[LINK] VDSL/FTTN alternative to NBN-GPON - variation in temperature and reliability

[Nick Ross]

Hi Robin,

Did anyone follow up on this? It's a very interesting point. I've asked the
question of some FTTN cabinet specialists.

I've heard enough about datacenters to know that some flavours of the newer
ones are designed to run very hot indeed - the electronics can cope with it.

However, I asked about cooling i cabinets before and was told that they
have several fans in them. But that's surely not going to cut it in an
Australian summer? Or any perma-hot parts of Australia???

What other hot countries have FTTN (at least in their hot parts)?

N





On 11 April 2013 14:26, Robin Whittle <rw at firstpr.com.au> wrote:

> Further to what I wrote in the thread: "No batteries required in new
> broadband plan", placing a lot of electronics (VDSL transceivers, 10GBps
> or greater WDM (Wavelength Division Multiplexing) fibre optoelectronics
> and switches between them) in street cabinets, with backup batteries, is
> a really bad idea.
>
> In addition to being bulky, flood-prone and subject to being hit by
> cars, the cabinets and their electronic equipment will be subject to
> daily temperature cycling and being frequently extremely hot due to high
> atmospheric temperatures and the impact of the Sun on the cabinet.  To
> have refrigerated cooling systems in these boxes would make them
> extremely expensive in terms of power, and would make them even noisier
> than if they were cooled by ambient air with fans.
>
> A quick look at:
>
>   "Temperature Management in Data Centres: Why Some (Might) Like it
>    Hot"  Nosayba El-Sayed, Ioan Stefanovici, George Amvrosiadis, Andy
>    Hwang, Bianca Schroeder
>
>    ACM special interest group for the computer systems performance
>    evaluation (SIGMETRICS/Performance) 2012
>
>    http://www.cs.toronto.edu/~nosayba/temperature_cam.pdf
>
> indicates that temperature cycling is a problem, at least for disk
> drives in large data centres, where the ambient air is often cooled to
> the low 20s, and where I assume that humidity would be carefully
> controlled too.  (There would be no disk drives in an FTTN node.)
>
>   Observation 2: The variability in temperature tends to
>   have a more pronounced and consistent effect on latent sector
>   error rates than mere average temperature.
>
> These are variations within a large data centre - not the more extreme
> variations, every day, which would occur in a street cabinet.
>
> Electronic devices are now full of ball-grid array packages, where the
> balls are tin-based solder, which is more brittle than ordinary tin-lead
> solder.  BGA solder joints can't be visually inspected or repaired in
> any reasonable manner.
>
>   http://www.google.com/search?q=%22ball+grid+array%22&tbm=isch
>
> BGA devices themselves typically get hot due to their own power
> dissipation.  Other electronic devices such as regulators and power
> transistors dissipate more heat, and while they don't use BGA solder
> joints, they still use (typically, though not necessarily in Australia)
> lead-free tin-based solder which is more subject to fracture than
> lead-based solder.
>
> High temperatures, uncontrolled humidity variations, temperature cycling
> will all contribute to failures, including intermittent faults.  The
> cabinets might work OK for a few years, but who wants to have their
> Internet and phone services dependent on 10 or 15 year old electronics
> sitting out in a street cabinet?  Dust, insects, plant material etc.
> might also be problems.
>
>   - Robin
>
> _______________________________________________
> Link mailing list
> Link at mailman.anu.edu.au
> http://mailman.anu.edu.au/mailman/listinfo/link
>