11

LTC1588/LTC1589/LTC1592

1588992fa

APPLICATIO S I FOR ATIO

Op Amp Selection

Because of the extremely high accuracy of the 16-bit

LTC1592, careful thought should be given to op amp

selection in order to achieve the exceptional performance

of which the part is capable. Fortunately, the sensitivity of

INL and DNL to op amp offset has been greatly reduced

compared to previous generations of multiplying DACs.

Tables 2 and 3 contain equations for evaluating the effects

of op amp parameters on the LTC1592’s accuracy when

programmed in a unipolar or bipolar output range. These

are the changes the op amp can cause to the INL, DNL,

unipolar offset, unipolar gain error, bipolar zero and bipo-

lar gain error. Tables 2 and 3 can also be used to determine

the effects of op amp parameters on the LTC1589 and the

LTC1588. However, the results obtained from Tables 2

and 3 are in 16-bit LSBs. Divide these results by 4

(LTC1589) and 16 (LTC1588) to obtain the correct LSB

sizing.

Table 4 contains a partial list of LTC precision op amps

recommended for use with the LTC1592. The easy-to-use

design equations simplify the selection of op amps to meet

the system’s specified error budget. Select the amplifier

from Table 4 and insert the specified op amp parameters

in Table 3. Add up all the errors for each category to

determine the effect the op amp has on the accuracy of the

LTC1592. Arithmetic summation gives an (unlikely) worst-

case effect. A root-sum-square (RMS) summation pro-

duces a more realistic estimate.

Op amp offset will contribute mostly to output offset and

gain error and has minimal effect on INL and DNL. For the

LTC1592, a 250

µVopampoffsetwillcauseabout0.65LSB

INL degradation and 0.15LSB DNL degradation with a 10V

full-scale range (20V range in bipolar). For the LTC1592

programmed in a unipolar mode, the same 250

µV op amp

offset will cause a 3.3LSB zero-scale error and a 3.3LSB

gain error with a 10V full-scale range.

CS/LD goes high, the mode changes and the DAC output

goes to a value corresponding to the data code.

Examples using the LTC1592:

1. Using a 24-bit loading sequence, load the unipolar

range of 0V to 10V with the DAC output at zero volt:

a) CS/LD

b) Clock SDI = 1001 XXXX 0000 0000 0000 0000

2. Using a 24-bit loading sequence, load the bipolar range

of

±5V and the DAC output at zero volt:

a) CS/LD

b) Clock SDI = 1010 XXXX 1000 0000 0000 0000

3. Using a 32-bit load sequence, load the bipolar range of

±10V with the DAC output voltage at 5V initially. Then

change the DAC output to –5V:

a) CS/LD

b) Clock SDI = XXXX XXXX 1011 XXXX 1100 0000 0000

0000

Next, the bipolar range of

±10V is retained and the DAC

a) CS/LD

b) Clock SDI = XXXX XXXX 0010 XXXX 0100 0000 0000

0000

OPERATIO