Pre-charge anti-spark MOSFET switch for Li-ion battery

Question

I am looking for some advice on the direction of this project, I am a mech eng, who likes to tinker with electronics as a hobby.

I want to build a on/off MOSFET switch for a 20s4p li-ion battery pack, including a pre-charge circuit to limit the inrush current and preventing sparking on the connectors.

Most of the examples I have found are only rated for 12s (50.4 V) batteries, but I need to switch ~84 V at 150 A continuous, ideally with the lowest possible leakage current in the off state to prevent draining the battery when not in use.

examples - https://github.com/msglazer/Anti-Spark_Switch

https://github.com/VinFar/High-Side-NMOS-Antispark-Switch

I would also like to implement a soft-start circuit to use a momentary push button, push to turn on, then push for 3 seconds to turn off.

Maybe also an auto switch off after 30 mins if no current has been drawn. I have looked at using an ATtiny with the sleep mode, which consumes very little power.

I am looking at using this Directfet - IRF7769L1TRPBF, putting a few in parallel to have a good headroom of current capability.

Any advise on how to design a circuit to switch 150 A at ~84 V and how to approach this would be great!

Edited with proposed circuit diagram without MCU control yet.

Edit 2

Replaced the toggle switch with push buttons. I found this LTC7001EMSE#PBF gate driver, which can work at over 100 V, so can I just use the main Batt voltage to drive the gate with this? It says it's a high side driver, will this work?

https://everycircuit.com/circuit/6035721594601472

Answer 1

Nothing wrong with your approach, but there is a reason we don't do it that way in the large battery industry: the standard solution is far simpler and it adds galvanic isolation (which your solution does not).

It uses two contactors, a precharge relay, and a precharge resistor (typically 10 Ω, 50 W).

Answer 2

I am looking at using this directfet - IRF7769L1TRPBF putting a few in parallel to have a good headroom of current capability.

I wouldn't recommend putting FETs in parallel if you can avoid it. I've done this when desperate but it can cause all kinds of problems with load sharing and makes your system less robust.

I need to switch ~84 v at 150 A continuous

Digikey has this 110A, 100V FET available to ship immediately for ~$4USD/each. Keep in mind if you use NMOS you will either need a gate driver to hook up to the high side of the load or to hook it up to the low side of the load. If you want a high side switch with no gate driver, you can use this PMOS (pricey but simplifies your system) with this circuit topology:

^{simulate this circuit – Schematic created using CircuitLab}

I would also like to implement a soft start circuit to use a momentary push button, push to turn on, then push for 3 seconds to turn off. Maybe also an auto switch off after 30 mins if no current has been drawn. I have looked at using an attiny with the sleep mode, which consumes very little power.

For soft start, this can be achieved using a hardware RC circuitry depending on the sophistication of the behavior you want there. For more sophisticated soft start, you may want to PWM into the RC. For some of these other features like 3s to turn off, 30min auto switch you are better of implementing those in software. The ATTiny is an excellent choice. I've shipped Li Ion products with them and you can get their consumption well below 20uA. They also have solid timing features that can wake them up from sleep to do tasks periodically like turn off your auto switch after 30mins.

Last but not least, whatever FET you select you'll have to make sure you have robust heat sinking. There are a lot of ways to approach this. If you use a TO-220 you'll have a through hole to install a heat sink. Since you're a mech E I'll trust you can figure that part out ;).

Edit: Adding circuit requested by OP with ability to turn on micro at t=0 w/ pushbutton (It can be done w/ RC but there's a trade-off of user ergonomics for each imo this is superior). Upon button press the micro starts up and holds up the NMOS to VCC+ until conditions are met such that the micro no longer wants to draw current and resets GPIO to 0V (commits suicide).

^{simulate this circuit}

Edit #2: Common Comparator Circuit w/ Hysteresis that's suited for this application. Simply voltage divide down the battery voltage, carefully tuning the reference against the input voltage and put the output into the ATTiny's digital input to sense when your Undervoltage threshold has been reached.

Answer 3

This is a non-trivial problem, and not one that can be answered here, unfortunately.

For perspective, consider the mechanical analogy.

At 150A, we're talking wire about 9mm dia. So say we have a shaft 9mm dia. It's spinning at 6000 RPM with 20 N.m torque. You have, well, about "this much" space to construct a clutch between shafts. You must use materials rated maximum 150°C; exceed this and... the oil burns, materials melt or gall together, whatever. How do you do it?

This is probably an unfair comparison, because the active area inside the transistors is much smaller than the active area of a clutch. Well, the clutch does experience surface friction, but it can be made of conductive materials like copper (at least, immediately behind an arbitrarily thing friction surface, if copper is unsuitable); it can also be directly fluid cooled (or indeed fluid as the active element, as in a viscous clutch), while transistors can't. On the other hand, there's likely more inertia in a fully mechanical system (shafts, bearings, gears..).

Point of reference, transistors can dissipate up to a few joules of energy (essentially the single avalanche rating), on time scales of 10s of µs. At long time scales (approaching steady state), up to some hundreds of watts -- if fully heatsunk. Somewhere inbetween lies the pulse width and peak power that this application will experience. So, maybe more than a few joules can be handled, and a peak power more than a few hundred watts, but somewhere they meet in the middle, you can't have unlimited of both.

You're probably precharging on the order of 10 or 100J of capacitors in the ESC, which needs a comparable dissipation in the transistors. So, it doesn't look good. And you certainly can't take too long to start (or stop), if it's under load at the same time (which is implied by giving the switching current): you then have to dissipate some large fraction of full load power, on top of charging the capacitors.

---

I have created a circuit for my own purposes, which is able to switch up to 30V 20A, using a single transistor, and can do so for an extended period of time, about 100ms -- that is, while dropping significant voltage, without overheating the transistor. (Another, more robust element dissipates the power, and a temperature sensor stops early if it overheats.) But this is much too complicated of a circuit to describe in an answer here (it uses high frequency switching), and scaling it up to higher power levels is even less trivial.

What you seek is possible, but it will require real engineering time to create.

Answer 4

I think it is possible to use several NMOS devices in parallel to provide a controlled pre-charge of the capacitors in the ESC from the 85V battery pack. The following simulation shows just two 100V 137A MOSFETs with 4.2 mOhm RgsOn. This circuit assumes 500 uF in the ESC, and the two MOSFETs in parallel exhibit a voltage drop of 315 uV at 150A. A simple DC-DC converter provides about 24 VDC for gate control, which is divided down to 12V by two 5k resistors, and a 1 uF capacitor provides about 200 uSec turn-on delay. If desired, a microcontroller could apply PWM to the gates with an optocoupler. Exact values will need to be adjusted as needed, consulting the SOA curves for the devices. Although the MOSFETs can handle 137 amps, it would be better to use four to six in parallel to avoid having to use very heavy copper traces on the PCB. And these MOSFETs are only about $3 each.