Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 01:01pm 03 May 2018
Copy link to clipboard
Print this post
The ferrite choke with 3 or 4 turns of heavy wire does a great job of reducing the standby current of the inverter.
But it does nothing about the primary current limiting that ordinary chokes usually are used for.
The Aerosharp inverter has two smaller and one larger C core non saturating choke in in which can be re wound for the higher currents our inverters run at.
I did convert both of these chokes but found two of the smaller ones assembled into an E core did make a very useful non saturating choke without becoming too big physically.
Using the two cores side by side allows for a bigger cross section coil which is much easier to wind.
First job is to remove the core halves undamaged from the original. The core material is quite fragile so great care is required. I started by removing the stubborn screws bolting the endplates together. The nuts must be heated quite hot - I used a small blow torch - and then its easy to unscrew them.
Next, using a soft faced hammer and holding the choke in the hand tap one protruding core half side on. Repeat with other side and other core end until it is loose and can be pulled out of the winding. The cores have a 2.5mm plastic 'air gap' spacer so they separate relatively easily.
My cores had matching marks (faint white lines) on one side, these aid in re assembling them the same way. The mating surfaces need to be cleaned carefully, also the core part that sits inside the coil.
Then they were epoxied onto a plywood base with hardwood guide strips on each side.
Make sure the mating surfaces line up flat & clamp them lightly that way until the epoxy hardens. There should be 2 identical E cores with the matching lines aligning.
Now a winding jig needs to be made. I wound my coils 3 in hand simultaneously, this requires a sturdy jig. Most importantly, the wire end needs to be able to be clamped very well or they will pull out.
To stop the jig slipping in the lathe chuck I used a bolt head for the chuck jaws to clamp on.
To clamp the 3 wires with some friction in the tool post I used this:
To start, the wire length for 9 turns plus the ends is calculated and three pieces prepared thus:
Soldering them together greatly improves the clamping in the jig. This also readies this coil end for the terminal. Keep in mind that each successive layer requires the 3 wires to be slightly longer. Easily found out by how much after the second layer.
Winding is done by turning the chuck by hand and using the nearest pitch of the threading screw to advance the wire to the next turn.
The friction of the threading gears prevent the chuck from backlash so the wining can be stopped at any time without the wire unwinding. I managed to fit 9 turns (3 in hand) on one layer. These now need to be epoxied in place before trimming the toolpost clamped ends. Do place some plastic under the coil to prevent epoxy messing up the lathe bed .
Only one layer is done at a time, there is no way the wire ends can be held securely unless fixing them with epoxy, so its a one layer a day job.
Before the next layer is added there should be some wooden wedges placed at the beginning and end of the 3 in hand winding to close the gap between the outer wire and the jig cheek. If this is not done the next layer *will* slip off the top there and result in a messy layer, multiplied at each following layer. So, each layer is wound exactly on top of the one below - quite tricky unless the wedges are good.
Here is the finished product and the jig:
A few tips: take care that the lot is not epoxied to the jig! It should slide out without using force. For that the jig was waxed very well and a plastic film layer placed under the first turn. The windings end up quite tight - so they should be - and getting them off the jig core can be tricky. Patience and well hardened epoxy is the ticket here.
To assemble I cut 2mm thick 'air gap' spacers from some acrylic I had. These when tested with my core saturation rig and a 150 turn test coil (to keep the current low) showed this core would saturate above a 12 KW load, giving a generous marging for normal operation.
Lastly a picture of the various chokes I made & tried. The ones in the lower right corner are the ones used.
Klaus
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 06:45pm 03 May 2018
Copy link to clipboard
Print this post
Excellent description Klaus.
I have wound a few similar chokes over the years for various projects, and can confirm that a bit of work making a massively strong steel winding jig is well worth the effort involved.
If you have a four jaw chuck, I find it easier to use a rectangular solid block of steel as the main supporting core, then build that up to the exact rectangular dimensions required with some flat metal packing.
