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: 10:26am 08 Aug 2018
Copy link to clipboard
Print this post
Now the test results for those who might make an E core from the two smaller Aerosharp choke C cores.
The picture shows this was built the same way as the large one described above. The coil shown has 9 turns only of 45mm sq. I will try later if more turns (possibly 13) could fit if cross section is sacrificed.
The two half dowel sections shown were used to wedge the rounded ends of the coil in place. The coil must be locket firmly in place but, for experimenting, epoxying it in solid would make it no longer possible to change the gap spacers or to disassemble the lot.
For this test I only made 3 different spacers and I could not test the 'no gap' part as my test coil was just too wide for that.
Small E core made from 2 sets of small Aerosharp C cores. Core cross section 26x35mm = 910mm sq Available coil area 15 x 56mm
I think I mentioned elsewhere that the the 9 turn coil I have for this, with a 2mm gap, only yielded 43.5uH. With more turns and a thinner gap this size core might be suitable for a smaller inverter, perhaps in the <5KW range. Klaus
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 10:45am 08 Aug 2018
Copy link to clipboard
Print this post
The start is not a problem, and neither is joining lengths together, if you do it at the side of the winding where it bulges outwards to get around to the other slot.
I do a lot of foil windings, and here is a demo of a silly one turn winding with crazy thick wire. Subsequent turns just go over the bump at the inner start of the winding. The sides can stick out a fair way without being a problem.
When you do the next saturation test, can you record the dc voltage, and microseconds from zero current up to the peak, and the number of turns. From that I can work out a few things.Edited by Warpspeed 2018-08-09Cheers, Tony.
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 12:08pm 08 Aug 2018
Copy link to clipboard
Print this post
I can do that. The DC voltage was close to 40V, I remember watching that as my 2 big capacitors are only 40v rating. There might be a chance to do a test tomorrow. My test coil for the big E core choke has 125 turns. Now you got me wondering what the "few things" are
I determine saturation as soon as the sharp peak emerges from the apex of the triangular wave shape. I think you said so some time ago when we discussed building that saturation tester. Saturation is quite dramatic, just a small voltage increase shoots the peak way up. The noise from the coil also changes then.Klaus
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 10:42pm 08 Aug 2018
Copy link to clipboard
Print this post
With the good steel, saturation is fairly dramatic you get a dead straight incline, then it suddenly shoots straight up like rhino horn. No mistaking where it is.
We want to keep that saturation point well above any maximum current surges the inverter may ever see.
If we know the dc volts and microseconds slope, that tells us the inductance with 125 turns, and of course the ampere turns to reach saturation. Microhenries = applied dc voltage x microseconds, divided by current increase in amps
If we know the inductance with 125 turns we can then work out the theoretical inductance for only one turn on that core, with that gap. Simply divide by 125, and then divide again by 125 to get the "Al" value usually expressed in nanohenries for one turn only.
From that we know that inductance equals turns squared, so we can work out what inductance, and what saturation current any number of turns would produce on that core with that gap. If you can supply the numbers I will show how this works.Cheers, Tony.
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 08:24am 09 Aug 2018
Copy link to clipboard
Print this post
I see warpspeed has his engineer hat on again so I hasten to provide the numbers.
Firstly I must admit that I had forgotten about half turns are doable with E cores . So my test coil actually has 125.5 turns since the wire ends emerge from opposite sides of the coil. The gap for the tests, pictured below, was 1.5mm I used two large (33,000uF/40V caps in parallel. They have a 23.6A ripple rating each.
The first picture shows just before saturation occurred, the voltage to the caps was 39.5V. Rise time for the trace was 5mS.
My Hall sensor gives 40mV per Amp so the 600mV shown correspond to 15 Amps with the 125.5 turn coil.
The second picture shows after saturation has occurred, the voltage at the caps was 47.3V.
If you would like to do that test differently just let me know.
Today I managed to wind on the first layer of my new choke coil. It could accommodate 18.5 turns. Wound 2 in hand at once of 1.8mm dia wire. There should be room for 10 layers of that, depending how neatly they stack on top of each other. Klaus
zaphod Regular Member Joined: 03/06/2018 Location: United KingdomPosts: 93
Posted: 05:50pm 09 Aug 2018
Copy link to clipboard
Print this post
Just in case anybody wonders why people are crazy enough to use ferrite it's down to core loss that is a function of frequency and flux swing. At higher frequencies the core losses of steel become progresivly untenable hence the use of ferrite in high frequency power transformers. This application as a choke (as I understand it) between a high frequency pwm that is modulated to emulate 50hZ is an interesting mixture and I would be very interested in peoples experiences of core temperature rise. I don't know your switching frequency nor ripple current at that frequency but I am suprised to see these referred to as DC chokes but then I have no experience of your inverter design so please excuse my ignorance.Cheers Roger 1Kwp DIY PV + Woodburner + Rainwater scavanger :)
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 10:34pm 09 Aug 2018
Copy link to clipboard
Print this post
Roger,
This surprises a great many people, but you CAN use laminated steel cores and very thick wire sections at (relatively) high switching frequencies, but only if the high frequency ripple current component is kept very low.
