Thread Rating:
  • 2 Vote(s) - 5 Average
  • 1
  • 2
  • 3
  • 4
  • 5
48V 5000W Oil cooled inverter build
Output filters... how to stop your inverter from being a rather large and unfriendly radio transmitter..

These type of units typically have one or two filter stages and can be wired in parallel (as I'm told - but make sure the wiring is well balanced....) so I'm going for 3 x 20A units all wired in parallel as they are less than £6 each (20A version) and will then give 60A capacity.

Specific unit I have ordered is : CW4L2-20A-T
4 x 0.5mH     3 x 0.1uF     2 x 3300pF        1.0Mohm

They must be earthed as they are designed to "leak" around 0.6-0.8mA to earth as this is partly how the reflected radio interference / harmonics are dissipated. This earth leakage must be taken into account if your adding residual current brekers, either way these filters should be connected to the transformer output within the transformer housing (metal).

With this arrangement you may then end up with something like :
Note to mods : how can I increase the display size a little larger... ?

Now that my inductance meter has arrived (and a few quick tests) I can return to the input chokes... the unit's I'm running with at the moment have 11 turns of 9 x 1.6mm wires, which is a squeeze... and they also buzz quite a bit under load as I have not glued them yet because I intend to re-wind them and then glue them now that I have a meter and can check each choke separately...

So far I have run the unit with 95A input for an hour and the transformer bearly warmed up. The inverter board had 1W of fan output blowing over it and warmed up 12C.. I did run the unit up to around 130A for a few seconds but it was on a 100A breaker so was playing a bit of breaker roulette... I have a 160A breaker wired up so after the chokes are re-wired I will then run a full test at 6kW for a few hours and then wind it up a bit... Big Grin

Unfortunately, for a cheap build the chokes are something that is looking likely will be a DIY component, unless you are happy to spend around 5x more on them.
Korishan likes this post
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
not sure you can resize the image with embed images. we just click on the image to make it full sized.
Proceed with caution. Knowledge is Power! Literally! Cool 
Knowledge is Power; Absolute Knowledge is Absolutely Shocking!
Certified 18650 Cell Reclamation Technician

Please come join in general chit-chat and randomness at Discord Chat (channels: general, 3d-printing, linux&coding, 18650, humor, ...)
(this chat is not directly affiliated with SecondLifeStorage; VALID email req'd)
Ok, so been trying a few more choke combinations (over 50) and logged quite a bit of data, which might be of general interest brecause it also relates to any H-bridge driven inverter.

The higher your battery voltage supplying the inverter the higher the idle losses will be. This might be relevant for those winter months when your trying to conserve power and do not have an excess because in those months it pays to have the batteries always sat at a lower state of charge. This reduces overall idle losse...

Being able to change the output voltage on the inverter also helps but the return may decrease significantly beyond a certain point and may cause issues if you drop the voltage to an excess (i.e. dropping to 190V on a nominal 230V system). The return may be subtle but can add up to quite a few watts over a winter period and many kWh over a 10 year timeframe.

Going back to the inductor/choke which is the "magic loop inductance", as per one Aliexpress explanatory diagram, not all chokes are identical nor deliver the ideal outcome. The choke has to be like the goldilocks story and just right for your combination of inverter board and transformer.

I had tried out well over 50 different combinations comparing various aspects, including the net effect on output voltage because this impacts the resultant voltage under load and the output behaviour when your battery gets to a low voltage and you have a high load on the inverter, where the output voltage starts to fall off a cliff. There are no outliers in this chart, the 4 outlying points just represent a different choke combination....

Adding a rough (ok very rough) trend line in accordance with the "magic loop" theory, something else starts to stand out... goldilocks...

