On Thu, 16 Nov 2017 11:57:22 +0000, Martin Brown wrote:

The 300Lm/W was achieved on a near infinite heatsink under perfect
laboratory conditions with a fabrication technique that doesn't easily
scale up - basically a one off research device made of unobtanium.

Around 130Lm/W is the current production device which is still
impressive and better than HPS (but still short of 220Lm/W LPS).

As far as what is on the shelves you're hard pushed to get over
100lm/Watt. I quite like the newer stick filament type, in standard
sort of clear envelopes not the fancy "period" ones. You have to
watch the power and lm numbers as not all are 100lm/W. The warm
white colour temperature is good, and light dispersion very similar
to perl tungsten. They don't appear to have any electronics either.
Not had this type long enough to comment on life.

The LED failures I've had have been the chips falling off the
board/heatsink...

Also recently swapped some 5' 58W flory tubes to LED. The first was a
bit of a distress purchase and is a Philips 20 W 2000 lm that cost
around £15. Two others came from Aldi at £4.99 each (flogging 'em
off), 22W 2000 lm. Now a 58 W flory is supposed to chuck out 5000 lm
so at 2000 lm was a little concerned that the LED tubes would be
rather dim. In practice straight after swapping think, "hum it's not
quite as light". But after using them for a while I no longer notice
or think there isn't enough light.

--
Cheers
Dave.