Ciara McCarthy: Hi, my name is Ciara McCarthy and I'm the Chief Marketing Officer at EPS Global. Today we're diving into the world of optical networking with John Lynch, VP of Sales for EMEA at EPS Global, and Kevin Affolter from Coherent. Kevin, the Product Line Manager for 100G ZR DCO, is here today to discuss his latest product. Let's jump right in. John, could you please introduce yourself? Afterwards, I'll hand over to you to discuss the technical aspects of the 100G ZR DCO optical transceivers with Kevin. 

Kevin Affolter: Certainly, John. The 100G ZR family of products is quite unique in the market. It enables IP over DWDM applications to be realized using 100G routers and switches. For those unfamiliar with IP over DWDM, it essentially means plugging the colored optics directly into switches and routers. IP over DWDM results in a smaller footprint and lower power dissipation because it allows for the removal of transport equipment typically found in the network, as well as short-reach optics. This saves on cost, power dissipation, and footprint, all contributing to lower overall costs. Additionally, with less equipment in the network, reliability is improved. The development of this product family was significantly enabled by co-developing our optics with our in-house Coherent DSP capability, allowing us to create the most efficient solution possible. 

Kevin Affolter: The key differentiator is the form factor. It's a QSFP 28, a widely available form factor, though typically associated with older technology. This is the first time a Coherent solution has been available in that specific form factor. This enables IP over DWDM for 100G switches and routers. Crucially, this was made possible by significantly reducing the module's power dissipation. Achieving a level around 5 watts was a Herculean effort. That reduction in power dissipation is the major achievement, enabled by our custom-designed DSP and optics. 

Kevin Affolter: We developed the DSP with our internal team, partnering with external fabs, of course. This approach allowed us to focus on a few specific applications and remove any unnecessary features from the DSP. Typically, when working with a commercial DSP partner, the significant investment required means they often include many features to address the broadest possible customer base. In our case, the application space was quite narrow. Constraining it to these few applications was key to reducing power consumption and maintaining a small size. Fewer features mean less silicon real estate and lower power consumption, all contributing to the overall design. 

Kevin Affolter: It varies slightly by part. The -8 dBm version, designed for the commercial operating temperature range (0 to 70°C), consumes a worst-case end-of-life 5.5 watts. Typically, however, power dissipation is closer to 4.5 watts. 

Kevin Affolter: A couple of comparisons come to mind. One could be a 400G ZR or ZR+ type product, which uses a larger QSFP DD form factor. It's larger because it consumes more power—typically more than double what our module consumes. Another example is the CFP2 form factor, a more historical option for 100G solutions. Its power consumption is likely three to four times higher, and it's a much larger form factor that doesn't plug into a router or switch. Consequently, it doesn't enable IP over DWDM, so the previously mentioned savings are not realized. 

Kevin Affolter: Yes, a couple of factors influenced this decision. The primary one is the large existing installed base of 100G ports in switches and routers—millions of units—though currently used in non-IP over DWDM applications. This represents a huge market opportunity. This, combined with guidance from network equipment companies seeking more dense, lower-power dissipation solutions at 100G, led us to the QSFP 28 form factor. 

Kevin Affolter: Yes, both product versions are available in C-temp or I-temp modes, supporting applications from 0 to 70°C, or outdoor applications from -40°C to +85°C. The -8 dBm, or standard power product, is intended for ZR applications. It can be non-amplified, reaching up to 80 km, or amplified, extending to about 120 km. These are typically simple point-to-point networks. Currently, customers might deploy ZR4 to satisfy the same link, but that allows only one such link per fiber. With our solution, you can have up to 96 on the same fiber. The 0 dBm product, however, is intended more for metro regional networks, typically ROADM-based (Reconfigurable Optical Add-Drop Multiplexer). The higher power is required for consistency with the power levels in these networks. We also offer a version of the 0 dBm product with an integrated tunable optical filter on the transmit side. This filter cleans up out-of-band noise in the optical spectrum and is targeted for a specific application: colorless ROADM networks. In these networks, ROADMs are based more on couplers and splitters than filters, meaning there's no inherent filtration in the ROADM itself. 

Kevin Affolter: Both the high-power and standard-power parts have the same chromatic dispersion compensation capability, which is standardly 120 kilometers of reach. They also feature built-in configurability to enable additional chromatic dispersion compensation, extending the specification up to 300 kilometers. In reality, we've conducted demonstrations in our labs and at events like OFC exceeding 400 kilometers. This refers to chromatic dispersion tolerance. From a loss budget perspective, the standard device has a 22 dB loss budget, and the high-power device has a 30 dB loss budget. 

Kevin Affolter: Yes, I think the primary use case, which I hinted at earlier, involves customers who have a 100G router or switch. They are currently using it with short-reach optics, connecting from the switch to a transport box (a transponder or muxponder), then over the line system, with a similar setup—transponder, muxponder, short-reach optics—into the router at the other end. By transitioning to these 100G Coherent optics, they can remove that transport equipment. The transponder or muxponder is eliminated. The line system remains, but the muxponder and short-reach optics are removed. This results in fewer network elements, meaning fewer potential points of failure. There's also associated power dissipation and footprint with the removed equipment, all of which represent real costs to service providers. 

