cheap DIY LED

Slippertalk Orchid Forum

Help Support Slippertalk Orchid Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Ray & Tyrone, yes, although PPFD of PAR is closer to what plants want than lux/footcandle, and we use PPFD in publication of horticultural (and plant) science, it is not perfect as you mentioned. If the emission spectra are known, Yield Photon Flux (YPF) can be calculated, and it incorporates the action spectrum of plants (i.e. red light is more useful for photosynthesis than blue). Even with YPF, it is not a complete predictor of plant growth (Fig. 8a of the poster below).

This poster is an easy demonstration, and there are lots of things we can learn from it:
http://cpl.usu.edu/files/publications/poster/pub__2576523.pdf

There are several reasons. Action spectrum of plant photosynthesis is calculated under monochromatic light (measure PS rate with a light with narrow range of wave length). So it doesn't quite match with photosynthetic rates when light with several different wave lengths coexists (e.g. white LEDs). One example is the usefulness of green light depends on the intensity of light (when there are lots of red and blue light, the usefulness of green light increases, Fig. 8b). Also different spectra influence the photomorphogenesis (e.g. blue light makes more compact leaves, reduces stem elongation etc, top fig. of Fig. 7 and Fig 8), which is not reflected by PPF or YPF.

Although PPF is not a perfect measure, but it is pretty decent estimate of "usefulness" of light. Take a look at the bottom fig of Fig. 7. 4 blue diamonds (500 micromol/m^2/s) have higher dry biomass than 4 red squares (200 micromol/m^2/s) regardless of proportion of blue light. More PPF gave more growth in this case (obviously the relationship between PPF and growth isn't linear). The proportion of blue light changes almost no blue to mostly blue, and yes, red light (less blue) gives more overall growth, but the effect is not as dramatic as what you expect in terms of total dry mass. So from my measurement, since 3000K and 5000K COB LED showed similar PAR, I guess that neutral to warmer LED would be the sweet spot. But the shape of plants could be somewhat different.

Ideally, it would be nice to do the experiment similar to the lettuce with several species of orchids, but it would take much longer than with lettuce!

Alla, XHP isn't available yet.

Tyrone, yes, I didn't know about the Apogee issue until DavidCampen pointed it out. Percival Growth Chamber comes with a Apogee sensor, but they may have been calibrated for florescent light which is used in the chamber. There are several sensors with nice spectral response other than Li-Cor (which is commonly used among scientists).

http://www.apogeeinstruments.com/content/Quantum Sensors-LEDs.pdf

But all (except Apogee) is expensive. Mine is an old Li-Cor. Even Li-Cor has the problem with sensor shift (blue response become weaker).

http://cpl.usu.edu/files/publications/factsheet/pub__6622553.pdf

So they need to be calibrated periodically (and you have to specifically mention blue shift calibration). Calibration costs almost $200, so I haven't calibrated mine yet.

For those, who are interested in this kind of topic, I put bunch of related link to this thread, which I should update at some point.

Similarly, there is a HUGE variation in the spectral sensitivity of consumer level lux/footcandle light meter. So values from one meter may not be comparable values from another meter. But they are still useful to give some rough idea.
 
Great links! Thanks. That poster has lots of cool data.

There is a "cheap" quantum sensor in the form of the Seneye for aquariums. It would work just as well if all you want is a relative measure. The Skye and Kipp & Sonen seem the best options for a "true" measure.
 
Great links! Thanks. That poster has lots of cool data.

There is a "cheap" quantum sensor in the form of the Seneye for aquariums. It would work just as well if all you want is a relative measure. The Skye and Kipp & Sonen seem the best options for a "true" measure.

I didn't know about this product, thanks! It seems that the response of Seneye sensor is worse than Apogee, and Seneye isn't much cheaper than Apogee, isn't it? If you connect Apogee to a multimeter, the cost is $155 or so.
 
Of course I did.
The instant message I got back was a link to the online-shop with some hundreds of drivers to choose from...
 
Tyrone, yes, you can just get the Apogee sensor (maybe attach an alligator clip?), and use a digital multimeter to read ppfd.

Fibre, does your source carry 1150mm version of XF-3535L? In the US, I think that it is not in stock yet (at least for a small scale order).

With regard to the driver, the first two mean-well drivers are constant-voltage drivers, which aren't appropriate. You need constant "current" drivers. The last one is CC, and it seems to be good. It output 1050mA (should work with XF3535L), and the range of voltage is 9-48V (CC driver gives constant current and adjust the voltage to achieve the target current). Max efficiency is 87% which is usually achieved when the voltage is close to the max (48V in this case). When it is driven at around 36V (XF3535L), the efficiency is a bit lower, so it is probably around 85% (just a guess). But see below for a similar (and probably better) option where you need only 1 driver instead of 2 drivers to drive 2 LED bars.

