
ADT7462
Rev. 2 | Page 47 of 88 | www.onsemi.com
Dynamic TMIN Control Programming
Because the dynamic TMIN control mode is a basic extension of
the automatic fan control mode, the automatic fan control mode
parameters should be programmed first (see Step 1—Configuring
the MUX through Step 8—Operating Points for Temperature
Channels). Then proceed with dynamic TMIN control mode
programming.
STEP 8—OPERATING POINTS FOR TEMPERATURE
CHANNELS
The operating point for each temperature channel is the
optimal temperature for that thermal zone. The hotter each
zone is allowed to be, the quieter the system, because the fans
are not required to run as fast. The ADT7462 increases or
decreases fan speeds as necessary to maintain the operating point
temperature, allowing for system-to-system variation and
removing the need for worst-case design. If a sensible operating
point value is chosen, any TMIN value can be selected in the
system characterization. If the TMIN value is too low, the fans run
sooner than required, and the temperature is below the
operating point. In response, the ADT7462 increases TMIN to
keep the fans off longer and to allow the temperature zone to
get closer to the operating point. Likewise, too high a TMIN value
causes the operating point to be exceeded, and in turn, the
ADT7462 reduces TMIN to turn the fans on sooner to cool the
system.
Programming the Operating Point Registers
There are two operating point registers, one for the Remote 1
temperature channel and one for the Remote 2 temperature
channel. These 8-bit registers allow the operating point
temperatures to be programmed with 1°C resolution.
Operating Point Registers
Register 0x5A, Remote 1 Operating Point = 0xA4 (100°C default)
Register 0x5B, Remote 2 Operating Point = 0xA4 (100°C default)
Operating Point Hysteresis Register
The operating point hysteresis register sets the value below the
operating point at which TMIN begins to reduce.
Register 0x64, Bits [7:4] Operating Point Hysteresis = 0x40
(4°C default)
STEP 9—HIGH AND LOW LIMITS FOR
TEMPERATURE CHANNELS
The low limit defines the temperature at which the TMIN value
starts to be increased, if temperature falls below this value. This
has the net effect of reducing the fan speed, allowing the system
to get hotter. An interrupt can be generated when the tempera-
ture drops below the low limit.
The high limit should be set above the operating point but
below the critical THERM point. An interrupt can be generated
when the temperature rises above the high limit.
How Dynamic TMIN Control Works
The basic operation of dynamic TMIN control is as follows:
1.
Set the target temperature for the temperature zone, which
could be, for example, the Remote 1 thermal diode. This value
is programmed to the Remote 1 operating point register.
2.
As the temperature in that zone rises toward and exceeds
the operating point temperature minus hysteresis, TMIN is
reduced and fan speed increases.
3.
As the temperature drops below the low limit value, TMIN is
increased and the fan speed is reduced.
Short Cycle and Long Cycle
The ADT7462 implements two loops: a short (or decrease)
cycle and a long (or increase) cycle. The short cycle takes place
every n monitoring cycles. The long cycle takes place every 2n
monitoring cycles. The value of n is programmable for each
temperature channel. The bits are located at the following
register locations.
Dynamic TMIN Control Register 2 (0x0C)
Bits [2:0] (CYR1) = Remote 1
Bits [5:3] (CYR2) = Remote 2
Table 29. Cycle Bit Assignments
Code
Short Cycle
Duration
Long Cycle
Duration
000
8 cycles
1 sec
16 cycles
2 sec
001
16 cycles
2 sec
32 cycles
4 sec
010
32 cycles
4 sec
64 cycles
8 sec
011
64 cycles
8 sec
128 cycles
16 sec
100
128 cycles
16 sec
256 cycles
32 sec
101
256 cycles
32 sec
512 cycles
64 sec
110
512 cycles
64 sec
1024 cycles
128 sec
111
1024 cycles
128 sec
2048 cycles
256 sec
The cycle time must be chosen carefully. A long cycle time
means that TMIN is updated less often. If a system has very fast
temperature transients, the dynamic TMIN control loop is always
lagging. If a cycle time that is too short is chosen, the full
benefit of changing TMIN is not realized and TMIN needs to
change again on the next cycle. In effect, it is overshooting. It is
necessary to carry out some calibration to identify the most
suitable response time.
Figure 69 shows the steps taken during the short cycle.