400G Direct Attach Copper (DAC) cables provide high-speed electrical connections between switches, servers, network adapters, and other data center equipment. Unlike optical cables, which convert electrical signals into light, DAC cables transmit electrical signals directly through copper conductors.
The electrical architecture of a typical 400G QSFP-DD or OSFP DAC is based on eight high-speed differential lanes operating with PAM4 signaling. Each lane carries approximately 50Gbps of nominal data, providing an aggregate 400G Ethernet connection.
Inside the cable assembly, multiple shielded twinaxial copper pairs connect the transmit and receive interfaces at both ends. The host devices provide the SerDes functions, signal generation, equalization, and data recovery required for reliable communication.
Understanding this architecture helps explain how 400G DAC cables achieve high bandwidth, why their transmission distance is limited, and how cable construction affects signal integrity.
1. Overview of the 400G DAC Electrical Architecture
A conventional 400G QSFP-DD passive DAC uses an eight-lane electrical architecture commonly associated with the 400GAUI-8 interface. Each direction has eight differential signal lanes, allowing both devices to transmit and receive data simultaneously.
Each differential lane uses two conductors: a positive signal (P) and a negative signal (N). The receiver detects the voltage difference between these conductors rather than relying on a single-ended signal.
The main architectural elements include:
Host SerDes: Converts parallel data into high-speed serial electrical signals and recovers incoming data.
Transmit differential pairs: Carry outgoing electrical signals from one device to the other.
Receive differential pairs: Carry signals in the opposite direction.
Twinaxial copper conductors: Provide the physical transmission medium.
Connectors and PCBs: Connect the cable conductors to the electrical interfaces of the host equipment.
Shielding and grounding: Help control electromagnetic interference and maintain signal integrity.
1.1 Simplified 400G DAC Signal Path
Device A Device B +----------------------+ +----------------------+ | Ethernet MAC / PCS | | Ethernet MAC / PCS | | | | | ^ | | SerDes | | SerDes | | | | | ^ | | TX Differential | | RX Differential | +----------|-----------+ +----------^-----------+ | | | 8 High-Speed TX/RX Lane Pairs | +===========================================+ Shielded Twinax Cable +===========================================+ | | +----------v-----------+ +----------|-----------+ | RX Differential | | TX Differential | | | | | | | | SerDes | | SerDes | | | | | v | | Ethernet MAC / PCS | | Ethernet MAC / PCS | +----------------------+ +----------------------+
The diagram represents a bidirectional connection. Device A's transmit lanes connect to Device B's receive lanes, while Device B's transmit lanes connect to Device A's receive lanes.
In a passive DAC, the cable itself does not perform optical conversion, clock and data recovery, or active signal regeneration. The host devices handle the main signal-processing functions.
2. Eight-Lane PAM4 Signaling
One of the defining characteristics of conventional 400G QSFP-DD DAC architecture is its eight-lane electrical interface. Each lane uses PAM4 (Pulse Amplitude Modulation with four levels) to transmit two bits per symbol.
2.1 How 8 × 50G PAM4 Works
In a common 400GAUI-8 implementation, the electrical interface uses eight lanes, each operating at approximately 26.5625 GBaud.
Because PAM4 represents two bits per symbol, each lane carries approximately 53.125Gbps of nominal line data rate.
| Parameter | Typical 400G QSFP-DD Architecture |
|---|---|
| Aggregate Ethernet rate | 400Gbps |
| Electrical lane count | 8 lanes per direction |
| Modulation format | PAM4 |
| Symbol rate per lane | 26.5625 GBaud |
| Nominal line rate per lane | 53.125Gbps |
| Aggregate nominal lane rate | 425Gbps |
| Electrical signal type | Differential signaling |
| Transmission medium | Shielded twinaxial copper |
The 425Gbps aggregate nominal lane rate accounts for the physical signaling and coding overhead associated with the 400G Ethernet implementation. The commercial interface is still identified as 400GbE.
2.2 Why Does 400G Use PAM4 Instead of NRZ?
NRZ (Non-Return-to-Zero) signaling represents two possible voltage levels and carries one bit per symbol. PAM4 uses four signal levels and carries two bits per symbol.
