View Single Post
Old 10-16-2010, 11:29 PM   #305
Joined: Jun 2006
Location: Noleta, CA
Oddometer: 53
Fixed Brightness 24 W Flood Tear-Down Report

A couple of weeks ago while spelunking in the vendors forum I came across this thread. The packaging was appealing and seemed well thought out while the specs and comments from other users seemed to speak well in regards to performance. And they're priced very well, $70 each for the fixed 24 W floods (and the coupon code of "advrider" gets an additional 10% off). I was interested since I've been holding out on auxillary lights until LEDs reached a decent level of performance and price, these appeared that they might achieve that level. But I had a question regarding how the LED's were biased, I posted it on this thread and emailed twice asking the same question, all inquires went unanswered. I finally decided it be a lot easier to just order a pair and find out for myself rather than wait for a response (BTW, none received as of this writing). Hence this teardown report.

I placed my order from the ADV Monster website, the order was placed on the 7th and I received the lights on the 12th via USPS. Each light was shipped in it's own cardboard box wrapped in a plastic bag, the 7/8" mount and connectors were also in their own plastic bags in the same box and separated from the light by some styrofoam, nothing fancy but effective pacaging. If you order a pair then the two boxes come taped together. Here's what you get in each box (whoops, one clamp too many in pic):

I took one of the lights and proceeded to disassemble it, which is quite easy to do. Unscrewing the front allpws you to remove the glass lens cover and o-ring and gives direct access to the LED's and their optics, unscrewing the finned heatsink off the rear reveals an aluminum backing plate.

Backing plate...

Removing three screws on the backing plate frees the LED/heatsink/optics assembly. On the light I chose to disassemble the heatsink grease that was applied to the LED heatsink (to conduct heat more efficiently away from the LED's and into the light housing) was not particularly well applied, there two spots covering maybe 15% of the circumference with a fair amount of grease, the rest of the heatsink edge was free from grease. I ended up spreading it around more evenly.

Removing the LED assembly revealed the printed circuit board (PCB) that biases the LEDS. There are two surprises here, one pleasant, the other not so. The PCB is an aluminum core board, very good at dissipating and spreading heat, a very good choice for this application. Inexplicably, the PCB is not fixed in place, it's only supported by the wires suppling the power to the unit and the wires to the LED's, basically it's just flapping around on the inside of the light. In a high powered LED design one of the biggest challenges is in removing the heat generated, in this light the designer went to the expense and extra effort of using a metal core PCB to help spread the heat but then did nothing to help get the heat off the PCB. There's heat conducting adhesives for situations such as this, I will most likely be attaching mine to the backing plate with a thin coating of metal filled epoxy.

LED/heatsink/optics Assembly
The LED assembly consists of three LED packages, a backing heatsink and a multiple element lens (3, one for each LED). The LED's are supposed to be made by Cree, 500 lumens each but I've been unable find these particular LED's on their website.

The three LED packages are wired in series while each LED package consist of three LED's, apparently wired in parallel (judging from the measured forward voltage). With separate LED's it's a rookie mistake to wire the LED's in parallel since the forward voltages will be different resulting in one LED robbing current from the others. In this case it's OK since the three LED's within the package are on the same die their forward voltages will all be the same. Wiring the packages in series certainly an OK way to wire the LED's however, from a reliability point of view, if one LED package burns they'll all go out. There are components on the market that address that issue providing outputs to drive several LED's but that approach was not used here, most likely because it would drive the cost.

PCB Assembly

As previously mentioned the PCB uses an aluminum core board on which the LED bias circuit is constructed. The first component you notice is the big two terminal device dangling off the board by a couple of wires with a silicon sleeve over it. This is a bimetallic switch and is typically used to protect assemblies from getting too hot by either shutting them off (they're used in a lot of blow dryers in this manner) or by throttling back the power when a certain temperature is exceeded as in the case of this light. It has a trip point of 65 degrees C. I believe the silicon sleeve is to prevent the switch from sensing the case temperature as it would if there was direct switch to case contact, it's meant to sense the air temp inside the cavity the PCB occupies. The sensor is a nice touch and one that the seller is either not aware of or just didn't bother mentioning, thermal protection is a good feature to have and as we'll see- this light needs it if you're not moving.

crappy pic of bimetallic switch with silicon sleeve removed...

