Mailing Address: PO Box 11400 鈥?Tucson, AZ 85734 鈥?Street Address: 6730 S. Tucson Blvd. 鈥?Tucson, AZ 85706
Tel: (520) 746-1111 鈥?Twx: 910-952-111 鈥?Telex: 066-6491 鈥?FAX (520) 889-1510 鈥?Immediate Product Info: (800) 548-6132
an arduous task. There are a plethora of converters available
in the marketplace from a wide variety of manufacturers. The
organized by importance. This bulletin is intended to walk the
these factors and how to interpret those specifications.
The selection checklist can be broken up into two areas鈥?/div>
primary facts which cannot be compromised, and secondary
factors which may allow the designer some flexibility.
Primary
鈥?What is the required level of system accuracy?
鈥?How many bits of resolution are required?
鈥?What is the nature of the analog input signal?
鈥?How fast must the converter operate (conversion speed)?
鈥?What are the environmental conditions?
鈥?Is a track-and-hold circuit required?
Secondary
鈥?Does the system have multiple channels?
鈥?Should the reference be internal or external?
鈥?What
鈥?What
鈥?What
鈥?What
are the drive amplifier requirements?
are the digital interface requirements?
type of digital output format is required?
are the timing conditions?
if they are critical in the system鈥檚 performance. Understand the
test conditions of the specifications, making certain they closely
match the operating conditions in the digitizing system.
PRIMARY FACTORS
ACCURACY
The accuracy of the overall system must be considered. In
most cases, the A/D converter is only one element of the
digitizing system. For example, if the overall system accu-
racy requirement is 0.012% (12-bits), the A/D converter
usually needs to be more accurate, perhaps 0.006% (13-bits)
which would allow the analog front end an error budget of
one LSB (least significant bit). One good rule of thumb to
follow when selecting analog components for a digitizing
system is for each component to be five to ten times more
accurate than the total system accuracy. Resolution should
not be confused with accuracy鈥攖hese two terms are some-
what exclusive. Resolution simply refers to the theoretical
number of states (2
n
) which the analog input can be resolved
into, where
n
is the number of bits of resolution. Accuracy
indicates how close the converter comes to the theoretical
limits. For this section, accuracy will be examined in the
time domain, where specifications are static (DC). The four
key static accuracy specifications are differential nonlinearity
(DNL), integral nonlinearity (INL), offset error and gain
error. The ideal transfer function of a 3-bit A/D converter is
shown in Figure 1.
OFFSET ERROR
The transfer function of a 3-bit A/D converter with an
offset error of +1LSB is shown in Figure 2. Output code
is shown on the Y-axis and analog input amplitude is
shown on the X-axis. Offset error is the amount by which
the first code transition (from 000 to 001) deviates from
the ideal position at an input equivalent to 0.5LSB.
Another way to consider offset error is to examine the
distance between X-Y intercept on the graph and where
the straight line drawn through the actual transfer func-
tion intercepts the X-axis. Offset error will cause the
entire transfer function to shift along the Y-axis (output
code is the dependent variable). Offset error is commonly
expressed as LSBs, volts or percentage of full-scale
range (%FSR).
3
4
5
6
Not shown in the A/D converter checklist is the most
important element in the process. The first step is the
complete definition of the design objectives for the digitiz-
ing system. The converter checklist can be used as a guide-
line to insure all important factors are considered. The
objective is to find an A/D converter which fits into the
system design, not vice versa.
Secondly, it is very important once the converter selection
process has begun to have a complete understanding of the
specifications. Understand what the absolute requirements are,
and insist on guaranteed specifications for those parameters.
Although most reputable manufacturers鈥?typical specifications
have some measure of validity, never depend on those values
漏
1995 Burr-Brown Corporation
AB-098
Printed in U.S.A. April, 1995
1
2
3
4
5
6
7
8
9
10
11
12
