Hi All,

We have been working with the Link-switch-ph in a number of projects and have had a number of issues, most issues we have solved except for digital dimming.

Firstly you might say why not use analogue dimming? Well not all LEDs can be dimmed in this manner, Osram for example cannot guarantee predictable operation of their LEDs below 100 and sometimes 200mA so in this case analogue will not work.

The issue we are having with Digital dimming is with the light quality. We can get the LEDs to dim right down to 0.3% reliably however we introduce a problem which at different PWM frequency's looks to appear differently but after a lot of probing seams to be caused by the same problem.

At 100hz PWM dimming we could see that the light level would vary level up to approx 50% over a 0.1-0.05 hz cycle.

Increasing the frequency reduced the variance in light level until it looked to disappear however @1K- 4K PWM we would see slight flickers in the light level. This might not be seen on a test bench but is enough to ruin the quality of the light and would give people eye strain if used in general lighting.

We have determined the cause of this to be related to the drain voltage, PWM level and Current limit.

-At 100hz PWM Dimming on a system that is running @240VAC 50HZ we have determined that the frequency's are slightly not matched, so every cycle the dimming waveform is propagating along the mains frequency. At different points in time you will have the PWM on cycle over the mains Zero crossing or close to. Because of the input voltage(low) and the current limit (on the Link-switch) the LED string cannot be run correctly during this cycle, Over time (as they are not in sync) the brightness changes.

-At higher frequency this effect becomes less until it is not noticeable. At 1khz and higher this problem looks like occasional flicker, but it looks to be the same issue.

It looks like Triac dimmed and Analogue dimmed applications don't have this issue as they are synced to the input waveform anyway.

We have also removed this issue with the use of electrolytic s on the input line, however this will alter the power-factor and this design was chosen to remove the electos.

Our current test set-up is designed for 9-12LED, 700mA using the linkswitch 419 ( we have tried with 10uf(low) and 100uf(high) output mode selection)

We have invested a lot of time and money on this design and it would be good to have a PI engineer evaluate this problem and verify that our findings are indeed correct.

Any other help in finding ways around this issue would be appreciated

Nidan

(to restate if it didn't make sense)
-The Frequency of the mains input and the PWM are not the same and are not synced
-When the PWM turns on the the Drain voltage is too low the device cannot pull enough current to turn the leds on correctly)
-the output and led power varies overtime because of this
-At low PWM frequency's it looks like changing LED power (brightness)
-At high PWM frequency s it looks like slight flicker