Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > sci.electronics.design > #458612 > unrolled thread

Patch-panel fabrication

Back to article view | Back to sci.electronics.design

Contents

  Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-03 12:54 -0700
    Re: Patch-panel fabrication Lasse Langwadt Christensen <langwadt@fonz.dk> - 2017-05-03 13:23 -0700
      Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-04 00:10 -0700
        Re: Patch-panel fabrication Lasse Langwadt Christensen <langwadt@fonz.dk> - 2017-05-04 11:27 -0700
          Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-04 11:43 -0700
          Re: Patch-panel fabrication Clifford Heath <no.spam@please.net> - 2017-05-05 08:57 +1000
            Re: Patch-panel fabrication Lasse Langwadt Christensen <langwadt@fonz.dk> - 2017-05-06 05:47 -0700
          Re: Patch-panel fabrication krw@notreal.com - 2017-05-04 19:37 -0400
        Re: Patch-panel fabrication Neon John <no@never.com> - 2017-05-04 17:56 -0400
          Re: Patch-panel fabrication Lasse Langwadt Christensen <langwadt@fonz.dk> - 2017-05-06 05:46 -0700
        Re: Patch-panel fabrication Jasen Betts <jasen@xnet.co.nz> - 2017-05-11 05:57 +0000
          Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-11 10:37 -0700
    Re: Patch-panel fabrication Martin Riddle <martin_ridd@verizon.net> - 2017-05-03 20:32 -0400
      Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-03 23:59 -0700
    Re: Patch-panel fabrication Jasen Betts <jasen@xnet.co.nz> - 2017-05-04 06:34 +0000
      Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-04 01:45 -0700
    Re: Patch-panel fabrication whit3rd <whit3rd@gmail.com> - 2017-05-04 01:52 -0700
      Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-04 11:42 -0700
        Re: Patch-panel fabrication whit3rd <whit3rd@gmail.com> - 2017-05-04 21:22 -0700
          Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-04 23:04 -0700
            Re: Patch-panel fabrication whit3rd <whit3rd@gmail.com> - 2017-05-04 23:24 -0700
              Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-05 00:19 -0700
                Re: Patch-panel fabrication whit3rd <whit3rd@gmail.com> - 2017-05-05 00:45 -0700
                  Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-05 02:17 -0700
    Re: Patch-panel fabrication jrwalliker@gmail.com - 2017-05-04 08:40 -0700
      Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-04 10:12 -0700
        Re: Patch-panel fabrication jrwalliker@gmail.com - 2017-05-04 10:27 -0700
          Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-04 11:01 -0700
            Re: Patch-panel fabrication makolber@yahoo.com - 2017-05-04 11:18 -0700
              Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-04 11:27 -0700
        Re: Patch-panel fabrication Don Kuenz <g@crcomp.net> - 2017-05-06 05:33 +0000
          Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-06 15:40 -0700
            Re: Patch-panel fabrication whit3rd <whit3rd@gmail.com> - 2017-05-06 20:15 -0700
              Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-06 23:47 -0700
                Re: Patch-panel fabrication whit3rd <whit3rd@gmail.com> - 2017-05-07 03:51 -0700
                  Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-07 05:20 -0700
            Re: Patch-panel fabrication (POE) Jan Panteltje <pNaOnStPeAlMtje@yahoo.com> - 2017-05-07 17:30 +0000
              Re: Patch-panel fabrication (POE) Don Y <blockedofcourse@foo.invalid> - 2017-05-07 11:49 -0700
                Re: Patch-panel fabrication (POE) Jan Panteltje <pNaOnStPeAlMtje@yahoo.com> - 2017-05-07 18:58 +0000
                  Re: Patch-panel fabrication (POE) Don Y <blockedofcourse@foo.invalid> - 2017-05-07 12:24 -0700
                    Re: Patch-panel fabrication (POE) Jan Panteltje <pNaOnStPeAlMtje@yahoo.com> - 2017-05-07 19:48 +0000
                      Re: Patch-panel fabrication (POE) Don Y <blockedofcourse@foo.invalid> - 2017-05-07 13:29 -0700
                        Re: Patch-panel fabrication (POE) Jan Panteltje <pNaOnStPeAlMtje@yahoo.com> - 2017-05-07 20:54 +0000
    Re: Patch-panel fabrication Allan Herriman <allanherriman@hotmail.com> - 2017-05-07 10:35 +0000
      Re: Patch-panel fabrication Don Y <blockedofcourse@foo.invalid> - 2017-05-07 05:26 -0700

Page 3 of 3 — ← Prev page 1 2 [3]

#459277 — Re: Patch-panel fabrication (POE)