Korishan likes this post
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
anxiously waiting for any update.
Just curious - in your testing (excellent graphs & testing there by the way :-) , is the inverter board varying the transformer drive with load change? Eg widening the PWM with the load or is the PWM "static"? (obviously it varies to make the AC cycle). Wondering if feedback = better voltage regulation with load?
Running off solar, DIY & electronics fan :-)
The inverter works by PWM modulation (varying pulse widths through the sine cycle). The input to the transformer is effectively a 48V square wave at 23.4kHz with varying pulse widths to provide the right energy level into the transformer and then the transformer porovides a sinewave out (smoothed via the 5uF capacitor on the output - also need to measure the resonance to check 5uF is the right ideal value, but it works ok at the moment..) The output voltage is then fed back into the controller on the inverter to regulate the PWM modulation to correct any needed voltage change. By what I can gather the regulation is carried out once per cycle as it is not actually monitored throughout the full waveform but that does not make much of an impact.

Volatage regulation is showing around 2V drop at 5kW output (at 2.3kW <1V) but not certain where the drop is occuring, so far I have only run it to 6.1kW input (128A) with about 5.3kW out, but that is with meter readings I would not rely on. I have not had chance to sort it out and partly waiting to see if there are any issues that occur, plus I need to add a couple more 10mm2 cables to prevent the cables from heating things up before going above 140A..... the breaker is 160A so 6kW output is about the 100% duty limit. Will be ordering another board before trying long loading at 8-9kW and switching the breaker for 200A.

The input chokes I ended up with are 4 E70 ferrites in 2 pairs with 10/11 turns each and a gap to give 120uH per inductor (60uH in parallel), although I'm not convinced they are ideal (too high uH and higher idle current as a result, separate story) as I need to put the scope on under load and a few other tests under load to see how they are working. They are still not glued so they buzz a little under load I can tell the difference between appliances from the sound. I ended up 3d printing a special custom former to wind the wire on before taking the former out and slotting the coils over the ferrite core and clamping them together. Winding could otherwise be a bit of a challenge.. will post some pics.

Low load levels (upto 4kW) are no issue at all, and runs cold, only starts to warm up around 4kW. Warms up about 6C with 2.3kW and a small 12V fan.

Been running it for the last month with a couple of extension leads back into the house as a temporary test. The microwave, 3kW kettle, toaster and whatever else gets plugged in is on one lead (limited to 3kW / 13A @ 235V) and then the other lead has the TV and a few other bits on, which add up to about 150W 24x7, with a electric iron or whatever plugged in at the time. The kettle only draws about 2.6kW with the 235V source and the volt drop on the extension lead...

Filters arrived, wired up, but yet to add them to the outut.... need to switch everything off to hook the board up. Ended up with 3 filters in parallel and will post a picture.

Conclusion so far... very, very happy I built this unit rather than spending over 2x as much on a pre-packed inverter. If it fails, I can fix it and with the lack of parts involved it is easier to repair. I can upgrade it when a larger inverter board is brought out if needed. The transformer is good for life, so a real investment still worth its weight in 25yrs time and can be used over and over.

I know I can run the unit with well over 5kW spikes and constantly at 5kW. Startup spikes from the lawn mower (>2.6kW) or the vaccume cleaner (3.2kW spike for 800W cleaner !) are no issue on top of whatever is plugged in at the time.
CarelHassink, Korishan, Hanssing And 1 others like this post
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
That's great work there :-) Don't stop!
Running off solar, DIY & electronics fan :-)
This is the progress (or lack of) to date...

Been running now for 761kWh.
Nominal output 234V

Input at times...

Inductors (2 x E70 cores) wound with 3 x 3 x 3 (1.4mm - need to check) wires on a 3D printed former and the the E70 cores are slotted in to place.. 10 turns, gap is around 1.5mm, was using sheets of paper as a rough proxy for 0.1mm... 15 sheets.

Three wires were twisted together (with a drill) and then three of these were used to wind the core over a 3D printed former, which can then be taken appart so only the core is left.

The 3D printed former is then held together with a single bolt.

The windings are then wound with an overlap (turn when they reach a side) so that they remain flat and equal length....

The extra (buy separate) display for the control boards in my instance was a bit useless, appart from the temperature.

Voltage reads 219V when it's 234V, current reads 6A when it's 8A...
Hz, should never change so just a confirmation the basic setup is not 60Hz...
Korishan and Wolf like this post
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
These are the filters I ended up with

Rated at 20A each and have 3 in parallel

Very much a work in progress (hence bare 234V terminals....) so not a good example to follow.