I suppose it could also be done held stationary in a basic workshop vise if the required turns are few. But before you begin you absolutely must provide a very robust method of securing both ends of the winding or it will just unravel into a tangled mess when the winding tension is released.Cheers, Tony.
renewableMark Guru Joined: 09/12/2017 Location: AustraliaPosts: 1678
Posted: 10:26pm 03 May 2018
Copy link to clipboard
Print this post
Thanks Klaus, just curious why you didn't use two of the bigger ones? Just conscious of size and fitting everything in the case?
That leads to another question, can you have a choke that is too big?Edited by renewableMark 2018-05-05Cheers Caveman Mark Off grid eastern Melb
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 09:31am 04 May 2018
Copy link to clipboard
Print this post
Well, I did make a choke from two of the bigger ones - the one with 18 turns...
When testing, the smaller one gave me less sine wave distortion so that was chosen.
And, it fits easier too .
About fitting a choke that's too big, perhaps warpspeed is better qualified to answer that.Klaus
renewableMark Guru Joined: 09/12/2017 Location: AustraliaPosts: 1678
Posted: 11:12am 04 May 2018
Copy link to clipboard
Print this post
You answered that yourself, no technical explanation, but it would appear the smaller one matched your machine.
So basically the choke is A designed to lower the no load draw, thus reducing power drains overall.
B creating a nicer sine wave with lower distortion.
A is very to mildly important depending on power production and storage volume.
B well, how important is that considering how most appliances operate fine off messy mains power.
If a choke is operating before saturation at say 3kw load does it make the inverter more efficient or just make the sine cleaner and just look nice?Cheers Caveman Mark Off grid eastern Melb
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 12:55pm 04 May 2018
Copy link to clipboard
Print this post
Mark, I like to do things differently and what I build I do the best way I can possibly do it with what tools/materials are at hand. That other choke was a challenge I could not resist...
You will need the ferrite (4 turn) choke on your inverter. The other one is a bit like the icing on the cake, the inverter works fine without it.
What happens when a choke saturates is, it no longer restricts the max current that could flow. The ferrite choke saturates *all* the time when power is drawn from the inverter.
Look at the non saturating choke as a kind of electrical safety throttle. It just does not 'trip' like a C/B or fuse when a power spike occurs.
I did observe my different chokes having different influences on the sine wave form but why that is so someone better qualified could perhaps answer you here.Klaus
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 11:22pm 04 May 2018
Copy link to clipboard
Print this post
The choke is there to turn the high frequency PWM into a low frequency 50Hz sine wave.
It does that by storing and releasing energy to average out the fast and very violent on/off high frequency PWM square wave switching, and turn it into a nice slowly varying sine wave with very limited high frequency ripple current.
Rather like the springs on your car. The wheel can bounce up and down fairly fast with ripples in the road, but that is not transferred to the body. The springs store and release high frequency energy, and isolate the body from the nasty high frequency shocks and vibrations.
De we need a choke at all? Do we need springs on a vehicle?
If we couple our mosfet bridge straight into a large transformer (that also typically has a very low self resonant frequency) the transformer will look like a big capacitor to any frequencies above the self resonant frequency. It all becomes vastly worse if the transformer also has a high voltage step up ratio.
Every time the mosfet bridge switches, it has to almost instantly charge or discharge the very high capacitance of the transformer secondary winding.
That causes massive short peak surges of currents in the bridge mosfets every time it switches. Imagine the stress in the suspension of your car with solid steel wheels and no springs. If our transformer step up ratio is 9:1 the distributed capacitance of the secondary appears 81 times larger in the primary than it really is in the secondary. And we have either a full charge or a discharge occurring twice every 23.5Khz cycle.
Everything runs hot, and the idling current will be higher than it should be.
We desperately need to fit a series choke to effectively isolate the transformer at 23.5 Khz, but allow through all the 50Hz energy with negligible loss.
There are some other bad effects of having no choke, quite apart from the extreme transformer capacitance loading problem at the switching frequency.
