During the initial design stage, we were looking at multiple options for sensing current that would flow in and out of the battery pack.
As many of you know, there are mainly two approaches to sense current, these are can be classified as:
- Invasive – Sensing currents using shunt resistors or similar
- Non-Invasive – Sensing using hall effect sensors or current transformers
Since we were working with a DC System and needed accuracy down to 50mA resolution, we decided to go with an Invasive technique.
Non-Invasive current sensing is good when you want isolation but comes at the cost of reduced accuracy.
When it comes to current sensing using Shunt Resistors, the following disadvantages come up:
- Heat dissipation in the current shunt: This can especially be a problem when dealing with very high current and a small board size.
- Lack of Isolation: Since we are technically just measuring the voltage drop across a resistor, there is practically very little isolation. Furthermore, isolating analog signals completely is complex and expensive.
Other than these points above, sensing using a shunt is very accurate, is low cost and straightforward to implement.
Choosing a shunt resistor:
After deciding that we were going to use a shunt, the next thing we needed to do was find an appropriate part.
The resistor needed to be small enough in value so as to keep heat dissipation at a minimum(Heat= Square of Current x Resistance).
It also needed to be in an appropriate power rating value and physical package size.
And the most important thing was to choose a part with a very low-temperature coefficient.
The temperature coefficient governs how much the value of a resistor would change for every degree rise in temperature. This is usually expressed in parts per million with typical numbers being 50-500ppm.
Current shunts can be expensive so there is a trade-off between cost and accuracy. We decided to go with high-quality shunts from ISABELLENHÜTTE since costs were not a primary parameter. Our chosen shunt had a temperature drift of less than 30ppm per degree K.
Where should you place the shunt? On the supply side or near the ground node?
A very common question everyone has is whether to go with high side or low side sensing.
High side sensing is when the shunt is placed between the V+ and load. (Figure 1A)
Low side is when it is placed after the load but before the ground point. (Figure 1B)
Add alt text
The problem with low side sensing is that itis incapable of detecting load shorts and also causes GND disturbances. The advantage being a low voltage to be sensed which can be done with a single-ended input.
High side sensing is a little more complex in terms of implementation but was a much better technique in our case.
Our BMS was supposed to handle high input voltages all the way up to 80VDC, this meant that the potential on one point of the shunt would be 80 Volts!
What we need now is to find a high side current sense amplifier that is capable of handling voltages up to 80V on its differential inputs while also having a high CMRR.
A BMS also needs to be able to monitor the charging current. This meant that the current sense amplifier would need to have bidirectional sensing capabilities as well.
This is where current sense amps like the INA200 series come in
Add alt text
In the diagram above, the chosen IC features differential inputs that can handle common mode inputs from -6 to 90V. Additionally, the PWM Rejection feature filters out any kind of noise in the current waveform which would have otherwise caused false overcurrent triggers.
There are many such ICs available on the market, the ultimate decision of which one to get depends on cost, availability and board space constraints.
This article sums up techniques of sensing current and comes up with a practical design example of a high side sensing application. Accurate current sensing, especially in noisy environments such as EVs is a challenging problem with many ways to approach it, some of which are shown in the article above.
For an example of a current sense amplifier, consider looking at the datasheet for the INA240
Jaideep Sharma, Lead Hardware Engineer, MakerMax Inc.