Here are some ways for a microcontroller to produce analog signals:
- Built in D/A. Some have these, although they are not common.
- Filtered PWM. If your bandwidth * resolution requirements are low enough, this is a easy way to get analog signals. Many microcontrollers have PWM generation hardware built in, and you can trade off the resolution with the PWM period. A couple of poles of passive R-C low pass filtering after the PWM is usually good enough to attenuate the PWM frequency below the level you care about.
Advantages of PWM are that it is low cost, largely built into the micro, linear, no "glitch energy", and doesn't suffer from uneven steps at major powers of 2.
- External D/As. There are chips that do this over various digital busses, like SPI, IIC, IIS and other interfaces. This will give you a higher bandwidth * resolution product than filtered PWM.
There are many chips out there. Which one is appropriate for you depends on the resolution, bandwidth, and accuracy you need, and money you are willing to spend.
Each of these schemes produces independent analog signals per channel. A 8 channel D/A chip, for example, will be able to drive each channel with a independent signal.
However, you should step back and examine whether you really need analog signals. What will they be used for? Many things are more easily and more efficiently done with discrete pulses. If you want to control a motor, brightess-adjust LEDs, etc, most likely it is better to drive them with pulses. Even audio is commonly handled with pulses right up to the speaker, which is what class D amplifiers are about.