AN780

© 2002 Microchip Technology, Inc.

DS00780A-page 6

REFERENCE VOLTAGE CALCULATION

Now that the timing has been determined, eq3 can be used to

calculate the reference voltage:

= 1V 200mS/490mS

The reference voltage does not need to be calculated very precise

since it will have to be trimmed during calibration. A

±25% adjust-

ment range is enough to make up for just about any minor

calculation error.

MICROPROCESSOR PROGRAMMING

The PIC16C62A 8-Bit microcontroller was selected but any

reasonable processor/controller will suffice. The PIC16C62A is a

28-pin part that has EPROM programmability.

CONCLUSION

The scale works extremely well. The 8X oversampling makes it

very smooth and noise-free. The response time is within one

conversion (

less than 2 grams are accumulated in an integrating register until

it gets to either +1 gram or –1 gram. When this happens, the current

conversion is allowed to "get through" and a new base is estab-

lished in the accumulator.

There is also a facility in the programming that allows the raw data

to be displayed. These displays show the full 17-bit conversion

results. The basic converter noise is as predicted, typically 1 to 2

counts of flicker (16-bit accuracy) with an intermittent jump of

about 3 or 4 counts (1/f noise). One count is equivalent to 1/8 gram.

The actually 60Hz power line rejection ration of the TC500A was

not measured, but judging from the 6 to 8 counts of "rolling noise"

The effect of the differential ratiometric reference was tested by

changing the supply voltage from +4V to +6V. Although there was

a 1 – 2 second delay due to unmatched time constants between

the reference and the strain gauge, the final readings were exactly

the same. This shows that the power supply rejection is better

than 100dB.

Read Format

OUT

LOAD

MSB

READ

LSB

Load Format

OVR POL MSB

LSB

FIGURE 5: Serial interface protocol.