Ground advice needed for large nautical installation

Thread Starter

bordonbert

Joined Feb 21, 2012
40
Hi guys. A curiosity here. I am working on a PCB design for a piece of kit which will handle fast rise accurate (50nS) but slow frequency timing pulses between pieces of equipment well spaced out from each other. Pulses are 5V and 12V. This is on a large vessel at sea, (cross channel ferry size). Grounds on these vessels are notoriously "non-earthlike". In a worst case they can measure more than 40V difference from place to place and are very noisy.

The PCB is powered from an external switched mode PSU at low voltage, 5V/12V. Circuitry draws so little current there has to be a hi wattage load resistor across it to guarantee a PSU will stabilise and drop to the correct voltage. It is enclosed in a rack mounted metal case in a server room in the bowels of the ship and connects to a PC nearby as well as to equipment in other areas of the ship. Inputs are all opto-isolated so effectively low impedance differential mode. Outputs are not opto-isolated and could go to other parts of the boat.

The inputs take care of themselves, just keep them isolated totally. Outputs, while single ended, have also taken care of themselves in the past so I don't see them as a problem. I have the incoming PSU/circuitry 0V, case ground and PCB plane grounds to consider. I'm not sure what is the best way to handle this from the point of view of signal integrity, with a low voltage external supply and rack mounting like this safety is less of an issue. I am tempted to just connect them at a single point, or maybe attach the case to the others via a resistor.

Can anyone offer any advice?
 

studiot

Joined Nov 9, 2007
4,998
Fibre is a good suggestion.

Back in the 1970s/1980s we did not have this luxury and I remember taking any equipment manufacturers digital equipment offshore invalidated the guarantee.
But we still did it in the oil fields and we had lots of fun.
It was not only the grounds that played up.
Even powerful vessels generators suffered significant dips and surges as the vessel made way against load.
 

Externet

Joined Nov 29, 2005
2,202
Agree with optical link.
If good solid grounding connections are chosen; suggest to grind a proper metal beam and with oxyacetylene brazing, make a puddle of bronze at that clean spot and weld there a bronze screw head for maximum intimate connection. Even directly, a copper wire into the bronze puddle.
 

DNA Robotics

Joined Jun 13, 2014
647
There should be a bonding system in your boat that is attached to zinc anodes on the outside of the hull. Those zincs are the closest you will get to ground. Bonding systems get bad connections with age. If a boat next to you in a marina has electrical problems, it can electrify the water in a large radius.
 

cmartinez

Joined Jan 17, 2007
8,220
There should be a bonding system in your boat that is attached to zinc anodes on the outside of the hull. Those zincs are the closest you will get to ground. Bonding systems get bad connections with age. If a boat next to you in a marina has electrical problems, it can electrify the water in a large radius.
Wow... I'm learning a lot from this thread... seriously...
 

Thread Starter

bordonbert

Joined Feb 21, 2012
40
Hi guys. Thanks for all the responses but I think I may have worded the original question badly.

We are not talking about a boat in a marina here, we are talking about an ocean going deep sea seismic survey ship as big as a car ferry! The setup is in a dedicated server room down in the hull. The sensor and other devices we are connecting to are elsewhere in the wheelhouse, up on the masts, or on deck in the work area at the stern. I wasn't meaning what is the best way to design the circuitry, if I had that choice you are right, I would have preferred opto-isolation for all outside connections (which would be awkward with the PSU arrangements), or at least differential. I'm stuck with the circuitry as it stands.

I was considering that the best approach is as follows:

PSU in, (remember this is an external floating 5VDC+12VDC with low current drain) to 0V at point of contact on PCB, (GNDref).
All ground paths in circuitry link back only to GNDref, not to surrounding ground areas.
Surrounding isolated ground areas on PCB direct to GNDref only.
Case, (grounded in rack), connected to GNDref through low value resistor or resistor//capacitor.

The last point is the only one I'm not sure about. It has been suggested elsewhere that, as it is using an external Low Voltage PSU the case need not be grounded at all. I just don't like the idea of leaving it like that in case it ever gets run up on a bench out of the grounded rack. It isn't so much a safety aspect as a noise issue.
 

marcf

Joined Dec 29, 2014
288
Hello:

Please be advised that I have absolutely NO experience with marine voltage issues. I do have quite a lot of experience with large commercial Fire Alarm (Simplex) Installations though.

These systems handle fairly serious amounts of power and fairly advanced data transmission schemes with sensors (smoke detectors) and signaling devices (strobes and speakers) that 'snake' all over the place.

I think that they even have such systems on large ships.

One of the primary requirements of a good Fire Alarm system (see NFPA-72) is that it needs to be totally isolated from ground. The entire system needs to 'float', and quite a lot of effort is put into detecting ground fault conditions.
 

cmartinez

Joined Jan 17, 2007
8,220
Hello:

Please be advised that I have absolutely NO experience with marine voltage issues. I do have quite a lot of experience with large commercial Fire Alarm (Simplex) Installations though.

These systems handle fairly serious amounts of power and fairly advanced data transmission schemes with sensors (smoke detectors) and signaling devices (strobes and speakers) that 'snake' all over the place.

I think that they even have such systems on large ships.

One of the primary requirements of a good Fire Alarm system (see NFPA-72) is that it needs to be totally isolated from ground. The entire system needs to 'float', and quite a lot of effort is put into detecting ground fault conditions.
Question, I think I understand what you mean by the entire system being required to 'float' by not being grounded to anything external to the system, but does this system share a common ground among its participating devices?
 

studiot

Joined Nov 9, 2007
4,998
A good digital system need not share a common ground since it is counting coded pulses, with wide specifications on the voltage and waveshape.
 

Thread Starter

bordonbert

Joined Feb 21, 2012
40
My requirement is only for timing and synchronisation purposes based on reasonably fast clean edges. Nothing much is at high transfer rates as in, quantities of digital data being sent at high speed. Ship environments don't lend themselves to too much of that I'm sfraid.

It's 1PPS stuff at best to be able to synch the clocks of other devices and an accurate fire pulse based on position maybe every 8secs that is exactly approximate, (can't predict exactly when it will happen but when it does the time must be measured to 50nsec to know exactly when it did). There is some transfer of data to real time systems talking to the boat like pitch and roll sensors, depth sensors, autopilot etc but these are at pretty standard baud rates. Some of the other devices will be set up with their own opto-isolated or just differential inputs which would take care of things for us, but I can't guarantee this, or that others will connect the devices up correctly! Just wanted to cross as many Ts as I could ahead of time.
 

marcf

Joined Dec 29, 2014
288
Question, I think I understand what you mean by the entire system being required to 'float' by not being grounded to anything external to the system, but does this system share a common ground among its participating devices?
In a single large building, the entire system usually does share a common ground but it is only common to the FACP .
(Fire Alarm Control Panel)

(Think birds on high voltage transmission lines)

The common ground (called 0 volts) is completely isolated from any building grounds or electrical utility grounds. One of the primary 'jobs' the FACP has is to detect when one occurs and report it. If you ever see a yellow light on a FACP, hopefully it will not stay that way for more than a couple of hours, and it needs to be reported to somebody by the system in the form of a 'trouble alarm' within 2 mins.

A 'ground fault' alarm will occur with any leakage greater than about 1 meg ohms to ground (They are 24VDC systems), and are sometimes very difficult to find.

Peripheral devices 'talking' to the FACP could be isolated from the FACP and report a ground fault via an isolated communications link (fibre optic cable).

My experience was at the University of Washington in Seattle.
This building is typical:

http://pcparch.com/project/physics-and-astronomy-building

I guess (without any knowledge of what I speak), I am somewhat surprised, that this is even an issue with electrical control and communication on a large ship. (Other than finding ground faults when they occur, that is.)
 
Top