Peter Baxter
EDN
This Design Idea shows how to use a 63¢ (Q=1) microprocessor reset voltage detector IC in a relay coil drive circuit to greatly reduce the coil’s hold-in current.
When asked to estimate the coil current of a small Omron G5V-2 DPDT relay, most guessed around 25 mA. It actually measured 100 mA. Yet once activated, most relays require a hold current that is only 5% of the pull-in current. Even the tiny Omron G5V-1 SPDT relay shown in the examples below has a coil current of 30 mA at 5 V.
There have been plenty of circuits designed to reduce holding current. A number are based around discrete parts, with timing controlled by high value electrolytic capacitors. Surprisingly, the approach presented here has not been described elsewhere.
Use two I/O port lines
If available, using two outputs to control one relay may be the most sensible approach: Use one I/O to pull in the relay and the other to hold it in (Figure 1). Initially, set both high together, but about 20 ms later, return the PULL-IN signal low, while leaving the HOLD signal high – until it’s time to release the relay.
 |
|
| Figure 1. | A simplistic approach: use a second I/O port line to switch a lower drive current. |
Between the HOLD transistor’s collector and the relay coil is a current-limiting/voltage-dropping device such as a resistor, Zener diode, or diode string. These parts will dissipate very little power (between 10 mW and 25 mW).
Use a reset voltage detector
Often, only one control line is available. Here, a pull-in pulse (monostable) solution is desired (Figure 2).
 |
|
| Figure 2. | Use an active-high reset chip to create the pull-in current pulse. The MCP101 and ZVN3306F worked well with a range of relays; similar parts should work fine (e.g., MAX810). |
Small SOT-23 µP reset controllers are ideal. Active-low-output parts can be used directly if open-drain and able to sink enough current and withstand the relay voltage. Otherwise, use an active-high-output chip to drive a transistor.
Some considerations
All of the current to drive the MOSFET and BJT comes from the control line. Therefore, keep the NPN base current low enough to prevent high-level voltage sag. If the control voltage can’t reach the voltage detector’s threshold level, the reset pulse won’t happen. A reset IC that can be powered by a significantly higher voltage than its threshold level will ease the design requirements.
Most relay data sheets show a pull-in time no greater than 10 ms. Many µP reset chips generate a 100 ms or longer pulse, which isn’t an issue here unless absolute-minimum power consumption is desired.
Status LED
During bench concept development prototyping, it can be beneficial to include LEDs on specific nodes to allow the software developer to immediately sense that code is activating circuitry (Figure 3). This reduces the need to go probing those points with an oscilloscope.
 |
|
| Figure 3. | If including status LEDs, locate them on either side of the current-limiting/voltage-dropping device. |
LED manufacturers often don’t specify whether their LEDs will function on 1 mA or less – yet a number of LEDs will. Smaller (0402 size) LEDs tend to be what is considered. The following LEDs have been proposed as 1 mA types – but have not been confirmed:
| Table 1. | Low-current LEDs | ||||||||||||
|
|||||||||||||
Should this be a purchasable part?
A slightly expanded reset IC would make a one-chip relay-driving solution that I think would be a commercially successful part (Figure 4). After all, we’ll still be driving relays for a long time.
 |
|
| Figure 4. | A commercial IC implementation of this design might prove quite desirable in the marketplace. |
