On 9/27/2020 2:37 PM, Grant Taylor wrote:
> On 9/27/20 2:05 PM, Don Y wrote:

>> I note that industrial kit implicitly relies on the "user" being properly 
>> educated and/or the devices kept out of "mainstream" access.
> 
> Preventing incidental access is very important.

In an industrial/commercial/medical environment, you can probably count on some
minimum level of competence around the kit.  That's not necessarily the case in
a consumer environment (or SOHO).

> You can't prevent determined people.  But you can make it obvious that they 
> went out of their way to defeat safety measures meant to prevent incidental 
> access.
> 
>> And, I don't see any practical way that any additional safety mechanisms 
>> could be put in place ("presence detect" to inhibit emitter in the absence of 
>> a plugged cable, etc.)
> 
> Some of the optical amps (a.k.a. pumps) that coworkers work on can easily cause 
> severe harm if the beam ever gets loose.  Be that an improperly connected 
> connector or break in fiber somewhere.

But, you would assume (in that environment) competence to take SOME safeguards
against this happening.  I.e., you'd be protecting against the TRUE "accident"
and not outright carelessness/ignorance.

> My understanding is that some of this high power equipment has detectors that 
> require it to receive light from the remote end before it will electrically 
> engage the amp.  The premise being that there must be low power (of an 
> acceptable variety) received from the remote end before engaging the full 
> transmission power.  With the idea being that the only way to receive that 
> proper power is if the two devices are properly connected.  Thus if the fiber 
> does break somewhere mid-span, each end will stop receiving the requisite from 
> the other end and shut down their transmitters.

Yes, that's what I was suggesting, below.

>> As I'm not constrained to using existing implementations, I might be able to 
>> enhance safety in my application with non-standard design choices.  (e.g., 
>> presence detect for BOTH ends of the intended cable followed by a brief, 
>> tentative probe of that expected connection before accepting it as 
>> "intact"/safe)
> 
> That's why the transmitter must receive something through the medium to enable 
> full power.
> 
> Simply detecting the physical presence at one end is not sufficient.

My thought is that detecting the presence of a specific connector (I'm not
wedded to using COTS "standards") in a specific place could signify to the
local node that it APPEARS that a cable is in place.

That node could query the intended remote node for verification of the
similar condition on that far end.

Once both nodes THINK they are connected by a cable, they can attempt to
pass a "test signal" down the cable for verification.

Because they would only try this when they suspected a cable to be plugged
AT BOTH ENDS, *and* would know exactly when that trial signal was being
delivered, it need only make "one" attempt to verify the connection -- instead
of sitting in a loop persistently retrying while awaiting "success".

As such connections aren't intended to be transient in nature, a fault can
require some specific methodology for reestablishing the link; the nodes
need not "self heal" (cuz you don't want to risk them getting into a
loop continually probing for a cable that obviously WASN'T there!)

On 9/27/20 2:05 PM, Don Y wrote:
> And, of course, inquisitive users peering into transmitters or cable 
> ends and risking damage to their eyes.

Darwin awards exist for a reason.

> [I'm canvasing the warnings manufacturers of audio kit that support 
> "optical" ports but suspect that is much lower power.]

I expect that those light emitters (LASERs or LEDs) will be very low 
power.  Probably Class I (?), much like CD players.  Even if you can get 
to the light source, it probably won't do permanent damage.

> I note that industrial kit implicitly relies on the "user" being 
> properly educated and/or the devices kept out of "mainstream" access.

Preventing incidental access is very important.

You can't prevent determined people.  But you can make it obvious that 
they went out of their way to defeat safety measures meant to prevent 
incidental access.

> And, I don't see any practical way that any additional safety 
> mechanisms could be put in place ("presence detect" to inhibit emitter 
> in the absence of a plugged cable, etc.)

Some of the optical amps (a.k.a. pumps) that coworkers work on can 
easily cause severe harm if the beam ever gets loose.  Be that an 
improperly connected connector or break in fiber somewhere.

