Rotary transformer for power transfer to moving shaft

[mikeselectricstuff]

I'm currently looking at ways to get power into a rotating (30rpm) chandelier for a 1-off project. We're concerned about long-term wear issues with slip-rings ( as well as the high cost of good quality ones).

I'm thinking about using a rotary transformer, but don't know much about magnetics.
Requirements :
Max size about 65mm dia x 50mm high
Power about 100W, continuous duty.
Output voltage ideally 48ish, and load regulation of about 25% would be OK
Input voltage : completely flexible.
Don't care about efficiency as long as nothing melts!

One possibility is Medum power Qi, which can supposedly do up to 65W, but this looks like being at the low end of what we need, and there doesn't seem to be much hardware out there above 30W at the moment. 

Suitable magnetics seem a bit thin on the ground.
Something like this would be perfect, but struggling to find anything off-the-shelf at a sensible price
  

The next best thing would seem to be a standard circular pot-core like this

Obviously the gaps at the sides would have some effect, but how much ?

What sort of core material should I be looking at?
I'm thinking frequency in the 50kHz range ? 
The  gap would be as close as we can get within mechanical tolerances - guessing 1-2mm
It would be nice to avoid the need for any feedback for regulation ( i.e. output determined by transformer ratio) - is this realistic for, say, 25% regulation?

This paper has some useful info, though I don't understand teh function of L1/CR1 in fig. 1909
https://coefs.uncc.edu/mnoras/files/2013/03/Transformer-and-Inductor-Design-Handbook_Chapter_19.pdf
[ Specified attachment is not available ]


Any thoughts welcome.

[Wolfram]

Normal pot cores will probably work pretty well for this.

The rotary transformer will likely have a lower (and somewhat variable) magnetizing inductance compared to a traditional transformer.

You can reduce the required reactive energy that the bridge needs to provide by using a resonant topology, but dealing with parameter variations in the transformer, both due to manufacturing tolerance and due to rotation, movement and temperature is not a trivial task, so I would suggest starting with a non-resonant scheme to begin with.

The simplest approach that is likely to work well is a plain isolated charge pump. Drive the primary from a full bridge, and rectify the secondary into a capacitor. Load regulation is given mainly by the leakage inductance of the transformer, but 25 % should be achievable. The easiest is to run it open loop, note that the output voltage might rise at light loads, mandating some minimum loading. You can close the loop by sensing the secondary voltage and regulating the supply voltage to the primary bridge. Duty cycle will have a small impact on the output voltage.

The next step up would be adding an inductor between the secondary side rectifier and the cap, making it into a forward converter. Load regulation will likely be worse open-loop, but now you can regulate the output voltage by changing the duty cycle of the primary bridge, saving the pre-regulator in case you want to run it with closed loop voltage regulation.

50 kHz seems to be a reasonable operating point, depending on your desired input voltage. Most normal power grades of ferrite will be good for this. N87 or equivalent is a classic. There are better grades available these days, but maybe not in the pot geometry. Regular component distributors often have an enormous markup on ferrite cores, I usually get them from a dedicated magnetics supplier like semic.info these days.

[langwadt]

https://nanopdf.com/download/rotating-transformer-for-a-wound-rotor-synchronous-motor_pdf ?

[mikeselectricstuff]

but now you can regulate the output voltage by changing the duty cycle of the primary bridge

The reason I want to avoid feedback is that any feedback has to pass back over a rotating joint! Obviously there are ways to do this but want to keep it as simple as possible. the output will eventually be going into a DC-DC buck converter down to 5v or so ( after passing through a resistive structure, hence 48-ish volts) , so regulation isn't a huge issue.

There will be some minimum load of a few watts

[voltsandjolts]

Devil's advocate: Sounds like fun, but could it be a bit too much faff for a one off unit?

https://www.bgbinnovation.com/online-shop.slipringpackage/spb14-6-way-300mm-26awg/

https://www.ebay.co.uk/itm/153597143918

Off the wall idea; maybe a car alternator could be the slip rings and motor in one unit.
https://hackaday.com/2020/01/16/car-alternators-make-great-electric-motors-heres-how/

[NiHaoMike]

Perhaps start with a SSTC design, but with a secondary that has a few turns with a parallel cap?

