One area of new IC design where it is tough to get lots of functionality in a small space is industrial/automotive applications. The ambient temperature requirements make it difficult in general, especially for the automotive sector. Further, if you put 10 pounds of circuitry in a five-pound IC package, it will probably run pretty warm. Put that IC in a 125°C environment, and you'll find out pretty quickly just how good it really is.
Let's assume that the design team knows what it's doing and can create devices to work in this environment. What functionality do you need under the hood? I'm fond of control systems in general and controls for internal combustion engines in particular, so let's look at some of the pieces in typical setups.
The starting point (pun intended) is the ignition control. The simplest electronic ignition system would use a magnetic pickup to sense crankshaft position and a power transistor (FET or bipolar) to replace the points and drive the ignition coil primary. In between would be an amplifier stage, a low-pass filter (LPF) network, and a comparator. Put an adjustable reference voltage on the comparator, and you can tweak spark advance.
We can make that more sophisticated by adding a microcontroller unit (MCU) to tweak spark timing based on data from various sensors. The likely sensors needed measure engine temperature, ambient temperature, barometric pressure, air flow rate, water flow rate, and engine knock. We will need the circuitry to interface between the sensors and the MCU — amplifiers, LPFs, and ADCs.
If we need higher performance, we can beef up the spark by using a capacitive discharge topology. These charge up a capacitor and then dump it into the ignition coil primary with either an SCR or a large power FET.
There would still be a distributor in this setup functioning as a motorized high-voltage rotary switch. We can eliminate that by using multiple ignition coils. Some versions bring out both ends of the coil secondary and fire spark plugs in two cylinders simultaneously. In this scenario, one cylinder is approaching TDC for the power stroke, and one cylinder is in the exhaust stroke. This saves on the number of coils needed. The coil primaries are driven with multiple power transistors as above. The additional drive circuitry is trivial.
If the engine is fuel injected, we can drive the injector solenoids with more power transistors. Again, that MCU can control and adjust timing as needed. Regarding the handful of power transistors, these could be integral to the IC, but maybe that's not such a good idea.
The typical switching configuration here has the transistor as a low-side switch, so its drain or collector sees the battery voltage plus whatever load dump and ignition coil transients are present. In automotive systems, that's typically a lot. Surge specs range from 45V to 70V, depending on which section of which design standard you read.
If the transistors are integral to the IC, the IC needs to be built on a high-voltage process. That drives up the cost. It's cheaper to fab the chip on a 3.3V or 5V process and then add discrete power transistors.
You could put battery charging and status monitoring circuitry on the same device, but it's best to keep that separate. The functionality is different, and the voltage requirements are higher. As detailed above, put that on a separate IC.
Have you worked on any electronics for internal combustion engines? What problems did you encounter? How did you work around the problems?