Voltage Regulation for LED Lighting

[geoffschultz]

I also am not impressed that Dr LED by way of backing their bulbs is just offering you 1/2 price innards on a product that has obvious manufacturing defects.

We're certainly not in any disagreement here! The fact that I had 6 of 8 units fail within a short period of time indicates to me that there's a problem with the unit. I expected better customer service.

I've also been doing some more research on the thermal issue. Cree makes 5mm LEDs like the ones in the DoctorLED MARS dome that I have. Their web site has a document, http://www.cree.com/products/pdf/LEDlamps/LO566TWN1-70H-A1(104).pdf which, among other things, contains a graph showing the maximum forward current vs the ambient temperature. I've attached that graph as an image to this post.

When we took the measurement, the temp was about 70F in the room and we saw about a 50F rise after 10 minutes, but the case wasn't mounted to anything which allowed much better air cooling. Let's assume that on the boat that the temp could hit 90F and that there could be a 60F of heat rise after hours of use. That would put the temp at a maximum of 150F or 65C (which seems much hotter than I've ever felt it get). That translates to a maximum forward current of 13mA, and we measured 15mA, so that would be a problem. Thus, the thermal argument has a lot of merit.

All that would have to happen is to have one of the LEDs in a bank to fail and the increase in voltage would cause the others LEDs to fry.

-- Geoff

 

[walt]

How do you get a "whole department full" of engineers to work on your broken $60 VCR for free (not even any beer involved)- and enjoy it? Post your problem on the internet..

Even if its bad diodes (seems your measurement of 4 to 6 volts would confirm its the diode and not the associated circuit), it seems at least a little odd that 6 of 8 units would fail at the about same time. If its poor heat sink grease or the unreliable diodes, I think you would "generally" get more of an even distribution over time of the failures.

Was there any possibility of a one time event stressing the parts?

Not important, but Im also curious about the 1.67 watts burned in the diodes in the good unit. The spec shows the light using 3 watts. If only 1.67 of the 3 watts is being delivered to the diodes.. not very efficient. But more than likely, when you measured 1.67 watts to the LED, the fixture was also using much less than 3 watts. Did you guys happen to also measure what was going into the fixture? I would not think its difficult to get 80 or 90% of the input power delivered directly to the diodes providing the light output.

 

[tsmwebb]

We're certainly not in any disagreement here! The fact that I had 6 of 8 units fail within a short period of time indicates to me that there's a problem with the unit. I expected better customer service...

I didn't realize that you were looking at two-thirds of the units failing... I still think that a bad batch of LEDs, poorly assembled and run in tropical temps is a pretty compelling case for early failure. But 75% failure in a couple of years? Thats intolerable. Got any further recollections on the history of the failures? Maybe they did get spiked... Have you had problems with any of your other electronics?

Just by way of contrast, I reported one faulty 9 yr old solar panel to Kyocera hoping to get some advice on fixing it. They responded by replacing my entire array, sending me new panels before I returned the old ones, all at their expense (including shipping and taxes to and from Mexico). That pretty much sets the gold standard in customer care in my book! I'm thinking Dr Led has a long way to go.

--Tom.

 

[geoffschultz]

Was there any possibility of a one time event stressing the parts?.

Of course there is that possibility as we're on the boat for 6+ months out of the year. However, none of the electronics has had any problems, but the voltage dumping circuitry in the rest of the electronics is probably a lot better than a single transistor per bank of LEDs.

Not important, but Im also curious about the 1.67 watts burned in the diodes in the good unit. The spec shows the light using 3 watts. If only 1.67 of the 3 watts is being delivered to the diodes.. not very efficient. But more than likely, when you measured 1.67 watts to the LED, the fixture was also using much less than 3 watts. Did you guys happen to also measure what was going into the fixture? I would not think its difficult to get 80 or 90% of the input power delivered directly to the diodes providing the light output.

We measured the total current to the light at 180 mW once the voltage got above about 9.6V.

-- Geoff

 

[geoffschultz]

Here's a photo showing a unit with only 6 of the 36 LEDs burning brightly. The other 30 just have a dull glow.

 

[walt]

FYI, I have almost the same LED light but only use it maybe 10 to 20 hours per year. Mine has the red/white light option.

You mentioned that the architecture is 12 strings of three LED's. Each transistor powers 2 strings each. When a diode fails, its voltage goes to 4-6 volts from 3.1 when new. If one diode in a string of three goes bad, the string of all three will likely be dim.

Given that you said only 6 diodes are bright (out of 36) and that it takes all three diodes in a string working to be correctly bright, I would take that to indicate that only 2 of the 12 strings of three diodes are still working. So you must have had at least one diode go bad in each dim string - for a total of 10 diodes..

Yikes!!! Ive had no problem with mine after two years (but much less usage than you likely have)

 

[geoffschultz]

FYI, I have almost the same LED light but only use it maybe 10 to 20 hours per year. Mine has the red/white light option.
[clip]
Yikes!!! Ive had no problem with mine after two years (but much less usage than you likely have)

It very well may have been the fact that I got the first shipment and they had problems with manufacturing which were corrected later. It may be that I spend a lot of time in the very warm Caribbean. Who knows? Clearly there's an issue with the thermal paste. We'll see what happens. If I don't get a satisfactory response, I'll give the SensaBulbs a shot.

-- Geoff

P.S. I have 2 of the R/W lights also, and they too are exhibiting the problem in the white section.

 

[dvideohd]

It is not so obvious that there is a manufacturing defect... Possible, sure.....

Voltage spikes are the ENEMY of all portable equipment.. especially things with motors.... You can see this with even ordinary light bulbs... put the light bulb on a normaly, low load - no motors - circuit, and the bulb lasts a long time.. put it on a cicuit shared by multiple motors - then prepare for a short life...

Car equipment would be wise to prepare for 60-70V spikes... Boat equipment is not so well defined... I would not be supprised to see 100V spikes on a DC circuit - with motors present - and some corrosion...

So the idea to protect the LEDs is a good one... A reasonable series resistor - and a good "spike limiter" is not a bad idea...

--jr

BTW... . having had designed multi-LED circuits for commercial use, it is not so clear to a lot of designers as to what to look for... I worked to control the LED voltage AND the LED current.. . need to look at both pieces... LEDs will decay in output over time - but not what you are seeing.... For a auto/marine application - I would design to at least 60-65V - and provide circuit protection for higher voltage spikes.... But then again, I am paranoid - and don't like to fix things 6 months after deployment.... I might also state that the RATING was to 35V.... just to make sure that I was 100% in a safe operating zone...