Eddy current loss, and hysteresis losses in the core, and skin effect in the wire are all produced from magnetic effects, which in turn are produced only by the ac ripple current component.
If the inductive reactance is kept high enough, the ac ripple current component can be made very low compared to the huge 50Hz (almost dc) component.
Anyhow, a few watts, or even a very few tens of watts of loss in a multi kilowatt inverter are not really all that significant.
If you forget about trying to make something really small that can fit onto a circuit board, a big ugly steel choke can be made large enough and have enough inductance to reduce the ripple current far enough, so that the steel in the choke, and the steel in the transformer are both happy.
We have to get rid of all the high frequency PWM component anyway, and a choke with significant inductance goes a long way towards that. The only down side of using a high inductance choke in a buck regulator is poor transient response, which is why most switching power supplies have minimal choke inductance. A sine wave inverter is a very different beast to a fast responding switching power supply, and the choke requirements are also rather different. Cheers, Tony.
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 11:00pm 09 Aug 2018
Copy link to clipboard
Print this post
Klaus, O/k we have lift off....
Inductance in microhenries = 39.5v x 5,000uS divided by 15 amps peak = 13,167uH
Now our test winding has 125.5 turns, and inductance is proportional to turns squared.
So if we divide 13,167uH by 125.5 = 104.9uH and if we again divide by 125.5, we get 0.83598 microhenries for one theoretical turn.
We also know that it saturates at 15 amps x 125.5 turns = 1882.5 ampere turns.
From that we can now work out what we could have for ANY number of turns on that core, with that gap, as it now is.
If we have already decided that 18.5 turns will both fit physically, and carry the required current without overheating, we can work out what 18.5 turns can do.
One turn will give us 0.836 microhenries. We multiply that by turns squared. 0.836 x 18.5 x 18.5 = 286 microhenries.
Saturation occurs at 1882.5 ampere turns, so if we divide that by 18.5 we get 101.75 amps.
We have more inductance than we probably need, but its going to saturate a bit early.
So we rinse and repeat. Perhaps increasing the air gap from 1.5mm to 2mm and go through the whole procedure again, starting from the beginning.
Cheers, Tony.
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 08:34am 10 Aug 2018
Copy link to clipboard
Print this post
Thanks Tony, I printed your post and will warm up my calculator shortly. Had planned to use a 1.9mm gap but could also use the 0.9 + 1.2mm spacers for 2.1mm gap.
I should be busy for a few days winding one layer a day...
It will also be interesting to measure my finished choke and see how the calculated uH compare to the measured.Klaus
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 09:12am 10 Aug 2018
Copy link to clipboard
Print this post
The measured and calculated inductances should be quite close, the main source of error will probably be eyeballing the microseconds measurement off the CRO screen.
If you already have a 1.9mm spacer, try that first. Its an iterative process anyway, about the third guess should put you just about where you want to be.
When doing all this, always aim to have the maximum number of turns that will fit the core, and increase the air gap to reduce the inductance. You will always end up with "more" doing that, than by removing turns.Cheers, Tony.
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 10:45am 30 Aug 2018
Copy link to clipboard
Print this post
So this is the latest of my choke creations, 1.6kg of copper, 18 1/2 turns of 45mm sq wire. I ended up using a 2.4mm gap to get the peak saturation near my planned 200A range. Saturation is at 196.7A peak, giving it 3639.5 ampere turn. The inductance by warp's method calculated at 193 uH, my LC meter shower about 20uH less. I suppose this is because of it using a different frequency for the reading.
Tested it in the inverter today after first subjecting it to 125A AC from my stick welder to see if it got warm - no chance.
With a 2.5KW load (fridge & fan heater) the sine wave was as good as I hoped, better than any I had before. Perhaps tuning the secondary capacitor helped there too.
Since that I made a similar choke but using the smaller Aerosharp C cores, its for use in my 3Kw inverter. This one has 14 turns of 30mmsq wire. A 1.2mm gap saturated it at 160.7A peak, 2250 ampere turn. Inductance is 106.85uH calculated and 10uh less measured. Still well within Warpspeed's 100uH suggestion.
I think I have pushed the limits with the wire x section/turns ratio but it was worth it There are two spare coils to suit the smaller C core assembly. 9 turns only but of 40mmsq total. And one spare to suit the larger C cores, 18.5 turns of 40mm sq.
