Determine Polarity of AUX 53v DC input jack on Cisco Ethernet Switch?

[kyle]

Hello,

tl;dr: I'm trying to figure out polarity of a weird 53v DC input jack. I want to make my own DC adapter instead of paying ~$80 for one from Cisco.

I have acquired a Cisco switch that I'm trying to use to learn Cisco networking and commands. The switch that I have is model 3560CG and it doesn't come with a PSU, instead it is meant to be power via PoE (power-over-ethernet) or an AUX port on the back of the unit can be used to power it up. I do have a PoE adapter but it is not sending enough power to the Switch, it only sends 15.4w instead of 25w (I debugged the issue down to adapter being 'passive' and being able to negotiate with the Switch to agree on feeding 25w [this is part of how PoE protocol works])

I'm looking to get 48v 2A DC adapter for $15, I plan to solder my own plug. Here's how this AUX input jack looks:




I'm not sure how to figure out which pin is positive and which negative. Notice that one of the holes has slanted edge? Any idea which is which?

[helius]

The power supply in a telecom rack is normally -48 V (negative). When an isolated supply is used, like most linear and switching power bricks, that doesn't matter too much.
If the polarity isn't marked outside the case, I would suggest opening it and checking for polarized capacitors on the input traces. The polarity of the capacitors will tell you the polarity needed at the input.

[Silverio]

I trick I've used in the past to find the polarity in situation like this was to measure continuity between the case of the instrument and the DC jack. Usually the metal case and the negative pole are connected. The pin that has continuity with the metal case will be the negative and the other positive

[kyle]

Thank you, those are 2 solid tips! I will try testing if one of the pins is grounded to the chassis, and if not I will disassemble and look at polarity of filtering capacitors near input jack.

[tooki]

Thank you, those are 2 solid tips! I will try testing if one of the pins is grounded to the chassis, and if not I will disassemble and look at polarity of filtering capacitors near input jack.

Given that it could require negative voltage with respect to ground (in which case ground would be the positive lead) I would look at the capacitors inside, since that leaves no uncertainty.

[stevelup]

Cisco redundant power supplies are -52V with respect to ground.

As suggested above, I would suggest you take this device apart and look at nearby capacitors rather than relying on testing for continuity to the chassis!

[mikeselectricstuff]

Open it up - should be obvious.
Alternatively, it's quite likely there is a reverse polarity protection diode, so another quick test is to apply the supply through a current-limiting resistor that's high enough  that smoke won't happen if there isn't ( I;d try 10K). If you find that no current is drawn in one direction, that's the wrong direction.
 

[CJay]

Why not just use the 48V supply you're getting with a POE injector?

http://www.instructables.com/id/Power-Over-Ethernet-PoE-Adapter/

[kyle]

Thanks guys!

None of the pins were grounded to the chassis. I opened up the Switch and found +/- terminals. I tested negative side of nearby capacitor and found that it is connected with pin on the input jack that has 'slanted' edge.

I tested the Switch with my bench PSU at 48v and it works!. Relevant log line from serial connection:

Code: [Select]

Mar 30 01:28:33.354: %PLATFORM_ENV-6-PWR_OPTIMAL: Switch is fully operational. Current power source is Auxilliary
@CJay,
I did try PoE injector (Ubiquiti GP-H480-050G), see 1st post, but it appears that it is a passive device and would not negotiate proper wattage with Cisco Switch (the injector does not support 802.3at or 802.3af standards perhaps?). The switch would power up but it would be some sort of limp mode where none of my downlink Ethernet ports worked:

Code: [Select]

Mar 30 01:28:33.579: %PLATFORM_ENV-1-PWR_LOPWR: Switch is operating in low power mode, switch is not fully operational.

[dj831]

Hi, just my 2 cents on that old topic, in order to help others, just in case. It seems that ports down (or not connected) is an issue experienced by some. I've asked same question months ago on Cisco forum, but no one had an answer. And Cisco bug toolkit was to no avail

So, Cisco CPD series like this one (WS-C3560CPD-8PT-S) are mainly intended to be powered from uplink, i.e. Gi0/9 or Gi0/10, but they can be also powered from a power brick. Here's what gives a 'sh env all' from a working 3560CPD powered from interface Gi0/10 (output is a bit self-explanatory );

(...)

PoE Power - Available:0.0(w)  Backup:0.0(w)

Power Source   Type           Power(w)  Mode
-------------- -------------- --------- ---------
Gi0/10         Type2          30(w)     Available

Available : The PoE received on this link is used for powering this switch and
            providing PoE pass-through if applicable.
Back-up   : In the absence of 'Available' power mode, the PoE received on this
            link is used for powering this switch and providing PoE pass-through
            if applicable.
Available*: The PoE received on this link is used for powering this switch but
            does not contribute to the PoE pass-through.
Back-up*  : In the absence of 'Available' power mode, the PoE received on this
            link is used for powering this switch but does not contribute to
            the PoE pass-through.

NWX03#

These devices are VERY sensitive to power used. Of the two devices I own, only one worked with the power supply of a WLC2504 (same pinout, same supply, more wattage), and none worked with same power supply from eBay China (supply that works though on the same WLC2504). You got the message:

*Jan  2 00:00:39.436: %PLATFORM_ENV-1-PWR_LOPWR: Switch is operating in low power mode, switch is not fully operational. Current power source is Gi0/10-Type1
*Jan  2 00:00:39.442: %PDPSE-5-PWR_SRC_DISC: Power source change from   to Gi0/10. Current power source is Gi0/10

And all interfaces are NOT connected. But when you get it works, you have these messages:

*Jan  2 00:00:55.880: %PLATFORM_ENV-6-PWR_OPTIMAL: Switch is fully operational. Current power source is Gi0/10-Type2
*Jan  2 00:00:55.889: %CDP_PD-4-POWER_OK: 30 W power - NEGOTIATED inline power source on port Gi0/10

And then, interfaces are up:

*Jan  2 00:00:59.627: %LINK-3-UPDOWN: Interface GigabitEthernet0/2, changed state to up
*Jan  2 00:00:59.669: %LINK-3-UPDOWN: Interface GigabitEthernet0/1, changed state to up
*Jan  2 00:00:59.674: %LINK-3-UPDOWN: Interface GigabitEthernet0/7, changed state to up

When powered from PoE, it ***may*** works from a Cisco device, but you must have CDP enabled to get it works - I enabled CDPv2, a timer of 15s (I think this is what solved the issue but I may be wrong) and 'power inline port 2x-mode' on interface of 3560CG PoE that powers it to get it works. I've tested with a reload of the device to confirm.

@Kyle: if I remove CDP from port where device is connected, interfaces never come up ...

Not sure if that this is a device that I would purchase again ...

[paulca]

Generally with Cisco you work out whatever everyone else does, what makes sense and is logical and then pick the other way.

Cisco take proprietarization to a whole new level.  They deliberately make their stuff obtuse, unique and awkward so they can make lots of money selling training courses £1000+ a pop!  It's so backward it's unbelievable, but they have created their little empire now and making their stuff sensible, standardized or easy to use would ruin 100s of thousands of engineers hard spent investment in their training and certifications.

In the past people have written web UIs, GUIs and helper apps and Cisco invariable buy them and make them go away to maintain their training course empire.

They are worse then Apple and Microsoft combined.