Now that the critical elements of the Wireless RH/Temp Sensor Module have been chosen, it’s time to determine the power architecture.
We’ve chosen the devices such that both the STM32 microcontroller and the TI CC1175 transmitter will be powered by the same voltage – 2.5V. One thing I’ve learned in doing mixed-signal designs is that you generally don’t want to share a voltage rail between the analog and digital submodules. The rationale is that in the digital domain IO lines are switching constantly, causing transient current requirements at the regulator. If your PCB is poorly laid out (interruption in the return path for a signal, coupled traces causing cross-talk, analog and digital traces adjacent to one another, mismatched trace impedance, etc.) then there is a possibility that these transients can cause noise in the analog circuits.
In low-power designs there may be no issues. In designs where high switching currents are required, or the power regulators are sensitive to changes in current, then it could easily become an issue.
One way to remedy this is to use separate regulators from the main system voltage rail. Another approach is to place EMI filters, spec’d for the correct frequency and impedance, in series with the regulator output to filter the voltage rails each to the analog and digital circuits. Having bigger output capacitors also helps.
Before I discuss the choice of regulators, let’s talk about the powerhouse, the battery. In the past I have had success with using single-cell Li-Ion packs, and coupling them with a TI Battery Charger IC. I have already designed circuits with the TI BQ24073, so there was an opportunity for some design re-use. This device has a power path circuit for using the battery and the DC input together to provide power to the circuit. The device also monitors the battery and has external resistor-based settings for the charge parameters. It allows the designer/user to select between USB 100mA, 500mA, or externally limited charge currents up to 1.5A. The output voltage is boosted to 4.4V, which will provide the source for the rest of the power section.
So we need to knock 4.4V down to 2.5V, and for this design I opted for using two separate regulators for the digital and RF circuits. Here, I opted for two TI TLV70225 LDOs in the SOT-23-5 package. They provide 300mA of current each, are simple to design in, and only require ceramic output caps.
For the battery, I chose a single-cell, 850mAh capacity, Li-Ion pack from Sparkfun. I have used these in the past and they work well. Some quick calcs based on estimations:
- The CC1175 consumes ~26mA in transmit mode (@ 0dBm), 170uA in XOFF mode, and 0.5uA in power-down mode. Let’s assume it spends most of its time in XOFF mode. This is also spec’d at 3.0V, not 2.5V, so it will probably be lower.
- It’s hard to estimate the dynamic consumption of the STM32, but in other projects with the mcu in stop mode with the RTC running, I’ve seen currents as low as 15uA, so I’ll use that for this estimation.
- Quiescent current of the LDOs is spec’d as 30uA. There are two, so 60uA.
- Quiescent current of the battery charger IC is spec’d at between 50uA and 1.5mA. Let’s assume worst case.
These are the major current consumers. There are various pull-ups and downs in the circuit, which can be considered as well, but I’m trying to get a rough estimate of standby life of the battery.
So, on the 2.5V rails, we can assume:
170uA + 15uA = 185uA
The power requirement on 2.5V then becomes:
2.5V * 185uA = 462.5uW
If we assume 85% efficiency from the regulators:
462.5uW / 0.85 = 544.1uW at the input
At 4.4V this becomes the input current estimate (including the q-current) of:
544.1uW / 4.4V = 183.7uA
Let’s assume the efficiency of the boost regulator in the charging IC is also around 85%:
544.1uW / 0.85 = 640.1uW (from the battery) 640.1uW / 4.2V (fully charged) = 152.4uA
With a capacity of 850mAh:
850mAh / 152.4uA = ~5577 hours of standby time (232 days).
That’s a significant amount of time, and of course this is all based on estimates. The battery will cut off at some point, reducing the available capacity. Also when the voltage begins to drop, the current increases through the boost regulator. Also, this doesn’t take into account the dynamic current when the device is awake and transmitting. So this would be a best case scenario. I think we will find that measurements of current are actually less, but we’ll test that at a later time.