Dimmer / Power Supply Noise Filtering

[orinocopaul]

I am powering two LED strips via a 12V switched mode power supply. In-between the LEDs and the PS, lies an rf dimming controller. The problem is that when I am half dimmed, I get a 7.8 kHz audible hum (see attached spectrum and oscilloscope snaps). What is the best way to target and inhibit this 7.8 kHz hum?

[John B]

I'm assuming that the dimmer works via PWM control of the output? If so, drawing pulses of current from a switchmode supply is an unfriendly load. Usually you'll get that whine, and in the worst case it destabilises the PSU control loop.

A low side LED driver configuration with an inductor, cap and diode is the better way to go for filtering a pure PWM controller, but assuming you can't change that, a CLC filter (a "pi" filter) with bulk capacitance in the mF range between the PSU and controller should help.

[orinocopaul]

PWM controller. Yes. Could you please elaborate on the low-side LED driver configuration?

[John B]

Falstad Link

This link gives you an idea of the current loops involved in a PWM controlled circuit. I forgot that the method I was thinking of will also require a (C)LC filter on the input, as you can see how the current waveform drawn from the PSU looks. Not a bad idea to filter the LEDs too, as I've always found PWM controlled lights to cause too much interference.

[orinocopaul]

wow John! Great visual simulation. I think I got the idea. A low side driver switches directly to the ground.

[orinocopaul]

I cannot modify the dimmer so I'm going to try filtering the voltage between the SM-PSU and the dimmer.

I have 12V LED and the current varies from 1A to 4A at maximum intensity. The 7.8 kHz whining is reaching maximum amplitude at ~2A current. Could you please provide some coordinates for the selection of capacitors, inductors and resistors? What values should I pick for the CLC Pi filter?

[John B]

I would start with several parallel 1mF low esr electrolytics on each leg of the pi filter, with a ~5-10uH toroidal inductor. Do you know what the PWM frequency is? If it's a few hundred hertz, it may be too impractical to filter it on the PSU side. Give the pi filter a try though.

[orinocopaul]

Do you know what the PWM frequency is?

It seems the PWM frequency matches the audible frequency as well: 7.8 kHz. This is how it looks at different levels of intensity.

[Raj]

I am powering two LED strips via a 12V switched mode power supply. In-between the LEDs and the PS, lies an rf dimming controller. The problem is that when I am half dimmed, I get a 7.8 kHz audible hum (see attached spectrum and oscilloscope snaps). What is the best way to target and inhibit this 7.8 kHz hum?

Can't the PWM frequency be increased? Is it derived from a microcontroller? (if it is, maybe you can change it's clock speed for a faster one.

[orinocopaul]

Can't the PWM frequency be increased?

I'm going to try out the Pi filter. I have no idea how to re-program the MCU.

[orinocopaul]

I would start with several parallel 1mF low esr electrolytics on each leg of the pi filter, with a ~5-10uH toroidal inductor.

Here is my first attempt at filtering out the PWM noise. I used two 1.000 uF caps (EEU-FR1E102) and one 7.8 µH inductor (PA0431LNL), following John's advice. The second post-inductor cap seems to be quite effective at bringing the noise levels down. The pre-inductor cap can be taken out without any perceptible outcome. Here are some before and after snapshots. Any ideas for further improvement?

[orinocopaul]

I am powering two LED strips via one 12V switched-mode power supply. In-between the PS and the LEDs, there is an rf dimming controller. The problem is that I get a 7.8 kHz audible hum inside the power supply, which correlates to the PWM dimming frequency. What is the best way to target and filter this 7.8 kHz hum/whining?

I tried a pi filter recommended here and this is what I got. How can I improve this?

PS. I'm using 1mF low esr capacitors and 7.8 µH inductors.