 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. |
Forum Index : Electronics : Inverter Choke Design Discussion
Page 1 of 4   |
|||||
| Author | Message | ||||
wiseguy Guru  Joined: 21/06/2018 Location: AustraliaPosts: 998 |
I have been thinking a bit more about the choke in a Nano / EG8010 style inverter. For simplicity, lets consider them to be both 20kHz fundamental. The supply is 50V nominal (make the sums easy) output is 230VAC @ 6kW. For round figures again call the primary AC 30VRMS & no losses so efficiency is ~ 100%. I need a sanity check for the ball park values I have come up with. The 50V source will supply ~ 120A The primary will have the equivalent of 6000W/30V or 200A RMS flowing through it. The peak 50Hz inductor current will be 200A x 1.414 or ~ 282A. But the choke has the HF pwm switching current and the 50Hz current coincident. If the 50 Hz current is 282A, then if we allow a P-P HF ripple current of say 30% P-P then the HF current will be ~15% higher than this or around 325A at the sine peak. Can someone either agree or tell me where this is wrong - it was a (big) tad higher than I expected and I need to know the real value to help with the choke design using the sendust type ring cores I am working on. I hope this does not give anyone a headache on Australia day....... Edited 2024-01-26 17:27 by wiseguy If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
phil99 Guru Joined: 11/02/2018 Location: AustraliaPosts: 1790 |
It is! I think (perhaps wrongly) the key is the reactance of the choke at 20kHz, assuming the reactance of the primary, at 20kHz is much lower. It may in fact be significant. If the choke reactance is 5Ω at 20kHz the p-p ripple will be just 10A. The 50Hz reactance of that choke will be about 5Ω x 50Hz / 20000Hz = 12.5mΩ 12.5mΩ @ 282A ≈ 3.5V lost. Is this too high? If the choke reactance is 1Ω at 20kHz the ripple, will be 50A. for a total of 282 + 50 / 2 = 332A. The 50Hz reactance of the choke will be about 1Ω x 50Hz / 20000Hz = 2.5mΩ 2.5mΩ @ 282A ≈ 0.7V lost. So it's a trade-off of volts lost versus peak amps. Edited 2024-01-26 18:08 by phil99 |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 998 |
By my understanding reactance only applied to sinewaves. I see the choke as an inductance with very small resistive component. The inverter applies a square-wave to the choke resulting in a triangular current wave, providing it doesn't saturate. The triangular current ramp causes storage and release of energy from the inductor so there is a type of peak to peak "impedance" that is constantly varying. By my calculations the p-p current for a 16uH choke was 310 - 340A with an average of 325A. I think plugging in arbitrary resistance values is not too helpful, the 2.5mΩ @ 282A translates to 200W. I think the only resistance that should be considered is the actual winding resistance which will be very low in the order of 10s of micro-ohms. There is really little energy lost as the apparent series "impedance" is ramping current being translated into stored energy which is then released on the other half of the triangular wave with very low loss. Isn't this fun  If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| phil99 Guru Joined: 11/02/2018 Location: AustraliaPosts: 1790 |
Reactance (not resistance) dissipates no power as the voltage and current vectors are perpendicular. For the restive component of its impedance the vectors align so power is lost. As you say it is very small so I ignored it. Xl(Ω) = 2 x π x F(Hz) x L(H) An inductor has a separate reactance at each frequency it is operating at. A 16µH has a reactance of 5mΩ at 50Hz (double my second example) and 2Ω at 20kHz. For the harmonics of 20kHz it is multiples of 2Ω. At 50Hz it will drop 1.4V. It can be measured via a low pass filter that removes the 20kHz component. That can be measured separately via a high pass filter. The triangle wave you mention is composed of a diminishing series of these harmonics. I left them out as their effect is much less than the fundamental and I am too lazy to do the Fourier Transform. The reason for starting with the reactance is to decide how much voltage drop at 50Hz can be tolerated versus the amount of 20-kHz ripple current you must cope with. Then you can calculate the inductance that will achieve that by transposing the equation above. |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 998 |
Phil complex maths is rarely used by me for over 30 years - even then always found it a chore. Xl(Ω) = 2 x π x F(Hz) x L(H) When I knew this formula it only applied to a sine wave ? Yes a square wave is mathematically made up of a fundamental sine wave and all its higher harmonics. But we are not talking about resonances and phase shifts. We apply a DC voltage to an inductor, current ramps at a linear rate. When the applied voltage is set to zero all the energy flows back from the inductor (in a perfect inductor anyway). I am sure the mix of the 50Hz sine and 20kHz squarewave + a toroid and choke & resonating capacitor could fill a dozen blackboards of pure maths analysis but I don't really want to go there. My interest in the application of the inductor between the bridge and the toroid is conservation of energy and the only losses I can see in this scenario (for a perfect inductor) is the fundamental resistance of the winding. My posted query was to find out if my analysis of the current values I calculated for the inductance requirements as a storage device and to keep it out of saturation was miles off or in the right ball-park. Is there any chance you could help me a little with answering that fundamental question ? (preferably in layman terms - my head already hurts) Edited 2024-01-26 22:41 by wiseguy If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| phil99 Guru Joined: 11/02/2018 Location: AustraliaPosts: 1790 |
Just did a lengthy reply and it all vanished. Too late now, way past my bed time. |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 998 |
Happened to me earlier tonight - extremely frustrating- there should be an autosave text feature until posted or deleted by the writer - is admin listening  If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| nickskethisniks Guru Joined: 17/10/2017 Location: BelgiumPosts: 416 |
Interesting question Mike, are you ready for more headache? Are you after the peak AC+DC current? My thoughts on this: I think you are close and are missing just the last piece of the puzzle, If I recall it correctly we can see our inverter a little bit as a buck converter. So in an ideal situation we can remove also the transformer, we just end up with an LC filter like in a buck converter or hf inverter filter right? In a buck converter but also in an inverter peak to peak is highest at 50% dutycycle with nominal power, so IF the circumstances are right you are right and end up with 50% dutycycle at the top peak. And your calculation is right in my opinion. BUT that is when the ACpeak+DCpeak current is Imax. So I think your assumption/feeling is right and the peak current will be lower because of the fact your dutycycle will not be 50% and will be higher at high current levels. Otherwise your turns ratio is to high, or right if it was intended that way. For example: If your battery is low at 48V your dutycycle will be close to 100% at nominal power, at 56V it will be more like 85%. I think the modulation index and margin will make those percentages lower, but you get the idea. It's very difficult to calculate it because of all the factors, those factors you are now leaving out the equation to represent it more simply. So to make a good calculation you need to start to include for example the transformer ratio and all the losses and all the things I now forget. Already said, The reactance for a 16uH inductor is only 5mR for 50Hz but at 300A that's 1.5V (3-4%) you are missing at the top, could this explain the somewhat flat tops we are seeing in some of the sinewave pictures? With hf inverters that really counts because of the higher inductor values involved, sure current levels are lower but still, so you really need to take that in account when designing. Anyway this could be all nonsense what I said, I'm going to sleep about it now and wish you a lot of fun or headache.  Edited 2024-01-27 07:49 by nickskethisniks |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 998 |
Ok many hours later, I will present real worked examples for each 22.5 degrees of the 50Hz sine wave and what is going on with choke current at these points. In over 30 years of switchmode design apart from resonance controllers or looking at parasitics, I can't recall ever seeing reactance used in typical calculations for determining a square wave source and an inductor when discussing currents voltages or power or referring to the fundamental input or outputs of a properly designed solution . I found this answer from another engineer about this very question. BASc in Electrical Engineering & Computer Science, University of Waterloo (Graduated 1987)1y "Reactance is most typically used to characterize the small signal behaviour of inductors and capacitors. Power conversion, on the other hand, relies on the large signal behaviour of the inductor. However, the small signal behaviour — or reactance — of an inductor is of relevance when evaluating parasitic effects such as ringing at turnoff. In this case the reactance forms part of the tank circuit that causes the ringing. Now, getting back to the crux of the question, does the switching frequency affect the reactance? The switching frequency determines the peak minimum and peak maximum inductor current. If the inductor is ideal — or, at least, it is assumed ideal for the purpose of analysis — no real inductor is ideal — then the switching frequency does not affect the reactance. However, if the inductor exhibits any change in inductance due to even mild core saturation then the switching frequency does indeed affect the reactance as a lower switching frequency results in increased maximum and minimum peak currents. Sometimes this effect can be observed in the parasitic ringing at turn off. As the current drops there may be a noticeable drop in the frequency of the ringing." For the full example, I present it below as a PDF - I created it in Word as I did not want to run the risk of losing a post half way through and it would be a lot more work to embed this in a post. I have already spent waaay too much time on this to illustrate and present my analysis - but I think it REALLY was worth it ! Choke Design Analysis.pdf To model this I had to break it into smaller chunks I could visualise & deal with. This may not be the end of this choke question but I am comfortable to go with what I have so far, for estimating peak choke current - but still look forward to being educated further. Edited 2024-01-27 13:17 by wiseguy If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
