Interesting technique: frequency locked resonant wireless power receivers / Electromagnetic Radiation / Forums

Forums

4hv.org :: Forums :: Electromagnetic Radiation

« Previous topic | Next topic »

Interesting technique: frequency locked resonant wireless power receivers

Move Thread

LAN_403

BigBad

Tue Dec 30 2014, 09:29PM

Registered Member #2529 Joined: Thu Dec 10 2009, 02:43AM
Location:
Posts: 600

I noticed that with wireless power the receiver coil, if tuned for maximum distance, at close range tends to ring up to high voltage.

This wastes power, because the resistive losses in the receiver bleed off power as current flows through the coil and capacitor. It also tends to make the capacitors more expensive because they have to be able to handle higher voltages.

So if you can limit the voltage on the receiver coil, potentially you can improve efficiency.

So in LTSpice I simulated a circuit to put a feedback loop to control the coil voltage; basically I tried draining the coil down through a resistor when it went too high. I figured it would waste some power, but not as much as if the coil rung up to high voltage.

It kinda worked, but the delays through the loop were a bit annoying; it tended to damp the coil out entirely and then take a while to ring back up again. So the output voltage wasn't very stable. You could put a linear regulator on the output but it wasn't a very good system. The resistor had to be in parallel across the tank, but if it was too big it might as well not be there, because most of the current circulated in the tank, bypassing the resistor, whereas a small resistor didn't damp very well. I tried various things, I tried adding taps to the coil, I tried all kinds of things, it kind of worked, but mostly sucked.

Then I replaced the resistor with a capacitor in parallel with the tank circuit to basically detune the coil controlled with a TRIAC. Again this kinda worked and was more efficient. But it was still a bit bouncy because of the way the feedback loop worked.

Finally I worked out a better circuit adding a controlling capacitor in parallel with the main tank circuit through a triac, and driving the triac from an optoisolator controlled by a zener on the low voltage side. By closing and tuning the control loop it gave very fast feedback and this regulated the output very, very well. In fact it was pretty impressively good, none of the bouncy output at all, just a little ripple from the diode/capacitor.

I liked the capacitor in parallel like that, so then I tried to work out what the ideal behaviour could be, I figured if I controlled whether the capacitor was in or out per cycle I could actually fine tune the tank circuit, so that the long term frequency would exactly match that of the transmitter. Normally the receiver is always a bit out of tune, but in principle I could crystal lock it, set the primary inductor/capacitance so it was running a bit fast, and then pull it in. The TRIAC adds a bit of on-resistance, about 0.1 ohms, which lowers the average Q very slightly but it's quite low resistance and only active maybe one cycle in ten anyway and it's in parallel with the main leg, so it's not too bad at all. Sounded pretty good to me.

Then I realised- you can not only frequency lock, but actually phase control the receiver. If it's 90 degrees out of phase, then the receiver won't receive any signal from the transmitter it will average out, but if it's in phase, the tank circuit will grow, and if it's 180 out of phase the tank circuit will feed back to the transmitter. And all the bits in between.

Practically an ideal system, very low impedance, crystal locked resonance and controllable amplitude!!!

Then I looked at the circuit I'd designed with the unbelievably good regulation... and discovered that that's almost exactly how it worked! (In fact I think the TRIAC in the simulation was kicking in for parts of a cycle, so it was continuously varying the phase.)

So I'd sped up the feedback loop so much I'd achieved phase locking, I'd made a phase-locked control loop that did exactly that, and that's why it was so very effective!

Doh, but YES!

Still, it's only a LTSpice simulation, but it's looking really promising.

The other thing is, you can do the same thing on the transmitter side, so you can crystal lock the transmitter frequency.

Anyway from what I can tell from the published papers, it looks like a couple of others have had similar ideas, so I'm at least a few months behind on this, I only just thought of it, and haven't read any of the other papers. The Witricity people were using some dorky switcheable capacitive network, but this only needs one extra capacitor and some silicon to connect it in or out each cycle, or fractions thereof.

Here's an LTSpice model, you control the coupling coefficient K1 to set the range (between about 0.5 and 0.007 works fairly well) you can set the transmitter voltage and can look at the vout. I'm still fiddling with it, it's far from optimal, but it does seem to work to a reasonable degree.

]regulated_wireless_receiver.zip[/file]
edit: coupling coefficient worked down to 0.007, not 0.07

2Spoons

Tue Dec 30 2014, 10:34PM

Registered Member #2939 Joined: Fri Jun 25 2010, 04:25AM
Location:
Posts: 615

Having worked for one of the big players (not witricity) in the wireless power field i have seen many methods for secondary side regulation, including this one. I've even had a hand in inventing one or two. I just want to point out that the switchable cap network is not as 'dorky' as it first appears - its reasonably easy to implement on chip (with external caps), and has the advantage of being static once under control - no switching losses. In this game efficiency is everything - since these things are being stuffed into cell phones and thermal limits are pretty strict.

nice piece of work though.

BigBad

Tue Dec 30 2014, 11:01PM

Registered Member #2529 Joined: Thu Dec 10 2009, 02:43AM
Location:
Posts: 600

Well, I personally still think it's a bit dorky. ;)

There's always going to be switch losses with the capacitor network, because you're grounding the caps through the chip and it has a resistance. You're possibly saving control power, but it doesn't look to be very much.

Ash Small

Wed Dec 31 2014, 09:41PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Any obvious problems with this, BB?

Zeners are (1/2 Vmax)-Vf. The idea is that whenever the voltage reaches Vmax, one of the zeners switches the second capacitor into the circuit, so Vmax can never be exceeded.

Things should settle down almost immediately, or am I missing something?

EDIT: Or do the zeners need to be Vmax-Vf?

BigBad

Thu Jan 01 2015, 01:44AM

Registered Member #2529 Joined: Thu Dec 10 2009, 02:43AM
Location:
Posts: 600

Looks interesting, but Zeners can't usually take the (pretty big) current.

Other than that, should work I think.

edit: I simulated it, ignoring the current problem, it didn't seem to work particularly well as drawn for slightly murky reasons, but when I used the back-back zener motif in the gate trigger circuit for a triac to control the capacitor, that actually seems to work quite well, although so far, not quite as well as the opto circuit at medium-large range (seems to work much better at short range), although there's lower part count and it looks more reliable.

Ash Small

Thu Jan 01 2015, 06:54PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

I'm wondering what differences it would make if the zeners were inverted, but it is making my head spin a bit

Maybe I should simulate it and see.

BigBad

Thu Jan 01 2015, 08:30PM

Registered Member #2529 Joined: Thu Dec 10 2009, 02:43AM
Location:
Posts: 600

As you drew it, it's pleasingly symmetric, but the caps and the zeners are in series, so you can put them in any order, and inverting them both is the same as swapping them.

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob