UPDATE: Cree has just voluntarily recalled these lamps — all 700,000 of them — because of a problem with arcing on the spring-loaded connectors.
Like the original Cree T8, the new T8 TW series is a replacement lamp: You pull out the old fluorescent tube light and insert the new Cree, and voila, you have a linear LED tube light operating from your existing fluorescent fixture— no re-wiring, no pulling out of the old fixture ballast.
The new lamp is significantly cheaper than the original Cree T8: $22 vs $30. Easy to install, easy on the wallet — let’s take a look at it.
The new LED T8 TW series has a round cross-section and is visually indistinguishable from a fluorescent T8, unlike the old Cree lamp which had an elliptical cross-section. And, where the previous version had a solid aluminum heat sink along the back of the lamp, this version is all plastic, and is opaque on the side that faces the fixture. Getting rid of that big chunk of aluminum must have had a lot to do with the drop in price.
But how does it perform?
The table below shows the manufacturer’s specs for the lamp compared to a common fluorescent T8 lamp.
|Cree T8 TW series||T8 fluorescent|
|Color Temp||2700K, 4000K||3500K|
|Lumens||1700 lm||2470 lm|
|lm/W||92 lm/W||77 lm/W|
|Claimed lifetime||50,000 hrs||24,000 hrs|
While Cree claims the lamp has a power draw of “as little as 18.5W”, the ballast I used in my two-lamp fixture, a very common Philips Advanced, had a total fixture power draw of 49W, (24.5 W/lamp), for the 2700K version, and 52.6W (26.3 W/lamp) for the 4000K lamp. I’ve asked Cree with what ballast they were able to achieve 18.5W/lamp; we’ll see what they say.
On to the spectral power data for the lamp. Let’s look first at the 4000K version. As usual, we’re using the MK350 spectrometer with the MoreSpectra software from Moreland Lighting:
Next up is the 2700K version; I’ve superimposed the 4000K graph —the green line — over the 2700K lamp’s SPD, which is a handy feature of the MoreSpectra software:
You can see that while the 2700K lamp shows that the blue-green spectrum is relatively less in the 2700K spectrum, the R12 number shows 89, even higher than the 4700K’s R12 of 70, so you don’t need to worry about the 2700K version’s ability to render blue.
(I personally preferred the 4000K version for room lighting.)
In addition, the lamps are dimmable if your fixture has a dimmable fluorescent ballast, which are relatively rare. The lamps turn on instantly, with no hum or flicker in my non-dimming installation.
Next, let’s take a look inside the lamp.
The first and most obvious change in the new version is that it gets rid of the long aluminum heatsink backbone, a significant savings in cost and weight. What else?
It’s a bit of a pull, but you can get the end-cap to pop off, and see that Cree has again made good use of connectors rather than hand-soldering to make the connection from the fixture power to the internal LEDs. (This pretty much destroy the light, however.)
What’s equally impressive is the minimal parts count of low-cost, discrete devices. There are no power ICs, or even transistors — just a handful of capacitors and diodes and a single inductor at one end. The pc board itself is a single-sided, inexpensive board – no metal-core board.
There are 80 LEDs on the board: Two parallel strings in the same polarity of 20 LEDs each, that connect to two additional 20-LED strings in opposite polarity to the first two strings for a total of 80 LEDs.
You can see the wide traces on the board that also act as an inexpensive heat-dissipating device — a poor man’s heat sink.
Below are two photos from last year’s teardown article showing the heatsink and elliptical tube as well as some of the 120 LEDs it took to light up that lamp:
The original Cree lamp used 120 LEDs; the new lamp has 80. However, the old version claimed an output of 2100 lm or 17.5 lm/LED, while the new lamp produces only 1700 lm, or 21 lm/LED — a drop in total lumen output, but a 21% increase in lm/LED over the original version. Cree apparently believes the market will accept a lower rated lumen output for the lamps, and, combining it with the increased output per LED, it’s able to drop the number of LEDs per lamp.
I’m thinking that Cree actually has quite a bit more room to drop the price. Especially since, given the lack of electronic components, the main cost of the lamp, other than the 80 LEDs, is probably the plastic. I would not want to get into a price war with Cree on making LED tube lights, given that Cree makes its own LEDs.
(Actually, I don’t know for sure that Cree uses its own LEDs in this lamp. These LEDs measure 3.3 x2.9mm, and Cree’s portfolio of LEDs with that packaging dimension don’t show an LED that looks like this. But I may have missed it; Cree has a huge LED component portfolio.)
Now let’s step back and look at the bigger picture of the lamps — what market is this replacement lamp aimed at, and is it the best solution? The usual reason for changing to LEDs is the increase in efficient conversion of electric power to light, and this has been a strong reason to move from incandescent bulbs to LED bulbs. However, fluorescent lamps are pretty efficient already, with some fluorescent T8 lamps achieving efficacies of over 90 lumens per Watt (LPW). At best, when the Cree lamp is able to achieve a total power draw of 18.5W, a number which includes the fixtures legacy fluorescent ballast, the Cree produces 1700 lm and achieves 92 LPW — or about the same as a modern efficacious fluorescent T8.
However, the number that I saw on my Kill A Watt power meter hooked up to the fixture with a very common Philips Advance ballast, was 24.7W, or 67 LPW — which, if we were holding LED tube lights to the same standard as the directive issued last summer, would not qualify for the minimum efficacy numbers for a fluorescent T8. It seems a bit strange to pay more money to replace fluorescents with LEDs to achieve LESS light output.
So why would there be a market for LED replacement lamps? Two reasons: Light quality, and the “cool” factor.
The light put out by these lamps is a high CRI, which is not easy to achieve in a fluorescent lamp. If you are using this light in a residence, where you don’t want the crypt-keeper appearance conferred by high-efficacy, low CRI, ultra-cold fluorescent lights, these are a good fit. And, especially if you have a retail or commercial establishment where using LEDs allows green bragging rights, then these are also a good fit. But if first and foremost, you want good value for your time and money, these are not such a good fit.
This is not a slam against the Cree (or the Philips InstantFit) lamps — it’s more of a questioning of the replacement approach. Having to drag along legacy ballasts, which add not only power conversion inefficiencies but also their own shorter lifetimes, is not a good use of resources. My suggestion is that you bite the bullet, and re-wire your fixtures to remove the fluorescent ballast.
Cree is not in the retrofit lamps market as yet. I certainly have no special insight into Cree’s plans for lighting, but Cree’s policy in the past has always been to make it as easy as possible for the consumer to switch to LED lighting, and “easy-as-possible” does not match up to having to re-wire (retrofit) your fixtures. Nonetheless, I can see Cree getting into the retrofit LED tube market eventually, if only to get away from the lamentable inefficiency of the legacy ballast problem.