Lost port(s) after load shedding

[Dave]

As far as bonding DC negative to Earth ground my gut tells me no but I will ping Dave to answer that.

Correct, DC negative in a proper world should not be bonded to earth ground.

[jermudgeon]

The important thing is that the Electric Service Ground is bonded to the tower grounding. However if AC power is being supplied from a distance you are better to not use the service Earth Ground at all and just use the tower Earth Ground which you can read about in this post: http://forum.netonix.com/viewtopic.php? ... =30#p13447

Both service ground and tower ground are tied together — it's a grain leg and metal building.

No AC power is being supplied at a distance (not 100% sure what this means). I had read that post, will show it to my electrical engineer(s).

The other thing that is important is to make sure there is a dedicated Earth ground from all radios to Earth Ground that is a shorter and less resistant path than the Ethernet cable path which is where adding a larger service loop sometimes does wonders to save equipment. Likewise make sure the equipment has a shorter less resistant path to the same Earth Ground as the radios use so the equipment does not try and get Earth Ground through the Ethernet Cables.

There is a large service loop; is much longer than path to ground.

As far as bonding DC negative to Earth ground my gut tells me no but I will ping Dave to answer that.

If you RMA the switch we will put a NEW switch board in it for $150.00

Thanks.

We lost another ethernet transformer today, and the switch rebooted twice in a row (same incident). We're replacing it with a new switch. Not sure whether it's residual damage from yesterday or an inherent flaw in our overall DC design. Frustrating.

[sirhc]

Both service ground and tower ground are tied together — it's a grain leg and metal building.

OK so your saying there is no dedicated ground wire running from your antennas down to the ground rods and to your equipment and your relying on the steel structure as your ground path?

Q: What is Ethernet cable made of?
A: Copper

Q: What is the resistance value of copper?
A: 1.68 x 10-8 Resistivity ρ (ohm m)

Q: What is the resistance value of steel?
A: 1.43×10−7 Resistivity ρ (ohm m)

Q: Is the copper wire all one solid single piece of copper
A: Yes

Q: Is the steel 1 solid piece or many pieces bolted together.
A: DUH

Now I am not sure if you understand resistance but steel is a horrible conductor compared to copper. Now also my guess is the steel building and grain mill is not one solid piece of steel but is made up of "many" parts bolted together. Each of those connections is insertion loss or much higher resistance. Plus those connection are usually coated with paint or galvanized which has terrible electrical characteristics.

If you read my ground posts I warn people about thinking steel towers are adequate ground paths. I do not care if you have a 100 foot service loop as there is no way the copper wire will ever be a more resistant path than the steel so any surge from ESD or Static would always chose the Ethernet cable.

Now I know you checking on where the electric service is coming from, you might want to consider what I suggested in the post about not using the Earth Ground from the Electric Service and just using the Earth Ground from the tower.

But you need dedicated ground wires.

[jonb42]

I'm working on this same job with Jermudgeon and have a few questions / comments:

Re bonding negative to ground, this is a common practice in control cabinet wiring, though I know there are are also many people who swear by an isolated negative. It means that one can say that the +ve terminals are all +48V with reference to ground, which is a fairly standard idea. In an isolated ground power supply or demand (equipment) the positive and negative terminals could be floating hundreds of volts above ground, as long as they were only 48V with reference to each other.

There are -48V systems common to the telecom industry, -48V meaning by definition that the positive terminal is bonded to ground which puts the negative terminal at -48V (with reference to ground). We were actually tentatively planning on using a -48V electrical system for these installations because the EnodeB BreezeCompact installation manuals say that they require -48V. In actual fact inside the EnodeB there is no connection between negative and ground, and it does not care in practice whether the 48V it receives is +48 or -48 (with reference to ground). (I assume the mention of -48V in the manual is to reassure those telecom customers who need it to be compatible with their existing DC systems.) The reason we did not go with -48V was because I attempted to power a Netronix this way and received a defiant spark in response indicative of some internal connection between the negative (black) terminal and ground (the case). I had previously tested with an ohm meter while the Netronix was off to see if there was an internal jumper and got an open circuit - so I'm assuming the connection is to do with the design of the internal power supply circuit. In actual fact the Netronix appears to want to hold the negative at around 2.4V above ground. (Just to be clear, in a -48V system the +ve terminal is still 48V above the -ve terminal - it is just that the negative terminal is 48V "below" ground because the positive is clamped to ground. So as long as I didn't ground the Netonix it had no way of knowing that it was getting -48V power.)

