Ethernet Transceivers Beyond 100G

A quick guide to Ethernet Transceivers Beyond 100G

In our previous blog, we discussed how 100G Single Lambda technology paved the way for higher speed transceivers. However, with the rapid rise in demand for 200G, 400G and 800G interface speeds, the market has seen several technological solutions emerge, often leading to confusion and compatibility concerns.

So, what are the available options for getting more signals through a transceiver while still maintaining backward compatibility with the QSFP format?

Increase the capacity of the host side electrical interface.

QSFP-DD doubles the number of electrical contacts on the QSFP host-side connector while maintaining compatibility with the existing QSFP28 format.

Below is the electrical pinout of a 100G QSFP28, featuring a total of 8 lanes (Tx + Rx) using 25G NRZ signalling:

Examples of transceivers using this format include the GQS-MPO101-SR4C 100m short-range device from Gigalight and the EOLQ-851HG-02 from Eoptolink.

By making the module slightly longer, an additional row of connector pads can be added to both the top and the bottom of the PCB, providing an extra 4 x Tx and 4 x Rx lanes. This brings the total to 8 x 25G lanes in each direction for a 200Gbps link.

 

Examples of 200G QSFP-DD transceivers include the GQD-MP0201-DSR4C from Gigalight and the EOLD-852HG-E-02-X from Eoptolink.

As signalling technologies advance and lane speeds increase, new form factors such as the QSFP56 and QSFP112 and their corresponding double-density versions, have been introduced. This enables transceivers with data rates of up 800G.

 

Increase the capacity of the optical side interface.

Having seen how the electrical interface can be increased in density from 4 to 8 lanes in each direction through double-density host-side connectors, we now look at how this translates into connectivity on the optical side.

The example parts given so far, all use MPT/MPO connectors where each signal channel is carried as 25Gbps NRZ on a separate fibre.

The block diagrams for 100G and 200G transceivers from Gigalight are shown below:

100G QSFP28 and 200G QSFP-DD

GQS-MPO101-SR4C 100G QSFP28-SR4

GQD-MPO201-DSR4C 2 x 100G QSFP-DD-SR8

 

The MTP/MPO 12 fibre and 24 fibre connectors are used, and at 800G the 16 fibre, with the option for dual 12 fibre, is available, with the following optical pinout.

Note:

• The transceiver MTP/MPO connectors are male with built-in guide pins.
• Type B MPO trunk cables are the most common way to connect.
• Angled Physical Contact (APC) polished fibre is typically used above 40G (refer to the datasheet to confirm).
• Single-mode MTP/MPO cables are used in conjunction with Parallel Single Mode (PSM) transceivers for longer-reach applications. For example: GDM-SPO201-FR8C 200G-DD-PSM8

 

Whilst MTP/MPO cabling is well-suited for data centre rack-to-rack interconnects, dual LC fibres are more widely available and are often more cost-effective over longer distances. As a result, the optical multiplexing standards used at lower speeds – such as CWDM and LWDM – continue to be applied at data rates from 100G up to 800G. These multiplexing techniques enable high-speed Ethernet transmission over distances ranging from 2km to 10km and beyond.

When specifying a transceiver, it remains essential to understand which optical multiplexing method is being used in the link. For more details see our previous blog Understanding 100G QSFP28 Optics.

 

Increase capacity through use of higher-level coding

So far, we have considered transceivers using 25G NRZ signalling. However, as discussed in our article on 100G Single Lambda devices, PAM4 encoding enables twice the data rate compared to NRZ. This advancement has allowed DSP-based transceivers to double speeds without requiring additional channels.

Here’s a quick summary:

NRZ (Non-Return-to-Zero) and PAM4 (Pulse Amplitude Modulation with 4 levels) are two different digital signalling methods used in high-speed data transmission. NRZ uses two voltage levels — one for ‘0’ and one for ‘1’ — transmitting one bit per symbol. In contrast, PAM4 employs four voltage levels, enabling the transmission of two bits per symbol. This effectively doubles the data rate without increasing the symbol rate.

200G QSFP56 and 400G QSFP-DD

In practical terms, this means you now have the option of transceivers using a QSFP56 electrical interface with 4 x 50G PAM4 channels for 200G, such as the Gigalight GQS-MPO201-SR4C.

Or the 400G 8 x 50G PAM4 device  GQD-MPO401-SR8C

400G QSFP112 and 800G QSFP-DD

As mentioned earlier, 100G Single Lambda transceivers use 100G PAM4 signaling.  When combined with the capacity-increasing methods outlined above, this paves the way to 800 Gbps and beyond. To support these higher lane speeds, a new interface is required — QSFP112 — for 400G transceivers such as the GQS-MPO401-SR4C

 

and the double density version for 800G GQS-MPO801-SR8C

 

 

Conclusion

The demand for transceivers with ever-faster capability has led to the development of multiple technologies, resulting in a wide variety of electrical, optical and coding techniques. It’s no surprise, therefore, that users can find it challenging to determine which type of transceiver ensures compatibility across their network.

When selecting a transceiver, consider the following key questions:

• Speed: 100G / 200G / 400G / 800G
• Electrical Interface: QSFP28 / QSFP-DD / QSFP56 / QSFP112
• Optical Connector: MTP/MPO or Dual LC / multi-mode or single mode / CWDM or LWDM
• Coding scheme: 25G NRZ / 50G PAM4 / 100G PAM4