Funding for startups

Investors Pump $44M Into nLight

On Thursday, nLight Photonics announced that it has closed a $44 million funding round with investments from Menlo Ventures, Oak Investment Partners, Mohr Davidow Ventures, Adams Capital Management, Cedar Grove Investments, and several individual investors. This is the optical components maker's second funding round, bringing its total funding up to just over $50 million.

The startup is still keeping its cards close to its vest, but does say that it's developing pump lasers and erbium doped fiber amplifiers (EDFAs) -- devices that boost the power of an optical signal. NLight's CEO, Scott Keeney, says once nLight's products are ready they'll alleviate some of the bottlenecks present in long-haul networks.

Power is the key. Keeney says that unlike many pump lasers on the market today, his firm's lasers are strong enough for Raman amplification, which is fast becoming an important technology for the long haul (see Raman Amps: Key to Optical Future and CIBC Bullish on Raman). In Raman amplification, the optical fiber itself becomes the amplifier -- a very-high power laser is shined down the link, which shakes the molecules of the fiber so hard that it enables the transfer of energy from the pump laser beam into the signal. This effect is only significant above around 700 milliwatts of pump power, according to industry experts.

EDFAs also benefit big time from having higher-power pumps. Every time the number of channels in a DWDM (dense wavelength-division multiplexing) system doubles, the optical power requirement for boosting those signals also doubles. Since most of the cost of pump lasers is in the packaging, it's better to increase the power of individual devices, rather than simply multiplying the number of pumps inside the amplifier.

The bottom line is that network equipment manufacturers are looking to cut down the number of times optical signals need to be regenerated in long-haul networks (see The Ultimate Backbone ). When Raman amplifiers are used in combination with EDFAs, they could allow the signal to travel an extra 100 kilometers before it needs boosting.

NLight's obviously picked a product with a future. But at this stage it's hard to know whether the startup has the technology to stand up to the heavyweight competition or whether it has the manufacturing prowess required to simplify life for its customers. Though he’s clearly excited about nLight’s prospects, Keeney doesn’t want to stray too far from the press release just yet.

The competition looms large in the shape of JDS Uniphase Inc. (Nasdaq: JDSU; Toronto: JDU) and SDL Inc. (Nasdaq: SDLI), partners in a proposed megamerger (see JDSU and SDLI Delay Merger, Again). JDSU and SDL are the two most dominant players in the pump laser chip market, according to figures from Electronicast Corp.

"The [JDSU-SDL] deal is fundamentally about -- and don't quote me on this -- pump lasers," Keeney says. "What SDL really has are fantastic 980-nanometer pumps." (see Who Gets a JDSU Jewel?).

Furukawa Electric Co. Ltd. is also a significant player. Smaller fish in the pump laser pond include nimble startups such as Novalux Inc. (see Laser Startup Bags $109 Million), Princeton Lightwave, and Alfalight Inc. (See Alfalight Closes $28M 2nd Round.)

Keeney says he thinks the market will be big enough for all comers. Indeed, even as others were impressed with the size and price of the coming JDS-SDL merger, Keeney sees the marriage as a validation of his firm’s technology.

And, about a year from now, when nLight’s products will be on the market, Keeney hopes his firm will surpass what JDSU and others are pumping out.

Phil Harvey, senior editor, and Pauline Rigby, senior editor, Light Reading http://www.lightreading.com
Photonboat 12/4/2012 | 8:59:15 PM
re: Investors Pump $44M Into nLight Is this really accurate? Does anyone have a link to a study or paper verifying that Raman increases transmission distances fourfold?

<<...Regeneration...is typically performed every 400 to 600 km...Raman amplifiers can increase this distance fourfold...>>

Also, as for pumping, Furukawa isn't a "small player" in the 1480 nm market. (They are a smaller player in the 980, but Raman pumping is in the 14XX range.

Petabit 12/4/2012 | 8:59:07 PM
re: Investors Pump $44M Into nLight There are lots of points to make about this article, and most of them stem from fundamental misunderstandings about Raman amplifiers.

1) Raman amplifiers reduce the loss of a span of fibre. A typical span loss is 23 dB (to get one photon out, you need to put 200 photons in). A Raman amplifier can typically reduce that loss by 4-10 dB. This makes a 80km span of fibre behave more like a 60 km span. Reducing the effective span length increases the distance between regenerators. So Raman amplifiers can make your system go further.

For more details see the raft of papers at OFC this year. It also explains why Corvis can go further with 60km spans than Ciena can with 80km spans. It's just a shame that real networks have 80-100 km spans.

2)Raman amplifiers do not have thresholds. Depending upon the fibre type, 200 mW of pump will give you 4 dB of gain, 400 mW will give you 7 dB of gain, and 800 mW will give you 10 dB of gain. It is completely wrong that the Raman effect begins at 700 mW. The thing to bear in mind is that 499 mW is the IEC limit for how much power you can use before the entire system must be placed inside a shielded room. Class 4 lasers are scary objects.

3) Typical 980nm pumps go up to 220 mW, 1480 nm pumps to 300 mW. Except the ones that those power levels cost way more than the lower power ones.

4) Furukawa make really nice 1480nm pumps. And very very nice Raman pumps. They do outsell SDL and JDS.

5) The whole JDSU/SDL Zurich thing is all about 980 pump CHIPS. Between them, they make around 90% of the chips. Which are packaged by Nortel Opto, Corning Lasertron and SDL. Who sell them to the EDFA manufacturers. JDS sell almost no pump lasers.

6) Since most of the cost of the pumps in in the packaging, it is better to use more low power pumps than a few high power ones. Since cost has nothing to do with it. Try control, or reliability as the reason why most modern edfa designs use 6 to 8 pumps.

7) Raman is a cute trick, and will be useful in conjunction with EDFAS, but the poor efficiency, size, cost and safety of Raman amplifiers will prevent them becoming the dominant technology.


[email protected]
iprsvp 12/4/2012 | 8:59:01 PM
re: Investors Pump $44M Into nLight that's a good one.
Pauline Rigby 12/4/2012 | 8:58:59 PM
re: Investors Pump $44M Into nLight Petabit,

Thanks for your post. I've got some questions and comments about the points you raised:

1) Raman amplifiers boost the signal level above what it would normally be, but it's still loosing energy overall. That seems a kind of nitpicky point to me. It's still amplifying, but has more losses to contend with than a lumped amplifer.

So, to boost the signal -- or stop it from loosing it's energy so fast -- the extra energy comes from the pump laser beam, no? That's what I was trying to say. After all, where else can it come from.

Do you agree?

2) The sentence about needing 700 milliwatts was carelessly worded, so I changed it. Granted there is no threshold, but I think that this figure of 700 mW, giving a bit under 10 dB of gain, is what is generally considered a useful amount of gain. Correct me if I'm wrong, please.

3) Good point, though I think much higher powers are becoming available now. I've heard of 1 watt on chip (can't remember the wavelength though!)

4) Sorry, Furukawa. I've given it due recognition in the story now.

5) I know.

6) This doesn't make sense to me.

7) I pointed out that Raman amplification is used in combination with EDFA's, n'est ce pas?


Petabit 12/4/2012 | 8:58:59 PM
re: Investors Pump $44M Into nLight Hi Pauline,

Raman amplification is not quite how you describe.

The important thing to preserve though an optical line system is signal to noise ratio (SNR). An amplifier will boost the signal power, it will also boost any noise coming into it, and add some more noise of its own. So any amplifier degrades the SNR

Using a length of fibre followed by a lumped amplifier, allows both the signal and the noise to be attenuated by the fibre, before they are both boosted by the amplifier. The inherent noise of the amplifier is sort of constant, so if you put too long a length of fibre before the amplifier, the amplifer noise will swamp the signal and you lose all the SNR.

So putting your amplifiers close together introduces less SNR degradations than further apart. A distributed Raman amplifier reduces the loss of the fibre, so the SNR is still degraded by the span, but less than if the Raman amplifier was not present. Effectively shortening the spans increases the system reach.

The drawback is the distributed Raman amplifiers also add noise the the system. So you have to carefully balance the positive effect of increasing the Raman pump power to reduce the loss of the fibre and improve the SNR, with the negative effect of increasing the Raman pump power and increasing the noise it introduces.

So a small amount of Raman pump power will improve the SNR of the system, but adding too much will start to degrade the SNR.

For most of the common fibre types this leads to an optimum SNR point at about 700 mW.

Once you go past 499 mW you have to put a load of engineering controls at the site. The fibre must not have connectors, no patch panels, it must be shielded, you have to be really careful about cleanliness.

To get wideband gain from a Raman amplifier you need multiple pump wavelengths, up to 12. Which means you need lots of medium power pumps.

The SDL 1W pump is interesting, but only useful for low bandwidth systmes (like submarine).


Pauline Rigby 12/4/2012 | 8:58:57 PM
re: Investors Pump $44M Into nLight That was a really useful explanation.

Good one!

peterpan 12/4/2012 | 8:58:47 PM
re: Investors Pump $44M Into nLight test posting. pls ignore.
photons 12/4/2012 | 8:58:45 PM
re: Investors Pump $44M Into nLight testing
Sign In