Ignis Makes Price Promise
The company has plans to produce Sonet transceivers for a price that's comparable to enterprise optics. "Our mission is to provide telecom-grade, singlemode products at multimode price points," says Mike Lebby, Ignis's CEO. Lebby was formerly a VC at Intel Capital (see Intel VC Resurfaces at Ignis Optics).
The idea of bringing enterprise economics to Sonet components was apparently what motivated JDS Uniphase Inc. (Nasdaq: JDSU; Toronto: JDU) to buy IBM Corp.'s (NYSE: IBM) transceiver division last December, according to industry observers (see JDSU's Bid for the Enterprise).
But back to Ignis: Its first product, announced this morning, is an OC48 (2.5 Gbit/s) SFP (small form-factor pluggable) transceiver, targeting VSR (very short reach) and SR (short reach) applications at up to 2 kilometers. Ignis says it is already shipping beta samples of this product and plans to be in full production in Q2 2002.
SFP is a common type of package in enterprise networks. It has the advantage that modules can be fitted to boards after the rest of the board has been populated, so the optics don't get subjected to the high temperatures of soldering. It also allows customers to chose optics separately from the board -- therefore, systems vendors don't need to keep optics in their inventories. Devices may also be easily upgraded or replaced if defective.
However, there are several reasons why this style of module does not translate easily into a Sonet situation.
First, Sonet data rates are typically much higher than those in enterprise networks, and that leads to electromagnetic interference (EMI) issues. Second, enterprise networks usually only span a few kilometers and therefore can use multimode fiber. And optics for multimode fiber are a lot cheaper and easier to manufacture than those for Sonet.
Ignis has taken a long hard look at the packaging and manufacturing of transceivers and believes it has come up with something that will solve both these problems, according to Lebby. He's not willing to divulge many details at this stage, but he does offer an outline of the solution.
To solve the EMI problem, Ignis has eliminated the TO (transistor outline) can from its modules, Lebby claims. The TO can is a round metal package that was invented in the 1950s for packaging individual transistors, he explains. When optics came along, vendors simply put a window in the top of the can to let light in or out, and they've been using this to package lasers and detectors ever since.
The trouble starts as data rates go higher. The TO can has three or four legs sticking out of one end, and when operated at 1 Gbit/s or higher, the legs start to behave like antennas. "We believe most of the competition is using the TO can, and they have a lot of work getting 'round the EMI issues," Lebby contends.
It's worth pointing out that Ignis is not the only vendor to work around the TO can problem. Manufacturers of Xenpak modules for 10-gigabit Ethernet have already been forced to come up with various alternatives -- but they've ended up with a rather large, distinctly un-enterprise-like package (see Sizing Up Xenpak ).
The other issue -- manufacturing costs -- is related to the tolerances required in assembling singlemode products. Tolerances are much more relaxed in multimode fiber, which has a typical core diameter of 50 microns and tolerances of plus or minus 10 microns. In contrast, the core of a singlemode fiber is only 3 to 4 microns in diameter, and that pushes manufacturing equipment to its limits.
Lebby says that Ignis has devised what it calls "expanded and contracted beam optics," so that it can manufacture its products using equipment and processes intended for multimode gear.
But what about those low prices? At one end of the scale, Sonet transceivers can sell for several thousands of dollars or more, depending on data rate. At the other end, enterprise optics can retail for less than $100. Ignis won't release prices for its products, but they will lie somewhere between these limits.
However, there's one other factor affecting transceiver prices -- testing. Telecom-grade equipment needs more extensive reliability qualification than enterprise optics, and that could mean at the end of the day there will still be a price differential between otherwise identical products for the two markets. Lebby acknowledges that this could push up the price of his company's products.
It's worth noting that other companies are pursuing similar goals. Picolight Inc., for example, released a family of SFP transceivers earlier this month (see Picolight Launches Transceivers).
Picolight is using Vertical Cavity Surface Emitting Lasers (VCSELs) to get its manufacturing advantages. VCSELs help to lower costs because the lasers themselves are a lot cheaper to make and test than standard edge-emitting devices (see Laser Blazers). But again, it is the packaging that really makes the difference, says Warner Andrews, Picolight's VP of marketing. "We have a monolithic package based on a single alignment, and therefore it can be automated," he claims.
— Pauline Rigby, Senior Editor, Light Reading