High frequency PWM dimming of LEDs is OK?

[treez]

Hello,


We are wanting to do high frequency PWM dimming of the output of Resonant Inductive Couplers (which feed LED lamps) hooked up to a twisted pair bus.
Do you think the “gap toothy” resulting waveform (see below) will be a cause of any problems?

(This is a different question on a subject discussed elsewhere here)

Block diagram………………..
https://i46.tinypic.com/28151kz.jpg

….you can see the “shorting FETs which dim by stopping current flowing to the LEDs when the FETs are ON.
(please ignore the number of secondary turns, as you know, theyre resonant inductive couplers)

We do NOT want to do low frequency PWM dimming, (~200Hz) because the low frequency of this dimming causes problems for the current source. (it means the current source is switching from high power to low power output which causes problems)

Therefore, we wish to do dimming at a much higher frequency (83.3KHz). But do you think that the resulting current waveform (as below) is going to mean problems?


LED lamp current with no dimming (D = 1)
https://i47.tinypic.com/o9iv.jpg


LED lamp current with dimming with D = 0.083
https://i48.tinypic.com/3ctas.jpg


LED lamp current with dimming with D = 0.33
https://i50.tinypic.com/1zofdzp.jpg

LED lamp current with dimming with D = 0.92
https://i49.tinypic.com/np5d91.jpg

(Also, how would you reduce dissipation in the current source by correcting the power factor in the twisted pair?)

Resonant Inductive Couplers
https://en.wikipedia.org/wiki/Resonant_inductive_coupling

 

[FvM]

As a preliminary remark, the design can be expected to distribute fast switching transients from the individual PWM switches all along the supply bus and most likely showing disastrous EMI behaviour.

I also assume that there are lower part count solutions serving the same purpose. Apart form this points, it should work (after correcting the rectifier polarity).

For maximum LED efficiency, the rectified current should be filtered.

Regarding power factor respectively generator efficiency, it depends on how you intend to implement the current source property. That's one of the points that sounds simple at first sight, but involves a considerable effort. I imagine a PWM controlled bridge with output filter.

 

[treez]

Thanks FvM, rectifier polarity now corrected in first post.

 

[dick_freebird]

The phosphor dwell time will smooth any ripple quite a bit
in the optical domain.

You might be able to find phosphor optical decay time
info at the LED mgr or in industry publications.

 

[treez]

Thanks, the thing is though, the couplers are resonant inductive couplers , and i think it may be best to do the dimming by "slicing" (phase cutting) each 10us pulse, so that resonant frequency is still strongly in there...do you agree?

 

[kripacharya]

Thanks, the thing is though, the couplers are resonant inductive couplers , and i think it may be best to do the dimming by "slicing" (phase cutting) each 10us pulse, so that resonant frequency is still strongly in there...do you agree?

i'm surprised you can couple magnetically from a twisted pair. in my day a twisted pair was used to reduce em interference !!

secondly, you call it resonant coupled, but i don't see any caps anywhere. how do you achieve this resonance ? At ~83KHz you would need a largish cap which would then need to be electrolytic.

lastly - with multiple simultaneous couplings, the resonance frequencies would change. And with pwm switching going on simultaneously .. this could be a tough one to setup and align.

Having said all that, would love to see how your project works out !

 

[treez]

sorry i forgot to draw the cap in across the output coil that you see.

...Twisted pair....yes i see your point...but you just bend out a loop in the twisted pair, and hook the ferrite (resonant) coupler into it. I am not sure if there is a "buffer" stage between the resonant coupler and the load..maybe there is?..maybe its a current transformer acting as a buffer............i dont know this bit as i never saw it...its secret.