Grogster Admin Group  Joined: 31/12/2012 Location: New ZealandPosts: 9066 |
They are.  I will pass this on to Gizmo - perhaps he can put a feature like that in place. Smoke makes things work. When the smoke gets out, it stops! |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 998 |
Thanks for considering my suggestion, I do hope Gizmo likes it ! If I had a dollar for every post that got lost when working on it.... If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| Godoh Guru Joined: 26/09/2020 Location: AustraliaPosts: 379 |
Hi Mike, yes it can get very complicated very fast this stuff. My take on reactance is this. Reactance is the opposition to current changes in an inductor and opposition to voltage changes in a capacitor. It should not matter too much what shape the wave is the inductor does not like current direction changes. The magnetic field in the core charges up, when the current goes the other way the field brakes down. Opposing the change. It was taught to me in this way.. You can't have an instantaneous change of current in an inductor. You can't have an instantaneous change of voltage in a capacitor. I would think that a triangle or square wave would cause more reactance than a sinewave, as the change in direction is faster Pete |
||||
Gizmo Admin Group Joined: 05/06/2004 Location: AustraliaPosts: 5019 |
Yes it is annoying. To answer you question, no, admin is not listening. I get a notification of every new post, and if its interests me, I go have a look. But I would read less than 1% of the posts on the forum. I rely on others to notify me if there is a problem. Autosave, I'll look into it. There's a bit more to it that you would think. Glenn The best time to plant a tree was twenty years ago, the second best time is right now. JAQ |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 998 |
Ok I will start this next part by saying I do believe totally in XC and XL. Both those formulas involve ω or angular velocity as used in SINEWAVES. I feel that things are being introduced into this discussion about choke current where vague terms are thrown in with no real reference to the original question of calculating peak choke current only that maybe it affects it. OK, how and by how much is the choke current affected, I am seeking an answer not more questions. When I did electronics maths, sinewaves was it. Squarewaves came from NE555s some time later when they were invented. When I started with switchmode my head was rearranged considerably cause a lot of the stuff I learnt had to be unlearnt. You can't have an instantaneous change of current in an inductor. You can't have an instantaneous change of voltage in a capacitor. Both 100% true there was nothing wrong here I learnt that too. Not sure I can buy this though It should not matter too much what shape the wave is the inductor does not like current direction changes Well somewhat true but the wave shape matters in our application I submit, we are starting with a squarewave, introducing an inductive storage element and our current is then a linear ramp sawtooth (hopefully) and all the energy we put into the choke is returned (with a small delay) the resultant is changing incrementally at 200 discrete steps every 10 milliseconds. Where and why and by how much will "reactance" or other relevant effects change the current figures and to what result ? I am purposely ignoring the AC capacitor being reflected back and the effect it has on the circuit. I do intend to eventually put a current clamp over the inductor because there are too many influences going on that is beyond my maths abilities and more easily measured empirically. But first I need a reasonable value choke that was derived from nominal data in the inverter and that is where this sad saga all started. Edited 2024-01-27 15:26 by wiseguy If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
KeepIS Guru Joined: 13/10/2014 Location: AustraliaPosts: 1369 |
Mike, I really like what you are investigating here, I look forward to seeing where this ends up, even if it doesn't reveal all that you are looking for, it gives some a better understanding and for others, it refreshes the complex parameters at play in this part of the circuit. . Edited 2024-01-27 15:57 by KeepIS It's all too hard. Mike. |
||||
| phil99 Guru Joined: 11/02/2018 Location: AustraliaPosts: 1790 |
Reactanve is directly proportional to fraquency and inductance. It is of key importance. Xl(Ω) = 2 x π x F(Hz) x L(H) That is correct but it goes further. It not only opposes a change in direction but any change in magnitude as well. A high frequency has faster rates of change so sees greater opposition, ie greater reactance. That is correct, and as each of the component harmonics that make up a square, triangular etc waveform is a sinewave it can be applied to each of them. A bit of calculus gives the total reactance to that waveform. The reactance is higher for higher harmonics so they contribute little to the total current and can be ignored. The Fourier Transform lets you dissect any wave-shape to gat all the component sinewaves and their phase relationships. At a practical level the inductance value is a compromise. A high inductance minimizes the peak current but reduces the voltage available to the primary. That could be offset either by adding some turns to the secondary or, if the transformer core never gets near saturation, take a turn off the primary. Feedback would compensate. In any case a bit of flat-topping isn't the end of the world. The mains can be badly flat-topped but everything keeps working. If inductor saturation is an issue a bigger core is usually the answer. I note some constructors stack many cores together and that seems to work well. KeepIS used an incremental experimental approach that worked very well. PS got dumped off TBS again, just minutes after signing on. This time I had a backup copy. |
