Plessey took another stab at commercializing its proprietary GaN-on-Si process by making its next-generation mid-power LED available. It won't be enough.
The new chip is the "entry level" PLW114050, according to LEDInside, though as of this writing, there hasn't been a press release from Plessey since July.
We covered the background for Plessey's entry into the LED manufacturing fray in detail with Brian Coppa's piece on this site back in April. In a nutshell, the company develops GaN-on-Silicon with MOCVD, using a patented recipe for the epitaxial buffer layers that it says use only 2.5 um, versus competitors' epitaxial layers that are 8 um thick.
Management's bet
Plessey seems to be betting on better capital equipment utilization; it takes about three hours per micron for LED epitaxial growth to occur, resulting in triple the throughput per $2.5 million reactor. But does this penny-pinching really matter in the grand scheme of things? I'll submit it doesn't, and it will drive the company to insolvency.
Plessey is hell bent on using the thin-buffer, GaN-on-Si recipe that the company acquired from Cambridge University in February 2012 with the purchase of the commercialization spinoff CamGaN. One of the core claims is that GaN on six-inch silicon costs 80% less than GaN on sapphire or SiC. It's hard to believe the final product will be any cheaper. The company making the SiC LEDs has the cheapest packaged device I've seen in distributor channels that has a respectable and usable 120 lm/W, which can be immediately stuffed on a board or substrate.
Filling the fab
The manufacturing throughput for CamGaN's recipe looks promising and will deliver more LEDs per machine, but a semiconductor company needs a compelling product to fill the fab. This entails performance and customer system cost. It's arguable that Plessey could get by with a lower fab utilization percentage, but it doesn't appear that the number of fab moves has changed -- at all. If that's the case, all but one step in the fabrication of the LEDs have the same throughput and cost as their competitors, such as Toshiba/Bridgelux. So I'm guessing maybe Plessey can break even with a 50% utilization, vs. the more than 60% most fabs need.
Let's not look at Plessey's advanced PowerPoint presentation, which is sitting at 150 lm/W and shipping in volume. Instead, let's take a peek at the freshly announced PLW114050 data to see if the company can fill that fab with the next-generation product.
Being among an august band of rogue experts, I'll mount a direct comparison with another GaN-on Si-LED, Toshiba's TL2FK-NW1. The two appear to have the same production status.
| Part | CCT | Lumens | CRI | Vf | If | Lm/W | Footprint | Θ1/2 | Toperating |
|---|---|---|---|---|---|---|---|---|---|
| TL2FK-NW1 | 5,000 K | 23 | 80 | 3.3 V | 80 mA | 87 | 3.0x1.4 mm | 120° | -40~100°C |
| PLW114050 | 5,000 K | 12.3 | 70 | 3.4 V | 60 mA | 60 | 3.0x2.1 mm | 120° | -30~85°C |
Compelling? You be the judge.
Unsolicited advice
The only way for Plessey to win business, in my opinion, is by caving the pricing, but that also means its customer has to deliver almost 50% more power for the same light output in a system, which is not cheap. To me, this new device seems like a lame duck, though it is an advancement over the previous one. The CTO's roadmap projects 150 lm/W, but I question whether Plessey will be around long enough to see it happen. It has already missed its 1W white LED goal by six months -- an eternity for a startup. There's no reason it couldn't have made this device. It just needed to chop the wafer into a bigger die size (assuming the dislocation densities aren't fatal).
One of the largest strategic errors any company can make is making decisions that result in late market entry. Another is to stick to its guns due to sunk funds and with technology that simply is not ready for prime time. Were I the CEO of Plessey, I'd have licensed Bridgelux's epi recipe and put that MOCVD machine and fab to work. That would have given marketing a compelling product to sell, and I'd kick asses to get the word out, including having the still-absent press release on the company website.
If this company had its act together, it would be going head to head with Toshiba right now -- lumen for lumen, efficacy for efficacy -- while its boffins would be playing on the midnight fab shift to bring the cost reduced thin-epi into play. If I were Plessey's CEO, I'd play that throughput IP when my fab was at 90% utilization, giving my boffins the time to get to that 150 lm/W on thin-epi GaN-on-Si -- not at the time when there are tumbleweeds in my cleanrooms.
With these kinds of very dumb operational and executive decisions, I can only question how much longer this company can last. Sucking on the public teat (as Plessey has been) can last for only so long before momma slaps you away.
Good luck, Plessey. In my opinion, luck's the only thing that's going to keep you afloat unless you make a lot of changes -- stat.
Related post:
No comments:
Post a Comment