Table 35. Design Requirements

8.2.2.1 Programming the Calibration Register

For this example, assuming a peak current of less than 15 A, the Current_LSB is calculated to be 457.7 μA/bit using Equation 2. Selecting a value for the Current_LSB of 500 μA/bit or 1 mA/bit significantly simplifies the conversion of the returned value from the READ_IN and READ_PIN commands to amperes and watts. For this example, a value of 1 mA/bit is chosen for the Current_LSB. Using this value for the Current_LSB does trade a small amount of resolution for having a simpler conversion process on the user side. Using Equation 1 in this example with a Current_LSB value of 1 mA/bit and a shunt resistor of 2 mΩ results in a MFR_CALIBRATION register value of 2560d (or A00h).

8.2.2.2 Calculating PMBus Coefficients

The m, b, and R coefficients are fixed for bus voltage measurements returned by the READ_VIN and READ_VOUT and are available from Table 1.

For current and power measurements, the value for the m and R coefficients must be calculated. For current measurements returned by the READ_IIN and READ_IOUT commands, the value for m is calculated by inverting the Current_LSB used to set the MFR_CALIBRATION register and shifting the decimal location if needed to minimize rounding errors. In this example, using the Current_LSB of 1 mA/bit, the value of m is calculated to be 1000. Shifting the decimal location does not obtain higher accuracy because the value for m is a whole number. The value for R in this example is 0 because the decimal location for the value of m does not need shifting.

The POWER_LSB value is 25 times the value of the CURRENT_LSB, therefore, the value for m is reduced by a factor of 25. For this example, the value for the m power coefficient 1000 / 25 or 40. For this case, the R coefficient is also 0 because m is a whole number. If m is not a whole number, then shifting the decimal place is advantageous to reduce rounding errors while keeping the value between –32768 and 32767. Decimal shifts to the right result in negative values for R and shifts to the left result in positive values; the number of shifts is the absolute value of R.

The value of 0 can be used for b for both current and power measurements with very little loss in accuracy because the offset for power and current measurements is very low. The m, b, and R coefficients are fixed for bus voltage measurements returned by the READ_VIN and READ_VOUT and are available from Table 1.

8.2.2.3 Programming Warning Thresholds

Warning thresholds are set by converting the warning value from volts, amperes, and watts to the appropriate digital word using Equation 4 with the correct values for m, b, and R. For example, to set a bus voltage overvoltage warning at 5.5 V with the VIN_OV_WARN_LIMIT command, the correct value to write with this command is 4400d or 1130h. The least significant last three bits of the 16-bit word are hard coded to 0 because the warning thresholds only have 12 bits of effective resolution. For this example there is no change to what is written in the VIN_OV_WARN_LIMIT register because the last three bits are already zero. To set an overcurrent warning level at 15 A with the IIN_OC_WARN_LIMIT command, the correct value to write to the device is 15000d (or 3A98h).

8.2.2.4 Calculating Returned Telemetry Values

When the value for the m, b, and R coefficients are known, returned values can be translated to volts, amperes, or watts by using Equation 3 with the calculated m, b, and R coefficients. Alternatively, returned values can be calculated by multiplying the returned code by the corresponding LSB size as discussed in the Calibration Register and Scaling section.