Work doesn’t stop at the “Gough Labs”, and this week, another anonymous donation turned up for the review challenge. This is yet another LED downlight, from a company called BrighterLED Lighting Solutions. I have no idea where this came from, or who they are, as their website isn’t even working at the time of publication:
Unboxing and Features
The unit is packed in a glossy colour-print cardboard box, featuring a distinctly orange colour scheme. The top panel claims the unit is a 10W unit with an integral driver and IC-F abutted and covered insulation contact rating.
The unit is a warm white colour temperature, and requires a 90mm cut-out. It has a 58mm height. It is complete with 1.2m flex and plug, and claims to be compatible with “LED dimming technology.”
Another panel claims the unit to produce 800 lumens, model BLLS-DL10-I115 with IP44 “weather resistant” rating.
The “full” specs are provided on another panel. The CRI is claimed to be >80, exact figure not provided. The beam angle is slightly wider than most competitors at 100 degrees (compared with 90 degrees for most). It uses 2835 SMD LEDs with a matt white trim. It is covered by a three year warranty. The 800 lumen output results in a luminous efficacy of 80 lumens/Watt, which is pretty “average” in a market where manufacturers are aiming for 120 lumens/Watt, and products above 100 lumens/Watt are becoming more widely available. Unfortunately, no indicative lifetime is provided, nor is any colour-matching information (i.e. SDCM).
On another side of the box, there is an SAA logo, which is not normally printed on boxes, as well as the regulatory compliance mark (RCM) for Australia.
Once the unit is removed from the box, at first glance, the unit shares some visual similarities with the Atom Lighting AT9033, namely the coloured curved cap, the shape and size of the cable retention clip and the printed batch number. The same difficulties in having it sit in a stable “driver down” orientation also affect this unit. The unit comes with a protective cap on the mains plug pins, probably to avoid scratching in transit. It claims an ambient temperature rating up to 45 degrees Celsius.
Already, there appears to be an issue with the labelling of the product, namely the IC-F logo on the luminaire is only as small as two lines of text on the cap. This doesn’t meet the AS60598.2.2 standards, specifically clause 2.6.103b which requires “a minimum size of 25mm x 25mm.” I’d estimate it’s barely 10mm x 10mm as it stands, which is why most other manufacturers use a large label on the side of the unit to declare the insulation contact rating.
The unit has a wavy pattern on the plastic-coated heatsink that isn’t present on the Atom, and uses much more flexible and flimsy feeling spring wire for the retention clip, similar to old halogen downlight surrounds. It’s probably not too big of an issue, but it just doesn’t feel quite as high quality. At least it has a ribbed orange plastic end cap which improves the fit and prevents damage to the ceiling material.
Similar to the Atom and many other downlights, the driver is integrated behind the LED heatsink without any ventilation or spacing, which means that the driver is likely to operate at a higher temperature than necessary, shortening its lifetime.
The unit doesn’t have any included gasket despite its IC-F rating, which is likely to mean a poor seal around the recess hole, and some heat loss/dust marks in the future. It probably saves them a few cents though.
The front opalescent diffuser is flush to the surface, with a moderate matte white surround that looks very similar to other products. A small amount of excess adhesive can be seen at the seam towards the top.
The unit claims to be dimmable, but a list of compatible dimmers is not included. Trailing edge dimmers and universal dimmers are suggested. Clearances above the luminaire (HCB) is required to be 50mm, and side clearance (SCB) is 25mm.
Owing to the physical similarities between most products, the teardown procedure is pretty much the same.
We can see a double-grip design for the cable retention system that works quite well. The terminal block seem to fit well, although the stripped ends of the leads seem to be just the right length. With the cap removed, we have access to the PCB.
The PCB is a single sided PCB, although is of a fibreglass type. It has silkscreening on this side, and solder resist on the other. Clearances behind the terminal block are slightly tight, although not as bad as in the DETA product. The input is protected by a fusible resistor on the left corner, where the input is then “crossed” over to the right side of the board. As the fusible resistor is not otherwise contained, catastrophic failure will rely on the casing of the unit to contain any hot fragments.
RV1, a MOV from JNR Varistor (7D471K) provides some surge protection. A few inductors are also present for RFI suppression, and polyester capacitors for transient protection, although they are not safety rated. Not a major issue, as it is protected behind a fusible resistor I suppose.
A white printed zone with no components can be seen, and is due to the SMD IC controller on the opposite side. Two electrolytic caps are used in the design, which seems to be an all time low, both Pchicon RF series 105 degrees C rated caps. The capacitors are manufacturer rated for a 3,000 – 7,000 hour load life, which translates to about 24,000 – 56,000 hour lifetime at 75 degrees C, which should be enough.
