How best do I smooth out the PWM dimming output from an LED controller?

[maks_3]

I have a 12v 10A/channel LED controller and I am trying to eliminate the stroboscopic effect from the PWM dimming that is used to create the impression of lower brightness.

To give some more information the LED strips should use up to 90W each (~7.5A per channel), I experimented running a capacitor across the output and it did seem to alleviate the strobe effect but it increased the brightness substantially and made the controller give off a high pitch whine which I assume was because the capacitor was drawing too much current during the on cycle and why the overall brightness increased.

I then used a higher voltage electrolytic capacitor so that it wouldn't be storing up as much energy during the full 12v on phase and given that it was just a lot bigger I figured it would be able to handle more total power without getting hot like the lower voltage one, but this still produced some whine in the controller and increased the brightness which in addition to stressing the controller made it so I still could not use the dimming function properly.

I was thinking of running one of those high current resistors between the capacitor and one of the LED lines to stop the capacitor pulling too much current and also putting a diode parallel to that resistor so the capacitor can dump its energy unimpeded when the pwm output from the controller moves into the off phase.

Would this work and if so what values should I use for the capacitor and resistor?

I would appreciate any advice including pointing out if this is just completely wrong to begin with.

Thanks

[Konkedout]

A couple of drawings would be worth 1000 words....  I would need to see a schematic of what you have and what you are proposing.

But the human eye seems to integrate (average) brightness of a rapidly blinking light, and PWMing is the most efficient way to dim an LED.  You start with a current level at which the LED works efficiently and then reduce the duty cycle to dim from there.  In addition to efficiency, an added benefit is better brightness tracking between multiple LEDs.  If you were to scale back a steady state current you might find that different LEDs might dim differently, while PWMing would work better that way.

If the blinking is objectionable, is it possible to increase the frequency?  That might tend to reduce the available duty cycle range, but I would think that 1 KHz  blinking would be much less visible than 120 Hz for example.

[Ian.M]

LEDs are current driven, though LED strips include a series ballast resistor for every small group of series LEDs to allow the strip as a whole to be voltage driven.  A parallel capacitor is very little use as the LED current will drop rapidly as the capacitor voltage decreases, making it hard to maintain brightness during the PWM off time, it messes with the control range as the capacitor fully charges near the beginning of the PWM on time, and it risks blowing your LED controller, as from its point of view the capacitor is much like driving a dead short.  Cheaper controllers will blow immediately if connected to a large capacitive load, so yours must be one of the good ones with fast current limiting.  However it is under severe stress.

To smooth the current through the LEDs, you'd need a big  inductor, and a high current Schottky diode, across the series combo of inductor and LED strip, to allow the current to continue, recirculating during the PWM off time.

However the cost (and size and weight) of an inductor capable of passing 7.5A dc WITHOUT SATURATING, and having enough inductance to maintain the current through the worst case PWM off time, will most likely put you right off the idea as it will almost certainly be more expensive than replacing the controller with one capable of high frequency PWM.

[tooki]

I have a 12v 10A/channel LED controller and I am trying to eliminate the stroboscopic effect from the PWM dimming that is used to create the impression of lower brightness.

Just do higher frequency PWM.

As someone who is very sensitive to PWM flicker, I'm an unapologetic warrior for high PWM frequencies. Many people think the frequency adequate for film projection or CRT displays (60Hz) is enough for lighting, and it absolutely is not, because point light sources aren't the same as images. (And anything that creates specular highlights as it reflects light becomes a point light source.) And it's well-known that our peripheral vision reacts far faster than central vision, which is why we often find PWM flicker in the periphery to be more annoying than when we look straight at it.

There's research out there that stipulates that eliminating flicker to the most sensitive people in the absolute worst-case scenarios (including head and eye movement) requires upwards of 30kHz PWM!

Now, in my experience, about 3kHz is enough for me for household lighting. But because that's a frequency within the audible spectrum, it can result in audible whine in the power supply or dimmer. So it's often advisable to use something ultrasonic, so above 20kHz. When I built the dimmer for the lighting under my desk, I targeted 25kHz, which ended up at about 26 or 27kHz after component tolerances.

[DavidAlfa]

Yep, just increase the PWM frequency to >25KHz , should work great.
Alternatively you can use inductive constant current drivers.

[Zero999]

The obvious solution is to increase the PWM frequency, but it's possible the OP is using an off the shelf controller which is not adjustable. It might be possible to modify it by changing a crystal or resistor/capacitor value, but that could increase switching losses and reduce the power capacity of the controller.