The cheap chinese driver which I linked has a similar efficiency. I measured it to be 85%, but it was at 21.83V, so if you use it with 1150mm version (36V), then the efficiency should be higher than this. While the measured current was 906mA, the voltage range is a bit short 18-35V. It probably works, but no guarantee.

I just did quick search, and this ebay driver seems to work. Also, FastTech has this 30W driver, which can go to higher voltage, but it has a low power factor (I don't understand "power factor" fully, but higher is supposed to be better). I don't have thise models, though, so I don't know the actual current or efficiency. From my experience buying and testing cheap Chinese drivers, efficiency could be decent to surprisingly good, but the actual current could be a bit off. You can look for a driver, something around 800-1050mA, and the voltage range contains 36V (or slight higher than that). Usually, they are sold as 30, 40, or 50W drivers.

But if you prefer name brand drivers (Meanwell etc), here are a couple options (I didn't go through all possible ones).

LPF-40-42 (960mA) or LPF-40-48 (840mA), efficiency close to 90%

LPF-40D-42 and -48 have the ability to dim.

One of LPC-100-1050 can drive 2x XF-3535L at 1050mA. This is probably more economical option than 2x LPC-60-1050.

I agree that Mean Well site should have a better way to search for a driver. If you want to go with name brand driver,s I can probably look into more details about other MeanWell models.
 
Yeah, I found a source for LUXEON XF-3535L!
I will give it a try and buy a pair of L235-4080AULM5JAI0 and L235-5080AULM5JAI0.
Now I'm not sure what driver is the proper/best one.

This is my preselection:
APV-35-36,
LPV-35-36,
LPC-60-1050

Are these drivers suitable for the long LUXEON XF-3535L - 900mA - 36V?

The APV and LPV models are constant voltage drivers and I presume that you do not want a constant voltage driver but rather want a constant current driver like the LPC or LPF models.
 
... You need constant "current" drivers. The last one is CC, and it seems to be good. It output 1050mA (should work with XF3535L), and the range of voltage is 9-48V (CC driver gives constant current and adjust the voltage to achieve the target current). Max efficiency is 87% which is usually achieved when the voltage is close to the max (48V in this case). When it is driven at around 36V (XF3535L), the efficiency is a bit lower, so it is probably around 85% (just a guess). But see below for a similar (and probably better) option where you need only 1 driver instead of 2 drivers to drive 2 LED bars.
...
But if you prefer name brand drivers (Meanwell etc), here are a couple options (I didn't go through all possible ones).

LPF-40-42 (960mA) or LPF-40-48 (840mA), efficiency close to 90%

LPF-40D-42 and -48 have the ability to dim.

One of LPC-100-1050 can drive 2x XF-3535L at 1050mA. This is probably more economical option than 2x LPC-60-1050.

I agree that Mean Well site should have a better way to search for a driver. If you want to go with name brand driver,s I can probably look into more details about other MeanWell models.
The MeanWell LPC-100-1050 can drive up to 96 volts which makes the output significantly dangerous. I prefer drivers that will not exceed 48 volts.
http://www.trcelectronics.com/led-power-supply-class-2.shtml

I had been looking at the MeanWell LPC series for my next purchase. The LPF series is similar but more expensive because it has active Power Factor Correction. I think that most residential users do not care about Power Factor since they are not paying for reactive power.
http://www.nist.gov/pml/div684/power_121509.cfm

This place seems to have the best prices for MeanWell drivers: http://www.trcelectronics.com/Meanwell/power-supply-lpc.shtml
 
Last edited by a moderator:
Thanks a lot for your help, naoki and David!

Now I have to check all your links and so learn more about LED-drivers.

The Luxeon XF-3535L I like to use is the one of 1150 mm length, 36V, 900mA, 157 lm/W, 4536 lm - 5076 lm.
 
