DirectAC LED modules without driver and flicker

Dear friends,
just for giggles, I purchased a DirectAC LED module some time ago from this vendor.
http://www.ebay.de/itm/Direct-AC-22...hash=item35fd260b52:m:mxO2KL7R66dZNKBogeyjqzw

Then I got the idea about driving it with DC to get rid of the flicker which was very clearly present at 230V AC direct drive. When I waved my hand above it, I saw the typical flicker effect.

I saw that there is a rectifier directly at the beginning, so I soldered 3 pieces of 4.7uF 400V capacitors I had at hand to the output contacts.



The flicker was not noticable any more! Of course, adding a capacitor somehow defies the purpose of a solid state LED module. But only needing a capacitor instead of a driver has its merits:
- cheaper, caps can be salvaged from power supplies and such
- can be made bigger so that capacity loss is acceptable to some extent, thus ensuring a longer application life
- can be moved away from the LED module, so that it will not be heated, thus ensuring longer life
- film capacitors could maybe be used, with a potentially huge lifespan
- if caps fail, the LED module will still work, only with flicker, no worse than normal.

Some russians arrived at the same idea I just found out:

http://ledway.ru/topic8049.html

They see the problem that the module no runs at a higher effective voltage, thus the output increases, so they want to add current limiting resistors.

Provided I will find my IR thermometer, I will make a temperature comparison of running it with and without caps, and compare.
But for now, I have to say that this is a very fine option of getting to run a rather large LED module, flicker-free, with minimum cost and complication.

What do you think about it?

Subscribed. Interested in your results.
-increased voltage - how much of an increase are we talking about?

Well, I just measured the LEDs VF without caps, and its ~30V. With caps 40V, and the module gets uncomfortably hot. So instead of the 24ohm and 27ohm resistors, I soldered in 39ohm and 47ohm. Now I have ~19V VF. So the output power can be nicely dialed in by changing the resistors.
Edit: I just measured 28.5mA on the AC line, which translates to 6.56W, which seems about right. Clamped to a not-so-big aluminium block as "heatsink", I get ~45° over ambient, which is ok I think. I just ordered 2 pieces of the 15W version to put into another desk lamp like this. Seems like I will have a really bright solution for less than €10, surprising.

I have now finished my research with extremely good results.



The resistors of the modules must be swapped as indicated in the table:



Now I get pratically as much light as with the 2x 50W Epistar mod (run at 40W), but with only 30W input. Also, the lamp head temperature does not rise 45°C over ambient, as with the Epistars, but only 25°C.
So I have
- more efficient and cooler LEDs
- cheaper LED modules
- no flicker
- no cost for LED driver
- no driver lossses
- equal or even slightly better color temperature (~6000K instead of 6500K)
- easier to mount modules, although I had to reposition the center mounting hole due to sidewards area constraints
- no critical components despite the LEDs. Capacitor failure will only result in flicker, not darkness

I enjoy so called "filament" led bulbs, for fitting into existing chandaliers, but they have issues - low power ones, in E14 socket, have no driver, only capacitor ballast, and flicker too much. So I've designed a driver, which drives them with high frequency AC, so no flickering at all (and can be dimmed if desired). Currently, circuit is under development, but there is nothing special - all parts are taken from household CFL lamp ballast.