bond the two together, while it will disconnect the two when the inverter is in bypass mode and there is almost no difference between E and N.
 
Update: I have fitted a relay to bond the live and neutral. The relay is operated by Eskom mains power, so if there is power the relay breaks the bond and when there is no power the bond is re-engaged. The only drawback is that the output is still floating for about 10 seconds when mains has just come online. This is however way better than it floating all the time! Pity my inverter cant signal it without any dodgy modifications.