AD526

REV. D

–12–

FLOATING-POINT CONVERSION

High resolution converters are used in systems to obtain high

accuracy, improve system resolution or increase dynamic range.

There are a number of high resolution converters available with

throughput rates of 66.6 kHz that can be purchased as a single

component solution; however in order to achieve higher through-

put rates, alternative conversion techniques must be employed.

A floating point A/D converter can improve both throughput

rate and dynamic range of a system.

In a floating point A/D converter (Figure 42), the output data is

presented as a 16-bit word, the lower 12 bits from the A/D

converter form the mantissa and the upper 4 bits from the digi-

tal signal used to set the gain form the exponent. The AD526

programmable gain amplifier in conjunction with the compara-

tor circuit scales the input signal to a range between half scale

and full scale for the maximum usable resolution.

The A/D converter diagrammed in Figure 42 consists of a pair

of AD585 sample/hold amplifiers, a flash converter, a five-range

programmable gain amplifier (the AD526) and a fast 12-bit A/D

converter (the AD7572). The floating-point A/D converter

achieves its high throughput rate of 125 kHz by overlapping the

acquisition time of the first sample/hold amplifier and the set-

tling time of the AD526 with the conversion time of the A/D

converter. The first sample/hold amplifier holds the signal for

the flash autoranger, which determines which binary quantum

the input falls within, relative to full scale. Once the AD526 has

settled to the appropriate level, then the second sample/hold

amplifier can be put into hold which holds the amplified signal

while the AD7572 perform its conversion routine. The acquisi-

tion time for the AD585 is 3

µs, and the conversion time for the

AD7572 is 5

µs for a total of 8 µs, or 125 kHz. This performance

relies on the fast settling characteristics of the AD526 after the

flash autoranging (comparator) circuit quantizes the input sig-

nal. A 16-bit register holds the 3-bit output from the flash autor-

anger and the 12-bit output of the AD7572.

The A/D converter in Figure 42 has a dynamic range of 96 dB.

The dynamic range of a converter is the ratio of the full-scale

input range to the LSB value. With a floating-point A/D con-

verter the smallest value LSB corresponds to the LSB of the

monolithic converter divided by the maximum gain of the PGA.

The floating point A/D converter has a full-scale range of 5 V, a

maximum gain of 16 V/V from the AD526 and a 12-bit A/D

converter; this produces:

LSB = ([FSR/2

µV. The

dynamic range in dBs is based on the log of the ratio of the

full-scale input range to the LSB; dynamic range = 20 log

(5 V/76

µV) = 96 dB.

74–

LS174

74–

LS174

74–

LS174

AD7572

LSB

MSB

1/6

56

+

+

+

+

10k

S/H

AD585

2.5MHz

+

+

1 s

1/6

3

4

1/6

1

2

AD526

B

F

S

A0 A1 A2

+

+

10k

S/H

AD585

74-

123

1/2

CLOCK

125MHz

LM339A

1/4

10k

10k

10k

10k

1/4

1/4

1/4

1/4

+5V

1/6

+

+15V

–15V

10 F

AD588

10 F

+15V

–15V

+5V

D12

D11

D10

D9

D8

D7

D6

D5

D4

D3

D2

D1

E1

E2

E3

68pF

68pF

BUSY

47 F

10 F

10 F

+15V

–15V

10 F

10 F

+15V

–15V

10 F

+5V

+5V

50k

30pF

+5V

10 F

+15V

–15V

10 F

+5V

10 F

10k

5k

2.5k

1.25k

1.25k

1 F

1

2

4

5

9

10

3

6

8

12

13

11

1

2

3

11

10

74ALS86

A0

A1

A2

NOTE: ALL BYPASS CAPACITORS ARE 0.1 F

Figure 42. Floating-Point A/D Converter