Pull-up and pull-down are methods of processing digital signal lines by connecting them to the power supply side and ground side, respectively. Signals may be connected directly or via a resistor.
(1) Direct connection
Direct connection means fixing the input level of the dedicated input pin to either high or low. Connecting signals via a resistor not only increases the number of parts, but also may pick up signal noise. Direct connection is therefore more commonly used for the dedicated input pin.
However, in some systems (such as those that require a high level of reliability), anticipating the possibility of the input short-circuiting with the power supply or ground due to internal degradation, it is considered better to connect the signals via a resistor to prevent the entire system breaking down. Therefore, whether or not to insert resistors when fixing the input level should be determined based on the level of design quality required.
(2) Connection via resistors
In general, there are two cases in which signals are connected via a resistor.
One is when the input level of an I/O alternate-function pin which could be used as an output has been fixed to high or low, as mentioned before. In this case, a limiting resistor is inserted because if the signal is directly connected to the power supply or ground, a short-circuit will occur when the pin changes to an output pin. Even if the pin is not intentionally set as an output pin by the program, it might unintentionally become an output due to noise, so it is better to insert a resistor just in case.
The other case is when a pull-up or pull-down resistor is incorporated in order to absorb signal noise or reflected waves by reducing signal line impedance, to speed up rising or falling of signal, or to satisfy the specifications of input level. Particularly in specifications that regulate the interconnection of devices via a cable, use of a terminator (terminating resistor) is often defined.
[Tea break] |
In MOS-to-MOS connections, there is rarely a mismatch of voltage levels. However, when MOS devices first appeared, MOS-to-TTL connections were not uncommon. In this case, when the TTL output is connected to the MOS input, pull-up processing is often used because the high-level specification could not be satisfied.
Note that when both pull-up and pull-down resistors are incorporated, resistive division can cause the input to become an intermediate potential. In most MOS devices, therefore, only a pull-up resistor is connected to the input, as required.
Inserting a buffer is one way to improve both the high level and low level, but in this case, it is necessary to check whether the delay is within the allowable range.
(3) Resistance value
Let's look at what kind of pull-up resistance value should be fixed.
If this value is defined in the interface specifications, that is the value that should be used.
Generally, resistors with a value of several kΩ to several tens of kΩ are used. Using resistors with a very high resistance is meaningless, and anyway, too high a resistance may even lead to the introduction of noise.
(a) Fixed input
When the MOS input level is fixed, the load current becomes input leak current.
For example, when ILIH = +5 uA and ILIL = -5 uA with VDD = 5 V, the result is r1 = r2 = 5/(5 × 10-6) = 1 MΩ.
Assuming VIH = 0.7VDD and VIL = 0.3VDD, when r1 = r2 = 1 MΩ with VDD = 5 V, we can calculate that R1 = R2 = 750 kΩ from the above expression. In other words, if these values are exceeded, the potential difference between the pull-up side and pull-down side increases, and the input's high or low level cannot be maintained, making it an intermediate potential.
(b) Input signal processing
Since it is most common for the input signals of MOS devices to be pulled up, not down, we will focus on describing pull-up processing.
When a signal is compensated, while the output is changed to low level in the case of pull-up and to high level in the case of pull-down, current flows to the resistor. Therefore, in order to lower the current consumption, it is better to set as high a resistance value as possible within the range that satisfies the input characteristics.
However, since there is output capacitance, input capacitance, and floating capacitance on the wiring, when the resistance is too high, it is necessary to consider that the waveform may deteriorate according to the RC time constant.
The lower-limit value during signal compensation is determined, mainly according to the output device's drive capacity.
In a digital output, the minimum value of the high level (VOH) and the maximum value of the low level (VOL) are defined according to the drive current IOH and IOL, respectively. When the resistance is low, pull-up processing raises the low level and pull-down processing lowers the high level. As a result, the input becomes an intermediate potential, or the device is degraded due to a current that exceeds the drive capacity. For example, consider a case where the system is operating on 5 V, IOL = 5 mA, and an output with VOL = MAX. 0.5 V is pulled up to the 5 V power supply. The minimum resistance value is (VDD - VOL)/IOL = (5 - 0.5)/0.005 = 900 (Ω).
If a pull-up resistor is externally attached despite the device having an on-chip resistor, note that the resistance value becomes lower due to the parallel connection of the resistors.
Also, if the system includes two devices with different operating voltages and the output is pulled up to the higher power supply voltage, the absolute maximum rating of the device with the lower operating voltage is exceeded, or the current loops around, damaging the device or power supply. In this case, therefore, be sure to pull the output up to the lower power supply voltage.