Author Topic: Non-PWM LED power  (Read 5405 times)

0 Members and 1 Guest are viewing this topic.

Offline Richard CrowleyTopic starter

  • Super Contributor
  • ***
  • Posts: 4317
  • Country: us
  • KJ7YLK
Non-PWM LED power
« on: May 23, 2018, 12:14:01 pm »
In an audio forum (GeekSlutz.com) there was a question about severe EMI/RFI in a studio coming from  RGB LED strips. Common product that uses PWM to control color hue and light level.

I suggested that the EMI was coming from the PWM power supplies and indeed the interference goes away when the LED strips are unpowered.

I suggested a non-PWM power supply to allow continued use of the LED strips without the severe EMI.  However, here in the 21st century, are there even non-PWM solutions available off-the-shelf for end-users?  If it were me, I would just DIY, but what advice could we give someone who needs a commercial solution?

Ref: https://www.gearslutz.com/board/connectors-cables-stands-and-accessories/1216437-led-hum-xlr-cable.html
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 19345
  • Country: gb
  • 0999
Re: Non-PWM LED power
« Reply #1 on: May 23, 2018, 12:25:32 pm »
Perhaps you should design one and market it as a high-end audiophile product?
 

Offline senso

  • Frequent Contributor
  • **
  • Posts: 951
  • Country: pt
    • My AVR tutorials
Re: Non-PWM LED power
« Reply #2 on: May 23, 2018, 12:48:00 pm »
Transformer AC-AC, rectify, redo control system with linear current, make the light adjustable with 3 pots?

Call it a Class A led driver, prepare wallet for tons of monies.
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 19345
  • Country: gb
  • 0999
Re: Non-PWM LED power
« Reply #3 on: May 23, 2018, 12:51:29 pm »
Transformer AC-AC, rectify, redo control system with linear current, make the light adjustable with 3 pots?

Call it a Class A led driver, prepare wallet for tons of monies.
I would have thought the easiest way to convert a PWM to linear is with a passive filter, i.e. using inductors and capacitors. I don't know what frequency is typically used for LED PWM, but if it's in the audio band, I would have thought it would need fairly large inductors, for it to be effective.
 

Offline Miyuki

  • Frequent Contributor
  • **
  • Posts: 896
  • Country: cz
    • Me on youtube
Re: Non-PWM LED power
« Reply #4 on: May 23, 2018, 01:59:06 pm »
Most of commercial LED equipment is crap and put lot of EMI noise (this things shall not pass testing, but I think it was never really tested at all) and some of them also annoying visible flickering and buzz
I know just some industry grade led lights use linear supplies to eliminate dangerous strobe effect and other problems, but it is other world

Hope in future there will be color lights with smoothed output
Like early and cheap LED backlight at LCD was/is flicker, today better brands have smoothed output and produce steady light
 

Offline xani

  • Frequent Contributor
  • **
  • Posts: 400
Re: Non-PWM LED power
« Reply #5 on: May 23, 2018, 03:15:06 pm »
Well, filtering would probably help a lot considering that instead of having whole strip being antenna it would be limited to power supply+filter (that then can be put in farthest away). Or just making actually well designed SMPS instead of "the cheapest possible one that sends right current to leds".

Of course *real* audiophile solution would be using 3 audiophile-class amplifiers to drive the led strips  :-DD
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 14309
  • Country: fr
Re: Non-PWM LED power
« Reply #6 on: May 23, 2018, 03:58:45 pm »
Proper filtering should lower emissions to acceptable levels.

The ultimate solution would be to drive the LEDs with adjustable constant current for dimming (no PWM). Would be less power-efficient, but very limited noise.

I have looked a little and off-the-shelf drivers that would be entirely linear are hard to find. But Philips has what they call "linear drivers" in their Xitanium LED drivers product line:
http://www.usa.lighting.philips.com/products/oem-components/xitanium-led-drivers
They claim those don't use PWM and provide low-noise dimming. Might be your best bet outside of designing your own.

Or maybe you could ask treez's company to design one for you?  8)
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 6694
  • Country: nl
Re: Non-PWM LED power
« Reply #7 on: May 23, 2018, 04:45:57 pm »
I suggested a non-PWM power supply to allow continued use of the LED strips without the severe EMI.  However, here in the 21st century, are there even non-PWM solutions available off-the-shelf for end-users?