Anybody interested?Klaus
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 10:07pm 30 Aug 2018
Copy link to clipboard
Print this post
Great effort Klaus, that waveform looks perfect.
And I bet the idling current is pretty reasonable too.
[quote]The inductance by warp's method calculated at 193 uH, my LC meter shower about 20uH less. I suppose this is because of it using a different frequency for the reading.[/quote]
Chokes are notorious things for measuring any precise figure for inductance.
The problem is that the "BH" curve is very non linear. That is the B component, magneto motive force, versus the H component, actual magnetic flux produced is highly non linear. If plotted in both directions it looks something like a gum leaf.
Measured inductance of any choke that has anything other than an air core will be different with both changes of any applied dc bias and the amplitude of the applied ac component. The slope continuously changes, and also how much of the curve of the slope in use also effects the measured inductance.
A laboratory inductance measuring instrument probably works by applying some small ac voltage and coming up with some kind of inductance reading. If we test at a much higher applied ac voltage, the reading can be very different because we will be using a much wider part of the "S" shaped curve.
Now if we also add some dc bias to the core material, that will also change things. In fact if we bias it right up fully to magnetic saturation in one direction, most of the inductance will completely disappear.
So any estimates of what any choke with a lot of dc bias current is going to finally do will be a bit of a stab in the dark. We can get pretty close by testing under dynamic conditions as close as possible to how the choke will eventually be run, and then work from those figures.
If you planned to use your new choke as an inductor, maybe as part of a tuned circuit, filter, or something like that without any dc current flow, your inductance measuring instrument should come very close.
But if using that choke with maybe 100 amps dc flowing through it, and very little actual ac, your inductor tester will come up with a different but much more realistic figure. So don't be too concerned about getting different inductance readings from different testing methods. They are probably both right.
You will see this same effect if you measure the magnetizing current of a transformer at different applied voltages. Its far from linear, and its the inductance that is changing. The ohmic resistance certainly cannot change.
If the current in the choke can ramp up cleanly to something much higher than the choke will ever see in actual operation, without saturating, it will be fine. And the slope of that ramp will be a much more realistic measurement of true dynamic inductance under heavy dc bias than just measuring the inductance with a wimpy low power laboratory inductance bridge.
So if the numbers differ by 10% or 30% with different measurement methods that is not at all unusual. We do not need a precise choke inductance value anyway. As long as its big enough to do the job easily, without falling flat on its face, we are winning. More is always better for both inductance and saturation limit.Edited by Warpspeed 2018-09-01Cheers, Tony.
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 09:43am 31 Aug 2018
Copy link to clipboard
Print this post
Thanks Tony, I'm sure its "big enough" .
Had it powering over 4KW today, a bit of hum but all was fine.
Then I tried the same load on my 3KW (single aerosharp toroid) inverter which has the smaller choke which I describer above, in it. This one "did fall flat on its face", kept dropping power & re starting (but nothing blew up)with that load. It ran fine with a 3KW load. Its not the over current shut down, as that would lock off.
I'm not worry about that as that inverter is my spare unit. But I'm still curious by what mechanism it could drop power (voltage?) and immediately power up again then repeat that trick. The wiring is certainly up to that load and more. Klaus
Revlac Guru Joined: 31/12/2016 Location: AustraliaPosts: 1026
Posted: 10:10am 31 Aug 2018
Copy link to clipboard
Print this post
Tinker that sine wave is perfect, the bit extra work tuning everything up looks worth the effort. Cheers Aaron Off The Grid
Warpspeed Guru Joined: 09/08/2007 Location: AustraliaPosts: 4406
Posted: 10:33am 31 Aug 2018
Copy link to clipboard
Print this post
Klaus, Don't really know why it shuts down and restarts, some features in the software I suppose. There must be something happening that the software is not too happy about causing it to shut down and then restart.
Aaron, Getting the magnetics working properly is probably the least understood part of building these PWM inverters. The higher the power level the less "sins" you can get away with without there being consequences.
Klaus has pioneered the way, and we could do far worse than follow in his footsteps.
Cheers, Tony.
tinyt Guru Joined: 12/11/2017 Location: United StatesPosts: 438
Posted: 02:17pm 31 Aug 2018
Copy link to clipboard
Print this post
Klaus, Looking at your latest choke creation shows that you are an artist and perfectionist as well.
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 10:16am 01 Sep 2018
Copy link to clipboard
Print this post
Thanks, I am not so sure about the "artist" but have been accused of the second character.
I had a big stack of 600-900gram reels of recycled 1.8mm enamel wire, winding these chokes was a great use for it as only short lengths were required. Winding two in hand simultaneously might be better than the single bigger diameter wire you used. The 2x1.8mm pitch also corresponded closely to a screw feed setting on my lathe so all I had to do was to keep on the wire tension and turn the jack (with a lever). The large pitch also negated turns counting after the first layer, it is physically not possible to squeeze in another turn by accident. Each layer was fully epoxied all around and let go off before the next. That way the whole coil became fairly solid when the time cam to press out the center.
