Focus on the crosspoint switch

Switches and specifically crosspoint switches are the building blocks for telecomm systems enabling them to offer high performance and flexibility. Crosspoint switches can handle multiple data streams, several clock sources and many protocols. Crosspoint switches guarantee inputs can find a uncongested path to an output and that the connection paths can be changed without affecting other paths. Crosspoint switches are commonly used as a splitter or a multiplexer in networking and communications infrastructure equipment to provide redundancy for data and clock signals.

If you need speed, Texas Instruments offers two new 4 x 4 non-blocking crosspoint switches, the SN65LVDS250 and SN65LVDT250, that operate at more than 2.0 gigabits per second (Gbits/s). These devices come in a flow-through pin-out, which allows for ease in printed circuit board (PCB) layout. They use low-voltage differential signaling (LVDS) outputs to achieve a high-speed data throughput while using low power. These switches are ideal for clock buffering and multiplexing or data transmission in datacom and telecom applications such as high-speed network routing and wireless base stations.

Each output driver of the SN65LVDS250 and SN65LVDT250 includes a 4:1 multiplexer to allow any input to be routed to any output. Internal signal paths are fully differential to achieve high signaling speeds while maintaining low signal skews. The SN65LVDT250 incorporates 110-ohm termination resistors for applications where board space is at a premium.

2.0 Gbits/s operation with the crosspoint switch from Texas Instruments

“Fairchild's new FSUSB11 switch complements the company's portfolio of USB and ultra-portable solutions, meeting design challenges in the consumer, industrial, medical and other fast-growth hand-held instrumentation markets,” said Gerald Johnston, Fairchild product manager for analog switches. The company's analog switch portfolio includes audio amplifiers, LED drivers and LEDs for backlighting, supervisory products such as temperature sensors and reset generator circuits, and a variety of DC/DC conversion products including LDOs and MOSFETs. In essence, they are good building blocks.

The FSUSB11 is a high performance dual single pole double throw (SPDT) analog switch specially designed for the switching of both analog audio signal and USB 1.1 signals. The device features low R(ON) of 1.3-ohms maximum at 4.5-V VCC and 4.3-ohms at 2.7-V supply. High bandwidth and ultra low On-resistance make this switch able to pass both USB low and full speed signal with minimum signal distortion.

“Network equipment providers are looking to develop unified low-cost, flexible platforms that can scale from enterprise to edge aggregation and core applications,” said Jay Cormier, product line director for high-speed switches at Analog Devices. “The ADSX34 allows multi-protocol switching, while solving the difficult analog issues of signal integrity and power. This enables OEMs to develop equipment on time, on budget, and with system flexibility,” he said.

The ADSX34 is a complete crosspoint solution that offers:
* 34 highly integrated channels operating up to 3.125 Gbits/s each.
* Per-channel programmable receive equalization and transmit pre-emphasis that allows equalization over 30 inches of FR4 material, including two, standard high-density differential connectors.
* Supports time slots from 24- to 4,000-characters.

* Per-channel time slot synchronization first-in first-out (FIFOs) absorb up to 128 bytes of variation in packet arrival times, simplifying system timing.

Many crosspoint switches are commonly used as a splitter or mux function in networking and communications infrastructure equipment to provide redundancy for data and clock signals. The National Semiconductor SCAN90CP02 with its two-input, two-output LVDS non-blocking crosspoint switch that handles signals up to 1.5 Gbits/s, is just right for communications applications. It supports many different configurations, including a conventional crosspoint switch, a buffer/repeater, a 1:2 splitter, and a 2:1 multiplexer. It is one of the first LVDS (low voltage differential signaling) devices to handle a 1.5 Gbits/s on a 2×2 crosspoint.

National Semiconductor's two-input two-output LVDS crosspoint switch

Applications and design issues
What are the biggest applications for crosspoint switches today, and what are the key design issues in these applications?

For TI the applications driving crosspoint switches include communications backplanes; fault-tolerant telecom, datacom and digital video systems. “The issues for fault-tolerant systems are redundancy for continuity of service in the event of failure in the primary path, and built-in self test requirements that alert service operators of system malfunctions,” said David Mulcahy, product marketing manager for TI. “Most customers require loop back redundancy,” he said.

Crosspoint switches fall into two major categories; analog switches and digital switches. For example, video switchers and router applications use both analog and digital switches. Analog Devices offers both types. “The biggest applications for crosspoint switches today are in digital applications for networking switch and router equipment,” said Will Drachler, Product Line Manager for Analog Devices.

“The design issues for networking switch and router equipment include having an attractive feature set,” said Mr. Drachler. “For example, offering a system that can handle multiple traffic types (MSPP-multi-service provisioning platform) such as Ethernet, ATM, and SONET. ADI developed a 34×34 synchronous crosspoint switch, the ADSX34,” he said. The ADSX34 has provisions to switch Ethernet packets and can switch fast enough to accommodate TDM traffic. These provisions are addressed through key features such as individual channel FIFO buffer and in-band switch addresses.

“Another challenge is to increase the aggregate data rate (throughput),” said Drachler. It is typically addressed through:
* Increasing port density of a given networking box and/or
*Increasing data rate of existing ports of a networking box

Each approach presents its own problems, according to Drachler. “Increasing the number of ports is difficult due to space and power constraints. Additionally, increasing the data rate can cause signal integrity headaches due to transmission line loss, crosstalk and reflections,” he added.

Analog Devices solutions provide attractive options for customers facing these design challenges. To address varying number of ports ADI offers an array of sizes from 2 x 2 thru 160 x 160, operating at data rates from dc to 10Gbits/s. “The increasing data rate on existing ports in these systems is typically the more popular option because of field installations for legacy systems,” said Drachler. To address this issue ADI develops input equalization and transmit pre-emphasis to undo the signal degradation that occurs through transmission. Obviously, the higher the data rate the worse the degradation. The company's equalization developments range in data rate from 1Gbits/s through 10Gbits/s. To address both the density and the signal integrity issues, ADI's ADSX34 (as an example) offers a dense 34 channels operating up to 3.2Gbits/s. Each of the 34 channels includes receive equalization, clock and data recovery, FIFO buffer, Serdes and transmit pre-emphasis.

Other design issues for networking switch and router equipment include:
Low Power: — Trying to fit more functionality/throughput in the same box and in the case of legacy systems trying to do these increases in the same footprint and with the same power supply is a real challenge. ADI's ADSX34 addresses this issue by providing 34 channels (108 Gbits/s throughput) with all previously mentioned functionality on with only 4-W total power and all 34 I/Os active. ADI says this is accomplished by its core competence of developing the signal path electronics on a custom, transistor-by-transistor basis utilizing only the necessary amount of power.
Low Cost: — System cost is reduced through the minimization of external components to increase the channel count and provide the necessary signal integrity.

System requirements for crosspoint switches
“A crosspoint by it's very nature is designed to save the system designer space and provide him flexibility,” said Sameer Vuyyuru, Intersil's Director of Marketing for video and amplifiers. “The issue for analog crosspoints tends to be density of channels, and that needs to be traded off against the thermal characteristics of the system and the routability of multiple channels into a compact package. Additionally, the thermal capabilities of today's BGA and QFN packages are excellent and are not an impeding factor,” added Vuyyuru.

Another system level consideration is the programming interface. It is critical for double latching to achieve synchronous operation in a system. The interface standard is typically I2C, SPI or any simple data/clock bus.

“Additionally, there is an interesting system issue because the normal implementation of these crosspoints consists of multiple instantiations on the same board to form even larger crosspoints,” said Mr. Vuyyuru. “The turn-on and turn-off times need to be very fast to accommodate these architectures. One must also be able to place the outputs of the crosspoint into a high impedance mode and allow for daisy chaining of the programming interface,” he added.

ADI's Will Drachler, says that the key system requirements that an IC manufacturer must meet when designing crosspoint switches include, architecture, data rates, and compensation for the transmission degradation. Typically, that means the IC manufacturer must know what the channel looks like, including the transmission line type, lengths and connector type. The IC maker also must have standards compliance (for XAUI and others) foremost in mind when designing the crosspoint switches.

For TI's David Mulcahy the biggest challenges for today's crosspoint switches include:
* Signaling levels — the input and output signaling levels need to be compatible with the source outputs and destination inputs. Low-voltage differential signaling (LVDS) is now capable of providing a low power consumption, low EMI alternative to more common CML and LVPECL interfaces. TI's crosspoints have a large common-mode range on the input which means they can accept LVDS, CML or LVPECL levels. TI outputs are either LVDS or LVPECL.

* Dynamic Switching Performance — After a change in configuration is commanded, you should know how quickly the change be carried out. TI's SN65LVDS250 4X4 LVDS crosspoint has eight control pins that are used to configure the non-blocking switch. The tSWITCH (Select to switch output) specification at 1.6-ns max is nearly one hundred times faster than other components that use serial control interfaces.

* Packaging — TI's devices come in TSSOP packages, which many customers find are more suitable for solder-flow manufacturing and allow easier pin access for development and in-operation test activities than other solutions (offered in leadless leadframe-type packaging).

* Power — the contribution of the crosspoint switch to the overall system budget becomes significant if multiple signal lines are assigned to a crosspoint switch. LVDS outputs with a simple 100-ohm termination scheme offer lower power consumption than LVPECL and CML, which require more complicated and more current-hungry schemes involving a pair of resistors connected to a termination voltage. TI offers 2 x 2 and 4 x 4 LVDS crosspoints with best-in-class power consumption ratings.
* Channeling cross-talk — Modern crosspoints have differential routing internal to the chip. This enables higher speeds and also reduces the device's susceptibility to EMI and reduces the radiated emissions, which can lead to channel-to-channel cross-talk.

Recent Switches
You might want to warm up your printer or get a pencil and paper out to have easy access to this diverse shopping list of some recent introductions from a few companies. For example, Intersil will soon be announcing the world's densest configuration of an analog crosspoint and has been tailored to handle RGB/YPbPr video. In addition the company has high-performance crosspoints in small configurations, such as the HA4314 and HA4344.

The HA4314B is a very wide bandwidth 4 x 1 crosspoint switch ideal for professional video switching, HDTV, computer monitor routing, and other high-performance applications. The circuit features very low power dissipation (105-mW enabled, 4-mW disabled), excellent differential gain and phase, and very high off isolation. When disabled, the output is switched to a high impedance state, making the HA4314B ideal for routing matrix equipment.

TI unleashed several crosspoint switches including the LVDS250, which the company says is the lowest power, high-speed 4 x 4 LVDS crosspoint. A version with integrated 100-ohm termination for LVDS signals, the SN65LVDT250, is also available. It is being used for interfacing a baseband controller to channel cards in a basestation. Its non-blocking architecture allows any input to be passed through to any ouput. Both devices operate at more than 2.0 Gbits/s and offer best-in-class low power. Their flow-through pin-out allows for ease in printed circuit board layout. LVDS outputs are used to achieve a high-speed data throughput while using low power. The devices are ideal for clock buffering and muxing or data transmission in datacom and telecom applications such as high-speed network routing and wireless base stations.

The TI LVDS122 is the lowest power, highest speed (1.5 Gbits/s) 2 x 2 device with LVDS output. The device is the industry's fastest, lowest-jitter 2 x 2 LVDS crosspoint switch that includes LVDS outputs and either LVDS or LVPECL inputs. Offering speeds at 1.5 Gbits/s, the LVDS122 is well suited for gigabit Ethernet, OC-12 clock distribution, and other applications where system availability is critical. The device can also operate as a dual differential repeater/translator and a 1:2 splitter.

The SN65LVCP22 and SN65LVP23 2 x 2 crosspoint switches support 2:2 buffering (repeating), 1:2 splitting, 2:1 multiplexing and 2 x 2 switching. In addition, the devices incorporate wide common-mode (0 volts to 4 volts) receivers, allowing for the receipt of LVDS, LVPECL and CML signals. The SN65LVCP22 (1 Gbits/s) supports LVPECL/CML to LVDS level translations and the SN65LVCP23 (1.3 Gbits/s) supports LVDS/CML to LVPECL level translations. To establish accurate alignment of outputs in all applications, the output channel-to-channel skew is less than 10 ps (typical) and 50 ps (maximum).

TI's 1Gbits/s 2 x 2 crosspoint switch

Analog Devices recently introduced the ADSX34 to help resolve several challenges that customers were having, especially in the optical market. Each port of the ADSX34 34-port synchronous crossbar switch can be programmed for either in-band or out-of-band modes. It accommodates dynamic and static switching on a per-port basis and supports both cell and circuit switching. Further, power consumption can be reduced by turning off unused ports or modules that aren't needed in an application. The ADSX34 has provisions to switch Ethernet packets and can switch fast enough to accommodate TDM traffic. These provisions are addressed through key features such as an individual channel FIFO buffer and in-band switch addresses.

Crossing over
What will future crosspoint switches need to provide that aren't supported now? Increasing amounts of functions, which support higher layers of protocols, is one that ADI's Drachler sees as necessary.

TI's Mulcahy said it's all about speed. With ever increasing speeds in applications, newer crosspoints will be challenged to cater to 3.125 Gbits/s, especially for XAUI the 10-Gb attachment unit interface. In order to be able to transmit data at these high rates, crosspoints will have to use signal conditioning methods involving pre-emphasis on outputs and receiver equalization on the inputs. Both methods serve to counteract the inherent attention on the transmission media, be that backplane traces or cables.

Crosspoint switching may be the building blocks for several markets including telecommunications but manufacturers must continue to confer with their customers to meet their needs whether it is speed or added functionality. Regardless, it's an exciting time for crosspoint switches.

Company Information

Analog Devices, Inc.
Tel: 800-ANALOGD (262-5643)

Fairchild Semiconductor
Tel: 207-775-8100

Intersil Corp.
Tel: 321-724-7000

National Semiconductor Corp.
Tel: 800-272-9959

Texas Instruments Inc.
Tel: 800-477-8924, ext. 4500

0 comments on “Focus on the crosspoint switch

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.