Linear regulation gives mains module 85% efficiency

Supertex has got together with Osram Opto to make a mains-powered light engine using only linear regulation.

04sep13SupertexOsram 594

Linear regulation usually means dreadful efficiency, but Supertex has a neat trick where its CL8800 ‘sequential linear LED driver’: it switches series strings or LEDs in and out through the mains waveform.

The result is that the linear regulator only ever has 10 or 20V across it and LEDs and “an electrical efficiency of 85% is achieved with up to 1,600 lm output, while reaching >120 lm/W”, said Osram on its ‘LED light for you’ website.

As it presents essentially a constant current load to the mains, power factor is >0.95 with <20% THD line current.

04sep13Supertex 543The diagram simplifies things a lot and in practice there are more LEDs in some strings than others.

There are 11 of Osram’s recently-released Duris S8 LEDs in the reference design, which are 5.6×5.2mm devices with either six or eight die inside (18.6-22.2V or 24.8-29.6V, 390 or 500 lm (typical 25°C 3,000K))

“The module is very compact [50x50mm] and requires no external components, it fulfils all EMI standards and conducted emission is near to zero. It is fully isolated, can be directly mounted on a heat sink and has an automatic current derating at high temperature,” said Osram. “With selected dimmers, phase dimming is possible, too.”

Measurement results including driver:

4000K, 11W: 1,435 lm, 131lm/W

3000K, 11W: 1,289 lm, 117lm/W

4000K, 17W: 2,210 lm, 131lm/W



  1. I was also wondering how constant current, as opposed to constant resistance, could give a high power factor.
    I notice taps 1 and 2 draw 90mA, and taps 3,4,5 and 6 draw 115mA in the spec sheet, so there is a slight attempt at raising current as voltage rises.
    With all the rectifier-capacitor loads out there on the grid, you would think the power companies would be happy to see a bit more current drawn out in the wings of the sine wave, away from the peak.
    Steve B

  2. I’ve seen the CL8800 datasheet and thought it looked like a great product. I had considered building something with it, but the package is just not suitable for a hobbyist. As such, it’s good to see Osram design a product with it. The specs look very good too… 113lm/W is excellent, especially compared to the Seoul Acriche 2 modules I used in a recent design.

    I’ve wondered about the use of a constant-current regulation scheme, though, in terms of achieving a PF near unity. It seems that a PF = 1 would result in the current varying in proportion to the mains voltage. I haven’t calculated PF since college.. my old text says that PF is average power over apparent power, and then goes on to consider sinusoidal voltages and currents. In this case, the load is not reactive in any way, but is clearly not simply resistive. Interesting problem….

Leave a Reply

Your email address will not be published. Required fields are marked *