Unfortunately, I can’t vouch for the reliability of Pchicon capacitors, being a relatively unknown Chinese product, and it seems only time will tell if they will meet their claimed lifetime ratings. In the past, other capacitors of questionable reputation had failed causing many product failures of computer hardware such as motherboards, LCD monitors and graphic cards, and was termed the “capacitor plague”.
The main MOSFET is an InPower Semiconductor ISU04N65A, a 650V 4A N-Channel MOSFET with 1.9 ohm Rds. It appears the board was designed for a larger, possibly higher rating MOSFET with a DPAK package, and instead was later substituted for a smaller unit resulting in the legs being splayed and the unit sitting proud of the board. A smaller unit often comes with a higher Rds, resulting in higher heat generation and lower efficiency due to energy loss.
But most concerning is that the transistor is sitting with its back case tab facing towards the electrolytic capacitor, meaning that the capacitor is likely to be “heated” by the transistor and thus have a shorter lifetime. This was probably a mostly unavoidable compromise owing to the need to “squeeze” everything into the limited space with a single layer board to keep costs down.
A look at the underside of the board appears to show a date code of Week 13, 2016 (or the other way around, I’m not sure). There is silkscreen as well on this side, which appears to show the board ID as AU-FB01A REV1.2. Unfortunately based on this information, the original manufacturer could not be reliably determined.
We can see the incoming mains crossing from one side to the board to the other. On most competitors’ products, the board is “cut” to increase the separation and avoid any potential for acing, whereas on this board, no such feature is present. I suspect it’s probably not a major issue, as there is just enough creepage distance and solder resist for insulation.
Several interesting characteristics can be seen on this board. The terminal blocks are like those on the HPM DLI9002, of the two-studded type. On this particular board, there was quite a bit of solder applied, so the connections look relatively good, and it withstood several wire connecting/disconnecting cycles without damage. But to achieve this, they seemed to have used so much solder and heat that it had set with a “rippled” appearance implying that the board was moved before the solder had fully set.
The remainder of the soldering looks mostly acceptable, although the alignment of the SMD components is significantly to the left of where they are expected. This may be due to poor machinery alignment at the factory. The exception to this is the joint of C11 to the bridge rectifier DB2 where there is a “thick” solder bridge along the trace. I wonder why this was done – was there a fault in the PCB trace or solder mask?
Anyhow, the heart of the current driver is the Dialog Semiconductor iW3688-01, an AC/DC Advanced Digital Power Controller for Single-Stage Dimmable LED Drivers, as used in the DETA DET190. This particular IC claims wide dimmer compatibility with leading-edge, trailing-edge and digital dimmers.
The wires to the LED array are tack soldered to the bottom, but they didn’t do a half-bad job of it. Next to it seems to be a position for a MLCC ceramic capacitor (C13) over the LED array that may have reduced the “ripple” across the LEDs (and hence stress, and potential flicker) somewhat, but I suspect it wasn’t fitted for cost reasons. It’s not absolutely necessary, but it’s probably a nice-to-have option.
From the other side, it looks very similar to other products with a white diffusing plastic cone. The cone isn’t quite centered, with a slight gap which is larger on the right side.
The ring is not adhered into the heatsink, which is another dog-bowl shape heatsink. I’m sure you probably know the drill if you’ve read my other reviews. The MCPCB wires enter from the corner, and are soldered to the board directly, again, fairly well compared to their competitors.
One difference of this unit is the much larger MCPCB and wider spread of the LEDs. This may be advantageous for heat reasons, although probably increases costs as MCPCBs aren’t that cheap.
The PCB has a total of 24 LEDs of the 2835 variety, and is marked with AU-FTE 15 16.
But alas, things are not perfect, as it seems this MCPCB may be a bit cheap. The board appears to show bubbles in its construction, suggesting possible delamination of the copper foil and insulator layers from the aluminium backing due to overheating or poor manufacturing. If these bubbles expand, support and thermal path for the LEDs may be compromised, and the trace may eventually crack if sufficient stress accumulates. It’s not something I’ve seen before.
Also, a closer look at the left side in-between the two LEDs shows a speck, which upon closer examination, appears to be a dead insect which got trapped inside the LED during manufacture. It’s not the first time I’ve encountered such things, but it’s something shouldn’t be happening …
Despite the large area MCPCB, the connecting traces don’t take full advantages of the heat transfer ability, as they only have thin “lines” between LEDs rather than large areas of copper (i.e. etch as little as you can). As a result, it’s likely that the LEDs will operate at a higher temperature than absolutely necessary. Because of the poor contrast of the traces even with contrast enhancement in post-processing, a highlighted version on the right which traces the PCB trace path in green is on the right.
The PCB utilizes a single-series arrangement which ensures all LEDs have even current and even stress. It’s the arrangement I personally find most preferable.
Continuing the disassembly, it was confirmed that the unit had white thermal grease applied to the interface between MCPCB and heatsink. The paste quantity was appropriate, and judging from the pattern, it had spread quite effectively to cover the majority of the surface with the exception of the areas just immediately around the screw holes.