I would consider rolling my own controller, with a comparator IC, or MCU, depending on the requirements.

[maks_3]

I made sure to test it with fewer LEDs than I intend to use eventually and with a power supply that can't supply that much current so that might explain why it didn't break the controller.

Given that I already have the controller and quite a few components from old electronics I would rather try to smooth the PWM than buy a higher frequency controller but I'll bear that in mind if this doesn't work, maybe it's not rational but I just don't like the idea of looking into full brightness LEDs let's say 20% of the time, with pupils that are much wider than they would be if it was full brightness 100% of the time.

I've included a picture of what I was thinking of doing, but even if it could work I don't know how to calculate what size resistor I would need in combination with what size capacitor to ensure the LED + capacitor current draw does not exceed 10A.

Can you tell me more about how I would go about calculating what sort of inductor Schottky diode combination I should use?

Thanks again everyone

[Zero999]

That won't work. The capacitor will charge slowly via the resistor, then very quickly thought the LEDs, so there'll still be a lot of flicker.

An inductor, with a diode connected in reverse parallel with the output of the controller, forming an open-loop buck converter will work. The problem is, the lower the frequency and higher and smoother the output current, the larger the inductor. You haven't said what frequency it is, and how smooth the current needs to be, so it's impossible to calculate, but at 10A and a frequency low enough for there to be visible ripple, it will need to be very big, like a decent size transformer, probably larger and more expensive than the controller.

Here's an application note dealing with how to size an inductor for a buck converter, which is similar to what you're doing.
https://www.ti.com/lit/an/snva038b/snva038b.pdf

[maks_3]

So the power supply that I will be using has the ability to vary the output voltage in something like the 10-14v range so while it would have theoretical max power (though I think the seller exaggerated a bit) of 180w for across both strips (so 7.5A/channel*2) I was thinking that if I ran the power supply at the lower end of its voltage range it would hopefully give some more headroom between the current each strip of LEDs draws and the 10A that it can supply per channel.

Unfortunately the controller I am using does not have any branding on it for me to look up more information such as the frequency it does PWM dimming at. It will take me some time to get my head around the document you linked so can I ask, assuming I can run each channel at ~5A and it has a fairly average PWM frequency (maybe 500Hz?) and I want the output from each channel to be smooth enough for the LEDs to vary imperceptibly in brightness, can you give me an idea on what sort of values for inductors and other components I would need to use?

The main reason I bought this LED controller was so that I can vary the colour temperature (and brightness) by varing the intensity of warm and cold white LEDs via a remote, so if could figure out a rough sense of the cost of the components that I'd need to use, and it turns out to be quite expensive to average out the voltage in that way I can then decide to just buy two of those RF controlled motors that rotates a potentiometer for controlling for example volume on audio equipment instead, and just solder those potentiometers in place of the potentiometers on two suitably powerful DC to DC step down converters and achieve wireless dimming and colour temperature control that way, assuming I can match the values on the potentiometers.

[Vovk_Z]

You are ignoring facts that LEDs are current driven and an inductor must suite better to smooth a PWM. Ideal capaitor may consume indefinite large current to charge. Real capacitors and real circuits have resistance so you have finite but still large charging currents. But inductors tend to smooth all currents which makes a life of all componets easier.

[Zero999]

LED strip contains built-in resistors and requires a constant voltage source, not a current regulated supply.

The LEDs look like an approximate voltage source, with a series resistor. 12V white strip consists of many strings of three blue phosphor covered LEDs in series, each with a a single resistor. The voltage drop of each LED is roughly equal to 3V and doesn't vary much, depending on the current. The three LEDs in series have a total voltage drop of 9V. If the voltage is varied, the LEDs won't turn on, or be extremely dim, until the voltage reaches 9V, then the brightness will linearly increase towards 100% at 12V.

If the PWM is filtered with an inductor and capacitor, it will convert the PWM into an approximate voltage source. The brightness will no longer scale linearly with the duty cycle, although it won't flicker.

[maks_3]

So that's one of the things I was hoping to achieve by doing this, as I would sometimes like for these LEDs to be able to run at a very low brightness, I don't know if increasing the value of the inductor will increase or decrease brightness at the same duty cycle, but I was thinking of just getting plenty of quite cheap matching smaller inductors and running them in parallel until the current across each indicator matches the specifications and just increasing the number in parallel depending on how it affects the overall brightness.