Current shunt design issue

[Terabyte2007]

I may answer my own questions here but maybe I can get a few tips!

Designing a voltage monitor for an 80 Amp 5V supply + a 10 Amp 12V supply which will be switchable between the outputs. Using a current shunt with a precise value to get my voltage drop which then feeds my digital panel meter. Problem is, the higher current is causing the shunt to heat up changing the resistance thus changing my reading, not by a lot, but not acceptable by my standards. My thought was to use a temp probe attached to the shunt and feed that back to an op amp circuit for error correction in the meter readout. Essentially, I need to compensate for the change in resistance and adjust the reading according. Are there shunt materials which are more stable over a wider range of temps? Or, is there a better way to do this? This was a shunt from China, not sure of the quality or material properties.

[ejeffrey]

For such a high current / low voltage I would strongly suggest the use of a hall effect sensor.

Other than that, good shunts are made from manganin or a similar low thermal coefficient.  They should also be beefy enough so they don't self heat too much.  I don't think temperature compensation is a great plan: the thermocouple is going to lag the shunt temperature, and not account for temperature gradients.

What is the value of your shunt resistor?  How much is it heating?  How much error do you see?

[LukeW]

a) Use a smaller shunt resistance to reduce its power dissipation. Change the shunt amplifier gain accordingly.
b) Use a higher quality shunt resistor with lower tempco.
c) Use a bigger shunt resistor, well over-rated in terms of its power dissipation, so it has more thermal mass.
d) Consider a Hall-effect device, such as (to pick one pretty much at random that is rated for 100A) the ACS759LCB-100B.

[Richard Head]

It sounds like your shunt is overheating. Stop treating the symptom and treat the cause. Get a bigger shunt or cool the existing one.

[mzzj]

It sounds like your shunt is overheating. Stop treating the symptom and treat the cause. Get a bigger shunt or cool the existing one.

One of the oddities in "this is the way it has ever been and now its a industry standard"  is that bigger current measurement shunts shouldn't be used at rated current continuously.
Usually continuous use rating is 2/3 of the nominal current rating and in some cases even less.


Other than that I would suspect that the chinese pingpong-brand current shunt is less than perfect. Look for a manganin-based shunts from brand-name manufacturer and derate it to 1/2...2/3 of the nominal rating, check that the shunt temperature stays below 80 degrees celsius and you should be good to go.

[David Hess]

The easiest solution is to use a better or larger current shunt.

One way besides temperature to compensate for changing current shunt resistance is to inject a small AC current into the shunt, synchronously demodulate the AC voltage produced, and use that to measure the actual current shunt resistance while applying that correction to the actual signal via an analog multiplier or digitally.  Some high end bench multimeters can do this allowing them to measure the in-circuit current to high accuracy through an unknown resistance like a printed circuit board trace without cutting it or using a current probe.

[turbo!]

It doesn't seem like the "quality from China" is the issue as much as the design practice of the person designing this system. 

What is the resistance of your shunt? Per ohm's law, dissipation is I^2R. Even a 1mohm shunt will dissipate 6.4W at 80A, or 10W at 100A. What is "within acceptable"?  What resolution do you need with what accuracy of reading? What is it that you're trying to accomplish with this?

[SArepairman]

I would not use a thermometer with a shunt unless it was designed to accept one.

Does your shunt bolt to a heat sink? If you get a TO-220 shunt, bolt it to a heat sink that has a temperature sensor attached this might be enough to get a bit more accuracy.
Also, maybe you can just get the heat sink big enough so that the shunt drift is reduced @ max temperature.. or just put a simple temperature activated fan on the heat sink.

regardless of what you do a heat sink will help you. I bet you would actually get pretty good performance if you put a TO style temperature sensor against the heat sink....


OR use a over sized regulator with a thermal pad on it, and attach a sensor to the thermal pad (imagine back to back to-220). You can assured there will be a low thermal resistance between the sensor and the resistor at this point.. so long you can handle the dissipation without the heat sink (it might cause your resistor to actually drift if its really hot all the time)...
Or you can take a heat sink which accept your shunt, drill a small hole through it, enough for the temperature sensor to slide through and contact the thermal transfer pad on the back of the shunt, so you can have direct thermometer contact and a heat sink (to possibly  reduce shunt drift due to temperature). Then use some thermally conductive glue to seal it in there, should be decent.