At the same symbol rate, PAM4 can transmit twice as many bits as NRZ. This makes it possible to achieve higher bandwidth without simply doubling the symbol rate.
| Feature | NRZ | PAM4 |
|---|---|---|
| Signal levels | 2 | 4 |
| Bits per symbol | 1 | 2 |
| Data capacity at the same symbol rate | Lower | Twice that of NRZ |
| Signal-level separation | Larger | Smaller |
| Signal integrity requirements | Less demanding in comparable conditions | More demanding |
However, PAM4 has smaller voltage separation between adjacent signal levels. This makes it more sensitive to noise, distortion, crosstalk, and other electrical impairments. High-quality cable construction and host equalization are therefore essential.
3. Differential Signaling and Twinaxial Copper Pairs
The electrical signals in a 400G DAC travel through differential copper pairs. Each pair contains two conductors carrying complementary signal waveforms.
The receiver measures the voltage difference between the two conductors. This approach helps reject common-mode noise and improves signal transmission in electrically noisy data center environments.
3.1 Differential Signal Transmission
For each lane, the positive and negative conductors carry complementary signals. The receiver detects the difference between them rather than measuring either conductor independently.
This is particularly important at high data rates because even small amounts of noise, impedance discontinuity, or signal distortion can affect the receiver's ability to distinguish PAM4 signal levels.
3.2 Twinaxial Cable Construction
A 400G DAC typically uses multiple shielded twinaxial conductors inside a common cable jacket. Each twinaxial pair is designed to maintain controlled impedance and reduce unwanted coupling between adjacent signal paths.
A representative cable construction may include:
Copper conductors: Carry the high-speed differential signals.
Individual insulation: Separates the two conductors within each pair.
Pair shielding: Helps reduce electromagnetic coupling between signal pairs.
Overall shielding: Provides additional protection against external interference.
Drain wire: Supports the grounding and shielding arrangement in applicable cable designs.
Outer jacket: Provides mechanical protection and insulation.
The exact layer arrangement varies by cable manufacturer and conductor design. Some products use combinations of aluminum foil, braided shielding, insulation, and protective polymer materials.
3.3 Differential Impedance
Many high-speed DAC designs use a nominal differential impedance of 100 ohms. Maintaining consistent impedance across the cable, PCB traces, and connector transitions is important for minimizing reflections.
Impedance discontinuities can cause part of the transmitted signal to reflect toward the source. These reflections may distort the received waveform and reduce the available signal margin.
4. How Are the Electrical Lanes Connected?
A typical 400G QSFP-DD passive DAC contains eight transmit differential pairs and eight receive differential pairs. This provides eight electrical lanes in each direction for full-duplex communication.
The internal wiring connects the transmit interface at one end to the receive interface at the opposite end. The reverse connection is made for the other direction.
| Signal Group | Quantity | Function |
|---|---|---|
| TX differential pairs | 8 pairs | Transmit electrical data from one end |
| RX differential pairs | 8 pairs | Receive electrical data at the same end |
| Ground and shield connections | Design-dependent | Provide grounding and shielding continuity |
| Management connections | Design-dependent | Support cable identification and monitoring |
The total of 16 differential signal pairs represents the two transmission directions across the cable assembly. It should not be confused with 16 independent 400G channels.
4.1 QSFP-DD Connector Interface
QSFP-DD uses a double-density electrical connector architecture that provides eight high-speed electrical lanes in each direction. The cable ends contain connector PCBs and contact interfaces that route the differential signals between the twinax conductors and the host ports.
Connector PCB design is important because the signal must pass through several transitions between the cable conductors, PCB traces, and connector contacts.
These transitions must maintain controlled impedance and minimize insertion loss, return loss, and crosstalk.
4.2 Cable Management and EEPROM
Many DAC assemblies include an EEPROM or cable identification memory in the connector housing. The host can read information such as cable identity, vendor details, and supported operating characteristics through a management interface such as I2C.
This management function is separate from the high-speed data path. An EEPROM does not make a passive DAC an active signal-conditioning cable.
5. What Happens to the Signal Inside a Passive 400G DAC?
In a passive 400G DAC, the electrical signal travels directly from the transmitter in one host device to the receiver in the other host device.
The cable does not contain a retimer or redriver in its high-speed signal path. Instead, the connected equipment must provide the equalization and signal-processing capability required by the electrical channel.
5.1 Transmitter Operation
The host SerDes converts parallel data into high-speed serial electrical signals. The transmitter applies the required signal levels and launches the PAM4 waveform into the cable's differential copper pairs.
5.2 Signal Transmission Through Copper
As the signal travels through the twinax cable, it experiences insertion loss, reflections, crosstalk, and frequency-dependent attenuation.
The cable design attempts to minimize these impairments through controlled impedance, low-loss materials, shielding, and carefully designed connector transitions.
5.3 Receiver Operation
At the other end, the host receiver detects the incoming differential signal. Its equalization and signal-processing circuitry compensate for channel impairments and recover the transmitted data.
Both host devices must support the required electrical interface and signal-integrity conditions for the link to operate reliably.
5.4 FEC and Error Correction
Forward Error Correction (FEC) is an important part of modern high-speed Ethernet systems. It adds redundancy to the transmitted data so that the receiver can detect and correct certain transmission errors.
For a conventional passive 400G DAC, FEC processing is performed by the host platform rather than by the passive cable itself. Cisco's 400G QSFP-DD passive DAC documentation explicitly identifies host-side FEC processing.
FEC improves the reliability of the complete link, but it does not replace the need for proper cable design and electrical signal integrity.
6. Electrical Architecture of Passive DAC vs Active DAC vs AEC
Although passive DAC, active DAC, and Active Electrical Cable (AEC) all use copper conductors, their internal signal-processing architectures are different.
| Architecture | Signal Path | Internal Electronics | Primary Function |
|---|---|---|---|
| Passive DAC | Direct electrical connection | No active signal-conditioning IC in the high-speed path | Transmit signals through copper |
| Active DAC / ACC | Electrical connection with signal conditioning | Typically redrivers or linear equalizers | Compensate for signal loss |
| AEC | Retimed electrical connection | Retimers with CDR | Recover clock and data and regenerate signals |
In a passive DAC, the host SerDes must handle the electrical channel. In active copper designs, some signal-conditioning functions are moved into the cable assembly.
A redriver primarily applies analog equalization and signal conditioning. A retimer performs clock and data recovery and retransmits regenerated data. These functions should not be treated as equivalent.
7. Signal Integrity Challenges in 400G DAC Cables
Maintaining signal integrity is one of the most important challenges in 400G copper connectivity. The combination of high lane rates and PAM4 signaling leaves less margin for electrical distortion.
7.1 Insertion Loss
Insertion loss describes the reduction in signal amplitude as the signal travels through the cable and connectors. It generally increases with cable length and frequency.
At 400G lane rates, excessive insertion loss can make it difficult for the receiver to distinguish the four PAM4 signal levels.
7.2 Return Loss
Return loss describes the amount of signal reflected toward the source due to impedance mismatches. Poor connector transitions, PCB design, and cable impedance variations can increase reflections.
7.3 Crosstalk
Crosstalk occurs when signals in one differential pair interfere with neighboring pairs. Effective shielding, pair separation, and connector design help control unwanted coupling.
7.4 Skew
Skew refers to timing differences between signals. Excessive skew can affect signal sampling and reduce the margin available to the receiver.
Both intra-pair skew and lane-to-lane skew matter in high-speed multi-lane connections. Cable manufacturing and assembly processes must control these timing differences within the relevant interface requirements.
8. How Does the Electrical Architecture Affect 400G DAC Reach?
The maximum reach of a 400G DAC depends on the electrical performance of the complete channel, including the cable, connectors, PCB traces, and host SerDes.
As cable length increases, insertion loss and other signal impairments become more difficult to compensate for. This limits the distance over which a passive cable can maintain the required signal integrity.
For example, Cisco specifies passive 400G QSFP-DD DAC lengths of 0.5, 1, 2, 2.5, and 3 meters for its QDD-400-CUxM product family.
These are product-specific supported lengths, not universal maximums for every 400G DAC. Different host platforms, cable constructions, and electrical channel budgets can produce different results.
When a longer copper connection is required, an active copper solution may be considered. For connections beyond practical copper reach, an AOC or separate optical transceiver and fiber cable may be more appropriate.
9. 400G QSFP-DD vs OSFP DAC Architecture
QSFP-DD and OSFP are two major form factors used in high-speed data center networking. Both support eight-lane 400G electrical architectures in common 400G implementations.
| Feature | 400G QSFP-DD DAC | 400G OSFP DAC |
|---|---|---|
| Form factor | QSFP Double Density | Octal Small Form Factor Pluggable |
| Common electrical lane configuration | 8 × 50G PAM4 | 8 × 50G PAM4 |
| Signal type | Differential electrical | Differential electrical |
| Transmission medium | Twinaxial copper | Twinaxial copper |
| Connector design | QSFP-DD MSA interface | OSFP MSA interface |
| Typical application | 400G switches and servers | 400G data center and HPC platforms |
Although the two form factors can use similar lane signaling, their mechanical dimensions, thermal characteristics, connector interfaces, and host compatibility requirements differ.
A QSFP-DD DAC should not be assumed to fit or operate in an OSFP port. The cable connector must match the device interface or use a specifically supported breakout configuration.
10. How to Select a 400G DAC Cable
Selecting a 400G DAC requires checking more than the aggregate data rate. The cable must meet the electrical, mechanical, and compatibility requirements of the connected devices.
10.1 Confirm the Form Factor
Identify whether the switch and server ports use QSFP-DD or OSFP. Check the exact connector configuration at both ends of the cable.
10.2 Verify the Electrical Interface
Confirm the supported lane configuration and electrical interface. A common 400G QSFP-DD DAC uses eight 50G PAM4 electrical lanes, but other 400G electrical implementations exist.
10.3 Select the Appropriate Cable Length
Measure the actual cable route and choose a supported length. Avoid assuming that every 400G DAC can operate at the same maximum distance.
10.4 Check Host Compatibility
Verify that both connected devices support the cable vendor, coding, electrical characteristics, and operating requirements. Matching connector form factors alone does not guarantee compatibility.
10.5 Review Breakout Requirements
Some 400G DAC products support breakout connections, such as 400G to 4 × 100G. Confirm that the cable wiring, connector configuration, and host port support the intended breakout mode.
11. Conclusion
The electrical architecture of a conventional 400G DAC cable is based on high-speed differential signaling over shielded twinaxial copper conductors. A common QSFP-DD implementation uses eight 50G PAM4 electrical lanes in each direction to provide a 400G Ethernet connection.
Each lane uses a differential pair, and the cable assembly connects the transmit and receive interfaces between the two host devices. The cable itself provides the electrical transmission path, while the host SerDes performs signal transmission, equalization, and data recovery.
Controlled impedance, shielding, connector design, and low-loss cable construction are essential to maintaining signal integrity. Because passive DAC does not contain retimers or redrivers in its high-speed signal path, its supported reach depends heavily on the complete electrical channel and host capabilities.
For data center engineers, understanding these architectural details is important when selecting 400G DAC cables for switches, servers, and high-performance computing systems.
Frequently Asked Questions
Q1: How many electrical lanes does a 400G DAC cable use?
A common 400G QSFP-DD DAC uses eight electrical lanes in each direction, with each lane carrying approximately 50Gbps of nominal data.
Q2: Does a 400G DAC use PAM4 or NRZ?
Common 400G QSFP-DD DAC implementations use PAM4 signaling, which carries two bits per symbol and supports approximately 50Gbps per electrical lane.
Q3: What is the lane rate of a 400G DAC?
A typical 400GAUI-8 implementation operates at approximately 26.5625 GBaud per lane, corresponding to a nominal line rate of 53.125Gbps per lane.
Q4: How many differential pairs are in a 400G DAC?
A typical eight-lane full-duplex design has eight transmit and eight receive differential pairs, for a total of 16 high-speed signal pairs across both directions.
Q5: What type of copper cable is used in a 400G DAC?
400G DAC cables typically use shielded twinaxial copper conductors designed for high-speed differential signaling and controlled impedance.
Q6: What is the differential impedance of a 400G DAC?
Many 400G DAC designs use a nominal differential impedance of 100 ohms. The exact tolerance and measurement requirements depend on the applicable interface specification.
Q7: Does a passive 400G DAC contain a DSP?
No. A conventional passive DAC does not contain a DSP or retimer in its high-speed signal path. The host equipment handles the required signal processing.
Q8: Does a 400G DAC perform FEC?
FEC is generally performed by the host platform in a conventional passive 400G DAC connection. The passive cable itself does not perform error correction.
Q9: What is the difference between 400GAUI-8 and 400GBASE-CR8?
400GAUI-8 describes an eight-lane electrical attachment interface, while 400GBASE-CR8 identifies a 400G copper physical-layer application. Their specifications address related electrical link requirements but are not interchangeable terms.
Q10: Why is the reach of a 400G DAC limited?
High-speed copper transmission is affected by insertion loss, reflections, crosstalk, and signal distortion. These impairments increase with cable length and must remain within the supported electrical channel limits.
TEL:+86 132 6656 7067




















































>
>
>
>
>
>
>
>