JDIY9 : Low Power


Ever since I have started building “things”, I have always wanted to build a project that uses as little power as possible, while still being intelligent. I opted for a data logging device that will read a pressure (VTI 1000) & temperature sensor (Sensirion SHT11) and calculate those readings by using 32-bit floating point math, and then save those readings to a microSD(µSD) card.

I started using the parts that I have always used for any Propeller project to get things going, and then started removing parts, mainly the voltage regulator, for parts that were more power efficient. When I was selecting sensors and parts, I picked the VTI sensor because I wrote the demo code for it, and was familiar with it's power usage. I looked up and most microSD cards take little to very little power to run, and that was the main two components to the project. I might add a wireless communication at a point, but at this point I am keeping the power consumption as low as possible. That is the main goal for this project, is how low can I go.

I had a couple options for the voltage regulator replacement, a MC3326 and a MCP1703. I ended up going with the MCP1703 because the whole project took less than 30mA at start and less than 20mA while running. The MCP1703 can sink 250mA and has by far the lowest quiescent current at typically 2µA, which helps while the Propeller waits.

Testing at first allow a battery to run 4 days with almost 35k readings with little battery drain; there is current testing with the new voltage regulator. The testing results went well and I have opted to keep the MCP1703; now to build the project box.

Instead of wiring up another 40-pin DIP Prop and supporting circuits, I opted to use a Protoboard with a couple edits; mainly removing the voltage regulators since there was more quiescent current than wanted. Once I removed the voltage regulators, I added a MCP1703 and used all the existing Vdd and Vss to power the rest of the project. It worked out great, and will be doing that in the future for similar projects.

I put the project in an acrylic box instead of building another one out of wood allowed me to show off the sensors in a cool way and easily portable. I took the project out while driving around town to get some results. I put it in front of the AC and left it in the sun where the temperature quickly jumped up to 115 degrees to get a good range.

After a day of testing, I concluded that humidity and dew points were something I wanted to include to make it more like a data logging weather station. That way, I can put in out in nature area and maybe offer some citizen scientist project. I went with a well known temperature and humidity sensor from Sensirion, the SHT11. There is great code for it and I'm running floating point for the VTI sensor. It was a breeze, however I ran into one issue while computing Celsius to Fahrenheit, and the reading was correct but there was an apostrophe (') before the values were saved, which was odd. To solve it, I converted the dew point value from C to F before saving it and was back on track.

Open Office Screenshot
This extra apostrophe is messing with my graph
Now I have pascals, temperature in Fahrenheit, atmospheric pressure in PSI, relative humidity, and dew point all being saved ever iteration, which currently is set for 1 minute. I have 1 second, 10 seconds, 1 minute and 10 minute iterations in the code for easy changing. I'm in debate on adding an XBEE module so I can gather the data on the SD card without disturbing the project as a whole. A solar panel to recharge the battery is, I think, the last piece so there is no need or worry that the battery will die. I read up on some simple recharging of a Li-Ion and I do not foresee a great deal of issues there. For now, it's just battery (3.7V @ 6000mAh) powered at the lowest power consumption I can get. Right now it runs µA's while in stand-by and about ≤ 15mA while pulling sensor data and writing to the µSD card. So it's been 5 days and the battery finally died, I accidentally reset the project and lost all the data saved (haha opps). I researched Li-Ion charging and there was a IC that stood out from the rest and that is the TP4056. There are plenty of pre-built boards available on Ebay, or build one from scratch. I ran the numbers and it costs less to buy one already built. I ordered some that should arrive around July 10th. The solar panel I have will more than recharge the battery, and the TP4056 board I got as over-charge protection, so everything is looking good. I got the TP4056 module and solar panel installed and connected. It's been a couple of day cycles and it's running strong; I think I could add the Xbee and still not come close to draining the battery. I have been taking readings throughout the days and from what I see on Vin I think I'm in really good shape. The TP4056 has over-charge protection and so do the batteries, so I'm not worried about them being damaged; there is as solid 2.99V on Vout on the regulator, and that can get down to 2.7V before anything gets “funny”.

Voltage Readings

Date Time Results
Sunday 7/8/2018
1:38 AM
4.06V @ Vin & 2.99V @ Vout
Sunday 7/8/2018
2:00 PM
4.09V @ Vin & 2.99V @ Vout
Monday 7/9/2018
12:30 AM
4.07V @ Vin & 2.99 @ Vout
Monday 7/9/2018
3:17 PM
4.10V @ Vin & 2.99 @ Vout
I will be sure to check the voltages during cloudy days, but from previous testing, there is about a 5 day buffer, which I feel is plenty enough. SUCCESS!! Log this one in the books as a total success, and my first low power and solar powered project. I have already took it on outings to get different readings, such as hikes and drives. I can unplug the solar panel and have 5 days before it needs a recharge, or just let it recharge the next day so I dont have to take the panel with me if I dont want to. I completed a final 5 days of testing to find out how much data the 2GB µSD Card could hold. 5 days of data came out to about 300KB, which puts the data collection limit to about 90+ years; I think that should be satisfactory since the battery life is about 10 years.

Difficulty: !!!!
I give this a 4 !!! because of the soldering work, the voltage regulator was really small, and needed a special board so that I could use it. It was really important to bring the project into µA power use, component and programming was key to keeping current low.

The video below is the the project with the low quiescent current voltage regulator being used. The voltage @3.1V simulates a battery that is getting low.