Still doing a bit of research to confirm these are ok in parallel as I'm not entirely convinced that when in parallel they do not alter the filter frequency range.

Separately as well, when using RCD / RBO / Residual current breakers the whole issue of earth loops and neutral return through an earth loop can occur if your wiring is not a nice standard house setup...

Shed 50A --- shed 50A --- 80A breaker --- Blocks --- Meter :::: common non RCD wiring
    |            |                           |
    |            |                      House boards (x2)
    |            |                       |  |  |  |  |
    |            |                       |  lights, etc...
    |            |                       |
  Earth        Earth               Shed 32A  

With my setup I intend to back feed the house from the shed and disconnect the mains (appart from a single winter charge circuit).

In my instance I have 2 x SWA cables down to the sheds and 3 separate additional earth spikes 2m into the ground.

For earth leakage trip devices I need to feed into the main (larger) cable rather than feeding back into a circuit in any for the distribution boards as this would make the earth leakage devices trip out of sequence, unintended or not at all for some. Will add a full circuit in a later post.

With my setup, on the DC side I have 3 x MPPT controllers pushing out just over 100A at 46-54V at times, along with what were 125A breakers (one at the battery end and the other at the remote end 2m away outside the shed). I like to add breakers at the source and destination ends of a wire...

Anyhow, this is the insde of the second of the failed breakers....

What I found out when stripping them down and having a closer look at the design there are a few things, which stand out a little, like the poor clearances for the circuit and bad contacts.

These overheated through running at high load, which through many metal expansion / contraction phases results in the terminal contact pressure reducing, increasing contact resistance and at 100A a little resistance can make a lot of heat. They ran fine for a while until the terminals worked loose and that caused the additional heating to cascade back into the breaker and overheat the whole lot.

Curiously they did not trip (as I thought they would on thermal overload) and one contact ended up open circuit.

The other breaker is still partly wired in and have not stripped that one down yet (it's a lot more burnt than this one) but neither failures resulted in any worrying flaming burn the place down failure scenario.
Korishan likes this post
If you can't quantify how much they cost, it's a deal, I'll buy 5 of them for 3 lumps of rocking horse ......
Thinking out loud here: with the EMI filters, are they seeing much of the 23kHz after the 5uF cap?
The filters probably have iron powder type core material for the inductors.
Would this be the least lossy at idle current if there's some 23kHz to damp out?
Would a variant of your e70 cores help on the output side maybe as a pre-filter to the CW4L2 units?
Also wondering if there are ways to assemble eg the 5uF output cap (or other caps) cap that might help efficiency, eg 5x smaller caps?
Running off solar, DIY & electronics fan :-)

