How to check efficiency of a chinese panel (driver and led)

[pirg420]

Is this possible with a multimeter and other tools? Be nice to see how efficient my drivers are, and if possible how efficient the actual led is. Possible? Anyone explain how? Im gunna guess you measure voltage and amps before and after the driver? a detailed explanation would be awesome, if this is possible, thanks guys.

 

[churchhaze]

That's basically it. Then once you have power in and power out, efficiency is always Pout/Pin

Is this possible with a multimeter and other tools? Be nice to see how efficient my drivers are, and if possible how efficient the actual led is. Possible? Anyone explain how? Im gunna guess you measure voltage and amps before and after the driver? a detailed explanation would be awesome, if this is possible, thanks guys.

 

[SupraSPL]

It is easy to check the voltage of the string because you connect your meter probes in parallel, although it is important that you check voltage as close to the driver as possible.

As far as checking current, that can be trickier because your meter probes need to "bug in" to the string, connected in series, but you can bug in at any point on the string. If you have a pair of multimeters that is even better so you can check vF and current at the same time.


Now you have voltage and current so volts * amps = total Watts on the DC side AKA dissipation Watts.

Regarding heat, driver efficiency changes (usually increases 1%) as the driver warms up. Also, driver output tends to decrease as it warms up, dropping the current and Vf of the string. Finally, the Vf of the LEDs will decrease as they warm up. So the idea is to wait until the whole system is thermally stable and the driver thoroughly warmed up before to take your driver efficiency measurements.

On the AC side, you need an RMS Watt meter to check power consumption. Then divide DC-side Watts by AC-side Watts and you have driver efficiency. The driver efficiency will vary depending on the Vf. It tends to get more efficient as you approach the maximum of the Vf range.

 

[SupraSPL]

If you want to take it a step further, you can test the temp droop with any cheap light meter. You take the light measurement and check the dissipation watts by pulsing the light for each reading. Calculate dissipation W at room temp. Then let it run until it reaches thermal stability and take the same readings again, calculate dissipation W warmed up. Most often, the dissipation wattage will be lower warmed up. So if dissipation W is 96%, apply a 4% penalty to the pulsed light measurement. It is critical that neither the light meter or the lamp is able to move between measurements.

Then divide the adjusted room temp measurement by the warmed up light measurement and you can see the % decrease (temp droop). For all-white grow lights, an awesome temp droop would be 2%. A good temp droop is 5%. 7-10% is poor (running hot). Anything above 10% means the lamp is not cooled sufficiently IMO. Lamps with red diodes are a different story, they experience larger temp droop, but they make up for it with lower current droop.

Example math:
Pulsed light measurement 40000 lux
Pulsed power dissipation 52W

Warmed up light measurement 37000 lux
Warmed up power dissipation 50W

50W/52W=96.1%

40000lux * .961 = 38440 lux (new pulsed light measurement with power adjustment applied)

37000lux/38440lux = .963 = 3.7% temp droop

 

[pirg420]

Thanks for the info supra. IS there a way to find out how efficient the led is? or just the driver?

thanks

 

[Greengenes707]

If you want to take it a step further, you can test the temp droop with any cheap light meter. You take the light measurement and check the dissipation watts by pulsing the light for each reading. Calculate dissipation W at room temp. Then let it run until it reaches thermal stability and take the same readings again, calculate dissipation W warmed up. Most often, the dissipation wattage will be lower warmed up. So if dissipation W is 96%, apply a 4% penalty to the pulsed light measurement. It is critical that neither the light meter or the lamp is able to move between measurements.

Then divide the adjusted room temp measurement by the warmed up light measurement and you can see the % decrease (temp droop). For all-white grow lights, an awesome temp droop would be 2%. A good temp droop is 5%. 7-10% is poor (running hot). Anything above 10% means the lamp is not cooled sufficiently IMO. Lamps with red diodes are a different story, they experience larger temp droop, but they make up for it with lower current droop.

Example math:
Pulsed light measurement 40000 lux
Pulsed power dissipation 52W

Warmed up light measurement 37000 lux
Warmed up power dissipation 50W

50W/52W=96.1%

40000lux * .961 = 38440 lux (new pulsed light measurement with power adjustment applied)

37000lux/38440lux = .963 = 3.7% temp droop

Great info, but I'm not understanding the 4% penalty...where is that from?

 

[SupraSPL]

As the driver warms up typically its current output falls, so the warmed up number is being penalized by the driver if we tried to make a direct comparison. Also, the Vf of the LED drops from the reduction in current and from the heat in the junction, further reducing dissipation Wattage. So we have to penalize the pulsed number by the same percentage to adjust for that. There are some drivers that will not change their output as they warm up or even increase their output, so in that case the pulsed number gets adjusted upward to make up for it. Of course that introduces slight error due to current droop, but if you have an adjustable/dimmable driver, you can reduce that variable.

@pirg420 unfortunately the cheapo light meter is no good for absolute measurements. It is highly precise but utterly inaccurate, meaning you get great repeatability but the meter cannot see each wavelength equally, so there is no way to convert the readings into PAR W accurately. In particular it cannot see blue and red/deep red very well. Even expensive PAR meters have this problem, although GG has reported success using them when comparing their reading to a spectroradiometer reading.

For example when I compare a Cree CXA3000k and Vero29 3000K and really push them hard (2.5A), the CXA shows higher readings (per Watt) than the Vero, which is not what we would expect from the manufacturer data. But I cannot declare that the CXA is more efficient at that current because the Vero has more red/deep red output which is probably throwing the meter, so I have no confidence in the comparison. We would need a calibrated spectroradiometer to really know what is going on with the spectrum.

Another problem, in order to check the output efficiency we would need to measure all of the light that is emitted, which requires the use of an integrating sphere or at the very least tracking the PPFD on a PAR grid and trying to convert it to PPF.

So for the most part I try to estimate junction temps by monitoring the Vf as the LED warms up and I estimate efficiency based on the data sheets. But that only works if you know the part numbers/bins of the LEDs used.

 

[dr.medecine]

Thanks supra, good info.