CyOptics demonstrated that its modulator developments, based on indium phosphide, have stolen a significant performance march on modulators made with the usual material for such devices, lithium niobate.
The big question is whether it’s a big enough performance advantage to convince system vendors that it’s worth taking a risk on what amounts to unproven technology. If it is, indium phosphide could end up displacing lithium niobate as the material of choice for high-performance modulators.
Let’s start with CyOptics’ performance claim. It says it’s been shipping samples of its 40-Gbit/s electro-absorption modulators (EAMs) to selected customers since the beginning of this year, and these devices are achieving bandwidths in excess of 36 gigahertz (GHz). That’s significantly more than the record 30 GHz that Agere Systems (NYSE: AGR) claimed for its lithium niobate modulator last month (see Agere Claims 40-Gig First).
These figures are probably quite confusing to folk that are used to bandwidths being quoted in terms of gigabits-per-second rather than gigahertz (like me), so let’s back up a bit and try and put things into perspective.
Modulators are used in conjunction with lasers to translate digital signals from their electrical format -- a stream of different voltages -- into a stream of light pulses that can be carried over optical fiber. The laser sends out a continuous stream of light, and the modulator acts like a shutter, interrupting it. Various encoding schemes are used to translate the 0s and 1s of digital data into different voltages and light pulses, so the performance of the modulator is quoted in terms of how many times per second it can interrupt the light coming out of the laser, rather than the amount of data carried by the pulses it creates
Now for the tricky bit. When a modulator is dealing with enormous flows of digital data, like 40 Gbit/s, it has a tough job keeping up. The modulator gets a signal telling it to open its shutter, and while it’s still in the process of opening it, it gets another signal telling it to close its shutter, and so on. (It’s not an actual shutter but it’s easier to visualize things like this.) The upshot is that the shutter never manages to get fully open or fully shut. The key thing is to get a big enough difference in the light power so that detectors at the other end of the fiber can distinguish between the two states.
This is where the performance issue comes in. The faster the modulator works, the further it can get in opening and closing its shutter and the bigger the difference it can achieve between the two light levels. The usual way of measuring things is to work backwards and give the cycles-per-second that will ensure that no more than three decibels of power is lost by the light as it passes through the modulator -- which in the CyOptics case is 36 GHz or 36 thousand million cycles a second. Rather a lot, eh?
Now let’s return to the issue at hand: indium phosphide versus lithium niobate modulators.
CyOptics says its 40-Gbit/s EAM isn’t just better on bandwidth. It’s a lot smaller -- less than one inch long, compared to more than fives inches for the lithium niobate equivalent. It also consumes about one quarter of the electric power. The power issue is important because system vendors can get away with lower-cost electrical components as a result, according to Hava Volterra, CyOptics’ VP of marketing and business development.
In addition to this, EAMs are more suitable for volume manufacturing because they’re semiconductor devices, unlike lithium niobate modulators. That translates into lower prices, according to Volterra. Right now, however, it’s too early to say how much lower prices might be.
Of course, manufacturers of lithium niobate modulators contend they’ve got a significant edge over companies developing indium phosphide EAMs, which include not just CyOptics but also Alcatel Optronics (Nasdaq: ALAO; Paris: CGO.PA) and OKI Semiconductor.
In fact, Codeon Corp., a startup developing lithium niobate modulators (see Codeon Bets on Clever Crystals), dismisses indium phosphide as an unsuitable material for modulators in anything other than metro networks. Ganesh K. Gopalakrishnan, Codeon’s CTO, lists the following reasons for taking this stance:
- High return loss. Gopalakrishnan says EAMs reflect back a lot more light than lithium niobate modulators, which causes problems.
CyOptics’ Volterra rejects this. “The return loss on our EAM is under 28dB, which is well within the Bellcore spec and the same as LiNb [lithium niobate] modulators,” she writes in an email to Light Reading.
- High insertion loss. This makes EAMs unsuitable for long-haul transmission networks, according to Gopalakrishnan.
Volterra says “this isn’t a problem in long-haul networks because they’re always amplified.”
- Low on/off extinction ratio. This measure of the clarity of signals limits EAMs to metro applications, according to Gopalakrishnan.
Volterra says this is plain wrong. “It’s actually high with EAMs (which is good). In fact, a major advantage of EAMs is that you can get the same extinction ratio as you would with LiNb modulators, but with much lower drive voltage. Conversely, if you run EAMs with the voltages you ran LiNb modulators, you'd get an extinction ratio of over 20dB -- which is extremely high."
- Chirp. Another phenomenon that distorts signals, chirp isn’t fixed, as it is with lithium niobate modulators, which makes it much tougher to deal with.
Goal! Not a peep from Volterra on that one.
- Different EAMs needed for different wavebands. Lithium niobate modulators, in contrast, can be put to use on the C, L, or S wavebands.
Volterra acknowledges this but says it’s not an issue in today’s systems, which only use the C band.
— Peter Heywood, Founding Editor, Light Reading