tween 0 and 3/16 of V

mator decoder instructs the comparator multiplexer to select

the eight tap points between 8/256 and 2/8 of V

nects them to the eight flash comparators. The first flash

conversion is now performed, producing the five MSBs of

data.

The remaining three LSBs are generated next using the

same eight comparators that were used for the first flash

conversion. As determined by the results of the MSB flash, a

voltage from the MSB Ladder equivalent to the magnitude of

the five MSBs is subtracted from the analog input voltage as

the upper switch is moved from position one to position two.

The resulting remainder voltage is applied to the eight flash

comparators and, with the lower switch in position two, com-

pared with the eight tap points from the LSB Ladder.

By using the same eight comparators for both flash conver-

sions, the number of comparators needed by the multi-step

converter is significantly reduced when compared to stan-

dard half-flash techniques.

Voltage Estimator errors as large as 1/16 of V

will be corrected since the flash comparators are connected

to ladder voltages that extend beyond the range specified by

the Voltage Estimator. For example, if 7/16 V

9/16 V

tap points below 9/16 V

decoded by the estimator decoder to “10”. The eight flash

comparators will be placed at the MSB Ladder tap points be-

5

each side of the Voltage Estimator’s span will automatically

correct an error of up to 16 LSBs (16 LSBs = 312.5 mV for

V

Voltage Estimator’s output code will be corrected by sub-

tracting “1”. This results in a corrected value of “01”. If the

first flash conversion determines that the input voltage is be-

tween 8/16 V

5

tor’s output code remains unchanged.

After correction, the 2-bit data from both the Voltage Estima-

tor and the first flash conversion are decoded to produce the

five MSBs. Decoding is similar to that of a 5-bit flash con-

verter since there are 32 tap points on the MSB Ladder.

However, 31 comparators are not needed since the Voltage

Estimator places the eight comparators along the MSB Lad-

der where reference tap voltages are present that fall above

and below the magnitude of V

Comparators are not

needed outside this selected range. If a comparator’s output

is a “0”, all comparators above it will also have outputs of “0”

and if a comparator’s output is a “1”, all comparators below it

will also have outputs of “1”.

2.0 DIGITAL INTERFACE

The ADC08161 has two basic interface modes which are se-

lected by connecting the MODE pin to a logic high or low.

2.1 RD Mode

With a logic low applied to the MODE pin, the converter is set

to Read mode. In this configuration (

Figure 1), a complete

conversion is done by pulling RD low, and holding low, until

the conversion is complete and output data appears. This

typically takes 655 ns. The INT (interrupt) line goes low at

the end of conversion. A typical delay of 50 ns is needed be-

tween the rising edge of CS (after the end of a conversion)

and the start of the next conversion (by pulling RD low). The

RDY output goes low after the falling edge of CS and goes

high at the end-of-conversion. It can be used to signal a pro-

cessor that the converter is busy or serve as a system Trans-

fer Acknowledge signal.

2.2 RD Mode Pipelined Operation

Applications that require shorter RD pulse widths than those

used in the Read mode as described above can be achieved

by setting RD’s width between 200 ns–400 ns (

Figure 5). RD

pulse widths outside this range will create conversion linear-

ity errors. These errors are caused by exercising internal in-

terface logic circuitry using CS and/or RD during a conver-

sion.

When RD goes low, a conversion is initiated and the data

from the previous conversion is available on the DB0–DB7

outputs. Reading DB0–DB7 for the first two times after

power-up produces random data. The data will be valid dur-

ing the third RD pulse that occurs after the first conversion.

2.3 WR-RD (WR then RD ) Mode

The ADC08161 is in the WR-RD mode with the MODE pin

tied high. A conversion starts on the rising edge of the WR

signal. There are two options for reading the output data

which relate to interface timing. If an interrupt-driven scheme

is desired, the user can wait for the INT output to go low be-

fore reading the conversion result (

Figure 3). Typically, INT

will go low 690 ns, maximum, after WR’s rising edge. How-

ever, if a shorter conversion time is desired, the processor

need not wait for INT and can exercise a read after only 350

ns (

Figure 2). If RD is pulled low before INT goes low, INT

will immediately go low and data will appear at the outputs.

This is the fastest operating mode (t

sion time, including data access time, of 560 ns. Allowing

100 ns for reading the conversion data and the delay be-

tween conversions gives a total throughput time of 660 ns

(throughput rate of 1.5 MHz).

2.4 WR-RD Mode with Reduced Interface System

Connection

CS and RD can be tied low, using only WR to control the

start of conversion for applications that require reduced digi-

tal interface while operating in the WR-RD mode (

Figure 4).

Data will be valid approximately 705 ns following WR’s rising

edge.

3.0 REFERENCE INPUTS

The ADC08161’s two V

define the zero to full-scale input range of the A to D con-

verter. This allows the designer to vary the span of the ana-

log input since this range will be equivalent to the voltage dif-

ference between V

outputs that minimum output voltages above GND can also

be compensated by connecting V

equal to this minimum voltage. By reducing V

V

can be increased (i.e., if V

The reference arrangement also facilitates ratiometric opera-

tion and in may cases the power supply can be used for

transducer power as well as the V

operation is achieved by connecting V

necting V

The ADC08161s accuracy degrades when V

less than 2.0V.

The voltage at V

digital output of all zeroes. Through V

www.national.com

12