The MeanWell LPC-100-1050 can drive up to 96 volts which makes the output significantly dangerous. I prefer drivers that will not exceed 48 volts.
http://www.trcelectronics.com/led-power-supply-class-2.shtml

I had been looking at the MeanWell LPC series for my next purchase. The LPF series is similar but more expensive because it has active Power Factor Correction. I think that most residential users do not care about Power Factor since they are not paying for reactive power.
http://www.nist.gov/pml/div684/power_121509.cfm

This place seems to have the best prices for MeanWell drivers: http://www.trcelectronics.com/Meanwell/power-supply-lpc.shtml

Thanks, David. This guy also says that PF is not so important. It mentions that low PF for residential appliance has only small impact environmentally (in addition to no impact with electricity bill). So the FastTech (or other eBay) driver with low PF and no PF correction should be ok. But with the price of LPC-60-1050 linked by David ($17.5), a bit extra cost for the name brand might be a better option. I don't know how well Octopart covers European sources, but this place has it for 14 euro. Octopart couldn't find LPC-100-1050, so it might be hard to find it any way.

With regard to voltage, we all know that higher voltage is scarier, but I wonder how much more dangerous it is in practice (both medical and fire hazard). I put my fingers into gel electrophoresis (accidentally), but I don't remember the voltage (I guess it wasn't so high because it was just mildly ouch). I was thinking that HLG-185H-C1400 ($69.10) + 4x Vero 18 ($13.41 ea), whose total is $122.74 (+ heatsink if you don't have free heatsinks), would be a sweet deal; about 170W actual consumption to cover 4x4' area. The driver shows one of the highest efficiency (max of 94%), but it goes up to 143V, so it may not be for everyone.

fibre, if LPC-60-1050 is within your budget, that seems like a nice and solid choice. For the cheap eBay drivers, this thread contains lots of good info:
http://www.rollitup.org/t/diy-leds-how-to-power-them.801554/
It is probably too much info, but I learned quite a bit from it.
 
...
With regard to voltage, we all know that higher voltage is scarier, but I wonder how much more dangerous it is in practice (both medical and fire hazard). I put my fingers into gel electrophoresis (accidentally), but I don't remember the voltage (I guess it wasn't so high because it was just mildly ouch).

You just got a tiny voltage gradient; if you had touched both electrodes you likely would not be posting here.
http://ehs.unl.edu/sop/s-electrophoresis_safety.pdf

I would have no worries about working with an electrophoresis unit in a laboratory but I would not operate one in my home. I do work on household wiring (120 & 240 v) without shutting off the power but I do wear insulating gloves and safety glasses and take precautions to exclude anyone else from my work area. I have also worked in chemical laboratories and manufacturing plants with explosives and potent toxins. Still, for a few dollars more, in my greenhouse, I prefer that the DC voltage to LED fixtures is kept below 48 volts.

...
I was thinking that HLG-185H-C1400 ($69.10) + 4x Vero 18 ($13.41 ea), whose total is $122.74 (+ heatsink if you don't have free heatsinks), would be a sweet deal; about 170W actual consumption to cover 4x4' area. The driver shows one of the highest efficiency (max of 94%)

I keep finding good deals on LED PAR38 spotlights, $10 for a 17 watt input LED:
https://www.1000bulbs.com/product/114019/GE-63105-SP.html
These, unlike the ones that I purchased from Home Depot, are rated only for dry locations but where I plan to mount them in the greenhouse should be in a place and high enough up that the dry location only rating should be sufficient (or I will put them in PAR38 housings.)
 
... But with the price of LPC-60-1050 linked by David ($17.5), a bit extra cost for the name brand might be a better option.

I don't know how well Octopart covers European sources, but this place has it for 14 euro.

Thanks a lot for this link! Looks like a good source. And yes, LPC-60-1050 is within my budget.

fibre, if LPC-60-1050 is within your budget, that seems like a nice and solid choice. For the cheap eBay drivers, this thread contains lots of good info:
http://www.rollitup.org/t/diy-leds-how-to-power-them.801554/
It is probably too much info, but I learned quite a bit from it.

Puh, very much info there! The most difficult thing is that it isn't easy to understand all these terms in German but in English...
 
You could use a single LPC-60-1400 or LPC-60-1750 and for the same price as an LPC-60-1050 you could drive a pair of Luxeon L235-xxxxAULM in parallel.

Except, I just looked at the price for the Luxeon L235-xxxAULM strips and they are about twice the price of the LPC drivers so it is better to maximize the power from the LED strips at the expense of not maximizing the power from the drivers.

As a side note, one thing to be aware of when you are using a CC driver that can supply significantly more voltage than your LED assembly requires is that the driver should always be disconnected from mains power when the connection to the LED is made. For example, the L235-xxxxAULM LED strip requires about 36 volt to drive it at 900 ma but the LPC-60-1050 will supply up to 48 volts. If you put a switch between the driver and LED string so that you could open the DC side of the circuit and plugged the driver into mains power with the switch open then the driver will take its output to 48 volts in a futile attempt to drive 1050 ma and then when the DC side switch is closed the driver will not instantly reduce its voltage so for a while it while be driving the LED string at a very high current (because of the high voltage), possibly enough to destroy some of the LEDs.
 
