•  
• EDN Europe MAGAZINE
•  
• Free SUBSCRIPTION
• Subscription Renewal
• View Media Kit

• EDN
• EDN Australia
• EDN Asia
• EDN China
• EDN Japan

DAC fine-tunes reference output

Data converters must have a stable reference voltage to accurately measure or generate analog signals. Such references offer many guaranteed levels of precision and stability. Their variety of output-voltage levels is much smaller, which manufacturers specify as standard values, such as 2.048, 2.500, or 4.096V. You sometimes need to dynamically calibrate the reference, fine-tune its output value, or generate a slightly different value. For instance, when you measure a voltage with a resistive divider, you could adjust the reference voltage to compensate for an error in the divider.

You can adjust any three-terminal voltage referenceusing a resistor, a current sink/source, and a buffer amplifier (Figure 1). Sinking or sourcing current causes theresulting voltage drop across R₁ to subtract from or add to the nominal reference output, V_REF :

V_OUT=V_REF-I×R₁ .

The buffer amplifier isolates V_OUT from the load, so the only current flowing through R₁ is that from the current source. You can implement this idea with a highly stable voltage reference, a current DAC, and a low-offset op amp (Figure 2). If you choose a value of 4.7 kO for R_FS1, the full-scale DAC current is 0.981 mA, as the data sheet states. With a value of 10O for R₁, this current value yields a tuning range of ±0.981 mA×10O=±9.81 mV, divided into 31 steps of 0.654 mV each.

Depending on the performance grade of the reference and its package, the initial output accuracy can be as high as ±0.02%. The DAC's output-current accuracy is only ±6%, but the tuning range is small, so the large tolerance has only a small effect on the output error. Combining these values with a 1% resistor tolerance for R₁ and the maximum offset value for the op amp yields the following equation for the maximum initial output-voltage error:

0.0002×V _REF +√(0.06² +0.01²)×|I _DAC |×R₁ +V_OFFSET.

If you use a 2.048V reference, the following equation calculates the error:

0.4096 mV+0.5969 mV×|DAC|/15 +10 μV,

where DAC represents the DAC's decimal equivalent-output value (-15=DAC=15). Thus, the DAC introduces a maximum error of 0.5969 mV, yielding a total of roughly 1 mV when you combine it with the initial accuracy of the voltage reference itself.

Because the DAC has an operating-temperature range of -40 to +85°C, you use the voltage reference's drift specification in that same range, ±3 ppm/°C. IC makers often use the box method to specify temperature drift (Reference 1). You can then calculate this maximum reference-voltage drift over the temperature range of -40 to +85°C:

125°C×±3 PPM/°C×2.048V=±0.768 mV.

The drifts in the DAC, R_FS1, and of R₁ cause the drift in the second term (I_DAC ×R₁). The IC vendor specifies drift in the DAC as a typical value of ±75 ppm/°C. You assume ±25 ppm/°C for the resistors. These values yield a typical drift:

√[(125°C×±75 PPM/°C)²+(125°C×±25 PPM/°C)²+(125°C×±25 PPM/°C)²] ×I _DAC×R₁=±0.102 mV×DAC/15 .

The DAC and the resistors typically introduce only roughly ±0. 1 mV of drift, which is substantially lower than the maximum drift of the voltage reference. The IC vendor specifies the op amp's maximum input offset over temperature as 25 µV—also much lower than the maximum drift of the voltage reference. You can examine the DAC's output voltage as a function of its input code, using error bars to indicate the i ni ti al accuracy and temperature drift (Figure 3). The error increases slightly for the higher DAC values, mostly due to temperature drift. The meas- ured values are at room temperature and are close to the theoretical values.

REFERENCES

Fry, David, "Calculating the error budget in Precision Digital-to-Analog Converter (DAC) Applications," Application Note 4300, Maxim Integrated Products, Sept 25, 2008, http://bit.ly/rte0TR.