On a sunny day (Sun, 7 May 2017 12:24:25 -0700) it happened Don Y
<blockedofcourse@foo.invalid> wrote in <oens2e$reg$1@dont-email.me>:

>On 5/7/2017 11:58 AM, Jan Panteltje wrote:
>> On a sunny day (Sun, 7 May 2017 11:49:53 -0700) it happened Don Y
>> <blockedofcourse@foo.invalid> wrote in <oenq1n$kb3$1@dont-email.me>:
>>
>>>> So it could cause damage to somebodies stuff if they plug it in and has that connector or a similar one.
>>>> I am replacing all those connectors by HLJ6115ANL, also because it has 7,8 and 4,5 accessible and I need that to get the
>>>> POE.
>>>>  http://www.ebay.com/itm/271891560407
>>>> But that is not something you can do with just any laptop... you plug in.
>>>
>>> Hmmm... are you suggesting the problem is OTHER devices that might get plugged
>>> into my network?
>>
>> Exactly.
>
>There's a hardware handshake that transpires between the PSE and a
>potential PD.  The PSE looks for characteristic resistances on the signal
>leads and, based on what it senses, determines what class of PD device
>is being plugged.  If it senses a noncompliant device, it doesn't
>apply power to the conductors.


I think it is problematic unless there is some question - response, say 'what are you',
'I am an 48 V device', OK, power on.
Just from the load (test current at some voltage, say resistance check) it would be very hard,
so many possibilities.


>This is how you prevent things like (analog!) telephones misplugged into PoE
>jacks from wreaking havoc.


I am worried about house guest finding an ethernet socket and plugging in their 2K$ laptop,
hey even my own.
You can check though, with an ohm meter, 7,8 to 4.5 if it is infinite it is probably OK.

 
>> Yes, OK, I use these selectively to places where I am sure there is only my own equipment connected:
>>  http://www.ebay.com/itm/151643358383
>> with my own power supply of course.
>
>Yes, Cisco has a little device that impresses 48V on the unused pairs
>for their (older) AP's.  But, it's literally just an adapter from
>48VDC wall wart "barrel connector" to midspan injector.  Plug one
>of the devices you're worrying about, above, into it and it
>will gladly deliver 48V @ 380mA to the "load".

Yes that is what this thing is, but price is good.

[toc] | [prev] | [next] | [standalone]

#459290 — Re: Patch-panel fabrication (POE)

On 5/7/2017 12:48 PM, Jan Panteltje wrote:
> On a sunny day (Sun, 7 May 2017 12:24:25 -0700) it happened Don Y
> <blockedofcourse@foo.invalid> wrote in <oens2e$reg$1@dont-email.me>:
>
>> On 5/7/2017 11:58 AM, Jan Panteltje wrote:
>>> On a sunny day (Sun, 7 May 2017 11:49:53 -0700) it happened Don Y
>>> <blockedofcourse@foo.invalid> wrote in <oenq1n$kb3$1@dont-email.me>:
>>>
>>>>> So it could cause damage to somebodies stuff if they plug it in and has that connector or a similar one.
>>>>> I am replacing all those connectors by HLJ6115ANL, also because it has 7,8 and 4,5 accessible and I need that to get the
>>>>> POE.
>>>>>  http://www.ebay.com/itm/271891560407
>>>>> But that is not something you can do with just any laptop... you plug in.
>>>>
>>>> Hmmm... are you suggesting the problem is OTHER devices that might get plugged
>>>> into my network?
>>>
>>> Exactly.
>>
>> There's a hardware handshake that transpires between the PSE and a
>> potential PD.  The PSE looks for characteristic resistances on the signal
>> leads and, based on what it senses, determines what class of PD device
>> is being plugged.  If it senses a noncompliant device, it doesn't
>> apply power to the conductors.
>
> I think it is problematic unless there is some question - response, say 'what are you',
> 'I am an 48 V device', OK, power on.
> Just from the load (test current at some voltage, say resistance check) it would be very hard,
> so many possibilities.

No, it's not a "static" process.

The PSE applies a low voltage (~3 then 10V) and checks to see if there is a
nominal ~25K resistance detected in the PD.

If so, the PSE (typically) applies another "classification" voltage (~18V) and
verifies that the PD represents a particularly sized load (250-4000uA for the
default class 0; 9-12mA for class 1; 17-20mA for class 2 and 26-30mA for
class 3 -- beyond that, you're looking at PoE+, etc.).

[The classes determine how much power the PSE will allow the PD to consume
before signalling an overload.  This allows the PSE to budget its power
availability (cuz it often has less than enough power to supply ALL ports
at the maximum limit)]

At any time, if the PD consumes more power than allowed, the PSE disconnects
the device -- usually within 100ms of the fault level being exceeded.

