Inside TI’s Multiaxial Approach to Coreless Current Sensing

Texas Instruments (TI) recently released what it calls the industry’s first multiaxial coreless Hall-effect current sensors for traction inverters, using a crosstalk-rejection algorithm to suppress crosstalk from adjacent phases, along with signal conditioning and compensation for temperature, frequency, and lifetime reliability (Figure 1). “We measure in an X- and a Z-field, so we’re able to make measurements in such a way that helps cancel out the crosstalk, and also helps cancel out vibration and movement,” said Jason Cole, Vice President and General Manager of the Sensing Business Unit at TI, in an exclusive interview with Power Electronics News, “As far as I know, we’re the only one to have made that [multiaxial] solution.” TI’s sensing business team partnered with Kilby Labs—TI’s advanced research center focused on high-risk, disruptive semiconductor technology—to help develop this solution.

Interference rejection

The solution of choice for measuring automotive busbar current uses a linear ambient Hall-effect sensor placed in the air gap of an iron, silicon steel, or ferrite core that surrounds the busbar (Figure 2). “A core normally acts as a shield,” said Cole.  Without a core, the sensor has to reject interference from adjacent phases that are only 120° out of phase in a three-phase system. “I call it a ‘noisy neighbor’ problem,” continued Jason, “because it’s like a neighbor knocking next door.” 

The cores add weight, cost, and mechanical complexity: “They’re expensive to place. They can be problematic. They’re cumbersome. They’re definitely not light,” said Cole. He also notes the complexities associated with their physical installation: “It’s in plastic housing, and you have to somehow mechanically attach it: holding it still, holding it in place, getting it around the busbar, putting it through, and bolting it down. It’s a very manual process.”

Removing the core’s size constraint gives engineers more freedom to shrink the traction inverter and improve power density. “If you don’t have to attach these things mechanically and figure out how to route them, you have more design freedom to design the entire inverter—so you can [potentially] shrink it down,” said Cole. 

Alternative current-sensing approaches

Outside of Hall-effect current sensing, shunt-based current sensing also exists—often using precision resistors in the current path and inferring current from the resulting voltage drop via Ohm’s law—as well as tunnel magnetoresistance (TMR) sensors that convert a magnetic field into voltage via quantum tunneling. The applications for each of these sensors are broad given their sensed current range, bandwidth, linearity, and accuracy. However, instead of traction inverter boxes, these types of sensors might be found in the automotive BMS, motor drives, OBCs, or battery disconnect units (BDUs). 

The commonly employed shunt-based current sensing also comes with its own mechanical integration considerations. Automotive OEMs have shifted from high-conductivity copper to aluminum busbars for weight and cost-effectiveness; however, sensing and switching components have retained copper terminations. While connecting copper to aluminum via fasteners is viable, it can raise reliability concerns, which has led to welding copper directly to aluminum. 

Cole briefly spoke on some of the considerations associated with shunt resistors for current sensing, as TI also offers these solutions: “If you go back to certain early EV designs, you’ll see them braze, solder, or weld the shunt in line with the actual busbar. This is sometimes still done today, and TI has a system solution for this as well.” He also mentioned how these copper terminals could introduce challenges: “Copper is a very good conductor but a terrible resistor, so it drifts, which makes getting an accurate measurement a challenge if just using a copper solution.” 

Cole continued, “If you measure current with a magnetic field, you can get a more accurate answer than just using copper, and it is easier than inserting temperature-stable metal in the busbar. But there are challenges with any way you try to make that measurement.” This is the core challenge with all automotive current-sense solutions: balancing performance (sense current, measurement error/precision, linearity, bandwidth, drift) with practical integration (reliability, installation, and cost). There are trade-offs with each approach. 

Hall-effect sensors in traction inverters

The current-sense solutions on the market that are uniquely suited to traction inverters are core-based or coreless Hall-effect sensors that can be arranged in various orientations (e.g., vertical differential or horizontal differential) to cancel interference from adjacent phases. Figure 3 shows how traction inverter boxes handled current sensing within previous specific EV models. 

These can be cored solutions with C-core Hall-effect sensors or over-the-trace Hall-effect sensors, where an additional U-core is used to improve immunity to stray fields and amplify the signal. As shown in Figure 4, coreless solutions typically require notches or cutouts within the busbar to locally increase current density, more accurately measure the magnetic field (B-field), and boost sensor sensitivity along either the x- or z-axis. 

This naturally comes with thermomechanical considerations, as notching lowers the conductor’s cross-section and raises its local resistance, creating a potential hotspot; shorter notches are typically preferred, and some vendors offer simulation tools to help plan notch geometry. Furthermore, movement between sensor and busbar throws off calibration in single-axis coreless designs; multiaxial sensing reduces this error. Cole elaborated on this: “Vibration causes what we call a displacement error, in that you can’t hold something perfectly still. When it moves up and down, it can cause an error in the reading, making the current level measure wrong. The fact that ours is multiaxial helps eliminate that error and bring it down.” For instance, the solution offers <1% displacement error at 0.4-mm movement and 0.25% at 0.1-mm movement. 

“You can put it directly on your main PCB and just route the busbar over the top or put it on a separate sensor board,” said Cole. Figure 5 shows an example installation. When asked about the sense currents, Cole replied, “You could see a steady-state of 200 A, no problem. But if you look at some of the specifications called out in the SAE documents, you can get as high as 1,700 A.” He noted that the 1700-A peak level wasn’t seen often: “You can definitely see a lot of current in sports cars, depending on the voltage [400 V or 800 V].” 

This solution does require preliminary magnetic simulation and busbar layout tuning: “You may want to put the sensor closer or farther away based on the sensitivity level you’re trying to achieve… There’s some trade-off that has to be made based on the distance you place it,” said Cole. System-level characterization is possible with the TMCS2100EVM evaluation module, which includes a clip and nickel-plated busbar where users can test up to three busbars. “We have a tool that we’ve built that’s available,” said Cole, referring to TI’s simulation tool to assist designers.  

Cover image: Adobe Stock

Reference

1. J. Reimers, L. Dorn-Gomba, C. Mak and A. Emadi, “Automotive Traction Inverters: Current Status and Future Trends,” in IEEE Transactions on Vehicular Technology, vol. 68, no. 4, pp. 3337-3350, April 2019, doi: 10.1109/TVT.2019.2897899

The post Inside TI’s Multiaxial Approach to Coreless Current Sensing appeared first on Power Electronics News.