All this to say - when an inadvertent short between +48V and ground did occur elsewhere in the system, because the ground was not bonded to negative externally, and all other pieces of equipment (including the main 48V power supply itself) were ground/negative isolated, the power circuitry internal to the Netronix which contains some kind of link between negative and ground received all the current that flowed. Another way of saying this is, had the Netronix not been connected the act of touching +48V to the ground would merely have created what is called a -48V system by the telecom world (with reference to ground). I already know that current is passed internally by the Netronix in this state because of the earlier experience I mention where I got the spark attempting to power it with -48V.

If we actually bond ground to negative external to the Netronix this would mean that in any accidental short circuit between +48V and ground any current would flow through the bonding jumper rather than the internals of the Netronix, so a fuse would blow without having harmed the Netronix power circuitry. This seems like a safer way to run the system, but perhaps there is another more urgent reason not to bond negative to ground?

I'm documenting all this detail not to be dogmatic but rather to state things as I see them currently - I've worked with electricity a fair bit, but not with these specific DC systems and at bottom am just looking to have this explained to me in a way that can make sense of our experiences.

Regarding the resistance of steel vs copper while I agree that "copper has less resistance than steel" it is more useful here to talk about resistivity which is an intrinsic property - in other words it recognizes the fact that area of conductor matters. A thick copper wire has the same resistance as a thin one but much less resistivity. Resistivity = resistance * area/length. So a fat piece of steel may very well have low resistivity than a thin shield of an ethernet cable.

However I don't mean to quarrel with your conclusion, which includes the important point about poor quality bolted connections - I do expect our tower's potential to be a bit different to the power supply's in the building, and intend to remedy this with a piece of #2 insulated copper as mentioned in a post here viewtopic.php?f=30&t=188. (Its a fairly common practice to use copper bonding jumpers across bolted connections on things like cable tray but still use the main chunk of steel as a conductor.) Once I run it I'll measure the difference before actually connecting it:)

Following on from this, however, I'm curious to know how exactly the Netonix ports may be fried by current flowing on the ethernet cable shields from tower > equipment cabinet in the building. I admit that in the current config there is doubtless a bit of current flowing due to ground potential differences, but if this is a steady DC voltage it shouldn't transfer to the enclosed Data / POE wires - and if the Netonix POE circuits are isolated from ground it shouldn't matter to them either. I'm not arguing against installing the bonding between tower and equipment cabinet, just trying to understand how it damages the ports.

I appreciate any feedback and am interested to learn more about this!

thanks
Jon

[sirhc]

Regarding the resistance of steel vs copper while I agree that "copper has less resistance than steel" it is more useful here to talk about resistivity which is an intrinsic property - in other words it recognizes the fact that area of conductor matters. A thick copper wire has the same resistance as a thin one but much less resistivity. Resistivity = resistance * area/length. So a fat piece of steel may very well have low resistivity than a thin shield of an ethernet cable.

However I don't mean to quarrel with your conclusion, which includes the important point about poor quality bolted connections - I do expect our tower's potential to be a bit different to the power supply's in the building, and intend to remedy this with a piece of #2 insulated copper as mentioned in a post here viewtopic.php?f=30&t=188. (Its a fairly common practice to use copper bonding jumpers across bolted connections on things like cable tray but still use the main chunk of steel as a conductor.) Once I run it I'll measure the difference before actually connecting it:)

Following on from this, however, I'm curious to know how exactly the Netonix ports may be fried by current flowing on the ethernet cable shields from tower > equipment cabinet in the building. I admit that in the current config there is doubtless a bit of current flowing due to ground potential differences, but if this is a steady DC voltage it shouldn't transfer to the enclosed Data / POE wires - and if the Netonix POE circuits are isolated from ground it shouldn't matter to them either. I'm not arguing against installing the bonding between tower and equipment cabinet, just trying to understand how it damages the ports.

Answer to red text above: Because UBNT bonded the digital negative to earth ground in their radios (airMAX Rockets) which is why when you run a cable diagnostics on the 10/100 ubnt device you see short on pairs 3 and 4 instead of OK or OPEN. Also to achieve the ability to do a factory default with the POE brick they are tagged onto the power coming off the center tap of the Ethernet transformers on pairs 1 and 2 which is why if you try and power any 24V airMAX radio with 24VH it fries the switch port and radio as it can not deal with the power on pairs 1 and 2. They should have put in an isolated circuit in my opinion but they did not so there is no electrical isolation on their devices. So if you are using a Rocket then negative is bonded to the SMA connector which is bonded to the dish or sector which is bonded to the tower so any stray current on the antenna or around it on the tower will jump right on radio and follow your Ethernet cable down. READ THE POSTS I LINKED, it will tell you why low voltage high current not only kill equipment but causes Ethernet ERRORS like CRCs. And some of the information is from other companies and even people with PHDs in this shit.

No, the fact that the steel is a much larger volume does not mean that Static or ESD charge around your antennas will flow through the steel verses the Ethernet copper wire.

Look, there is a reason cell carriers spend so much money on grounding and I learned this lesson the hard way many years ago and more recently was reminded of this when Sprint left a tower and removed the copper wire on a "Steel" water tank which you can read the saga here on this post:
viewtopic.php?f=30&t=1429

When Sprint removed the dedicated ground wire I immediately started losing Ethernet ports and equipment on a tower that I have "never" lost any equipment on.

Basically here is the outcome of my loses when Sprint took out the dedicated ground wire:
In the fall / early winter of last year we lost Port 1 and 2 current sensors and port 1 Ethernet port of a WS-24-400A switch. We thought this was odd but there was a TERRIBLE storm that night and in fact lightening had struck a house about 1/4 mile in front of the radio attached to port 1 so we replaced the gear and chalked it up to bad luck.

A few weeks to a month or so latter the same thing occurred but the storm was mild and mostly rain and high winds. Once again we replaced the switch and thought this was odd but I asked my guys to service the ground. The way Sprint removed the ground cable that ran up the tower it was easy to miss as they cut the 1" copper wire 15 feet up from the main tower ground bus after it entered the cable tray and 15+ feet down from the main top ground bus.

We hurriedly installed a #2 green in that ran inside the water tank (Ball tank) and in the process added 2 ground buses and the #2 was spliced in the middle which adds insertion loss or resistance but again we lost another port 1 and radio. We next ran a home run non spliced #2 and since then have not lost another piece of equipment and we have made it through the spring and early summer lightening season.

The last switch that was damaged is still in service but port 1 is DEAD as seen in the picture below.

Look that is a Steel water tank that is welded together not bolted together steel pieced which would be an even worse ground path from all the connections.

Transient voltage/current will take the path of least resistance to ground and if you think your coated bolted together steel no matter how much mass will look more attractive than a 1 piece copper wire as a path to ground then you have already lost the battle and be prepared to lose gear on that tower every year.

I can tell a similar story to almost everyone of my 15 towers from before we learned proper grounding. I used to cringe and lose sleep on stormy nights but not including this tower I have not lost gear at any towers from storms in 5+/- years and that loss was a direct hit to a 150' water tank that blew the red light off the top of the tank and it landed in the parking lot. Then a year before that was a 175' tower we rent space on that blew off an AT&T sector and killed a 100 turkey buzzards that called the tower home. Before that you would go back several years when we first learning about proper grounding and before that we lost gear it seemed every other week in the Spring and Fall.

All I can do is try to convince people the importance to grounding. How important it is to have a dedicated ground path from any piece of equipment to ground that is shorted and less restive than taking the Ethernet cable. All pieces of equipment on the tower and in the box should be all bonded together and the path to ground from any radio to ground needs to be shorted and less resistant than the path to the switch/router and then to ground.

Bonding the tower grounding system to the electric service ground is a must but if the AC power is coming from more than 25-50 feet from the tower then only use the hot and neutral and substitute the tower ground for Earth ground else you are bonding to different ground potentials and more than likely the tower ground will be better than the service ground which means all excess current from the service will flow towards your cabinet and then flow through your equipment and then up your Ethernet cable through your radio and to the tower to get to ground.

Which here is an example of not using the Service Earth Ground: viewtopic.php?f=17&t=1786&start=30#p13447

Look back in 2009 I started RF ARmor and started trying to tell people the benefit of shielding and how it improved performance and it was an uphill battle the whole way. Finally 5+ years later every manufacturer including UBNT is now pushing shielded equipment. - DUH

I have been pushing grounding for almost the same amount of time starting back in the UBNT forums and even more now thinking people would finally listen to me more now that I manufacture electronic equipment.

To be honest I am growing weary so do what you want with your grounding system but all I can say is storms do not bother me or cause me to lose sleep anymore can you say the same thing?


Here are some good posts on grounding:
viewtopic.php?f=30&t=188
viewtopic.php?f=30&t=1429
viewtopic.php?f=30&t=1816

You should also do some Google searches on ground potential damage to Ethernet switches:
https://www.google.com/webhp?sourceid=c ... 0potenital

[jonb42]

Thanks for all the work that went into this reply.

Point taken re static / lightning / ESD on shield vs thick steel - an excerpt from Motorola R56 manual (which is a great reference for this whole subject) says "Relatively small copper strap has significantly less inductance (impedance to lightning) than large wire conductors. For example, 38.1 mm (1.5 in.) copper strap has less inductance than 70 mm2 csa (#2/0 AWG) wire." This is talking about copper vs copper I know but the same principle follows - we're talking about very high voltages here. From what I have read I think this is due to the high voltage skin effect.

For lower, DC voltages such as ground potential imbalances larger wire has lower impedance than small which is why we're bothering with the #2 bonding ground as you recommend, to remove the current flow from the shields and instead flow them through the #2. I think we're on the same page here and we will definitely follow through and install a better grounding system as you recommend.

Can you comment on the "bonding ground to negative terminal" portion of my earlier post - perhaps addressing my concern that any fault current (when +48V shorted to ground) flows solely through the Netonix switch internal power supply if we don't bond ground to negative, since there is an internal low impedance connection from negative to ground. Alternately you could just reassure me that this low impedance connection is built to take x amount of current and the switch won't be harmed. Are you the one who actually designed the internal power supply circuitry, or can you ask the guy who did? I can post a circuit diagram showing what I'm talking about if I'm not describing it clearly.

thanks
Jon

[jonb42]

Anyone have any feedback on this?

thanks
Jon

[sirhc]

Can you comment on the "bonding ground to negative terminal" portion of my earlier post - perhaps addressing my concern that any fault current (when +48V shorted to ground) flows solely through the Netonix switch internal power supply if we don't bond ground to negative, since there is an internal low impedance connection from negative to ground. Alternately you could just reassure me that this low impedance connection is built to take x amount of current and the switch won't be harmed. Are you the one who actually designed the internal power supply circuitry, or can you ask the guy who did? I can post a circuit diagram showing what I'm talking about if I'm not describing it clearly.

Dave did give his 2 cents?

Correct, DC negative in a proper world should not be bonded to earth ground.

[jonb42]

I did see this, but took it as his view of how things should be in principle - in a proper world, as Dave says.

If there was no internal connection / low impedance then I'd have no worries about it myself - it wouldn't matter. For instance the BreezeCompact EnodeB has an internal power supply which is completely isolated from ground - ground can be connected to positive or negative and the power supply doesn't care because it was designed this way. If the Netonix power supply was similarly isolated from ground I'd have no qualms about NOT bonding negative to ground - it would not matter and might in ways be preferable as Dave suggests.

However there is an internal high impedance and as I said this means if I don't bond any fault current flows through the Netonix internals, which seems to be something to be avoided.

If it helps I can make a schematic of the Netonix power supply and discover where the internal "connection" between negative and ground is, but this is probably something you guys already have. I know its not a direct wired connection because when the switch is off there is no connection. Only when the switch is on does negative approximate ground.

thanks
Jon