Paralleled Shunts

[tallgirl]

I have a client who has asked me to confirm that three shunts can be placed in parallel to measure the battery charging current for a 43KW DC array feeding a 48 volt nominal battery bank. The theoretical answer is, of course, "Yes", and the battery monitor that's been specified will properly measure amp-hours in this configuration. (FWIW, they are a large electrical contracting company located out on the west coast).

Obviously there needs to be "and be sure to have overcurrent protection for each shunt" as these are 1000 amp / 100 millivolt shunts with a maximum continuous rating of 600 amps, as I recall, but I'm not designing the system, just answering a technical question about a battery monitor.

And now for the real question -- the current through each fuse / shunt pair is inversely proportional to the resistance -- that's basic circuits. And because we're dealing with very low resistances, even a small difference in resistance will mean a large difference in current and possibly all three fuses blowing once one does. The maximum charging current is around 1,100 amps at 42 volts DC (low battery cut-off) from the array, and about 1,600 amps if they start the generator during broad daylight. The maximum continuous discharging current is around 900 amps, again at 42 volts.

Is it safe to assume that the difference in resistance between one paralleled shunt+fuse and the next is going to be small enough that even though there will be differences, they won't be so great as to blow any one fuse? Am I right in assuming that the resistance of a fuse is fairly constant for any given amperage?

 

[__dan]

How does the meter know the current is being divided by 3 ? The install could have problems.

I would call the shunt manufacturer to see if they like the idea. I'm thinking the shunts would have to bolt directly to a common busbar with no fuses. If one heats, its resistance will go up and load the other two shunts. Calibration could be off by a lot, more than 1%

 

[tallgirl]

Dan,

The install will be just fine. On the subject of whether or not the battery monitor is going to process the data there is no question -- it will sum all three shunt values and use it as the net current between the system and the batteries.

All three shunts are to be bolted directly to the bus bar on either side. The question is whether or not a 1000A (or 600A ...) fuse needs to be bolted somewhere in the middle.

 

[gar]

110725-1121 EDT

A shunt is designed to have very little temperature rise, and also uses a relatively constant temperature coefficient material. Probably could overload by a factor of 3, 9 times power, and not destroy the shunt. Time of overload might be a factor, short time no problem.

If you parallel three shunts, then how equally do you think the current will distribute? Or does this not matter. Shunts are 4 terminal devices for a good reason. Will you use three differential amplifiers, one from each shunt, and then sum the amplifier outputs? Or just measure the output of one shunt?

How do you plan to check calibration?

Have you considered a Hall current sensor?

Could you not get a single larger shunt, or make one?

.

 

[Besoeker]

Shunts are 4 terminal devices for a good reason.

Yes, that's the first thing that struck me when I first read the question.
The measurement terminals are not the same as the power terminals and it is (presumably) the power terminals that will get paralleled which means that any difference in current between the three legs will result in potential differences between the measurement terminals of different shunts. Seems a bit messy. I'm with you - a bigger shunt or a Hall-effect sensor would be a better option.

 

[Besoeker]

All three shunts are to be bolted directly to the bus bar on either side. The question is whether or not a 1000A (or 600A ...) fuse needs to be bolted somewhere in the middle.

I think using fuses to protect the shunts is going to be difficult. Given the 1600A in some situations, it would need say, minimum a 600A fuse in each leg. Suppose one blows for whatever reason, then the other two legs would each have 800A if the current shared equally. I don't have curves for US fuses but I do for British (BS88) fuses. I don't suppose the fuse technology is vastly different.
A 600A fuse carrying 800A has a pre-arcing time in excess of 20,000 seconds - the time axis doesn't go beyond 20,000s so it could be much longer. In fact, even at 900A it is still above 20,000s. That's about five and a half hours. Would the shunts survive that?

 

[S'mise]

The systems I have encountered used external methods for monitoring the current. Not fuses. Eg. A shunt that reads current leakage to ground or elsewhere.

 

[brian john]

Why not one larger shunt?

 

[petersonra]

There is something screwy about this whole deal.

 

[__dan]

I still don't see how this meets requirements. The third shunt could be a gum wrapper that vaporizes in the first second under load. The meter, full scale deflection at 100 mV, will continue to see the common busbar voltage and assume calibrated voltage drop under load. The two remaining shunts will have the current split between them and and raise the common bus voltage proportionally.

The meter may be averaging rather than summing and take the common bus voltage x 3. Maybe it will accumulate errors if the current does not split exactly, equally as intended ?

Isn't a standard shunt calibrated for either 50 mV or 250mV full scale deflection at rated load ?

 

[dereckbc]

Why not just use a 1000 amp DC 50 mv shunt? Pretty common practice in DC plants for 24 and 48 volts up to 10,000 amps. 1000 is fairly small and common value. As for protecting the shunt with a fuse, never heard of it nor do I think it would be needed. Shunts are very robust and never have I even heard of one being damaged from a fault even on 10,000 amp 48 volt plants. Can't say the same for the structural steel, tools, or personnel near the contact point though. :happyno:

 

[S'mise]

I have seen them (shunts) disapear like a glass fuse.