For low-voltage systems, PECL logic levels are referenced to the positive rail (V_CC), which is 5V in some systems, +2.5V or +3.3V in other systems. A problem with low-voltage design is sustaining high current levels with a high fanout loading.

The PECL used in high-speed telecommunications requires odd supply voltages: a positive VCC of +3.3 V, and a termination voltage (V_TT) equal to V_CC ' 2 V = +1.3 V. The V_TT supply is regulated with respect to V_CC and must be able to sink current.

The PECL interface can be modeled as a voltage follower added to the differential pair. The PECL output is more of a constant voltage source as the impedance seen looking into the output is reduced by the Beta of the voltage follower transistor (Figure 1). The voltage source of the PECL output can be modified to maintain the output impedance at 50 ohms and the output common voltage near 1.2V.

Figure 1. The PECL interface can be modeled as a voltage follower added to the differential pair. The PECL output is more of a constant voltage source as the impedance seen looking into the output is reduced by the Beta of the voltage follower transistor.

Click to Enlarge

This is done in LVDS by adding a current source and resistor in each emitter circuit of the voltage follower. The emitter current is then adjusted so that the impedance seen looking into the output will be 50 ohm. A feedback loop senses the common mode voltage and controls it to be near a reference voltage.

Without that feedback loop, a separate regulator may be required. Most positive low-dropout (LDO) regulators can't sink current, though negative LDOs are designed for that purpose. An earlier PECL-termination circuit (see Maxim app note 3198) used a current-sinking negative LDO modified for positive-voltage operation. The GND pin was connected to V_CC, and the input (IN) was connected to ground. Those connections let the negative LDO operate as a positive-voltage sink in which the voltage at V_SET equals V_CC ' 1.25 V.

But this circuit offered an output-current capability of 400mA ï¾ enough to terminate about 14 differential pairs of PECL outputs. That circuit is revisited here on behalf of designers who require more than 400mA. Adding a single resistor increases the available output current by 300 percent, and adding a transistor boosts the output capability to 4A.

Adding R_sink to the original circuit (Figure 2) increases the current-sinking capability as follows. Knowing the output voltage is 1.3V, set the resistance to sink current at the lower end of the desired 400mA range. If the mid-range current is 1A, for example, set the resistance to sink 800mA (1.3V/0.8A = 1.625 ohms), which allows the output current to range from 0.8A to 1.2A. Make sure the resistor is rated to handle the power. This method works well if the desired output-current range is known and is no more than 400mA. (If I_OUT is less than 0.8A the IC will go out of regulation, and if more than 1.2A it will go into current-limit-protection mode, which also results in a loss of regulation.)

Figure 2. Adding one resistor to the basic circuit shifts the output-current range higher, but the range remains only 400mA: from 0.8A to 1.2A, for example.

Click to Enlarge

A better method for increasing the output current is to add a pnp transistor at the output (Figure 3). Pin 5 of the IC sinks base current from the transistor, which maintains regulation by sinking current via the collector. Current is limited by the transistor characteristics. With the components shown (including a TIP42B pnp power transistor) the maximum output current is 4A. The Figure 2 circuit (unlike that of Figure 1) remains in regulation up to the maximum level of current. Choose a high-gain transistor for high output current, but avoid Darlington transistors: the VBE of a Darlington is about 1.4V, which is more than the 1.3V output voltage.

Figure 3. A better method for increasing output current is to add a pnp transistor to the output. (A TIP42B power transistor yields an output range of zero to 4A.).

Click to Enlarge

The equations for calculating the values of feedback resistors R1 and R2 are same as for the original circuit:

,

(For PECL),

(For PECL),

(For PECL).