Kevin Affolter: Yes, thank you for that reminder. Absolutely. We have customers using 10G DWDM—specifically 10G tunable parts—who have already deployed the 40 or 80 wavelengths permitted, effectively filling up the C-band. They need to expand capacity but want to avoid purchasing or leasing new fiber. They can reuse the existing line system—the optical fiber, amplifiers, muxes, and demuxes—and replace their 10G routers and switches with 100G. This provides a 10X increase in capacity on that fiber, a significant part of their capacity expansion strategy. 

Kevin Affolter: Correct. The line system can remain exactly the same. 

Kevin Affolter: Yes, the SFF-8636 standard was written many years ago and has undergone several revisions. However, it has functional limitations. To give one example among many, it doesn't support wavelength tunability. Within a system compliant with SFF-8636, there's no command to change the module's wavelength. Since we're discussing tunable modules, this is a significant limitation. CMIS offers many more features, including wavelength tunability. Within the CMIS management interface, you can change the module's channel wavelength. Typically, CMIS products will be used by network equipment manufacturers and may be sold as part of a system solution. The SFF-8636 part is more intended for legacy routers and switches already deployed in the field, where the service provider aims to eliminate equipment (like transponders or muxponders), increase reach, boost capacity, or achieve a combination of these goals. 

Kevin Affolter: Primarily, it's to access the extra functionality CMIS provides, such as wavelength tuning and the ability to choose between Gigabit Ethernet and OTN. These are features offered by CMIS but not by SFF. 

Kevin Affolter: The TOF, or tunable optical filter, is used to suppress out-of-band noise in the optical spectrum emanating from the laser. If you examine a laser's spectrum, it has a noise floor and a peak at the lasing wavelength. The TOF suppresses this floor to much lower levels. This is important for colorless ROADM applications (unfiltered ROADM applications) but isn't necessary for colored ROADM applications. It's primarily needed for those colorless applications, and it's super important for users with deployed colorless ROADMs. 

Kevin Affolter: Yes, I believe we even have something that works for the extreme environments of Dublin! We also have I-temp versions available, operating from -40°C to +85°C. These are typically for outside plant applications, such as street cabinets or pole-mounted installations. 

Kevin Affolter: Yes, a point-to-point network is relatively simple—conceptually, a straight shot between two locations. This represents the 10G to 100G upgrade scenario I mentioned earlier. A ROADM network is more complicated. It allows traffic to be added dynamically and reconfigured at multiple points within the network. You can almost think of a ROADM network like a road network, with highways, on-ramps, and off-ramps—a very similar concept. 

Kevin Affolter: Yes. We have parts that work in both. The one with the tunable optical filter is used in colorless ROADM applications. The main difference is that in an unfiltered, or colorless, ROADM scenario, the ROADM itself provides no filtration, whereas a colored ROADM does. That's why we add the tunable optical filter to the module—to perform that filtration, clean up the spectrum, and help overall system performance. 

Kevin Affolter: Yes, absolutely. We have two manufacturing locations for these modules: Wuxi, China, and Ipoh, Malaysia. Products can come from either location. 

Kevin Affolter: Yes, correct. It makes sense for us, especially with the current tariff situation and even prior to it, to have geographical diversity in our supply chain. 

Kevin Affolter: Yes. Considering the two applications we discussed: in the case of a 10G DWDM point-to-point network where capacity is exhausted, how do you add more? You either light up another fiber—meaning laying or leasing one, which is expensive—or you use the existing fiber, amplifiers, muxes, and demuxes, remove your 10G components, and install 100G. You've increased fiber capacity by a factor of 10. That's a huge improvement without needing to lay new fiber. This is the first example of an incredibly cost-effective way to upgrade capacity on an existing network. The other is the IP over DWDM example. A customer has a router at each end of the network with gray optics, and then transponder or muxponder solutions carrying the DWDM traffic. By eliminating those transponders, muxponders, and the interconnecting short-reach optics, they save a significant amount of money. Central office footprint costs money, as does the electricity for extra equipment. Removing these elements means fewer network components, improving network reliability. These multiple benefits all have a direct financial impact on the service provider. 

John Lynch: Yes, Ciara. We're already working with numerous customers across Europe on these parts. Typically, we provide them with proof-of-concept samples, allowing them to test the parts in their desired platforms. We assist with any necessary technical support, occasionally modifying settings if required. Ultimately, it's a try-and-buy system. Customers can test these parts, and thus far, everyone who has tested them has proceeded to purchase and place orders. The reaction has been very positive, and our goal is to get as many parts into customers' hands as possible so they can try them and form their own opinions. We have stock available at our locations in Dublin, the US, and China. We are happy to assist any customers worldwide who want to try these parts by lending them units for evaluation. If they are not satisfied, they can return them. For any customers wishing to test these parts, we are working closely with Coherent on these proof-of-concepts. If customers require input from Coherent's engineering teams, we are happy to collaborate side-by-side. We maintain an open-door policy, and if customers wish to speak directly with Coherent's application engineering team, we can facilitate that. All options are available to customers who would like to test these parts.