LED vs. MH/HPS

[nomofatum]

LOL fair enough I run 5000K COBs at 227lm/W and building a new 3000K that will run at 191lm/W. On top of that, white LED lumens carry a lot more energy than HPS lumens. Not bragging or hating on HPS, just pointing out what is possible.

To answer your question OP, don't do it, 99.9% chance that LEd is a junker. You can get a 1000W HPS DE for $450.

Yes, they would carry more energy than HPS lumen per lumen since HPS is of a higher wavelength than most LEDs, the lower the wavelength the more energy per lumen or per photon. I'm curious if you have measured that real world or if that is what you suppose you are getting based on specs? Do you include all watts consumed, including any wasted power for extra fans and power supply loss in those numbers? If you are getting those numbers real world, and they don't cost an extreme amount to build or require elaborate cooling they would be the stand out winner.

 

[SupraSPL]

So the issue can be examined when we look at the LER. LER is: how many lumens a particular light emits with 1 pure radiant Watt. For HPS it is 380-390 lumens. For warm white LED 325. For green laser it would max out at 683.

The reason HPS LER is higher than 3000K LED is because HPS output more closely matches the response curve of human vision. Therefore, 1 HPS lumen represents a lot less photons than a 3K LED lumen. And 1 lumen from a green laser represents a lot less photons than the HPS lumen. Actually 1 lumen from a green laser represent the minimum possible number of photons for any light source.

So this confusion is ALL due to the weighting of light to human vision. We use LER to remove lumens from the equation, so we can work with radiant efficiency, PAR W and PPF instead.

Yes, they would carry more energy than HPS lumen per lumen since HPS is of a higher wavelength than most LEDs, the lower the wavelength the more energy per lumen or per photon.

You are onto something here. The photon energy issue is a separate one from the lumen issue. 1 PAR W of HPS light is the same amount of energy as 1 PAR W of 3K LED, but the photon count is different and that is what drives photosynthesis. Red photons are less energetic and therefore require less energy to emit/create. On top of all that, we suspect that an even spread across the visible spectrum is ideal. So we want an even distribution as long as we are not sacrificing emitter efficiency to achieve it.

 

[SupraSPL]

I'm curious if you have measured that real world or if that is what you suppose you are getting based on specs?

Very good question. I have done testing with a simple lux meter and very accurate pair of multimeters to verify that the lux/W actually increases as I decrease the current. This verified that the current droop graphs are correct, even at very low currents I see large gains of efficiency. I have also done testing to check temp droop on variety of heatsinks and COBs and estimated junction temp. In practice, I measure almost zero temp droop when running soft in room temp conditions and no more than 5% when run very hard. This means when running very soft we can use the Tj 25C lumen figures in the manufacturer graphs and we can trust the temp/current droop graphs in the CXA/CXB PDF. Actually, output may be even higher because the Cree graphs are based on 25C case temp and we ha no idea what Tj they are assuming but it is hight than 25C.

Another data point. I used the luxmeter to check if COB truly do crush the performance of the single emitters that we are familiar and have been thoroughly tested in laboratory conditions. Short answer, yes they really do.

Next data point, @stardustsailor has had CXAs and Vero tested in a lab with calibrated equipment and he reports that the COBs do emit above the minimum lumen figures listed.

So with all that in hand, I count pixels on the graphs to extrapolate the lumens/W. So I have high confidence that they will emit the figures I mentioned above. Those are the "typical" figures. Some will be slightly higher, some will be slightly lower.

Do you include all watts consumed, including any wasted power for extra fans and power supply loss in those numbers? If you are getting those numbers real world, and they don't cost an extreme amount to build or require elaborate cooling they would be the stand out winner.

Those numbers do not include power supply losses or fan power. They are striclty based on LED dissipation power. I think that is important for comparison because some builds are run passive cooled. Some builds use 83% efficient drivers and some are 94%. Also, HPS lm/W figures do not include ballast losses or fan power if cool tubes are used etc.

The lamp that I claim 227lm/W is passive cooled and powered by a 94% efficient driver. When dimmed below 200mA, it is significantly higher than 227 lumen/W.