My unanswered question was, how are the LED's biased? I asked this question because there are several ways to do it, the cheap way is to use a resistor but then the light brightness will be highly dependent on whatever voltage the voltage regulator on your bike is putting out and that can vary greatly depending on running conditions and number of electrical farkles you're running. I was pleased to find that this light did not go this route, it uses a switchmode buck controller circuit to supply constant current to the LED's, provided the input voltage is a little higher than the voltage the string of three LED packages requires to operate at the desired current, in this case it's a LED forward voltage of 9.8 VDC @ 2.2A. This means that for VR/battery voltages greater than ~11 VDC you'll see no change in brightness of the light. Under that voltage the light loses regulation and the brightness will vary widely with the input voltage. And the switchmode controller IC used can actually work with up to 40 VDC at the input (and is rated over a -40 to 85 degree C operational temp, the typical industrial temp range). The bimaetallic switch is normally closed, when it opens it changes the set current for the LED's from 2.2 A to 0.52 A, or roughly 1/4 the current (and power). When this occurs the LED becomes noticeably dimmer, this is described in more detail in the following section.

in office ambient conditions an assembled light was connected to a lab supply set for 13.0 VDC, the lamp drew 2.2 A from the supply. Three minute 12 seconds after being powered the current dropped to 0.52 A and the brightness decreased significantly. So, the light had insufficient cooling and it's internal temperature escalated to the point that the bimetallic switch opened up (nominally 65 degrees C). The light was left on in this condition to see if it would cool down enough for the bimetallic switch to close, restoring normal operation. After approximately 35 minutes of operating in this state the light went out completely, cycling power did not allow the light to come back on however, allowing it cool unpowered for about an hour restored normal operation. The complete shutdown surprised me as the controller IC said nothing about overtemp protection. While this type of protection is common on power IC's usually the data sheet makes some mention of it, it's a selling feature; given the choice between having your part burn-up or shut down when it gets too hot most would choose shut down.

I realize bench testing is not a very fair test for a light that typically is mounted on a moving vehicle, so armed with an anomometer borrowed from work I set up an experiment that would subject the light to approximately 30 mph of wind using a leaf blower with the light mounted on a tripod, powered by my truck's battery.

The leaf blower was positioned until the anomometer indicated about 30 MH, this reading is more like 31MPH of wind...

The neighbors enjoyed the leaf blower at 9 PM...

Short video of test.

I ran the test with the light and leaf blower on for 20 minutes, the current never throttled back and to the touch the light housing was fairly cool (unlike in the lab where it was fairly toasty). At the end of twenty minutes I figured it would never heat up enough to trip the bimetalic switch so I shut off the blower and left the light on to see if outside at 65 degrees F on a slightly foggy night the bimetallic switch would open up... it did but this time it took a little more than twice as long as in the office, about seven minutes. But that's not a bad thing, it's doing what it's supposed to do.

I'd like to have been able to perform some useful evaluation of the light output from these LED's but I lack the equipment to do so, suffice it to say they are bright, very bright. Like similar to my 35 W HID headlamp bright, and that's just one of the lights, I bought a pair. After seeing the full wattage 24W version I'd order the three level dim-able version if I had it to do over again. As it is I will most likely modify the ones I have so I can at least have a 1/4 power mode, the bimetallic switch will prove useful for this, I'll just need to cut either lead and wire an external switch in series, when the switch is open it will provide the 1/4 power mode. But that's gonna be extra work and wiring to accomplish, I'd have been money and time ahead if I'd just bought the dim-able version in the first place. Aside from the PCB floating around internally I'm impressed with the design and the value you get for your dollar with these lights.
Transmongified Naranja '97 F650ST

patobravo screwed with this post 10-17-2010 at 11:17 AM
patobravo is offline   Reply With Quote