News For LED: Recent Patents: Deep Dimming

Sunday, November 24, 2013

Recent Patents: Deep Dimming

Here's a simple way to defeat flicker during extreme dimming. I like this patent a lot.

In my most recent patent blog, I wrote about a patent for which I couldn't see any practical use. As a counterweight, today I'm writing about a patent I like. It solves a rather challenging problem, and it does so without using a microprocessor or lots of circuitry -- just two transistors. It's the sort of design I like to do for my clients: simple and cheap. My motto is "You can fix almost anything with a resistor."

The title of patent No. 8492992 is "LED lighting device and illumination apparatus." (As an aside, I have noticed that "LED lighting device" is a real popular phrase in patents these days.) The patent was issued to H. Otake and K. Asami and assigned to Toshiba Lighting. As usual, we are going to look at what problem the patent is addressing, how the invention works, and where we might expect to see LED lighting embodying this patent.

The problem

This patent involves what it calls "deep dimming" of LEDs, which is dimming to very low light levels. It says that when an LED is deeply dimmed, "the visual sensation of a person tends to sense flickering of the LED more clearly." It's a surprise to people who are just entering the field that, even though an LED is rated at 350 mA, it's still quite noticeably on at 1 mA or less. That's because of the logarithmic response of our eyes.

Figure 1

The patent gives a really interesting reasoning for why the LED flickers. It says that at very low current, the LED is close to the knee of its I-V curve. At that point, even tiny amounts of variation in current result in significant variation in voltage, which turns off the driver. In short, the driver hiccups because even a single switching cycle of current changes its operating point.

On a dimmer, it's even worse, since the holding current of the triac is so high that the dimmer oscillates. One way this is handled in conventional circuitry is to add a dummy load at the front of the driver. It turns on when the LED load is light, and it pulls enough current to keep the triac happy. But the circuitry is rather complicated and expensive. It's usually done by using an IC with such a control built in.

Figure 2

The method

If you look at Figure 1 (part of which is seen up close in Figure 2), the inventors use the base-emitter junction of a transistor (Q5) across a resistor (R5) sensing LED current. When the current is down below some level, the BJT turns off. That turns on the other BJT (Q4), which starts pulling current from the driver in parallel with the LEDs. The inventors arrange it so that the amount of current it pulls is enough that the converter can deliver a full inductor's worth of current in each cycle, and thus the fluctuations are gone.

I like this. These devices can use an ordinary, inexpensive driver IC. The whole circuit that has been added takes two bipolars and three resistors. It doesn't save any energy compared with the dummy load method, but it's cheap and it's simple -- the whole patent has only two claims.

Likely applications

Since this was assigned to Toshiba Lighting, I'm betting that Toshiba bulbs have had a problem with flicker during deep dimming in the past. So I expect to see this technique used in new Toshiba bulbs, probably A19s -- if it's not already there.

Related posts: