Interconnect Solutions Blog

Sometimes there’s just not enough space to run a standard category 6 cable within a cabinet, enclosure, drawer or other device.

Enter flat modular cords from Siemon.

With Flat Category 6 Compatible UTP Modular Cords from Siemon, designers can stack network cables together allowing multiple cables to run side by side for many applications:

• inside cabinet connectivity
• board to board connections inside enclosures or drawers
• along the side wall of a cabinet

Flat Category 6 Compatible UTP Modular Cords from Siemon

Siemon’s flat UTP modular cords are only 0.090″ (2.34mm) thick and take 50% less volume than standard Category 6 cable, yet the plug geometry is compliant to IEC- 60603-7 specifications.

The snag-less boot features over mold strain relief.  The RoHS compliant flat mod cords are compatible with both T568A and T568B wiring schemes and are designed to terminate 28 AWG conductors.

Learn more about Flat Category 6 Compatible UTP Modular Cords from Siemon and contact Siemon Interconnect Solutions for assistance.

For OEMs in need of a 110D connecting block solution, I wanted to make you aware of a Cost Savings Analysis available from Siemon Interconnect Solutions that shows the time and materials cost reductions from Siemon’s unique EON (eye of the needle) press fit 110D style connecting block design.

Siemon’s solderless press fit connectors reduce costs by eliminating the need to solder the 110D connecting blocks to a PCB (printed circuit board) and the processing time associated with the soldering process.

Customers have the option to terminate either 22-26AWG (0.64mm-0.40mm) solid or 7-strand conductors using IDC (insulation displacement contact) technology.

Siemon’s Interconnect Solutions business unit (SIS) is a team of experts committed to solving industry and customer-driven interconnect challenges. SIS specializes in the development of copper and fiber high-speed interconnects for the communications, data center, medical, automotive and industrial markets, providing value-added solutions across a variety of markets and customer applications. With over 400 patents, Siemon invests heavily in R&D and development of industry standards, underlining the company’s long-term commitment to its customers and the industry.

Siemon, a leading global manufacturer of IT network cabling and infrastructure systems, announces the release of a new QSFP+ (Quad Small Form-Factor Pluggable Plus) to SFP+ (Small Form-Factor Pluggable Plus) hybrid passive copper cable assembly. These hybrid fanout cables break a single 40Gb/s QSFP+ connector into four individual 10Gb/s SFP+ connectors, allowing users to connect multiple ports of SFP+ equipment to QSFP+ based equipment with a single, easy-to-manage assembly.

Read Press Release: Siemon Announces 40Gb/s Hybrid QSFP+ to SFP+ Fanout Cable Assembly for High-Speed Data Center Interconnects

As with organic life forms, the nodes and links of the world-wide web seem to have a varying rhythmic process of differentiation and then integration. At certain inflection points in the process, one can see an intended integration effort cause some differential effects which in turn meld together after another natural cycle. More than any other IO interface, Ethernet has expanded well beyond the original LAN section of the web that it has dominate for many years since it overcame the  rival Token Ring and VG AnyLAN interfaces.

(more…)

Siemon, a leading global manufacturer of IT network cabling and infrastructure systems, announces the addition of a thumbscrew fastener option for their 40+Gb/s SFF-8470 4X passive copper cabling assemblies. The thumbscrew provides added physical security to these high-speed low-latency interconnects, eliminating the possibility of accidental disconnections in SAS storage, ruggedized and military data center applications.

Read Press Release: Siemon Announces Secure Connectivity Options for 40+Gb/s High-Speed Data Center Interconnects

For point to point connections that run between blades, boxes and racks, cable diameter has been an issue relative to airflow, PoE bundle heat build up, routing flexibility and weight. As a response, 30awg cabling has been preferred and users have gotten use to this dominate gage. Even 32awg cabling has been used in significant volume to address these issues.

As data rates have increased, we have seen a large growth of active copper cabling usage in order to keep the 30awg preference and maintain needed cable lengths and topologies. We have also seen the fast rise of active optical cabling to deal with data rate performance versus link lengths.

Now we are developing 24-28Gb/s per lane links in various standards bodies like Infiniband. Based on feedback from DataCenter engineers and managers, the new Infiniband EDR spec will not have 24awg in the specification, the maximum passive copper gage is likely 26awg. Some members are saying it should just have 30-28awg. At this data rate and gage size range, passive copper will only reach about .5-2 meters depending on PCB trace versus cable length trade offs.

Even using active copper at this data rate range, the wire gage and length story is not so great as it was in the old days of 2.5Gb/s per lane. So active optical cabling has become a lot more important in achieving optimized cable diameter and link lengths. Many OEMs and active equipment users do not like having so many media types and are strategizing to keep the link types and lengths options simplified.

Ultimately we seem to be heading towards using the smallest media size option, single mode fiber for many and most applications at any length and data rate. This seems like a major inflection point in the history of interconnection media types; just my opinion….

PS. In the old Fibre Channel days we had to use 21-22awg copper to achieve link lengths, but there was room to spare. Back then most of Fibre Channel links were copper which was embarrassing because this technology called itself Fibre Channel. Now there is no copper media being specified in the latest 28/32Gb/s specification so finally Fibre Channel is all about fiber. Got fiber in your diet?

Traditionally, IO interface roadmaps have stayed their own predictable path and stayed separate from each other for several generations and many years. After the converged Camelot of many 10G interfaces and links, many IO roadmaps are radically changing data rate progression and are often converging and competing with several other IO communities’ next generation standards.

The InfiniBand Trade Association formally introduced their new IO interface roadmap just a few weeks ago and it is quite different from their last revision. The older one had progressed predictability uniform from 2.5G to 5G to 10G data rates which have been implemented thru standards, plugfests and products. But now the next logical data rates of 20G and 40G per pair have disappeared off the new roadmap revision.

14G per pair is now the next generation of InfiniBand and this association is planning for it and their concurrently new 25/28G per pair generation to be both released in 2011. They use the acronyms, HDR and NDR without any date rate number as the two undefined placeholders in their roadmap’s fuzzy “Market Demand 2014” space. I never saw a high speed interface coming out with two data rates platforms at the same time and having no defined next generation data rates or dates going ahead. It seems they are trying to stay a little ahead of Ethernet’s progression, but ready to converge opportunistically.

InfiniBand also recently expanded its architecture by have 1x lane and 8x lane added to the already 4x and 12x lane options as PCI Express already has. PCI-Express also has 16x and 32x lane options.

FibreChannel Industry Association and the T11 standards group are just about to release their 16G SFP+ single lane interface and are starting to develop a new 32G optical single lane interface The leading FibreChannel 16G spec has been partially borrowed by the new InfiniBand FDR 14G standard group and allegedly, the newer PCI-E 3.0 16G standards group. The FCIA is working on a new roadmap revision due to come out soon and is alleged to go beyond 64G, 100G as single links and aggregated to a 1 Terabit SAN link. Looks like no more Hard Disk Drives at FC 32G, no more Hard Disk Controller Blades and no more Copper SFP cables, just optical SAN switch links and software per FCIA.

FibreChannel 16G and SAS 12G interfaces may be merged into one chip and will likely be used with converged next generation FC/SAS SSDs, (Solid State Drives) which are mounted on blades or modules. FibreChannel and Ethernet 10G interfaces are sometimes merged into one chip with two SFP+ ports and support converged LAN/SAN systems. So one wonders if there will be Unified SFP+ 16G and or SFP+ 28G link next generation Unified systems.

OIF group is leading and co-working thru liaisons on 25/28/32G per lane converged electrical and optical interfaces with FCIA, IBTA and IEEE802.3 groups. OIF is especially focused on a four lane QSFP++ proposal. This supports 4x25G IB server and 4x 28G FC SAN and Ethernet switch links. OIF also has been working on a 40G per lane specification. It is alleged that if PCI-E still exists a couple of years from now and is not replaced by the very new LightPeak interface, it would likely use the OIF/FC 28/32G electrical spec partially for a nascent PCI-E 5.0

Ethernet IEEE-802.3bg 40G optical single lane link specification development is moving right along and is allegedly looking to expand beyond the long reach applications. It appears that the Call of Interest for a new 4x25G=100 electrical spec and QSFP+ copper and active optical PMD will occur at the November 2010 plenary meeting. Ethernet Alliance business leaders are planning to talk about a new roadmap at the October meeting that may include 400G and 1T interfaces. So much for the Ethernet older orderly by ten data rate progress that did 10Mbs, 100Mbs, 1000Mbs 10Gbs …

SAS Trade Association and NCITS-T10 storage community are working on their new 4x12G spec for 2012 release. Some early work is being done on their subsequent 24G per lane spec. The 6-12G HD-MiniSAS connector does not work at 24G. So there is talk about borrowing the soon to be chosen QSFP++25/28G connector for inclusion in the developing SAS 3.0 spec as some OEMs seem to have a preference for using it for 12-24G designed ports in very large storage systems, especially for petascale and exascale DataCenters.

CameraLink2, RapidIO,  Myrinet and many other IO standards are still following with a year or two lag using FC/IB/Ethernet specifications and media options especially QSFP+ AOC.

Although HDMI 3.0, USB 4.0 (25G?), DisplayPort 3.0 standards are said to be developing, the looming Lightpeak 1.0 and 2.0 optical specifications/products may come sooner and supplant those three IOs and maybe CameraLink, PCI-E and FireWire 1394. It would be amazing to see Commodity IO 25G interfaces taking over the lead from High End IO 25G interfaces in a 2013 convergence/collision. Maybe this would be facilitated thru the acceptance of high volume commodity Intel Atom and various ARM microprocessors building blocks being used potentially in some of the largest categories of DataCenter systems that are lowest power and highest density focused.

Has the Great Recession had an impact on acceleration of standards and IO interface convergence? Combined standards/technology resources, fuzzy roadmaps, non-interoperable Unified architectures and lots of co-opetition may be thought as side effects of this recession.

As on the waterways, highways and flyways of the world, there are many more speed options, topologies, lanes and links than ever before. Connecting electronic equipment with these new IO interfaces is supporting the exploding internet universe that is being driven by virtualization and video and LAN/SAN convergence factors.

There is an expression, “let the chips fall where they may”. I guess I’m thinking in that metaphor because I am heading to another electronic equipment conference in Las Vegas. I am hoping to learn more about the future of Active Copper and Active Optical networking cable assemblies.

For many years, having active chips embedded inside electronic connectors and cable assemblies was a niche application solving unique electrical distance issues. Fast forward to 2010 and it appears that half of the SFP+ interconnect volume are Active Copper and Active Optical cables. QSFP+ and other next generation cables seem to have near future unit volume forecasting of 50% Active Optical versus Copper as a whole.

The current variety of embedded chip usage includes signal conditioners, EEPROMs, repeaters and optical/electrical converters. Next generation higher speed active embedded chip possibilities seem to be expanding due to solving electrical PCB trace distance degradation. Some developers are exploring whether to take retimer and four lanes times 10 lanes gearbox chips off the blade or box boards and put them on the usually higher performance embedded PCB cable plug. Others are looking at moving even the PHY controller chip on to the cable plug PCB. O/E chip developers are looking at bringing other primary functions within their new chips.

So for a few years we may see a boom of Very Smart Cabling. But is this active cabling complexity and cost increase going to be acceptable further in the future or are all these functions going inboard as two or one chip instead. Will optical performance and interconnects be necessary for as high as 75% of networking cabling shipped in 2013-2014? Will all optical systems including optical PHY chips be necessary by 2015 or sooner? If the optical interconnect becomes pervasive in the near future, then most cables may use some type of MPO photonic connectors that will provide further miniaturization, densification and increased port counts on blades and boxes compare to today’s connectors and cables.

I guess seeing where the chips may fall this week in Las Vegas may likely be different from where they might land at the same location in 2012 and especially 2014. So watch these chips carefully and be ready for more change.


High Speed Cable Link proprietary encryption handshake issues are often adversely affecting the interoperability of a wide variety of Ethernet, InfiniBand and FibreChannel OEM products being installed in various DataCenters. Many cables using PCB-based plugs are embedded with EEPROMs that have codes for cable assembly performance level, assembly length, manufacturing identification and OEM product interoperability limiter. The SFP+ EEPROM specification, SFF-8472, allows for standard coding and private coding fields. (more…)