This section will focus on the results from testing and subjective impressions of the product.
Output Quality and Impressions
The light output was about where I would have expected it to be, and as with the spectral quality. The current driver was acoustically quiet, and was fast to start up when directly connected, almost imperceptible and certainly below one second on application of full power.
The output uniformity was close to uniform, with a slight perturbation in the top right corner, possibly as that may have been where the cable entered in the orientation of the product when photographed. The diffuser is doing a mostly-good job of preventing significant bright hot-spots.
According to the camera, the colour temperature is 3000K exactly, spot on as specified.
Compared to other downlights, the outside of the unit felt a touch warmer (subjectively) during operation than the other units. This is probably a good sign that the heatsink and thermal grease is doing a good job of removing the heat.
Warm-Up Power Trend
Warm-up power was measured with a Tektronix PA1000 and a Variac to adjust the voltage to 230V +/- 1%.
Over a period of 20 minutes, the unit started at 10.375W and reduced to just under 10W (9.98W). This indicates the unit is “honestly” 10W as stated on the label.
Power Consumption and Power Factor vs Voltage
Again, using the PA1000 and a Variac, the unit was swept through the range of voltages to determine its power parameters.
In the intended operation range, the unit was just under 10W of power consumption at a power factor of 0.94 to 0.96, which is an excellent result, and shows the PFC of the controller works as intended. This is probably more important for commercial and large installations where power factor and reactive power is an issue.
The luminaire maintained regulation down to 150v AC, and thus can survive deep brownouts without any visible effect. This is consistent with most LED lights.
Below that, and down to about 100v, the brightness began to decrease until it abruptly extinguished. Below that, the luminaire attempted to cycle on, but failed to do so, resulting in erratic power readings. This result is very similar to the DETA, and is expected because of the identical choice of controller IC.
Inrush current was measured with the PA1000 in inrush mode, and the unit cycled 10 times in succession, with peak positive and negative currents reported (but no duration).
The measured results were +1.893A and -1.803A, very similar to what was achieved with the DETA, and a very good result showing good inrush current management. Conservatively speaking, a circuit can probably handle about 18 of these units with no hassle.
Current Driver Output and Efficiency
The PA1000 was used to sequentially measure voltage and current from the non-isolated output.
The measured voltage was 72.23V with a crest factor of 1.021 indicating very small voltage ripple. The measured current was 125.13mA with a 1.189 crest factor, indicating the output is DC with some level of ripple.
The power delivered to the LED array was 9.04W, and the efficiency is thus calculated to be approximately 87.12%, which is not bad and again, similar to the DETA.
Dimming compatibility was tested with the three dimmers I presently own.
With the IKEA Dimma Leading Edge dimmer cord, the fitting and dimmer had a slight acoustic noise and start-up was delayed to about two seconds at the minimum output setting. The dimming range was approximately 15% to 100%, and no adverse effect was recorded even after prolonged running. There was some temporary changes in brightness during periods of mains of mains signalling, and it appeared that leading edge dimming might have increased the heating in the driver unit slightly.
With the Nixon Universal Dimmer, the fitting had slight acoustic noise as well probably because it was being controlled in leading-edge. The dimmer allowed for dimming right down to off, with a full linear range, although at below 10%, flipping the switch on and off led to the unit not being able to start up. This is probably expected behaviour.
With the DETA Universal Trailing Edge dimmer, the fitting was quiet, and the dimming was smooth and linear, between about 5-100% of the output.
On the whole it seems that dimming compatibility fared better than the DETA despite the unit using the same IC because this unit was 10W and thus met the minimum load requirement of the Nixon.
The BetterLED Lighting Solutions downlight has visual similarities with the Atom lighting product but are otherwise completely different products. One has to wonder if these similarities are intentional to confuse consumers, or if they are merely coincidental.
Regardless, the product shows a cost-optimized approach to the current driver, with single sided PCB and Dialog Semiconductor controller. The performance of the controller was quite good, achieving respectable efficiency, good power factor and dimmer compatibility, and the construction of the driver PCB was, for the most part, acceptable. The unit had good thermal grease application and adequate screw torquing.
However, the product employs the relatively unknown and unproved Pchicon capacitors, which may prove to be a point of failure. The design also places a transistor with its “hot” side facing the capacitor, which isn’t optimal. It has no air gap between the LED heatsink and the driver, which is common but again, suboptimal. It doesn’t have a stated lifetime, and its IC-F logo doesn’t comply with Australian Standard requirements. The PCB had some component alignment issues, and the MCPCB had some delamination and relatively thin traces which could potentially impact on its lifetime. It also has a pretty “average” 80lm/W luminous efficacy, which isn’t very high especially if you consider 100lm/W was a target a few years back, and the target now is above 120lm/W.
Just another product with both good and bad points that we can learn from.