If the PSE senses the device becoming unplugged (or, its current drain
dropping below a ~5mA keep-alive level), it will also disconnect the
device.

How the system handles the power and data is up to the system; do you
deliberately stop talking to a device when it stops drawing power?  etc.

>> This is how you prevent things like (analog!) telephones misplugged into PoE
>> jacks from wreaking havoc.
>
> I am worried about house guest finding an ethernet socket and plugging in their 2K$ laptop,
> hey even my own.
> You can check though, with an ohm meter, 7,8 to 4.5 if it is infinite it is probably OK.

Then your power sourcing equipment (the midspan injector you mentioned) has to
have the smarts in it to realize that the attaching device (AS it is plugged
in) is not behaving like a "genuine PD" *should* behave.  And, limit the
impact you have on that device to the "detection voltages" that are
applied at the start of the process.

Likewise, when you sense <whatever> being unplugged, you have to reset the
process so you can replace a PD with a non-PD, then the non-PD with a PD,
etc.

The good thing is that the PD (and PSE) controller chips handle most of the
analog processing AND the signalling protocols.  So, you slap one in your
design and then figure out how to coerce it to behaving as you'd like.

>>> Yes, OK, I use these selectively to places where I am sure there is only my own equipment connected:
>>>  http://www.ebay.com/itm/151643358383
>>> with my own power supply of course.
>>
>> Yes, Cisco has a little device that impresses 48V on the unused pairs
>> for their (older) AP's.  But, it's literally just an adapter from
>> 48VDC wall wart "barrel connector" to midspan injector.  Plug one
>> of the devices you're worrying about, above, into it and it
>> will gladly deliver 48V @ 380mA to the "load".
>
> Yes that is what this thing is, but price is good.

The Cisco device is considerably larger -- about the size of half a pack
of (American) 100mm cigarettes -- and only handles one device.   It
relies on the inherent current limiting of the wall wart AND the user's
best practices to protect the downstream device.

You can, no doubt, design a little midspan injector that is more
compliant.  But, at a fair bit more cost (for low quantities).
I suspect there are also fully compliant midspan injectors (as
well as endspan switches) available OTS.  I just can't budget
for the number I'd need!

As I needed a "custom" switch, it made sense to design the PSE facility
in addition to the PD devices.  Hence my comment in my previous reply
stating that I exploit the components that are designed for this "standard"
way of doing things -- but apply them with my own "twists".

[toc] | [prev] | [next] | [standalone]

#459296 — Re: Patch-panel fabrication (POE)

On a sunny day (Sun, 7 May 2017 13:29:15 -0700) it happened Don Y
<blockedofcourse@foo.invalid> wrote in <oenvrv$8g9$1@dont-email.me>:

>The PSE applies a low voltage (~3 then 10V) and checks to see if there is a
>nominal ~25K resistance detected in the PD.
>
>If so, the PSE (typically) applies another "classification" voltage (~18V) and
>verifies that the PD represents a particularly sized load (250-4000uA for the
>default class 0; 9-12mA for class 1; 17-20mA for class 2 and 26-30mA for
>class 3 -- beyond that, you're looking at PoE+, etc.).
>
>[The classes determine how much power the PSE will allow the PD to consume
>before signalling an overload.  This allows the PSE to budget its power
>availability (cuz it often has less than enough power to supply ALL ports
>at the maximum limit)]

OK.