Can you dim them without PWM though? They need a bit of headroom for the resistor based current limitation to work. Maybe programmable LED strips would be better for them, WS2812 modulates at 800 KHz local to the LED. Can't see that doing much EMI wise.
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 19345
  • Country: gb
  • 0999
Re: Non-PWM LED power
« Reply #8 on: May 23, 2018, 04:53:59 pm »
The ultimate solution would be to drive the LEDs with adjustable constant current for dimming (no PWM). Would be less power-efficient, but very limited noise.
In the case of LED strip, PWM would yield exactly the same efficiency as linear constant current control, since the LEDs just have series resistors. The optimum solution, efficiency wise, would be a switched mode constant current source.

Thinking about this again, too much filtering the PWM could also cause problems, especially at the lower end of the brightness setting, since the voltage across the LEDs would fail to reach the diodes' forward threshold voltage. It would be equivalent to powering them off an unregulated, constant voltage, buck regulator.

The filtering should be good enough to reduce the rise/fall times of the squarewave to a level which doesn't emit too much RFI but not completely obliterate it altogether, leaving only DC behind.

I suggested a non-PWM power supply to allow continued use of the LED strips without the severe EMI.  However, here in the 21st century, are there even non-PWM solutions available off-the-shelf for end-users?

Can you dim them without PWM though? They need a bit of headroom for the resistor based current limitation to work. Maybe programmable LED strips would be better for them, WS2812 modulates at 800 KHz local to the LED. Can't see that doing much EMI wise.
What about the data bus? It might not be as bad as PWM but I can see it causing problems with some sensitive equipment, if it's not shielded.
 

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 8112
  • Country: fi
Re: Non-PWM LED power
« Reply #9 on: May 23, 2018, 06:07:01 pm »
PWM dimming in a LED lighting system is a complete misconception from the start. Without proper engineering decisions from the start, the EMI is probably further compromised by all other incompetency happening during design details.

Solution: install proper LED equipment with legal (EMC passing), adjustable constant-current drivers. No PWM dimming needed. (PWM signal can still be used to control the driver, if needed.)

The fact that 12V, resistor-limited LED strip, dimmed with PWM, is still installed semi-professionally in 2018, is just sad and ugly. Sometimes people tend to just like the solution which is:
A) inherently just the Wrong Thing To Do
B) more complex
C) more expensive
D) wastes energy
E) somehow seems "traditional" and "simple"

Although, we drive gasoline cars as well, so this is not surprising.
« Last Edit: May 23, 2018, 06:17:47 pm by Siwastaja »
 

Offline chris_leyson

  • Super Contributor
  • ***
  • Posts: 1541
  • Country: wales
Re: Non-PWM LED power
« Reply #10 on: May 23, 2018, 07:38:23 pm »
I think a linear driver with a PWM front end is probably the best solution, linear driver keeps the noise down and the PWM front end keeps it all reasonably efficient. Granted it's extra hardware but a well designed wideband linear driver should clean things up. Unless of course the LED strip is populated with WS2812s, then it's probablhy a case of changing the LED strip, I can't see a string of WS2812s being quite.
Maybe you could modify the PWM driver by diverting the ripple current back into the current sense resistor and bypass the LEDs altogether. I've used that technique with an LM3404, see datasheet, and got the LED ripple current down to a few mA.
 

Offline Richard CrowleyTopic starter

  • Super Contributor
  • ***
  • Posts: 4317
  • Country: us
  • KJ7YLK
Re: Non-PWM LED power
« Reply #11 on: May 24, 2018, 01:22:19 am »
I advised the OP that a fixed, linear supply would eliminate the ability to dim or fade the different colors. 
I was assuming that simply reducing the DC voltage would only work over a limited range.
 

Offline Cyberdragon

  • Super Contributor
  • ***
  • Posts: 2676
  • Country: us
Re: Non-PWM LED power
« Reply #12 on: May 24, 2018, 04:55:21 am »
Just have a three ouput current mirror style supply with pots to manually reduce the base drive (proportionally to the actual base drive) of each pass transistor. Note, the current sense would be in the common, so all three colors would have to be balanced.