Who read this thread?
100 User(s) read this thread:
hbpowerwall (07-25-2019, 05:05 AM), daromer (03-02-2019, 01:58 AM), Sean (06-07-2019, 01:29 PM), mike (06-07-2019, 04:26 PM), station240 (04-09-2019, 03:45 PM), Joaquinsfy (02-26-2019, 01:42 PM), Korishan (06-08-2019, 09:04 PM), A + A (05-06-2019, 10:07 PM), PAF (06-08-2019, 02:50 PM), Mike C (05-08-2019, 05:40 AM), jdeadman (07-12-2019, 03:50 AM), wim (06-08-2019, 06:44 AM), watts-on (06-08-2019, 12:10 PM), jesusangel (05-06-2019, 12:10 PM), Hanssing (06-09-2019, 07:06 AM), Stefaan De Ridder (06-08-2019, 02:17 PM), Chiptosser (05-09-2019, 01:27 AM), KtB (06-08-2019, 06:10 PM), Mazlem (02-23-2019, 10:54 PM), Geek (07-31-2019, 02:23 AM), harrispp (04-18-2019, 10:46 PM), Jon (04-05-2019, 07:58 AM), Ivo Staelens (03-31-2019, 07:48 PM), floydR (03-08-2019, 03:55 AM), DarkRaven (02-24-2019, 12:37 AM), Swake (05-09-2019, 07:42 PM), Grumplestiltskin (06-29-2019, 11:11 PM), SemaJG (03-08-2019, 10:40 AM), srk (05-27-2019, 01:06 PM), emerca (05-17-2019, 03:49 AM), rev0 (07-10-2019, 03:17 PM), setite (05-23-2019, 01:36 AM), CarelHassink (06-09-2019, 04:48 AM), AndyMc1970 (07-06-2019, 11:55 AM), bogptrsn (03-11-2019, 10:20 PM), karmi27 (03-05-2019, 09:25 AM), bigblue (03-13-2019, 05:08 PM), Charly144 (05-07-2019, 08:54 AM), BaronVonChickenPants (06-06-2019, 09:30 PM), spinningmagnets (05-06-2019, 11:10 PM), drchips (06-07-2019, 02:54 PM), HughF (07-08-2019, 12:56 PM), Ahmed_Ali (07-16-2019, 08:04 PM), gpn (02-27-2019, 12:33 AM), kanchana (05-30-2019, 05:50 PM), babish (03-21-2019, 01:39 PM), Dala (02-27-2019, 11:11 AM), Crimp Daddy (03-08-2019, 05:00 AM), Redpacket (06-08-2019, 10:19 AM), completelycharged (07-30-2019, 08:31 PM), CAB14s2p (05-11-2019, 05:33 AM), Zvonko000 (04-09-2019, 02:53 PM), Oliman (06-08-2019, 05:32 AM), sniperdude24 (04-10-2019, 04:33 PM), whisperbot (06-30-2019, 07:27 AM), farmerjohn (07-20-2019, 07:50 PM), cadric (07-07-2019, 12:53 PM), choncy12 (02-27-2019, 12:00 PM), Bubba (06-10-2019, 12:37 AM), JDish (05-14-2019, 09:38 AM), Cherry67 (05-04-2019, 01:11 PM), CBlack (07-10-2019, 03:19 PM), Walde (07-06-2019, 04:31 PM), Riplash (06-09-2019, 01:55 AM), Angelo (03-30-2019, 04:59 PM), 100fuegos (05-12-2019, 09:36 PM), Jim Jr. (06-07-2019, 12:45 PM), (02-24-2019, 09:52 AM), Katarn_89 (05-07-2019, 11:39 PM), albach (06-25-2019, 12:21 PM), ungp (02-27-2019, 06:47 PM), w0067814 (04-05-2019, 07:27 AM), Gummiadler (03-08-2019, 11:23 PM), HikingGuy42 (06-07-2019, 03:28 PM), michaell (05-06-2019, 07:23 PM), Sammm (03-10-2019, 07:31 AM), Dragora (03-18-2019, 07:13 AM), chuckp (03-10-2019, 07:13 PM), MGPTWEGB (04-12-2019, 06:15 PM), emuland-metroman (05-04-2019, 04:44 PM), hazSolar (02-24-2019, 01:40 PM), Klug (07-01-2019, 03:13 PM), Chass90 (05-29-2019, 10:29 AM), Beholder (02-25-2019, 07:05 PM), Truth (04-08-2019, 06:20 AM), jaeger-audiodesign (05-04-2019, 07:31 PM), Leipi35 (06-09-2019, 10:12 PM), runsnbunsn (02-24-2019, 11:23 PM), yuri (03-27-2019, 10:26 PM), cowpen (06-16-2019, 11:10 PM), singlecell (03-11-2019, 11:20 AM), sl1me (05-05-2019, 02:29 AM), Wolf (06-08-2019, 01:00 PM), Bastler_Ro (06-06-2019, 06:17 PM), iscience (03-09-2019, 01:02 PM), hyperborean1 (02-26-2019, 04:58 AM), camthecam (06-11-2019, 06:23 AM), Generic (06-08-2019, 01:43 PM), stuboy (03-19-2019, 10:40 PM), rettore (03-22-2019, 06:44 PM)

Forum Jump:

Users browsing this thread: 1 Guest(s)