I keep finding good deals on LED PAR38 spotlights, $10 for a 17 watt input LED:
https://www.1000bulbs.com/product/114019/GE-63105-SP.html
These, unlike the ones that I purchased from Home Depot, are rated only for dry locations but where I plan to mount them in the greenhouse should be in a place and high enough up that the dry location only rating should be sufficient (or I will put them in PAR38 housings.)

The price is amazingly cheap (about $0.6/W). But don't you think that the efficiency seems to be kind of low (I think I mentioned this in your OB thread, but I might have forgotten)? This GE is 39 lumen/W and the HD one was about 50 lumen/W, I believe. lumen/W isn't completely useful for plants, but it is about one-third lumen/W of the luxeon or one-half of COB. Vero 18 is a bit overdriven @1.4A, but I think it will be at least 100 lumen/W even with warm white.

You could use a single LPC-60-1400 or LPC-60-1750 and for the same price as an LPC-60-1050 you could drive a pair of Luxeon L235-xxxxAULM in parallel.

Except, I just looked at the price for the Luxeon L235-xxxAULM strips and they are about twice the price of the LPC drivers so it is better to maximize the power from the LED strips at the expense of not maximizing the power from the drivers.

Yes, these Luxeon seems to have high efficiency and ease of assembly, but the price is a bit high ($36 to get around 32 watt).

Actually LPC-60-1750 in parallel (875mA for each) may be a good idea to get higher efficiency (close to the driver's max, and "reference" current for diodes). I usually don't think of parallel circuit because if one entire strip breaks (unlikely), the other strip is getting 1750mA, which is below the maximum for the Luxeon, but close to the max current.

I usually don't put a switch since it gets plugged into a timer, but it is an interesting point that the switch should be in the AC side (I think most example circuits are like this). I guess that the driver could waste energy if the switch is in the DC side.
 
I took a decision. Tomorrow I will order 5 pairs of Philips Lumileds LUXEON XF-3535L 1150mm 5700K. I will use 2 Luxeon for each shelf of 60 x 120 cm at a distance of about 20-30 cm. I like the cool white more than the warm 4000K - not for the benefit of the Paphs but for my eyes. I have two T5 HO 6500K bulbs (4100lm each) per shelf in use with good success. Sometimes I evan think these are too bright. The Luxeon 5700 K reach 4536-4644 lm each. So I will probably get more light because the bulbs emit in all directions while the LEDs are focused downwards.
A pair of T5HO consumed 108 Watt, the Lumileds will consume 64 Watt. I think this is a very good progress!

I will use one LPC-60-1750 CC-driver for two Lumileds in parallel. But also I will give a try to one LPC-60-1400 just to see what happens.

Unfortunately I will have to wait more than 8 weeks for delivering. But as I will get the new seedlings from the lab not before Mai I will have enough time to prepare the new shelves.

Thank you again naoki and David for all your help and good advices!
 
Cool, it will be interesting how 5700K works with orchids. The emission spectrum of 5700K shows lots of blue. Note that the emission spectra are very different between florescent (several jugged peaks) vs LED (2 peaks).

With 5700K CRI 70, typical luminous flux is 5076 lumen (a bit less than this in reality due to dealing with heat). If you drive it at 1.4A, you probably will be getting around 7000 lumen (if the cooling is adequate). T5HO output decreases with age, so one strip at 1.4A is similar to 2 of T5HO bulbs.

The maximum current for this strip is 1.8A. But most people use LED at 50% or less of the maximum current because the efficiency decreases with higher current (total output does increase). So it it the compromise between the total output vs efficiency. Also, heat is an issue with LED. More substantial heatsinks can lower operating temp, and efficiency increases. Additionally, the longevity should increase with lower temp. If you can find a better heatsink (or use active cooling with a fan), it might be beneficial for overdriving. You can try it (I personally wouldn't do it), but you should be aware of these issues.
 
Good additional advices!
I will use CRI 80. Not so effective (4644 lm), but as an artist I like to have more 'true' colored flowers in my basement. ;)

If you drive it at 1.4A, you probably will be getting around 7000 lumen

No, I like to drive two stripes at 1.4A. This should work well for some new deflasked seedlings. If it isn't bright enough I could follow your suggestion and drive only one Luxeon at 1.4A...
 

Latest posts

Back
Top