Dimming MR16 LED

[P. Short]

As I mentioned in another topic, the group-buy MR16 LEDs arrived last week. They appear to be nice and bright. Other than that they take 12V AC or DC, I don't actually know very much about them technically. Since I only asked for 3 bulbs (one red, one green, and one blue) I'm not very inclined to cut them open to see what is inside them.

Since then I've been playing around with them trying to get good dimming performance. What I would like to achieve is a smooth, linear dimming curve. The AC results were pretty horrible (very narrow dimming range, somewhat flashy operation). The DC results were a little bit better, although not what I'd like. When I power them with a some wimpy 9V battery (or a 125 mA constant current source) they dim more or less OK (most of the action is at the low end, and they are fairly constantly bright over the high end of the range). With a stiff 15V power supply they don't behave too well (almost no dimming range at all, they are either off or on). In this latter case I see huge current spikes when the PWM duty cycle is low, as though a large capacitor is being charged up (like John W reported in another thread on this board).

The controlling waveform right now is a simple PWM technique at 120 Hz. Later on I will play with using a variable period waveform (constant on-time, try to control the brightness by varying the off-time).

Does anyone know how d-light, LOR, etc control the MR16 LEDs? Do they use a similar scheme to what I've described, or do they have more complicated logic (such as measuring the current, or similar closed-loop techniques)?

--

Phil

[P. Short]

Some representative measurements on the WireKat group buy LEDs

Lamp Current Voltage
Blue 150 mA 7.6V
Green 150 mA 7.8V
Red 150 mA 5.8V

These LEDs seem to have a fairly big cap inside them. So, it would probably be better to dim them by controlling the current than by using PWM.

--

Phil

[P. Short]

Here is a proposed circuit for driving one of the easy-lighting MR16 LEDs from Renard. This is for use with the PWM code. The resistor is placed on the emitter of the transistor in order to limit the current going to the LED. The current to the LED is limited to about 3.4V/R, or about 150 mA with the indicated resistance. The reason for limiting the current is to reduce the current surge when voltage is first applied, to make the current flow through the LED more controllable.

This is circuit hasn't been tested yet, and is put up just as an initial proposal.

--

Phil

Edit...it seems to work fine (although I used 2N3904 transistors instead of 2N2222A). The 2N2222A transistors are preferred, since they have a higher current rating.

[P. Short]

Incidently, that resistor should be a 1W resistor (it will end up dissipating 1/2W when the light is on continuously).

With Red LEDs the transistor will also dissipate 1/2W, although it would be much less with Green and Blue LEDs.

--

Phil

[P. Short]

I'm beginning to think that the issue is more with the response curve of the Mark-I eyeball than anything else. Since I suspect that the response is more or less logarithmic over some range, and that a linear ramp would appear to fully on for most of the range. This is after prototyping a linear current ramp generator, and seeing pretty much the same thing as with the PWM control ramp.

--

Phil

[P. Short]

Playing with an exponential curve (to counteract the supposed logarithmic response of the eye), the ramp looks a lot better. There is action over the full range, although the response at the low end of the scale is not as smooth as I'd like (some of the LEDs cut off before the others).

--

Phil

[tconley]

Phil does this mean there is a working design for an ssr for dc dimming?

[P. Short]

My testing was on a solderless breadboard. Wjohn has a design for a DC SSR, but I don't know how far along he is in testing it with those types of lights.

--

Phil