Mercator is an Australian owned lighting company, whose products are occasionally seen in local hardware and lighting shops. This article looks at the Mercator Retina 10W LED Downlight, model number MD499 in White, donated anonymously under the terms of the review challenge.
Unboxing and Features
The unit comes in a colour print cardboard box, and is available in white (the reviewed product) and silver. This product is advertised as having an IP44 rating which would make it suitable for protected outdoor installations. It also has an abutted only insulation contact rating, which means that the unit must not be covered with insulation. This can affect the thermal efficiency of the dwelling in which it is installed, and make installation a little more complicated. It’s interesting how it tries to tout that it has a heat resistant housing for cooler operation when many of its competitors carry an IC-F/IC-4 rating.
The unit is available in three different colour temperatures, the tested unit being a warm white 3000K unit. It claims that the downlight is compatible with selected models of dimmers on their website. The list of dimmers was not that easy to find, but is here and covers a few Clipsal, Lumex and HPM units. The disclaimer is interesting, as it says “Mercator does not recommend a specific dimmer for any of it’s[sic] products.”
The unit comes with a 1.2m flexible cord pre-fitted and is backed with a three year warranty. This panel seems to claim that it is both leading and trailing edge dimmable – the first product I’ve encountered that makes such a claim (aside from the datasheets for the ICs used). It has a CRI of 80 (most competitors claim >80, but without specifics), a 90 degree beam angle and fits in a 92mm cut-out (and has a 62mm height according to the datasheet).
The output lumens are claimed to be 700 lumens for 3000K (70 lm/W), 750 lumens for 4000K (75 lm/W) and 800 lumens for 6000K (80 lm/W). This particular figure is slightly low when compared to the LED market as a whole, where >100 lm/W is becoming common and products approaching and reaching 120 lm/W are appearing in the market. It is also less than all the other reviewed LED downlights to date, with the exception of the DETA unit.
On the inside cover of the box, a cut-out template is printed. I’m not sure how useful it is, given that most installers just use a standard sized hole saw with the size marked on it, but I suppose you could cut it out and trace around it to see where the size and positioning of the holes would be in your ceiling.
Also included is the datasheet and warranty statement.
The datasheet gives the unit lifetime as 30,000 hours, in line with the majority of its competitors. It also details the three year warranty, which surprisingly, despite the IP44 rating advertised, does not cover any form of outdoor installation whether it is protected or not.
It also details the clearances required, which is specified as 100mm above the unit or to structural members, 50mm to insulation (or abutted), and 25mm insulation to building element (air gap between insulation and structural element for ventilation). This is greater than necessary for IC/IC-F/IC-4 rated units.
The datasheet does not make any lifetime claims in regards to abutting insulation and whether that would impact the lifetime (which I would imagine, might, due to the increased operating temperature).
The unit has its approvals on a rear label, which includes double insulation, indoor usage, regulatory compliance mark, do not cover, and IP44 rating. It is Made in China in August, 2015. It claims an ambient operating temperature up to 40 degrees Celsius.
Again, we are reminded of its “abutted only” rating with the label on the side of the unit. No gasket is supplied, thus breathing of the roof space through the recess holes and heat loss is a possibility. The unit features spring-loaded rat-trap style legs for easy installation, made of metal and with the tips covered in heat-shrink material.
The unit features an elevated driver unit that has an air gap between itself and the LED heatsink. This reduces the heat in the driver unit, and prolongs its lifetime – a hallmark of good design.
The front of the unit has a fairly thin bezel, and a near-flush opalescent diffuser. A small blemish on the outside was due to the photo being taken after teardown because the original photo was not in focus.
The included flex and plug appear to come from Shanghang Jianrun Dianye Youxiangongsi Fund Resources Electric Industry Co. Ltd. It carries SAA numbers SAA110201EA and SAA101003EA, meeting Australian requirements.
As with the other units, teardown starts first by examining the driver. Removal of the cord grip cover screw is necessary to gain access to the driver. The cord grip itself is a double-grip design which is effective in retaining the cable, but because of the way the flex is stripped, the cable exits awkwardly on one side.
The whole plastic cover assembly can be removed with careful prying around the edge to release the clips. Inside, a double sided PCB with silkscreen and solder resist is seen. The terminal blocks have a good clearance around them to avoid components potentially contacting mains wire. The PCB appears to be dated Week 29 of 2015, and has 021417V1.3 marked on it. The primary side includes a fusible resistor for failure protection, although this one is not enclosed with heatshrink so the outer casing must contain any failures. It has a MOV next to it for surge protection, and a series of inductors and regular non-safety rated polyester capcitors for EMI and transient suppression.