EDIT: You would have to have some sort of small feedback from each channel dimmer to the main regulator to ensure it doesn't overshoot the other channels trying to compenstate when you dim a channel. IE, dimming each channel would reduce the overall current limit. Or you could have one sense circuit feeding three seperate references.
« Last Edit: May 24, 2018, 06:17:19 am by Cyberdragon »
*BZZZZZZAAAAAP*
Voltamort strikes again!
Explodingus - someone who frequently causes accidental explosions
 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 19345
  • Country: gb
  • 0999
Re: Non-PWM LED power
« Reply #13 on: May 26, 2018, 08:04:35 pm »
I advised the OP that a fixed, linear supply would eliminate the ability to dim or fade the different colors. 
I was assuming that simply reducing the DC voltage would only work over a limited range.
Yes, that's the problem with merely reducing the voltage.

Constant current is also a non-starter, since the LED strip can be any length and the maximum current would depend on how long the strip is. Perhaps the controller could momentarily apply the full 12V to all LEDs, when its first turned on, to determine the required current for full brightness, then use constant current control with the current limited, to a percentage of the full voltage value? The trouble with that is the forward voltage of the LEDs drops slightly, as they heat up, so it wouldn't be perfect.

Another solution could be to build a constant current controller for a short piece of LED strip, internal to the controller and apply the same voltage to the external strip. The problem with that is the LEDs inside the controller would have to be closely matched to the external strip.

I think the most convenient solution is to use a passive filter, as it would work with the existing controller circuit. The trouble is what lower cut-off frequency to use? It needs to be significantly above the highest PWM frequency used, yet still reduce the rise and fall times of the pulses to minimise EMI. A filter with just inductors and capacitors would be efficient but how well it works would depend on the loading. An RLC filter would be easier to design so it works over a wide range of loads, but will dissipate some of the power as heat.
 

Offline james_s

  • Super Contributor
  • ***
  • Posts: 21611
  • Country: us
Re: Non-PWM LED power
« Reply #14 on: May 26, 2018, 08:42:27 pm »
It might be worth trying a strip on a variable power supply to see if an acceptable dimming range can be achieved, certainly it does work to some degree. I'm not sure how else it could reasonably be done with the series-parallel arrangement these strips use.

Another possibility is to use strips of addressable LEDs, there's a variety that requires rather precise timing but another variety that is essentially SPI  so if a high update speed is not needed then the data could possibly be clocked in at a rate that minimizes EMI. It could also be sent in bursts as needed rather than continuously updating which might help.
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 8973
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: Non-PWM LED power
« Reply #15 on: May 26, 2018, 08:47:05 pm »
What about PWM at a low frequency like 200Hz with the edges purposely very slow to minimize EMI?
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline David Chamberlain

  • Regular Contributor
  • *
  • Posts: 249
Re: Non-PWM LED power
« Reply #16 on: May 28, 2018, 07:37:24 am »
What about PWM at a low frequency like 200Hz with the edges purposely very slow to minimize EMI?

Im thinking its low frequency EMI that is the problem. Last time i was doing recording i was up at 96kHz sample rates so the PWM frequency is going to want to be somewhat above that.
 

Offline Kjelt

  • Super Contributor
  • ***
  • Posts: 6459
  • Country: nl
Re: Non-PWM LED power
« Reply #17 on: May 28, 2018, 07:59:09 am »
Resistor based ledstrips are not a professional solution and are not tested as such (hence the missing UL or other approval).
You should actually want to remove the resistors and place the leds in series and drive with a current source, however the voltage will be too high for flex strip (dangerous high).
So that is not an option.

If you look at the pro solutions they usually use medium or high power leds in stead of low power, are in a U shaped grounded profile and have local shielded dc-dc converters (actually current sources).
A different option are so called wall washers, very broad (50cm - 100cm) led fixtures that enlighten an entire wall from below or from top.
Those can usually be controlled by DMX, DALI  or some other lighting protocol. Different pricetag obviously.
« Last Edit: May 28, 2018, 08:01:01 am by Kjelt »
 
The following users thanked this post: Richard Crowley

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 8973
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: Non-PWM LED power
« Reply #18 on: May 28, 2018, 03:55:39 pm »
Resistor based ledstrips are not a professional solution and are not tested as such (hence the missing UL or other approval).
Low voltage devices do not need UL unless they have something else (e.g. lithium battery) that does.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf