5

What is the importance of a 4-terminal shunt resistor? What is real purpose behind using two seperate terminal for the voltage measurement across the shunt resistor?

Marcus Müller
  • 88,280
  • 5
  • 131
  • 237
IamDp
  • 1,865
  • 4
  • 26
  • 40
  • 2
    Kelvin connection (no time to answer now but the term might be helpful). – Wesley Lee Jan 09 '18 at 10:33
  • 1
    Good related article which helps understand the issue and also shows ways to use this technique with cheap 2-terminal shunts: http://www.analog.com/en/analog-dialogue/articles/optimize-high-current-sensing-accuracy.html – Manu3l0us Jan 09 '18 at 12:44
  • 5
    Possible duplicate of [What's the difference between a shunt resistor and a four-terminal resistor?](https://electronics.stackexchange.com/questions/192079/whats-the-difference-between-a-shunt-resistor-and-a-four-terminal-resistor), [why two pair of terminals are used in a shunt?](https://electronics.stackexchange.com/questions/294764/why-two-pair-of-terminals-are-used-in-a-shunt) – Dmitry Grigoryev Jan 09 '18 at 14:29

4 Answers4

21

Take a look at this. It shows two terminals as the main current flow (thick tracks) and uses 2 more terminals for measurement of the volt drop generated by that current: -

enter image description here

The "better" way is on the left because it takes measurement connections from a defined place and at no point on those measurement connections is there load current flowing.

The picture on the right shows the measurement connections at some small distance from the shunt resistor terminals and therefore there is a small volt drop that forms an error - in effect you can't rely on the stated value of the shunt resistor in order to convert the measured voltage to an assumed current flow.

Andy aka
  • 434,556
  • 28
  • 351
  • 777
  • 2
    +1. For future readers, the image is present in page 24 of this datasheet: http://www.ti.com/lit/ds/symlink/ina240.pdf – Wesley Lee Jan 09 '18 at 10:42
  • 1
    Problem is, your resistor only has two terminals - not four. – pipe Jan 09 '18 at 10:43
  • @pipe I'm not sure what you are indicating? – Andy aka Jan 09 '18 at 10:47
  • 1
    I mean, OP explicitly asks why certain resistors have _four_ terminals, while you show how to get an accurate reading with only _two_ terminals. So why would a resistor have four terminals when it's not necessary? – pipe Jan 09 '18 at 12:07
  • 1
    @pipe I thought it might be obvious but I'll make an amendment to be clearer. – Andy aka Jan 09 '18 at 12:20
  • 2
    @Andyaka, your answer above clearly details on the connection of the measurement traces. Expected answer was to exaplain the need of additional two terminals on the Shunt resistor which is answered in the below answers. – IamDp Jan 09 '18 at 12:52
  • @Durgaprasad my middle paragraph I think explains that - no load current on the measurement connections means no extra volt drop that causes an error to appear. – Andy aka Jan 09 '18 at 14:04
  • @Andyaka Yes, i missed that point!That would exactly translate the need of them. – IamDp Jan 09 '18 at 20:47
  • 3
    In a 2 terminal resistor, the pad geometry is important, which is outside the manufacturer's control. In a 4 terminal resistor, the way the manufacturer measures/warrants its performance is the *same* as the user will get, regardless of pad geometry. I'm not going to make a separate answer just to say this, would you to add it to your already good one? – Neil_UK Jan 10 '18 at 06:49
  • With a 2-terminal resistor, even using the "correct" Kelvin sensing layout as in the left example, the load current and sense current share a path through the solder between pads and package. A 4-terminal resistor can avoid this. – pericynthion May 27 '19 at 23:36
8

I will just go out on a limb and say that it's probably because of this:

  • 2 terminals for the current to flow through.
  • 2 terminals for measuring.

By having 4 connections, the supplier of the shunt resistor can make sure that what you are measuring across will be the shunt resistor where the current will flow which you are trying to measure.

If there was only two terminals, then you would have to decide for yourself where you want to branch off for measuring. There are many viable positions to measure across, but which one will most correctly measure the voltage across the shunt resistor? When you are making high quality stuff, this is no simple question.

Harry Svensson
  • 8,139
  • 3
  • 33
  • 53
3

Shunt resistors are used where there are high currents. As such the point at which you tap of the voltage to be measured is important. Moving it by even 1mm may make a big difference especially if there are large currents.
The four tap shunt is built to have the voltage points there where they represent the correct conversion from amps to voltage. See it is part of the calibration.

On your question of putting them in parallel.
That would be possible if the two resistors where really equal in value. In that case you can measure one of them and multiply the result by two.
But we all know components are never exactly accurate. Thus a small difference in resistance will make the current split unequal. I would say you have to measure both voltages and add them up.

Oldfart
  • 14,212
  • 2
  • 15
  • 41
  • Clear. How do we connect them if we are using two such shunts in parallel connection? – IamDp Jan 09 '18 at 12:53
  • @Durgaprasad: if you connect 2+ such 4-terminal shunts in parallel, then just ignore that extra terminals and branch off first on the track before first resistor, and on track after the last resistor in paralell. The reason is, when you CONNECT them in parallel, you need some TRACK or JOINT that does the connection, and these kind of things is precisely what you usually try to avoid by using special terminals. If you parallelize them, do your thing and just ignore them.. OTOH, if you want that to be precise, buy a single 4-term resistor that has N times larger resistance. – quetzalcoatl Jan 09 '18 at 13:07
  • @quetzalcoatl you mean smaller - putting resistors in parallel will make total resistance smaller. – Arsenal Jan 09 '18 at 13:53
  • My, of course! I was writing having in mind "series" the time, I have no idea why. – quetzalcoatl Jan 09 '18 at 14:05
  • @quetzalcoatl I would try the proper parallel connection. According my requirements i am not finding the right value resistor with enough power rating. – IamDp Jan 09 '18 at 20:40
  • @Durgaprasad: while parallel conneciton is a much lesser disturbance than series, I'm a bit way about that. You have two pairs of terminals per part (let's call them A/B for main and C/D for measurement), and basically have two options. For 3 parts, it's fairly obvious that A1-A2-A3 go together, and so B1-B2-B3. What about C/D? If you connect them the naive way C1-C2-C3 and D1-D2-D3 and measure that as C-all and D-all, then **any imbalance** on A123/B123 will cauase current to flow through C1-C2-C3 link and through D1-D2-D3 link. That may damage parts. – quetzalcoatl Jan 09 '18 at 20:53
  • @Durgaprasad: therefore, to be on the safe side, I would **not** connect Cx and Dx together, and instead, I'd place 3 ADCs and connect each one to C1/D1, C2/D2, C3/D3 separately as 3 measurements, and then collect data and average them on digital side. That way imbalances will not flow between resistors via side pins. However, that rises the cost and complexity. If you know that the currents are not, eh, too large, then connecting them as C-all and D-all may actually not do any damage and imbalances flowing that or the other way will do the averaging for you.. – quetzalcoatl Jan 09 '18 at 20:57
  • @Durgaprasad: .. still, IMHO, risky. Side pins are for low current voltage measurements.. If you go this way, start diggin through the datasheet to see if there are any current limits on that measurement terminals. – quetzalcoatl Jan 09 '18 at 20:58
  • 2
    @quetzalcoatl: Connecting C1, C2, and C3 each to a common point via moderate-value (e.g. 1K) resistors, and doing likewise with D1, D2, and D3, should avoid the need for separate ADCs if the measurement device's input impedance is high enough that the added series resistance doesn't matter. Current flowing in the added resistors may lead to measurement errors, but should typically be small enough to be a non-issue. – supercat Jan 09 '18 at 21:55
  • A shunt may have a wide range of current sensing, so "high current" is a relative term. It could be mA to a 100 amps. I have bought and used a 100 amp shunt with 0.1% tolerance for UL testing. The current feed and the sense terminals were the same. –  Jan 09 '18 at 22:42
  • @quetzalcoatl I would not prefer to use three different ADC channels for three shunts connected in parallel. This would simply eat up my ADC channels(as i have already run out of all). Tying C123 and D123 still seems to be a good solution and would not damage any parts provided you precisely identical shunt resistors! – IamDp Jan 10 '18 at 19:54
2

Shunt resistors for measuring large currents have very low resistance so they don't consume excessive power and produce excessive heat. Shunt resistors are usually far less than 1 ohm, usually being 0.1 or 0.01 ohms or even 0.001 ohms. The problem with using resistors of such a low value is that the resistance of the solder joints and circuit board traces can be a similar order of magnitude and hence will create a voltage drop similar in magnitude to the voltage drop across the sense resistor itself. This makes accurately measuring the voltage drop through the precision sense resistor a bit tricky. The most robust solution to this is to separate the high current path from the voltage sense path with a 4 terminal shunt. The high current path may have a few extra milivolts of drop due to the extra miliohms of resistance from the traces and solder joints, but the voltage sense connections will not see this due to the different current path and can therefore much more accurately read the shunt voltage.

alex.forencich
  • 40,694
  • 1
  • 68
  • 109