AN2678

Calibrating the M41T83 using analog calibration

9/16

These are typical values. The curve in Figure 4 may be shifted up or down, left or right for a

given crystal. Because the curve is non-linear, an increment of capacitance at one operating

point will not have the same effect as an equal increment at another operating point. Thus,

table-lookup methods cannot be used with analog calibration. Instead, an iterative

procedure is required to ensure accuracy.

2.2

Analog calibration algorithm

The analog calibration algorithm described herein uses a binary tree approach for

adjustment. It utilizes repeated measurements as described in the scheme above. With

each adjustment /measurement sequence, the RTC's oscillator is fine tuned with

progressively smaller changes of the internal load capacitance array via the analog

calibration register.

As shown in Figure 5, the procedure starts with zeroing out the analog calibration register,

then waiting a few milliseconds for the oscillator to stabilize. Next, the frequency

measurement is made of the FT signal. If it is faster than 512 Hz, load capacitance is added

to the RTC's oscillator. If it is slower than 512 Hz, capacitance is removed. And if the FT

signal is exactly 512 Hz, no further adjustment is required.

Figure 5.

Analog calibration algorithm

Set ACAL (012h) to 0

Wait for OSC to stabilize

FT?

FAST

SLOW

=512

DONE

Set ACAL (012h) to +4pF

Set ACAL (012h) to –8pF

FT?

FAST

SLOW

=512

DONE

Sub 4pF

Add 4pF

FT?

FAST

SLOW

=512

DONE

Sub 2pF

Add 2pF

FT?

FAST

SLOW

=512

DONE

Sub 1pF

Add 1pF

FT?

FAST

SLOW

=512

DONE

Sub 0.5pF

Add 0.5pF

FT?

FAST

SLOW

=512

DONE

Sub 0.25pF

Add 0.25pF

DONE

DONE

ai13999