DC Power Supply Grounding

C

Thread Starter

cmdPowerSupply

I have seen systems both ways, the DC common:

1) Grounded
2) Un-grounded (left floating)

What are the pros and cons of each of these?

The reason I ask is because I'm fighting noise/grounding issues on a machine. I have a very good understanding about shielding/twisted pairs/isolation/filtering, etc. But this is the only question I have... to ground or not to ground the DC commons from my 24VDC power supplies?
 
R
This question seems to come up every couple of months. The general opinion seems to be ground the negative.

Perhaps I can raise the side question, how to ground it?
Solidly
Through a resistor to avoid reaching PS current limit.
Through a lamp to indicate a ground fault without tripping,
Other
How to protect (24DC)
Breakers
Fuses
One big supply or several smaller ones?
Best method of interconnection

Regards,
Roy
 
C

Curt Wuollet

Ground the low side. Observe what people who don't have noise problems or weird common mode issues do.

Regards
cww
 
M

Michael Griffin

In reply to Roy Matson: I can only comment on applications in small assembly
equipment. Other industries might have different practices.

> Perhaps I can raise the side question, how to ground it?
> Solidly

MG: For most applications, this is the normal approach. Some types of
instrumentation however need to float. Read the manufacturer's documentation
in detail. If something needs to float, give it its own power supply and make
sure that the power supply is actually an isolating type (a linear type with
transformer is often better for this than switch mode).

> Through a resistor to avoid reaching PS current limit.

MG: I'm not sure what the point of this would be. Any decent 24VDC power
supply will have current fold-back or current limiting built in to protect
itself against overload. If it doesn't, start using a different brand of
power supply which does.

> Through a lamp to indicate a ground fault without tripping,

MG: Who is going to look at the lamp and why? If this is for some sort of
special application, then I don't want to comment. This is not done for
normal applications in assembly machinery though. Typically, nobody is going
worry about lamps until the machine stops working.

> How to protect (24DC)
> Breakers

MG: Miniature breakers are almost useless unless there is a very large
difference between the breaker rating and the "real" limit. They are good for
things like keeping people from plugging electric drills into programming
receptacles (AC), but not much use in small DC circuits.

> Fuses

Fuses are reliable and inexpensive. Typically though, fuses on the output side
of a small 24DC power supply don't do much in the event of a short, as the
power supply will go into fold-back much faster than the fuse will respond.
Some people however do put flip-out fuse holders in just to make isolating
circuits convenient when troubleshooting faults. You do need a fuse on the
input side of the power supply though. If the power supply is very large
though, you might want protective fuses on the output and on branch circuits.

> One big supply or several smaller ones?

MG: It depends on what you mean by "big". It is nice to avoid unusually large
power supplies because they are harder to get replacements for and they can
make wiring more difficult. Sometimes is it also a good idea to give things
like instrumentation its own power supply to avoid feeding solenoid noise
into it (the PLC CPU may sometimes need its own as well for this reason). On
the other hand, too many power supplies add a lot of complexity and waste
panel space.

> Best method of interconnection

MG: Don't interconnect if possible. Separation keeps faults localised.

The above suggestions work in many applications, but be sure to read your own
national (or local) electrical codes and apply good engineering judgement to
each problem. There is no single guideline that fits all applications.
 
R
In response to Michael Griffin
The reason I raised the issue to start a discussion, i too have my preferences

Any good quality instrument will be isolated. Try to avoid the ones that aren't. The additional power supplies are a nuisance.

>Through a resistor to avoid reaching PS current limit.

You don't want the supply to fold back otherwise all your instruments will go down.
If you had a 10 amp power supply grounded thru a 12 Ohm resistor the fault current would reach 2 amps and pop just the protection on the faulty loop, (I have never done this but I know some do)

> Through a lamp to indicate a ground fault without tripping,

I have seen the lamp on the front of operator panel I have also seen 2 lamps in series connected to +/- with the center point grounded, both glow dimly until you get a ground fault on one leg. In a mining application where downtime was about $1,000 per minute.

>MG: Miniature breakers are almost
>useless unless there is a very large
>difference between the breaker rating
>and the "real" limit. They are good for

I use the Allen Bradley, typical 1/4 Amp for a 2 wire loop, they pop quickly with a fault and are obvious whereas you have to test for voltage with a blown fuse. I have been on many start-ups wher a fuse blows and none are readily available.
Avoid the indicating fuse-holders for 24 V. The fuse blows but the LED still draws about 12 mA. that can really mess up a loop where the set-point is say 75%, I do use fuses on some prototype skids for cost but where I have a choice always specify breakers.

>Some people however do put flip-out
>fuse holders in just to make isolating

Yes these are good, but with the breaker just push the button

>> One big supply or several smaller ones?
If you had a connected load of 4 amps, would you use 1x10 or 2x5
> Separation keeps faults localized

I used to couple 2 supplies (fed from different circuits) together with auctioneer diodes with a relay on each PS to indicate failure, but then in 30 years I haven't seen one fail
I agree with you on solenoids but many times it's not practical to have a separate supply, but in this case you should add diodes to suppress the noise.

> (the PLC CPU may sometimes need
>its own as well for this reason).

Agree, you don't want to crash the processor because of an intermittent ground fault in the field.

Thanks for your post
Roy
 
M

Michael Griffin

In reply to Roy Matson:
> Any good quality instrument will be isolated. Try to avoid the ones
> that aren't. The additional power supplies are a nuisance. <

MG: "Isolated" may mean only 2 way isolation, rather than 3 way. For some types of higher bandwidth instruments and signal conditioners you don't always have the choice of 3 way isolation. In these cases you have to be careful how you hook up the power supply. A good example of this is the RDP S7DC series.

> You don't want the supply to fold back otherwise all your instruments
> will go down.If you had a 10 amp power supply grounded thru a 12
> Ohm resistor the fault current would reach 2 amps and pop just
> the protection on the faulty loop, (I have never done this but I know
> some do) <

MG: I started off my previous reply saying that I can only comment on applications in small assembly machines. In those applications a fault in any part of the machine means the machine is not usually able to function properly anyway. The emphasis there is on rapid detection, troubleshooting, and repair. The most common cause of a 24VDC electrical short in those applications is a damaged proximity sensor. In fact, having the machine go down immediately would be an advantage as otherwise the machine may damage itself due to an improper sensor signal. 2 amps by the way, could easily be the total 24VDC sensor/actuator load of an entire small machine.

> I have seen the lamp on the front of operator panel I have also seen 2
> lamps in series connected to +/- with the center point grounded, both
> glow dimly until you get a ground fault on one leg. In a mining
> application where downtime was about $1,000 per minute. <

Ground fault lights are sometimes used on power feeds (e.g. 600VAC) in factories, but I have never seen one on a 24VDC circuit. If you have someone
tending a couple of dozen automated machines, they are generally too busy to go over to a machine unless it's actually stopped. Again, the emphasis is on quick repair as a failure in any sensor or instrument means the machine (and line) needs to stop anyway as there is no redundancy. Automotive assembly lines can have high downtime costs as well, but the equipment design emphasis is on eliminating the causes of failures. The environment is generally fairly clean and dry (except for things like painting and coating and some types of welding), so this is a practical approach to the problem.

> I use the Allen Bradley, typical 1/4 Amp for a 2 wire loop, they pop
> quickly with a fault and are obvious whereas you have to test for
> voltage with a blown fuse. <

You are using a very small circuit breaker with what is probably a very large power supply. If the circuit breaker and the power supply were closer in size, you would probably find that the power supply would go into fold back long before the circuit breaker would trip. Even fuses don't always blow fast enough.

This isn't a big problem for typical modern assembly equipment though, except perhaps for very large machines. In the early days of 24VDC equipment, assembly lines were typically still designed with a few large central control panels being wired to numerous machines via long runs of multi-conductor cable. You had lots of cables, connectors, junction boxes, terminal blocks, fuses, etc. The cabling and connectors were one of the major reliability problems.

Modern assembly lines tend to put a small PLC in an electrical panel directly on each machine. There is much less wiring involved and the machines tend to be far more reliable (electrically) than the older styles. Actual wiring faults are very rare provided you take some care in routing cables away from where they might be damaged. The end result is that the most common
electrical fault is a physically damaged proximity sensor. These may short if struck on the end or if the coil is exposed due to rubbing against something. Most other devices tend to fail open. Other than failed sensors though, electrical shorts in 24VDC circuits are rare compared to other types of equipment problems.

By the way, very few miniature circuit breakers meet CSA (Canadian Standards Association) standards for use as protection. This means that you can't use them to protect a load or wiring in Canada. The problem with them is that the
manufacturing tolerances for them are usually so wide that they are unreliable.

You can use them though if they are not relied on as protection. In your example, the 1/4 amp breaker would be OK provided there was also a fuse
somewhere sized to protect the wiring. If the wiring and fuse are large enough, the breaker might trip fast enough to save your fuse.

It is not uncommon for machinery imported from Europe or the USA to fail electrical inspection on arrival in Canada on this issue. The usual solution is to replace the circuit breakers with fuses (modern fuse holders are often similar size and shape). This is normally only a problem on AC power circuits. 24VDC circuits typically come under a low power classification (unless there's a lot of current involved). I believe the USA was going to adopt the Canadian standards on this matter, but I don't know if that has come into effect there yet.

> If you had a connected load of 4 amps, would you use 1x10 or 2x5 <

MG: If I had a connected load of 4 amps, I would probably use a 6 amp power supply. If I needed a 10 amp power supply I would use a single 10 amp supply. These aren't that uncommon and the wire gauge isn't a problem. If I needed 20 amps though, then I would look at how I could split the loads up so that I could use 2 supplies with some loads on one power supply, and some on another. With the control panel being mounted directly on the machine, the individual loads are all connected within the same enclosure as the power supplies are located in. If valve bank 1 is powered from one power supply and valve bank 2 from a different power supply, it's really no problem to wire it up.

> I agree with you on solenoids but many times it's not practical to have
> a separate supply, but in this case you should add diodes to suppress
> the noise. <

Suppression diodes are standard practice, but sometimes they don't seem to be enough. Some customers insist on always having a separate power supply for the PLC CPU, while others don't care unless a particular machine is having a problem. The better PLCs seem to have good filtering on their internal CPU regulators anyway.
 
R
Michael,

I see where you are coming from and agree that shutting down the whole machine is the way to go but for a typical chemical plant you try to keep running as long as possible allowing the process interlocks to use up any surge capacity before tripping.

Re: Isolated
From a process instrumentation point I wouldn't use something a RDP S7DC unless it was for a protoype skid. For a dedicated machine it probably makes sense.

We'll have to disagree on fuses/breaker issue, no I don't use a fuse on the secondary, that's where the foldback comes in, and the wire is sized for that. Yes my power supplies are large relative to the load so if you get a fault it trips quickly rather than pull down all the loops.

I also reside in Canada but work on the design of chemical plants all over. In short I have no problem with what you say, for your industry it makes sense. Lets see if anyone else has something to add.

Regards,

Roy
 
Top