Two transistors are seen, the main switching transistor (Q1) was identified as a reputable ON Semiconductor NDD04N60Z-1G 600V 4.1A N-Channel Power MOSFET in an IPAK casing. This is a smaller transistor, and has a slightly higher Rds (2 ohms versus others which are ~1.1 ohm) which may result in slightly more resistive losses and heating, as well as lower conversion efficiency. The second transistor (Q2) was identified as a Unisonic Technologies 1N50Z 1.3A 500V N-Channel Power MOSFET in a TO-92 package with 4.6 ohm Rds.
This particular design has no primary electrolytic capacitors. All electrolytics used are Rubycon ZLH series capacitors. These are also considered highly reputable capacitors from Japan, with a load life of 6,000 – 10,000 hours at the 105 degrees Celsius rating, surpassing many of their competitors. This implies that the lifetime of the capacitors should exceed the lifetime rating of the unit at reasonable operating temperatures (e.g. 75 degrees Celsius, expected capacitor lifetime is above 48,000 hours).
The wire connections to the LED appear to be kept very short, and are soldered properly “through hole” rather than tack soldered as some other manufacturers seem to do.
The underside of the PCB reveals the primary bridge rectifier is a surface mount device, and the main controller is a Monolithic Power Systems MP4056 TRIAC Dimmable, Non-isolated Offline LED Controller with Active PFC, as is in the HPM DLI9002. Because of this, no isolation boundary can be seen in the design, and the outputs are not designed to be user accessible due to the potential for danger, hence the double insulation rating. The terminal blocks are the sturdy long slotted type, and the soldering overall appears to be neat and relatively consistent.
From the front, we find a relatively similar story as with the other downlights, with a diffusing conical sheet placed around the side to improve light distribution. This one seems to have a bit of a cut-out for the wires, which enter from the side, although it was not aligned quite right in manufacturing.
The sheet is not adhered to the heatsink at all, which is the aluminium “dog bowl” shaped shroud covered by plastic coating on the outside. Three screws are used to hold the MCPCB, all of them on the sides. A total of 21 SMD LEDs are used to make the array. The MCPCB has only the markings 13.2023V2. The manufacturer could not be determined with the information gained so far.
While the driver construction was relatively neat, the MCPCB assembly could have fared better. It was observed that the negative contact was not very well soldered – it’s hard to see from the photo but the “land” for the negative terminal (in gold) is mostly not covered by solder, and the contact was made just at the edge of the land. A slightly less worse attempt can be seen for the positive lead. Soldering directly to MCPCBs seems to be a challenge overall, and some manufacturers do it better. The risk is that, over time, the connections could fatigue due to thermal cycling and break.
Another observation was stray solder balls on the sides of LEDs on the MCPCB, also seen in the above image. This might be a sign of improper temperature or improper paste/flux amount or usage. While the unit does function, this suggests that manufacturing quality control might be lacking.
A contrast-enhanced image shows that this unit uses a single-series connection scheme, as in the HPM and DETA products, and has enlarged copper trace areas to maximise heat transfer from the LED through to the MCPCB and heatsink.
When looking at the thermal interface, it seems that some quality grey-coloured thermal paste is used. This might indicate a high-metal fraction paste or one containing silver, as you might use on CPUs, which is a good thing. However, some patterning of the initial application remains, and uneven distribution is seen, suggesting that this unit may also suffer from uneven torquing of screws or a not-perfectly-flat mating surface which affects heat transfer from the LEDs.
In this section, I will discuss some of the test results and subjective opinions from testing this unit.
Output Quality and Opinion
Under testing, the downlight powered on almost instantaneously and certainly well below the one-second threshold. However, unlike other units, this unit has a start-up buzz that lasts about a second once powered on. It sounds a little like an electric razor, and is quite noticeable.
The output of light is slightly non-uniform, likely due to the mounting screw pattern which results in a visible triangular pattern. According to my DSLR and RAW white balance correction, the correlated colour temperature was fairly close to the claim at 2900K.
Looking at photos and objects under the light, the quality of the light appeared roughly comparable to other units. It is likely that the difference between this “80 CRI” unit and most of the “>80 CRI” units is only a few units at the most.
The light was, however, slightly dimmer than the comparison on a side by side basis. This is expected, because of its lower lumen rating at 700 lumens, and the warm white colour temperature which necessarily uses less efficient white LEDs. Whether it actually achieves the stated lumens could not be determined.
No flicker was observed, and aside from the start-up buzz, the unit operated quietly.
Warm-Up Power Consumption
The power consumption was measured with a Tektronix PA1000 with input voltage regulated to 230v +/- 1%.
Over a period of 20 minutes, from cold, the power consumption decreased as expected of most LED products. The power consumption settled at 8.4W, a long way below the 10W claimed on the package. This discrepancy is definitely concerning, as without an integrating sphere and calibrated radiometer, I cannot accurately determine whether this is the result of better efficiency and unit improvements (and thus, improper marking) or a fault causing reduced light output.
Regardless, if we take the 700 lumens rating as being “gospel”, and divide it by 8.4W, the luminous efficacy may be up to 83.3 lm/W, making it a little more respectable.
As to why a company might have a product that consumes less power than advertised and not actually say that, it may be because of a marketing strategy whereby some less informed consumers purchase lighting based on its wattage rather than the output lumens, and mistakenly think that a higher wattage translates to a brighter output (which is not necessarily true). As a result, boasting a higher wattage would potentially draw more customers. This is, however, potentially deceptive and dishonest. Or maybe their upstream supplier changed the product on them without them knowing.
Power Consumption and Power Factor vs Voltage
Again, measurements were made with the Tektronix PA1000 with a Variac used to vary the input voltage through the full range.
As is common with most LED products, a very wide voltage range was seen and excellent power factor was observed. Specifically, above 110V, the light maintained relatively “strict” regulation control, with near constant power consumption up to 277V. The power factor remained above 0.96 throughout the expected operating range, which is an excellent result, especially for commercial customers where large numbers are to be installed and reactive power is a concern. Below 110V, the unit could not maintain output, and by 68V, the output was completely lost. In light of this, the downlight is expected to put out full light for even deep brownouts.
Inrush current was measured using the Tektronix PA1000 in inrush mode. This particular unit seemed to pose a challenge, as it seemed to have some sort of inductive characteristic on switch-off, resulting in high peak negative currents and voltage spikes on turning off.
This resulted in readings of +1.071A and -15.686A. The discrepancy in inrush in both directions seems unusual, and without having the ability to see the inrush waveform or duration, it is hard to judge what is going on.
I’m more inclined to believe the lower value, which would imply about 30 units could sit on a single circuit without too much trouble, but it is unusual that the manufacturer offers no guidance in regards to this.
Ballast Output and Efficiency
The ballast output was measured with the Tektronix PA1000 sequentially measuring voltage and then current to derive an approximate efficiency.
Measured voltage was 62.32V across the LED array with a crest factor of 1.025, implying a relatively smooth low-ripple DC. Measured current was 123.40mA with a crest factor of 1.264, which shows that the current regulation does have some variation in it, but not completely crossing zero. As a result, the power analyzer could not report a frequency or show its waveform.
The array delivered power is 7.69W, resulting in an approximate calculated power efficiency of 89.26% which is quite respectable and similar to the recorded value from the HPM DLI9002.
Testing for dimming compatibility utilized the three “random” dimmers I have in my collection.
With the leading edge IKEA Dimma lead, it was found that any level aside from full brightness resulted in flickering and both dimmer and fixture noise. The dimming range was restricted to about 20% minimum.
With the Nixon Universal, there was noise and flicker at most ranges above 10%. Dimming could reach as low as 1%, although start-up times were lengthened to about four seconds.
The only dimmer it co-operated well with was the DETA 150W Universal Trailing-Edge dimmer, where aside from a slight increase in start-up times, it was able to operate the full range linearly without any flicker. The low end of the dimming range reached about 10-15%.
It seems likely, that with reputable trailing-edge and possibly universal dimmers, that reliable, smooth, quiet and consistent dimming could be achieved, but compatibility with cheap, uncommon “random” dimmers seems to be limited.
The Mercator Retina LED is a very unusual unit. It claims a 10W power rating, but in testing, it was shown to be an 8.4W unit instead. While I suppose nobody should complain about a light that uses less power than stated, without being able to verify the actual light output, I cannot be sure whether this “saving” is due to output below the stated claims, or due to improvements in LEDs used. Regardless, the discrepancy between the printed specification and test result should be cause for concern.
The unit itself, on paper, has a relatively low output and efficacy compared with its peers. It also has some installation frustrations with its abutted-only insulation contact rating, and an IP44 rating which is not covered by warranty for outdoor installations under any circumstances.
The unit’s driver comprises of quality components with a good build quality, although the main transistor is slightly less efficient than some others. The MCPCB LED array did show some cause for concern with uneven torquing of screws, solder balls and poor soldering of wires to the array. The thermal interface material was good, but its contact was also uneven.
The unit has a “buzz” on start-up and some compatibility problems with the “random” leading edge and universal dimmers I had to test with. It worked best with trailing edge dimmers. Aside from that, it performed reasonably, without any perceivable flicker and light output that was similar in quality with other units. The output uniformity was slightly affected by the screw positioning and the diffuser.
Again, it seems that each product has its own mixture of strengths and weaknesses, but regardless, it has been an educational experience having the opportunity to examine this unit.