My understanding is that some of this high power equipment has detectors 
that require it to receive light from the remote end before it will 
electrically engage the amp.  The premise being that there must be low 
power (of an acceptable variety) received from the remote end before 
engaging the full transmission power.  With the idea being that the only 
way to receive that proper power is if the two devices are properly 
connected.  Thus if the fiber does break somewhere mid-span, each end 
will stop receiving the requisite from the other end and shut down their 
transmitters.

> As I'm not constrained to using existing implementations, I might 
> be able to enhance safety in my application with non-standard design 
> choices.  (e.g., presence detect for BOTH ends of the intended cable 
> followed by a brief, tentative probe of that expected connection 
> before accepting it as "intact"/safe)

That's why the transmitter must receive something through the medium to 
enable full power.

Simply detecting the physical presence at one end is not sufficient.

-- 
Grant. . . .
unix || die

On Sun, 27 Sep 2020 11:39:56 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:

>I've got to assume that there is nothing physically that can be done
>to enhance the safety of optical fiber deployment and that you have to
>rely on education and "professionals" aware of current best practices
>in its use and maintenance.
>
>Or, are there certain tricks that can reduce the liability?

Far too little information is provided to allow us to answer the
question.

What is the purpose of the optical fiber system?  Communications,
metal cutting, or collapsing deturium-tritium pellets to cause nuclear
fusion?

How long is the longest cable?

What is the operating wavelength and power level?

Are the "users" the general population, engineers, or Physics PhDs?

Is there enough physical security to keep random people out?

And so on.

Joe Gwinn

On 9/27/2020 12:17 PM, Grant Taylor wrote:
> On 9/27/20 12:39 PM, Don Y wrote:
>> I've got to assume that there is nothing physically that can be done to 
>> enhance the safety of optical fiber deployment and that you have to rely on 
>> education and "professionals" aware of current best practices in its use and 
>> maintenance.
>>
>> Or, are there certain tricks that can reduce the liability?
> 
> Please elaborate on what type of safety issues you're talking about.
> 
> There are all sorts of things in dealing with optical fibers.  Some specific to 
> optical fibers, and others quite common.
> 
>   - fiber shards puncturing the skin
>   - chemical burns
>   - thermal burns from things like fusion splicing
>   - optical burns from high power lasers
>   - falling from heights
>   - confined spaces
>   - drowning
>   - trip hazard
>   - crush hazard

I dismiss as "effectively equivalent"  any issues that are associated with
the process of installing/routing cable as it would apply equally to copper
or fiber.

I also rule out "unauthorized (successful!) repair" due to the general lack
of appropriate tools and know how for folks not educated (and equipped) to
do so.

But, that doesn't mean there won't be ATTEMPTED (naive) repairs from
"accidents" that sever a "cable" or break a connector body.

And, of course, inquisitive users peering into transmitters or cable ends
and risking damage to their eyes.

[I'm canvasing the warnings manufacturers of audio kit that support "optical"
ports but suspect that is much lower power.]

I expect cables to be "factory supplied" and any on-site modifications to
be performed by knowledgeable professionals -- not wannabe electricians!

I note that industrial kit implicitly relies on the "user" being properly
educated and/or the devices kept out of "mainstream" access.  And, I
don't see any practical way that any additional safety mechanisms could
be put in place ("presence detect" to inhibit emitter in the absence of a
plugged cable, etc.)

As I'm not constrained to using existing implementations, I might be
able to enhance safety in my application with non-standard design choices.
(e.g., presence detect for BOTH ends of the intended cable followed by
a brief, tentative probe of that expected connection before accepting
it as "intact"/safe)

On 9/27/20 12:39 PM, Don Y wrote:
> I've got to assume that there is nothing physically that can be done 
> to enhance the safety of optical fiber deployment and that you have to 
> rely on education and "professionals" aware of current best practices 
> in its use and maintenance.
> 
> Or, are there certain tricks that can reduce the liability?

Please elaborate on what type of safety issues you're talking about.

There are all sorts of things in dealing with optical fibers.  Some 
specific to optical fibers, and others quite common.

  - fiber shards puncturing the skin
  - chemical burns
  - thermal burns from things like fusion splicing
  - optical burns from high power lasers
  - falling from heights
  - confined spaces
  - drowning
  - trip hazard
  - crush hazard



-- 
Grant. . . .
unix || die