Output current do vary with line voltage as shown on p.13 (line/load regulation) of DER337. Please check if the variation you see is close to the variation on the report. If so, then that's the normal design behavior. Please review your design requirements and find out if the variation is acceptable and if production testing can be adjusted instead.

I'm also working on DER-337

Everything work fine if we did startup on 190-265Vac

The output is around 70V 0.24A while input is 18.6W. Efficiency is around 90%

Problem happen when we try to lower the input to 160Vac after power up @ normal voltage (ie 230vac).

While we keep lowering the input everything still on the regulation. But if we raise the input voltage back to 230Vac again the output is more and more which can consume upto 23-24W @230Vac. If we keep raising the input voltage upto 260Vac the consumption still high until we lower the input back to 230Vac then the consumption is back to 18.6W as normal.

We also try to do the same test by turn on @230Vac with 18.6W consumption.

Then lower the input to 160Vac the consumption still 18.X watts

But raising up to 230Vac again we got 24 watts.

If we cut the power and turn it on again we got 18.6W @230Vac

Quite weird but don't have a clue. May be you can help me.  

Hi, this is expected in the device regulation characteristic. The device is a digital control wherein the steady-state switching frequency and on-time are determined by the line voltage, voltage drop across the LEDs and converter efficiency. 

PI-Troi

@PI-Troi,

Thanks for the answer. When you say this is expexted, does it mean that this is normal and we can not improve anything? Is DER337 can adopt in the commercial product?

 I'm still curios to remove this efffect. It seems like the hystersis is too wide. If we can lower the hysteresis then it helps. But I don't know how and where to modify. Otherwise the tube has tendency to consume 18 or 24W at the same normal line voltage (230Vac). Also the coil will be very hot @ 24W output. Or we have to overrated the coil.  

Hi, if you have modified DER-337 make the sure the device doesn’t operate in CCM mode. Please use the PIXI transformer designer software. Typically for buck topology the regulation would be around +/- 10% over line input (hysteresis factored in) , since the output is indirectly sensed through the freewheeling diode current expect a wider regulation tolerance. Flyback regulation would have better regulation since the output current is directly sensed. Not sure though why you got a wider data points. 

Hi Troi,

 I'm still working on DER337. Seem that L4 created lot of radiated noise. I'm using AM radio as an indicator. The CKT can interfere any AM radio sation a meter away. Changing L4 from magnetic shield to dog bone unshield get worst.

If we let the driver float in the air it is much better. But put the CKT into aluminium T8 tube is a nightmare. The AL tube act as an antenna. No matter L4 is magnetic sheild or not. Any recommendation to improve this?

Thanks