Diode selection for an alarm clock

I bought an LED matrix alarm clock, which has about 30 white LEDs. The clock has three brightness settings - high, medium and dim.

I found that even on the dimmest setting, the clock still gives out a lot of light, so I opened it up to have a look inside.

Basically the LEDs are strobed in rows (as you would expect). However, on the medium setting, two 1N5817 diodes are brought in series with the LEDs, and on dim mode four 1N5817 diodes are put in series.

To lower the brightness further, I decided to add another diode to each string - thus medium would be 3 diodes, and dim would be 6 diodes.

Measured with a Fluke 87v, the Vf of the 1N5817 diodes is 0.57v. I had some 1N5819s in my parts box, and these measured 0.12v (surprising!). I therefore just stuck in some 1N4001s I had lying about.

Obviously (when you see the If/Vf graph), the 1N4001s Vf changes massively with VI, so all the rows are different brightnesses. Not great.

Should I just buy some more 1N5817s? Or is there something better I can use? The stock 1N5817s do have some difference between row brightnesses, but it isn't as bad as the 1N4001s. Any ideas?

Thanks

Andrew

Probably a silly answer, but why not series resistors instead of diodes? For a given current, they'll drop a given voltage...

The LED matrix is being strobed, so using resistors would give uneven brightnesses to each row, as the current for each row is different.

Andrew

The 1N5717, 1N5718, and 1N5719 all have the same characteristics except for reverse voltage rating. These are Schottky diodes and have a low forward voltage, so the 0.12V you measured for the 1N5719 at the low measuring current of your multimeter is normal, not surprising.

Don't know why you measure 0.57 Vf on the 1N5717's. Are you sure that's the correct part number?