To find your CPU's temperature sensor you can leave your system idle for a few minutes, to let temperatures drop, and then go to 100% usage for a while. The temperature that rises faster is the one you're searching for. Other available temperature readings usually come from your sensor chip itself, from the southbridge, the voltage regulator, or even from an additional probe placed under the processor. This additional temperature sensor is not necessarily a duplicate. Some CPUs are not actually able to report the internal temperature from their die. To be able to read their temperatures, an additional external sensor (thermocouple) is used. In such cases, you will see two temperatures referring to the processor. The higher of the two is from the die. As a final note, please remember that not all available temperature sensors are actually connected to something. If you happen to read unusually high or low temps, they are likely to be from a disconnected (unused) temperature sensor.

Hardware sensor chips are generic devices that can be used to measure voltages from anywhere. The measured voltage must be converted to the range required by the sensor chip. Standard monitoring chips specify which external circuitry must be used in order to measure voltages outside some range. Voltages like 12V, -5V, -12V and some others need this external circuitry. Some manufacturers chose not to follow datasheets. If this is your case, then you will read unusual values from SpeedFan. Since this custom circuitry is not known, SpeedFan does not try to "guess" it as any reading wouldn't be safe. If you get really odd voltage readings, simply enter configuration and uncheck the relevant ones.

This can occasionally happen if you have an nVidia video card and you are using some software combination. Due to the lack of documentation, the only option is to use the /NONVIDIAI2C command line switch when running SpeedFan. You can find further info about it in

S.M.A.R.T. (or SMART) is a technology built into recent hard disks. If enabled, it lets the end user query the hard drive about its health and, eventually, performances. The end user (you) can access such info by using some specific software. SpeedFan can query those info for you. SMART reports, for example, if the hard disk has been used for too much time (thus reaching its expected end of life), or if it took too many attempts to start spinning, or even if too many read errors occurred. About every modern hard disk can report its temperature this way. They do have a temperature sensor that might be located inside the enclosure or somewhere outside it.
Please, note that SF might inform you that a reboot is needed in order to support SMART on your PC.

I asked for datasheets to some manufacturers, but with slow or no answer at all. Anyhow: since version 3.00 I'm ready to, easily, add support for new chips. I've got several chips to add. I will start adding detection for them: if you will get a message saying "PLEASE REPORT", please contact me and I will add support for it in a few days :-)

Even though the sensor chip might be able to change fan speeds, the hardware manufacturer still has to include some external circuitry. This is not always included. I'm able to know if a motherboard can change fan speeds, but not if it is unable for sure. You might try to contact the hardware manufacturer and let me know.

There are several ways to label available readings (temperatures, voltages, fan speeds). The first source should be the BIOS. Enter BIOS at boot, write down labels and readings and compare them to those reported by SpeedFan. You can use manufacturer's custom hardware monitors to match readings too. SpeedFan strictly adheres to available datasheets for each sensor chip. Please remember that hardware monitors chips have some pins (small connectors) that should be connected to some additional hardware (temperature probes, thermistors or thermocouples) in order to be able to read temperatures. Only a few hardware monitor chips do label their connectors with "CPU", "System" and the like. Most of them use labels like "Temp1", "Local" or "Remote". Hardware manufacturers connect available pins to different temperature sensors basically according to the physical placement of components on the motherboard. This means that the same chip, an ITE IT8712F, for example, might be connected to a sensor diode measuring CPU temperature on Temp2 and, on a different hardware, it might be connected on Temp1. If you have a "Local" sensor and a "Remote" labeled one, this usually means that "Local" is the temperature of the monitor chip itself and "Remote" is the temperature read from a "remote" probe. When you have properly identified which temperature sensor is which, try to lower the speed of each fan and look at reported speed and temperatures. This way you can match PWM controls (speeds) with fans. Please, note that if you do not allow SpeedFan to change any fan speed and set all the speeds too low, then SpeedFan won't be able to avoid overheating.