[mikeselectricstuff]

Devil's advocate: Sounds like fun, but could it be a bit too much faff for a one off unit?

https://www.bgbinnovation.com/online-shop.slipringpackage/spb14-6-way-300mm-26awg/

https://www.ebay.co.uk/itm/153597143918

Off the wall idea; maybe a car alternator could be the slip rings and motor in one unit.
https://hackaday.com/2020/01/16/car-alternators-make-great-electric-motors-heres-how/

The problem is most slip-ring manufacturers either don't spec lifetime at all, or are vague about it. This could potentially need 1M+rotations, and replacement would be very difficult

[voltsandjolts]

True, but alternator slips rings are robust for thousands of hours at 2000rpm
https://www.ebay.co.uk/itm/274884305266
Anyway, I'll shutup now!

[Cerebus]

Out of the box idea: What's the mechanical shaft power like (assuming that 30 rpm is continuous)? Enough that you could get your electrical power just by putting a generator coil on the rotating part and have some permanent magnets (or a coil) as the stator?

(I still suspect that "Find the right slip rings" is the real and ultimately cheapest answer.)

[Terry Bites]

Car alternator sliprings and brushes are battle hardenned. But maybe a bit too sparky when dirty, but I think you like that kind of thing...... what about a pair horn push slip rings? Cheap nasty and robust. Or the real thang.
https://www.ebay.co.uk/itm/265066949029?var=0&mkevt=1&mkcid=1&mkrid=710-53481-19255-0&toolid=10050&campid=5338358731&customid=16318898208085802554912021000008005

https://www.ubuy.hk/en/search/?ref_p=ser_tp&q=Taidacent+Through+Hole+Conductive+Rotary+Slip+Rings

I like the idea of a magnetic coulping but an uphill struggle for a one off I think.

[T3sl4co1l]

The slots in the pot core don't matter much. Air gap between faces is much more significant.  That's limited by mechanical runout and such.

A resonant converter is alright, depending on if you can afford some means of feedback, or primary-side sensing, or dissipate the excess in a shunt regulator (so the primary delivers constant power, ez pz).  This is the main difference with gap: if you can't afford a tight gap, leakage will be substantial and regulation will be poor, and resonant will be attractive to deal with the additional reactive power; with a tight gap it's just like any split-bobbin transformer, the regulation might not be great but it's manageable.

Also possible with enough gap, and a particular magnetic arrangement; it could be arranged as a ferroresonant transformer, I think, which would effectively be regulated by the saturation flux of the materials, and the driven frequency.  This would require a ferrite "washer", to return some flux (magnetic shunt) around the secondary side, I think?

An example of feedback might be light-housing an IR LED out from the shaft, and put some phototransistors around it.  Sum them up and that's your feedback signal.  Does need good shielding/filtering from ambient, and does need to be filtered to remove ripple caused by rotation.  Slow control has consequences for output regulation, and in turn, how much filtering (energy storage) you need onboard; an ideal converter has approximately zero energy storage and instant control response.

I mean, assuming you can't get an axial position of course, in which case the opto is trivial.

Tim

[sandalcandal]

I'd agree a resonant topology is likely what you want to efficiently deal with the extra and less controlled leakage inductance that your are likely to encounter. However, you could potentially go with a "simpler" push-pull system and just eat the excess switching losses depending on how much margin (in terms of money, size and weight) you have available to add the extra snubbers and thermal management. Poor regulation can be dealt with using additional stages which shouldn't be too hard with something off-the-shelf (universal mains voltage SMPS?) but again would need to mind margins.

I wonder about the mechanical aspects here.
* What are the loading conditions?
* Will there also be potential non-axial loading e.g. wind or drafts in the installation location?
* How big of a shaft do you need for this chandelier?
* What will the bearing arrangement look like?
* Can the bearing arrangement maintain sufficient runout to stop rubbing under the expected conditions? ​
Leakage flux could lead to inductive heating of the shaft if it is conductive and/or magnetic.
If runout can be well guaranteed then building a tighter and more efficient transformer/wireless power transceiver could be made easier.