Today I have taken apart my old Powerjack inverter conversion - it has been sitting here un-used since the big inverter came on line. The plan is to replace the ferrite choke in it with one of my later creations, tune that big toroid secondary and fit bigger fans. Luckily I have a spare case for it to do all the different metal work on. Klaus
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 10:29am 01 Sep 2018
Copy link to clipboard
Print this post
No doubt this will bug me enough eventually to search for a reason. Perhaps I could get Poida interested? This kind of phenomenon is something he's much better qualified to investigate .
Different loads do seem to influence the inverter's wave form. The nearly perfect one I posted above is seen almost all the time with my normal house loads. But when the compressor of the spare fridge (which is close to the inverter) runs I see a very noticeable 'wriggle' at the bottom of the sine wave. This goes away once that compressor stops.
So I could have lots more 'fun' testing each inductive device at my place to see if and how the sine wave is influenced.
Never a dull moment here Klaus
Madness Guru Joined: 08/10/2011 Location: AustraliaPosts: 2498
Posted: 12:13pm 01 Sep 2018
Copy link to clipboard
Print this post
I get this same shutdown happening occasionally and I see Mark has mentioned it also. It was also mentioned by Oztules when fixing Marks boards from Clockman and was put down to long wires to the switch connected to pin 6. My single core inverter was doing it under heavy load and replacing the wires to the switch with shielded wires fixed it. The inverter turns off and instantly does a soft start, I have seen it many times and it has never caused a problem with the inverter.
Another situation that I have had cause it with my big inverter was when I was doing some welding last week with my MIG operating at around 120 amps. The one case where it happens that really has me scratching my head and frustrates me is using my charge regulator to send a PWM signal to a SSR connected to my HWS. Most of the time when there is excess power the PWM works very well to divert power to make use of it to heat hot water. However when the inverter is not doing much more than providing 50 hertz to keep the GTIs happy it will do this restart also. ATM I am using a different method that turns on the hot water when the regulator pwm to the GTIs drops to 70% it turns on the HWS and turns it off if the battery volts drop 1 volt below the target voltage. This method does not cause any issue but I like the other method much better apart from the shut downs.
Things I have thought of trying is 1 remove the 10K resistor that holds pin 6 to ground if there is no switch. 2 Connect 5 volts to pin 6 from the PCB and use the overtemp to shut down as done previously. 3 Put a metal shield connected to ground under the PCB.
I have not had a chance to try any of these as yet. There are only 10 types of people in the world: those who understand binary, and those who don't.
Tinker Guru Joined: 07/11/2007 Location: AustraliaPosts: 1904
Posted: 01:16pm 01 Sep 2018
Copy link to clipboard
Print this post
Thanks Gary, we will get to the bottom of it eventually.
That shielded wire you mention got me thinking, my big double stack inverter uses a shielded wire and I can't remember that one doing the restart thing when it is powering things. It did do it a few times when it was just idling of the battery while the house was powered from the other, single toroid, inverter and I overloaded that one with the chop saw testing. The single toroid inverter does *not* have a shielded wire to the switch.
While I had my double stack inverter apart to replace its choke I ran the house on the single stack inverter. Had a few temporary blackouts when the induction stove cut in, dropping its power level settled it down again and nothing blew up.
So I got another job now, shield the wires and the switch panel on top of the inverter (which is located above the toroid & choke).
1. I would not be game to remove that 10K resistor at pin 6 - it stops things going 'bang'.
2. I use the overtemp to control one fan, the other fan has its own independent temp sensor. That works so well I'd prefer to leave that alone. Always thought oztules method to use it for shut down was 'odd'.
3. My control board has a complete ground plane on both sides. Its also about as far as possible located away from the toroid & high current wires.
So I think we should start with the shielding of sensitive parts. My control board is small enough to even fit into a metal box if I have to. BTW, I have a spare one if you like to play with it - the gate connections are separate though, no ribbon cable. I think there is a pic of it at my last (experimental) inverter build blog.
The other thing I will do is using the little digital oscilloscope to monitor the sine wave as I operate various items at my place. For example, tonight I happened to be at the inverter while the microwave was defrosting something. That one *really* distorts the sine wave badly at the peak and at zero crossing. Then later, when the induction stove was running, the distortion was hardly noticeable but the AC voltage shot up by 5V when the power (500W) cycled on and dropped back when off.
It ran fine though with the microwave, no performance loss, just horrible sine wave.
Mark would say leave it alone while its working , but I can't do that , have to know why it is so.Klaus
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
.