With no choke, the transformer primary sees raw PWM at 23.5Khz, and that produces something called skin effect in the wire. Current only flows on the outer most surface of the primary. The main bulk of the wire sees almost no current below a depth of only about 1 to 2mm. So the primary runs hot, ESPECIALLY at very high power, not so much at idling power.
That skin effect can be mitigated by using a large bunch of individual wires in the primary no bigger than about 2mm diameter, instead of one solid stranded cable, but that is only a band aid solution to the real underlying problem. A series choke solves that underlying problem at the source.
The laminated steel in the transformer core will also see circulating eddy currents at 23.5Khz, so the core runs hotter than it should. Its all very bad, but an inverter without a choke will certainly appear to run and work fine. But it will run hot and always be much more highly stressed for no useful purpose.
A small ferrite choke will work, but only at low power. If you are really serious about high power, fit a non saturating choke and your transformer will run slightly cooler, and your mosfets will run at lower peak currents.
Can a choke be made too large ? Not really, but a very large choke can introduce indirectly some other effects.
The choke effectively decouples the bridge from the transformer at 23.5Khz, which is what we want. But as we make the choke increasingly larger, the decoupling effect also gets larger, and starts to have some effect at the lower frequencies.
If we have the usual expected transformer self resonance in the low Khz range, that can be excited and introduce stationary wobbles and faint ripples into the output, especially at very light loads. This can be especially noticeable where sometimes there is a kink just after the zero crossings when using unipolar PWM.
The high frequency wobbles should go away with heavier inverter loading, but the zero crossing kink may remain.
The kink should go away completely and the wobbles much less with bipolar PWM drive. The reason being that with bipolar drive there should be very low harmonic energy at multiples of 50Hz to excite transformer resonance.
Unipolar drive switches one side of the transformer hard at 50Hz, and that square wave will have odd harmonics that may tickle the transformer into wobbling or kink action.
Its not the choke causing all the wobbles, although a larger choke will probably make the wobbles worse, but the transformer resonances. We can overcome that paradoxically by lowering the transformer resonant frequency with some shunt capacitance. A suitably large choke allows us to do that.
With no choke, any extra added shunt capacitance would probably be death to our mosfet bridge from huge current spikes. But with a non saturating choke, and suitable capacitance across the transformer we can design everything to have a pretty good overall result.
All the high voltage grid tie inverters use a non saturating steel cored choke in the primary, plus a capacitor of several uF across the primary to lower the transformer resonance and further attenuate any residual 23.5Khz ripple.
That is just not practical for a low dc voltage off grid inverter transformer. The capacitor across the primary would need to be very large and have an impossibly high ripple current rating. So we place the shunt capacitor across the 230v secondary where it can have 81 times less capacitance (with a 9:1 transformer ratio) but still work pretty much the same as if it were a much larger capacitor placed in the primary.
There is no magic in any of this, it just requires a slightly deeper understanding of what is really going on. At very low power levels you can get away with far more design "sins". One you get up into the multi kilowatt region, it all becomes a lot more important if you wish to build an efficient and reliable inverter.
Cheers, Tony.
Clockmanfr Guru Joined: 23/10/2015 Location: FrancePosts: 429
Posted: 06:41am 05 May 2018
Copy link to clipboard
Print this post
"There is no magic in any of this"
Yet again Tony another excellent explanation.
Looks like on my OzInverter re-write, your quotes will get a few pages.
Thanks again.
Everything is possible, just give me time.
3 HughP's 3.7m Wind T's (14 years). 5kW PV on 3 Trackers, (10 yrs). 21kW PV AC coupled SH GTI's. OzInverter created Grid. 1300ah 48v.
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 07:32am 05 May 2018
Copy link to clipboard
Print this post
You are most welcome Clockman.
The main reason I am developing my own rather radical (and obscure) stepped sine wave inverter design is that it contains a great deal of very original thinking, and its totally different to the conventional PWM way of doing things.
There are some really interesting challenges that I have gradually overcome over time, with some very satisfying results so far. A great way to spend retirement.Cheers, Tony.
Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.
Print this post
.
.