[bdunham7]

How large is this chandelier and what is driving it?  I would think about integrating a transformer into a multi-pole ceiling fan motor somehow by adding windings to the armature.  The 'inside-out' configuration of some of these seems like that might not be too terribly difficult.  Might be a bit of work for a one-off, but maybe have a look at a ceiling fan motor and see if that is workable?

[sandalcandal]

This paper has some useful info, though I don't understand teh function of L1/CR1 in fig. 1909
https://coefs.uncc.edu/mnoras/files/2013/03/Transformer-and-Inductor-Design-Handbook_Chapter_19.pdf

Looks like a differential mode choke (sort of?) with a flyback diode possibly to try force power back out onto the input bus capacitor? I've never seen that before in any modern designs. Normally people seem to just let it flow though FET body diodes directly to the input bus.

[sandalcandal]

Thinking of this more. If you use a P core with a shaft through the middle (or near the outside surface) and that shaft is magnetic then due to the gap in the cores, flux will be encouraged to pass though and be concentrated in the shaft. Even if you use a non-magnetic but still conductive material, fringing flux near the gap will induce eddy currents in the conductive material that will also cause loss/heating. With a larger but very low flux design at low switching frequency this could be ok but I'd suggest something else.

The coaxial core design in the paper you shared is probably the best in terms of a compact, magnetically efficient solution but I think you could do a design that gets away with using P cores if you can tolerate having a thick shaft, though you may have considered it already. You could use a large hollow with the core in the centre. If you make the shaft of plastic (or some other non-conductive non-magnetic material) then you have freedom to make it quite thick if it is an outer layer surrounding the core. Alternatively you could use a metallic shaft but have a large air gap (or some other non-conductive non-magnetic material) between the inner diameter of the metallic shaft and the transformer core in the centre. There is also the benefit of better protection of the transformer core. The downside is you potentially have a shaft that is visually too fat? though I guess you could potentially hide the wider part with the transformer in the "body" of the chandelier?

Edit: Since all you want to do is stable transfer power across a small air gap, without the need for much regulation, doing a series resonant design should be very simple. If you operate at the resonant frequency then gain should remain very close to 1. If the resonant point shifts then voltage will drop if load is high but it shouldn't randomly climb to an unreasonable level if you fail to control it as with an LLC design. You could even add an external "ballast" resonant inductance which would reduce the relative effect of leakage inductance changes.

[langwadt]

not exactly super cheap at ~32€, but if it is for a one-of

https://www.digikey.dk/product-detail/en/tdk-electronics-inc/B65945A0000X022/495-5356-ND/3914532

but then, sliprings with a rated life of tens of millions revs isn't much more

https://www.ato.com/miniature-through-hole-slip-ring

[uer166]

One big problem that I see is the large mechanical forces exerted by the fields within the gap(s). It may be impossible to properly align and mechanically secure the 2 core halves axially well enough so that one side doesn't suddenly collapse and touch the other side. It's the same issue you have when aligning and securing a steel/permanent magnet rotor inside a stator in a BLDC. I think a motor/generator pair would be easier to do, but a slipring would probably be more reliable. Of course the best might be a wireless power solution where the mechanical forces are benign and the alignment non-critical. Maybe you can get some devkit like https://www.we-online.com/web/en/electronic_components/produkte_pb/demoboards/wireless_power/design_kit_200_w/wireless_power_200wkit_page.php that can do 200W supposedly readily and don't have to deal with Qi?

[sandalcandal]

Edit: This design is a bad idea as there is no capacitor in series with the primary winding to limit the magnetising current. Never mind, just my sim messing up. Still the lack of capacitor on the primary might cause issues? Need to explore further.


LTSpice simulation of a SR-ish type converter with a leaky transformer. The .asc is attached. Idea taken from tank XV here: https://www.ti.com/seclit/ml/slup376/slup376.pdf


Effect of change load from 50 Ohms (310 W) to 100 kOhms (<0.2W)


Effect for leakage inductance change from 70uH to 130uH in steps of 10uH

Stability looks pretty good overall. 25% Regulation seems achievable with a fixed ~28kHz input switching frequency.

Qi wireless chargers are just resonant air gapped transformers much like what is being done here but with more overhead protocols and control.

