Disturbing DAC output short-term glitch impulse interference

By: Bonnie Baker, Senior Application Engineer, WEBENCH Design Working Group, Texas Instruments (TI)

 

在DAC基础知识：静态技术规格中，我们探讨了静态技术规格以及它们对DC的偏移、增益和线性等特性的影响。这些特性在平衡双电阻 (R-2R) 和电阻串数模转换器 (DAC) 的各种拓扑结构间是基本一致的。然而，R-2R和电阻串DAC的短时毛刺脉冲干扰方面的表现却有着显著的不同。
 

We can observe that the DAC's dynamics are not linear when running at the operating sampling rate. There are many causes of dynamic nonlinearity, but the most influential are short-term glitch impulse interference, slew rate/settling time and sampling jitter.
 

The user can observe short duration glitch impulse interference while the DAC is operating across its output range at a stable sampling rate. Figure 1 shows this phenomenon on a 16-bit R-2R DAC, the DAC8881.
 

figure 1

 

This 16-bit DAC (R-2R) output shows the characteristics of a short glitch impulse disturbance during code changes from 7FFFh – 8000h.
 

In the end what happened?
 

Ideally, the output of a DAC moves in the expected direction from one voltage value to the next. However, in reality, DAC circuits have undershoot or overshoot characteristics during certain code-to-code transitions.
 

This characteristic is not consistent for every code-to-code transition. The undershoot or overshoot characteristics produced on some transitions are more pronounced than on others. These characteristics are quantified by the glitch impulse specification. A DAC glitch impulse can momentarily output an erroneous voltage that disturbs the closed-loop system.
 

An example of a DAC with a single, sudden glitch impulse is shown in Figure 2. This type of glitch impulse is commonly generated by a resistor string DAC.
 

figure 2

 

Single-shot DAC output short-time glitch pulse interference characteristics.

 

In Figure 2, the code transitions from 7FFFh to 8000h. If you convert these numbers to binary, notice that each bit of the two hexadecimal codes either transitions from 1 to 0 or from 0 to 1.
 

The Glitch Impulse specification quantifies the energy of this glitch phenomenon in nanovolt seconds, or nV-sec (GI). The amount of this glitch impulse is equal to the area under the curve.
 

Single-shot short-duration glitch impulse interference is caused by the desynchronization of the DAC's internal switches. So what causes this DAC phenomenon? The reason is that the synchronization of the internal DAC switches is not always so precise. As the integrated switch capacitors charge or discharge, you can see these charge exchanges at the DAC's output.
 

The R-2R DAC generates two regions of glitch error (Figure 3). Because of the double-pulse error, the positive glitch (G2) is subtracted from the negative glitch (G1) to produce the final glitch specification.
 

image 3

 

A DAC with an R-2R internal structure exhibits a double-spurt short-duration glitch impulse interference

 

The code transition in Figure 3 is still from 7FFFh to 8000h.
 

In order to understand the source of DAC glitch impulse interference, we must first define a major carry transition. At a major carry transition point, when the most significant bit (MSB) changes from low to high, the lower bits change from high to low, or vice versa. An example of such a code transition is 0111b to 1000b, or the more obvious change from 1000 000b to 0111 1111b.
 

Some might think that this phenomenon occurs when the output of the DAC exhibits large voltage changes. In fact, this is not the case with every DAC coding scheme. See Reference 1 for more details.
 

Figures 4 and 5 show the effects of this type of glitch on an 8-bit DAC. To the DAC user, this phenomenon appears at a single least significant bit (LSB) step size, or in a 5V, 8-bit system, at a 19.5mV step size.
 

Figure 4

 

In this 8-bit DAC configuration, the internal switch has 7 R-2R pins connected to VREF and 1 R-2R pin connected to ground.
 

Figure 5

 

In this DAC configuration, the internal switch has 1 R-2R pin connected to VREF and 7 R-2R pins connected to ground.
 

There are two areas where output glitches occur when the DAC is loaded with code: switch synchronization and switch charge transfer, which triggers multiple switches simultaneously.
 

This resistor string DAC has a single switch topology. A resistor string DAC tap is connected to different points in the large resistor string. The switch network does not require multiple transitions on the major carry, so the potential for glitch pulses is low. The switch charge will produce a small glitch pulse, but it is insignificant compared to the glitch pulse generated by the R-2R structure DAC.
 

During code transitions, the R-2R DAC has multiple switches switching simultaneously. Any loss of synchronization results in a brief period of all switches high or all switches low, causing the DAC's voltage output to migrate to the rails. The switches then resume, creating a single, short-duration glitch in the opposite direction. The output then settles.
 

The voltage locations of these glitches are quite predictable. When using an R-2R DAC, the worst case scenario for a glitch error is when all digital bits are switching while still transitioning with a small voltage change. In this case, the DAC code changes with a major carry transition; from code 1000… to 0111…
 

Check the actual DAC operation status
 

Now that we have defined the alternative code transitions for glitch impulse errors, we can take a closer look at the R-2R and resistor string DAC glitch impulses for the 16-bit DAC8881 (R-2R DAC) and the 16-bit DAC8562 (resistor string DAC).

In Figure 6, the DAC8881 has a short glitch impulse interference of 37.7 nV-sec, while the DAC8562 has a short glitch impulse interference of 0.1 nV-sec. In both figures, the x-axis scale is 500ns/div, and the y-axis scale is 50mV/div.
 

Figure 6

 

R-2R and resistor string short-time glitch pulse interference performance 
 

The glitch pulse disappeared
 

If DAC glitch interference is an issue, users can use external components to reduce the glitch amplitude (Figure 7a) or eliminate the glitch energy entirely (Figure 7b.)
 

Figure 7

 

Use a first-order low-pass filter (a) or a sample/hold solution (b) to reduce short glitch errors.
 

An RC filter after the DAC reduces the glitch amplitude (Figure 7a). The period of the glitch determines the appropriate RC ratio. The 3dB frequency of the RC filter is ten times ahead of the glitch frequency. When selecting components, make sure the resistor value is low, otherwise it will produce a voltage drop with the resistive load. Since the glitch energy is never lost, the cost of implementing a single-pole low-pass filter is that the error is spread over a longer period of time while the settling time is increased.
 

The second approach uses a sample/hold capacitor and amplifier (Figure 7b). The external switch and amplifier eliminate the glitch pulses generated by the internal switch of the DAC, resulting in a smaller sample/hold (S/H) switching transient. In this design, the switch remains open during the entire major carry conversion of the DAC. Once the conversion is complete, the switch closes, setting the new output voltage on the CH sampling capacitor. This capacitor continues to hold the new voltage when the external switch opens when the DAC is ready to update its output. This solution is more expensive and takes up more board space, but it can reduce/eliminate the glitch pulse without increasing the settling time.
 

in conclusion
 

Glitch impulse interference is a very important dynamic nonlinear DAC characteristic that you will encounter when the device is running at the operating sampling rate. However, this is just the tip of the iceberg. Other factors that affect high-speed circuits are conversion rate and settling time. Stay tuned for the next article on this topic.
 

references
 

1. “Encoding Schemes Used by Data Converters, Jason Albanus, Application Journal (AN-175), Texas Instruments
 

2. “De-glitching a DAC Using an RC Filter”