>At any time, if the PD consumes more power than allowed, the PSE disconnects
>the device -- usually within 100ms of the fault level being exceeded.
>
>If the PSE senses the device becoming unplugged (or, its current drain
>dropping below a ~5mA keep-alive level), it will also disconnect the
>device.
>
>How the system handles the power and data is up to the system; do you
>deliberately stop talking to a device when it stops drawing power?  etc.
>
>>> This is how you prevent things like (analog!) telephones misplugged into PoE
>>> jacks from wreaking havoc.
>>
>> I am worried about house guest finding an ethernet socket and plugging in their 2K$ laptop,
>> hey even my own.
>> You can check though, with an ohm meter, 7,8 to 4.5 if it is infinite it is probably OK.
>
>Then your power sourcing equipment (the midspan injector you mentioned) has to
>have the smarts in it to realize that the attaching device (AS it is plugged
>in) is not behaving like a "genuine PD" *should* behave.  And, limit the
>impact you have on that device to the "detection voltages" that are
>applied at the start of the process.
>
>Likewise, when you sense <whatever> being unplugged, you have to reset the
>process so you can replace a PD with a non-PD, then the non-PD with a PD,
>etc.
>
>The good thing is that the PD (and PSE) controller chips handle most of the
>analog processing AND the signalling protocols.  So, you slap one in your
>design and then figure out how to coerce it to behaving as you'd like.
>
>>>> Yes, OK, I use these selectively to places where I am sure there is only my own equipment connected:
>>>>  http://www.ebay.com/itm/151643358383
>>>> with my own power supply of course.
>>>
>>> Yes, Cisco has a little device that impresses 48V on the unused pairs
>>> for their (older) AP's.  But, it's literally just an adapter from
>>> 48VDC wall wart "barrel connector" to midspan injector.  Plug one
>>> of the devices you're worrying about, above, into it and it
>>> will gladly deliver 48V @ 380mA to the "load".
>>
>> Yes that is what this thing is, but price is good.
>
>The Cisco device is considerably larger -- about the size of half a pack
>of (American) 100mm cigarettes -- and only handles one device.   It
>relies on the inherent current limiting of the wall wart AND the user's
>best practices to protect the downstream device.
>
>You can, no doubt, design a little midspan injector that is more
>compliant.  But, at a fair bit more cost (for low quantities).
>I suspect there are also fully compliant midspan injectors (as
>well as endspan switches) available OTS.  I just can't budget
>for the number I'd need!
>
>As I needed a "custom" switch, it made sense to design the PSE facility
>in addition to the PD devices.  Hence my comment in my previous reply
>stating that I exploit the components that are designed for this "standard"
>way of doing things -- but apply them with my own "twists".

Understood.

[toc] | [prev] | [next] | [standalone]

#459196

On Wed, 03 May 2017 12:54:49 -0700, Don Y wrote:

> I have 96 drops that need to terminate on a pair of 48 connector panels.
>  The panels aren't supported in a rack but, rather, as plates in/on the
> wall.
> 
> There are constraints as to their size and shape.  And, most
> importantly, cosmetics (this is a house, not a business!  I want these
> things to be as unobtrusive/classy as possible).
> 
> I need to start cleaning much of this stuff up for "public display"
> (tangles of wire tend to be intimidating to most folks -- can see the
> forest for the trees!)
> 
> I could *possibly* take a pair of 96 port 4U panels
> <https://www.cablewholesale.com/products/network-phone/patch-panel/
product-69bk-06096.php>
> and butcher them -- cut them in half to yield a 48 port "left side" and
> a 48 port "right side".  But, there's so little "extra panel" left over
> when this is done that you end up needing to augment it, just to get it
> mounted!  (e.g., you've lost the mounting ears on one side of each
> half).
> 
> Fashioning a "half panel" from heavy gauge steel is tedious -- esp if
> you want it to look nice (don't want irregularly shaped holes for the
> connector bodies).
> 
> The "easy" way of making something "cleanly/precisely cut" would be to
> do so on the laser cutter.  But, that limits you to acrylic and thin
> sheets of hardwood (60W CO2).  Neither of which is particularly stiff
> when those sorts of long slotted openings are introduced.
> 
> With "limited/personal resources", I see the only practical approaches
> to be adding layers -- for strength (behind) and appearance (before).
> 
> For example, cutting a "front panel" that provides for mounting (e.g.,
> ears, oversized) and then smaller panels (undersized)
> to stack behind this to provide more stiffness.  Finally, an overlay
> (depending on how "pretty" the front panel piece might be) to carry the
> legend and dress it all up.
> 
> Alternately, trimming the existing steel panel to provide the stiffness
> function and adding a mounting flange and cosmetic overlays in front of
> it.  I'd have to think hard about the durability of the cosmetic overlay
> as it will see a bit of abuse (plugging and unplugging, wiping it clean,
> etc.) over the years and I'd like not to have it looking shabby from
> age/wear.
> 
> (If you've ever examined one of these panels, the "circuit boards" on
> which the connectors are mounted are typically offset behind the panel
> with standoffs.  So, there's a bit of empty space (in Z) to play with,
> there, if you tweek the standoff lengths.)
> 
> Any other suggestions?  I'd like to make a dry run cutting/dry-fitting a
> sample, this weekend to see how well the idea (whichever!) will work.


Large scale patch panel in Kyoto, Japan:
http://imgur.com/HwC59CC

Allan

[toc] | [prev] | [next] | [standalone]

#459206

On 5/7/2017 3:35 AM, Allan Herriman wrote:
> Large scale patch panel in Kyoto, Japan:
> http://imgur.com/HwC59CC

ROTFLMAO!  They need jewel indicators (backlit stained glass?) on each
side of the porches to complete the image!

[toc] | [prev] | [standalone]

Page 3 of 3 — ← Prev page 1 2 [3]

Back to top | Article view | sci.electronics.design

csiph-web