||||
| wiseguy Guru Joined: 21/06/2018 Location: AustraliaPosts: 998 |
"KeepIS used an incremental experimental approach that worked very well." Like me he also used Sendust/Koolmu type toroids to good effect. These have published Al values Bsat figures Ae values that can be used to ensure that results are repeatable and predictable. Their characteristics possibly helped to avoid catastrophe ? I believe the better soft saturation characteristics of the sendust type materials is why KeepIS's inverter has survived some pretty horrendous loads. But guesswork and trial and error is rarely a recipe for success especially for newbies and people that want to build an inverter who don't have electronics in their blood. My interest in trying to create a better inverter stemmed from the many tales of woe on these pages sometimes from inverters that were just switched on or had no load applied but went bang spectacularly. I am also confident there is a better way to determine the inductor where the first iteration should work fine and we also know why and anyone can duplicate it 100%. Depending on the output power required and the DC supply voltage, the 20/30/40/50kHz frequency used and the RMS primary of the Toroid, there will be a method that can be developed that will yield a suitable inductor value for any inverter. That is my motivation for the original post - I am not knocking anyone elses work and feel somewhat justified that my approach will eventually yield the results. Edited 2024-01-27 18:16 by wiseguy If at first you dont succeed, I suggest you avoid sky diving.... Cheers Mike |
||||
| nickskethisniks Guru Joined: 17/10/2017 Location: BelgiumPosts: 416 |
It's always better to do it right at the first try, that's something I learned and I'm still learning, all People with some science background are thought to think and ask questions... No one really learnt something by just copycat something. If you want to do it right, you also want to look at the leakage inductance of your transformer, that's basically a series inductor. |
||||
| nickskethisniks Guru Joined: 17/10/2017 Location: BelgiumPosts: 416 |
It's me again. So Mike, you want to make some calculations to choose the right core and inductance for the inductor? But were do we start, on what base do you want to start the calculations, calculating the right inductance, there are several reasons for the inductor? Main reasons for the series inductor: --> Cancel out parasitic capacitance of the transformer? --> lower peak current and saving mosfets --> lower iron/copper losses in transformer --> therefore lower idle current --> Limit the hf ripple current through the transformer, 30%? --> also better for the capacitors --> ... For good understanding you want to base your calculations on the 30% allowable ripple current, or are there other constraints? (sorry if I missed the punchline, I blaim it on information lost in translation :) ) With your experience in the powersupply design, you think we should include the leakage inductance of our transformer, it probably has a saturation point too? It's the transformers leakage inductance that also limits the di/dt in theory (until saturation), maybe the parasitic capacitance has an influence also. Just for fun I measured it for a 180VA tranformer with a 18V winding and it was 10uH. For a 2000VA transformer it was around 11uH, but this was with double the amount of windings needed for making a super low idle inverter, so probably more like 3uH. (48-230V) So these are lower values then I expected (it was done with a meter so not tested with high currents, shorted 230V winding). And therefore in my opinion it will stay very difficult to work something out for everyone that uses different parts/components, I don't want to scare people but certainly if they don't have the right equipment to test all of this it will be difficult to build something reliable. At least it's good people are trying and with your help there could be following a rule of thumb. And use something better then "just use 50uH for 48V" :). Edited 2024-01-28 09:13 by nickskethisniks |
||||
| Godoh Guru Joined: 26/09/2020 Location: AustraliaPosts: 379 |
HI Mike, I hope too that you come up with an inductor design that is easy to get parts for and simple to put together. My home made inverters are using Powerjack inductors at the moment. They are just Torroids with a few turns of wire around them. They seem to work but I don't know how efficiently. Idle current is pretty low, they run well, but I am about to test them with a bigger load to charge the car. So I will have to keep an eye on them then. I did have trouble when we had 12 volt batteries with blowing up powerjack inverters but most of that was put down to dodgy connections on the battery and power boards. Since converting to 24 volts and rewinding the transformers from the junked powerjacks and using aliexpress boards I have not had a blowup. It will be good to see how your experiments go Pete |
||||
| Page 1 of 4 |
|||||
 Print this page |
