WISPSwitch and -48V (positive ground) supplies.

[simonnz]

Hello,

I see this topic has come up a few times, so apologies for dragging it up yet again. Just want to confirm that my understanding below is correct, and also get some context around the warning about -48v on the product overview pages.

Looking specifically at the WS-12-DC, WS-12-250-DC, and WS-8-150-DC . We have one of the latter here, which is what I am basing my questions / observations on.


Can you confirm if my understanding / conclusions below are correct, with regard to the switch models above.


1. There is no connection within the switch between frame ground and the switch +ve or -ve input, or any other internal part.

2. The ethernet port POE -ve pins (7 & 8, also 3 & 6 for 'high power') are permanantly connected (via diode / something) to the switch -ve supply input.

3. The ethernet port POE +ve pins (4 & 5, also 1 & 2 for 'high power') are 'switched' / connected to the internally generated 24v or 48v line as needed.

If above is correct, there is no issue using this switch in a -48V environment, with ONLY NON POE devices connected to the copper Ethernet ports.

If above is correct, there is no issue using this switch in a -48V environment, with POE powered devices that DO NOT connect either the POE -ve or +ve line to their frame ground, or any other external connector (eg, antenna port). In other words powered devices that galvanically isolate the POE lines (complying with IEEE 802.3af 33.4.1 Isolation).

If above is correct, in a -48V environment, POE powered devices that connect the POE -ve line to frame ground, or some other 'groundable' external port will cause a -48V to ground short via the POE -ve line, and WILL DAMAGE the port / switch.


I also see the console serial port GND pin is connected to the switch -ve supply input. I had been thinking we could use an optical isolator (such as http://www.nordfield.com/serial-rs232-optical-isolator ), but I noticed that the switch TXD pin only has any output voltage on it when the RXD pin receives voltage from the partner device serial port. Can you elaborate on what is going on here, as this would appear to make using a port-powered isolator like the one mentioned above impossible.


Finally, I see mention of an upcoming switch with an isolated DC-DC power supply and 36-72v input range. Is there any further detail on this ( eta, specs, physical form factor, price point..)


Thanks,
Simon

[Dave]

simonnz

Thanks for the ultra clear, perfectly formatted, description!!!!!!!!!!!!!

You are 100% bang on with your comments.

The main issue is some manufacturers POE devices (radio's as an example) tie the POE return to shield ground in their device. This is totally wrong and they should not do this, but they have done it, much to my screaming & totally insane comments like, "YOU MUST BE ^&()&*)*(&* KIDDING", why would anyone do that...sigh....

That is why we are making a 400 watt , isolated 24 port version. It will have 1500 volts isolation between input & output voltages.

Expect to have samples ready for field trials in late September, and am looking for volunteers to do some field testing for me!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Thanks.

Dave

[simonnz]

HI Dave,

Thanks for the confirmation. I too was somewhat surprised to find the existance powered devices grounding their poe input lines. Even with 'normal' usage, I;d envisage various interesting failures in the presence of ground faults, etc.

Did you have any thoughts on what I was seeing with the console port TXD only 'liveing' when RXD active.

The new switch sounds good. We're in New Zealand, don't know if that's a bit too far a-field for field testing..


Thanks,
Simon

[Dave]

I also see the console serial port GND pin is connected to the switch -ve supply input. I had been thinking we could use an optical isolator (such as http://www.nordfield.com/serial-rs232-optical-isolator ), but I noticed that the switch TXD pin only has any output voltage on it when the RXD pin receives voltage from the partner device serial port. Can you elaborate on what is going on here, as this would appear to make using a port-powered isolator like the one mentioned above impossible.

This could be correct, as the RS232 chip I use is probably going into sleep mode after not seeing any activity on it for a period of time. It will be active when you power up the switch, or do a hard/soft reset, as their is lots of serial activity when booting.

[mayheart]

You can sign me up for testing.

I have two sites that are in telecom co-location areas that I need to put in some Netonix switches. Not having a DCtoDC converter with it would make things a lot more tidy.

[yoder]

I have noticed problems with certain UBNT Rocket M5 radios but not with Nanostations, Nanolocos, etc. Anyone know about Air Fiber 5?

[sirhc]

I have noticed problems with certain UBNT Rocket M5 radios but not with Nanostations, Nanolocos, etc. Anyone know about Air Fiber 5?

Because the Nanos are inside a plastic insulated chassis and do no come into contact with the steel and any UBNT radio that connects to an antenna will have a connection to the steel tower and can short out of reverse polarity ir negative 48V