[T3sl4co1l]

One big problem that I see is the large mechanical forces exerted by the fields within the gap(s). It may be impossible to properly align and mechanically secure the 2 core halves axially well enough so that one side doesn't suddenly collapse and touch the other side. It's the same issue you have when aligning and securing a steel/permanent magnet rotor inside a stator in a BLDC.

I am extremely interested in your 1.5T+ saturation pot cores and wish to know more at your earliest convenience!

Tim

[sandalcandal]

One big problem that I see is the large mechanical forces exerted by the fields within the gap(s). It may be impossible to properly align and mechanically secure the 2 core halves axially well enough so that one side doesn't suddenly collapse and touch the other side. It's the same issue you have when aligning and securing a steel/permanent magnet rotor inside a stator in a BLDC.

I am extremely interested in your 1.5T+ saturation pot cores and wish to know more at your earliest convenience!

Tim

FYI typical power magnetics don't go much above 200mT even at "low" frequencies of <100 kHz. 50mT to 100mT seems common for most modern applications.

[uer166]

One big problem that I see is the large mechanical forces exerted by the fields within the gap(s). It may be impossible to properly align and mechanically secure the 2 core halves axially well enough so that one side doesn't suddenly collapse and touch the other side. It's the same issue you have when aligning and securing a steel/permanent magnet rotor inside a stator in a BLDC.

I am extremely interested in your 1.5T+ saturation pot cores and wish to know more at your earliest convenience!

Tim

I did say it was same issue, but didn't say it was of the same level 

[T3sl4co1l]

More just to say -- do the sniff test; assembly won't be an issue, it's only pulling while active (unlike those permanent magnets, pesky indeed!), and, if your bearings will have trouble with a few gramforce from a ferrite core, you have much, MUCH bigger problems.

(For those that aren't aware -- magnetic force is proportional to flux density squared.  So, in the low 100 mT range, you can probably pry things apart by hand.  Fridge magnets are in the 10s mT range, using a pattern optimized for close-up holding force.  Ferrites top out at 400mT or so, while NdFeB magnets do about 1.5T, and some iron alloys (soft) go up to 2T or so, depending on how hard you want to try to magnetize them.  Any higher, you REALLY have to want it -- typically needing superconductors, as you'll get no help from iron pole pieces.  These (up to 2T) fields easily cause finger-crushing accelerations, but it's nothing a properly built mechanism can't handle -- the pressure (B^2 / (2 mu_0)) is equivalent to just a few bar.  So if you think about shop (compressed) air equipment, that's about the most strength required for any ordinary magnetic assembly.)

(As a corollary, this is essentially why coil guns do, and forever will, suck.)

Tim

[james_s]

I would expect slip rings to be more robust and reliable than the electronics required to drive a rotary transformer. I mean there are the car alternator applications already mentioned, and then stuff like rotating signs that used to be common out front of businesses, those use slip rings and would run for years and years. Carnival rides are another application that gets heavy use.

[mikeselectricstuff]

I'd agree a resonant topology is likely what you want to efficiently deal with the extra and less controlled leakage inductance that your are likely to encounter. However, you could potentially go with a "simpler" push-pull system and just eat the excess switching losses depending on how much margin (in terms of money, size and weight) you have available to add the extra snubbers and thermal management. Poor regulation can be dealt with using additional stages which shouldn't be too hard with something off-the-shelf (universal mains voltage SMPS?) but again would need to mind margins.

I wonder about the mechanical aspects here.
* What are the loading conditions?
* Will there also be potential non-axial loading e.g. wind or drafts in the installation location?
* How big of a shaft do you need for this chandelier?
* What will the bearing arrangement look like?
* Can the bearing arrangement maintain sufficient runout to stop rubbing under the expected conditions? ​
Leakage flux could lead to inductive heating of the shaft if it is conductive and/or magnetic.
If runout can be well guaranteed then building a tighter and more efficient transformer/wireless power transceiver could be made easier.

I don't think the mechanics will be an issue - there will be a large ring bearing, leaving about 80mm dia in the centre completely clear. The thing will be about 1.5m diameter and weigh about 30 kilos (all carbon fibre) . Budget is not an issue 

[bdunham7]

there will be a large ring bearing, leaving about 80mm dia in the centre completely clear....Budget is not an issue 

Then a slip ring that size with multiple contacts in parallel and AC power would seem to be a pretty easy, reliable solution.