[sdiy] LEDs and noise

rsdio at audiobanshee.com rsdio at audiobanshee.com
Fri May 8 04:36:54 CEST 2015

All current flows in a circuit. This means the dirt will be on both sides of the LED, no matter how you connect it. Using a "dirty ground" scheme helps you isolate the LED currents from other devices that need ground as a steady reference. But if you have any circuits that depend on the supply voltage for accuracy, then you'll need to keep the noise off the power rail as well. That means you'll need a "dirty power" as well as a "dirty ground."

Op-amps have pretty good rejection of power supply noise, but it's not perfect. I wouldn't rely on the PSRR to clean up all power noise. It seems like a good idea to use unregulated voltages for the LED circuits, and use the regulator to isolate the sensitive circuits from any dirty power. But I think you still need filtering to prevent any noise from traveling from your module into the backplane rails. So be careful.

By the way, if you run your LEDs between the + and - rails, then your resistors will have to dissipate twice as much wattage. I'd suggest splitting the LEDs half and half to each rail, using ground for all of them, of course, but dirty ground. As long as your dirty ground has its own path back to the supply without any non-LED circuits using that ground path, you should be fine (don't be like the TR-606 headphone output and mix ground traces).

I'm working on a circuit that uses an analog comparator with an open collector output, so the gate LEDs run from ground through the collector to the - rail. The comparator can only sink current, not source, so I had to invert things. I had the LEDs tied to the + rail at first, but realized that I needed double the wattage for my resistors, so I switched the + to gnd.

Other LEDs in my design are running on positive voltages, so I'm using the + rail and gnd. It ended up about half and half.

I'm not using any LED PWM, that would surely dirty up the power. Instead, the LED drivers are all linear. This means that the "dirt" on ground and the rails is more of a slight DC shift than real noise. But when you add 10 mA to 40 mA to those rails, especially with ribbon connectors, there will be some voltage bias - think spikes and ground bounce.

Brian Willoughby
Sound Consulting

On May 7, 2015, at 4:32 PM, David Moylan <dave at westphila.net> wrote:
> If you're dealing with a lot of "blinkenlights" what's the best way to handle them?  Is it better to power them between the + and - rails or from one rail to ground?  Is it worth using a "dirty ground" scheme, or is the noise induced on the supply rails itself?

More information about the Synth-diy mailing list