Hi,

My name is Mike and I'm the person running the Kickstarter campaign mentioned on this topic (https://www.kickstarter.com/projects/340271897/whisper-node). Glad to see people are looking into my project as an alternative for low-power applications. Anyway, I'll try to give my 2 cents about the battery powered nodes and the regulators.

@Michael-Moebius The MCP1640 is a great chip, but you're correct saying that it's complicated to built the module itself. The problem of assembling it on a breadboard or protoboard, is that you will probably end-up having very long leads/traces, this will cause lots of inductance, preventing the switching regulator to work well. There's also the interference the circuit can cause to others and to itself as those regulators run at high-frequency.

I've a look on the Pololu product you mentioned and it seems to be ok for a step-up regulator to produce, let's say, 3.3v from a single or dual AA, but the efficiency chart there show a bit of problem at lower currents when using a single AA. That might cause some unexpected battery consumption when the board is sleeping... but only a real-life test would give the answer.

@jmd Could you pls provide some additional details about your project:

• Which board/MCU are you using?
• Have you already discovered the power consumption of your circuit when its active and when sleeping?
  *Collect the active current can be trick as it might be a very sort spike, in this case you can rely on datasheets OR change your code into a loop to, lets say, keep reading a sensor continually just for you to measure the current.
• Do you really need 5V?

Sorry to ask so many questions, but when designing things to run for long periods every single mA counts! Also, stay away from 5V if possible, use 3.3V or even lower voltage is possible, that will save lots of battery.

If you really need 5V to power-up a special sensor (which there's no substitute which could run at 3.3V), the best approach is to have 2 power rails: a 3.3V rail powering the board and MCU and a 5V rail to power the special components. This second rail you'll need to keep off and turn on using the MCU only when necessary using a Mosfet, Power Switch or a Regulator Enable Pin.

Another thing related to 5V is that you probably won't be able to get if from a single AA, you'll need at least 2 cells to produce some usable current from a step-up. Remember that the battery voltage will drop during its life, so make sure you always calculate things using the lowest voltage (expect 0.9V by the end of the battery life http://data.energizer.com/PDFs/E91.pdf)

If you decide go to the LDO solution, bear in mind you'll always need to supply greater voltage, so the regulator will "burn" the extra power and deliver the selected voltage. For example, if you need 5V you probably will need to have 5 Alkaline cells so you can use the batteries down to 1.1V each (the extra 0.1V is for the LDO drop-down itself). People normally do the mistake of using 4 AA thinking they'll have 6V, but that's only true for the first 25% of the battery life

As a rule of thumb, if your project will spend most of the time sleeping you need to chose a regulator that is more efficient at lower currents. Always check the datasheets for "quiescent current" as well the current X drop-out charts. That will give you an idea about the minimum consumption when nothing is running.

I've posted some details how I tested the power consumption of the Whisper Node board on this forum: https://www.element14.com/community/thread/49699/l/new-ultra-low-power-arduino-board-with-built-in-rf-rfm69, you should be able to adapt some of that for your project.

Regarding price... switching regulators are not the cheapest part of a circuit. Depending of your needs it can easily be more expensive than the MCU itself

@Pavel-Hrudka Great options, although I couldn't find much information on the MAX856 and LT1300CS8 datasheets when running under 1.5V... The MAX seems to be better for lower currents, while the Linear has a good output power. Anyway, both manufacturers usually produce very good quality components.

For the Aliexpress option, there's no silicon model, but byt the look of the module design I would say it'll not have the best performance... I can see a diode which can add some undesired voltage drop as well a huge (22uH) non-shielded inductor, which doesn't look ideal to for efficiency. At the end you get what you paid for...

Cheers,
Mike M.

Hello @hek I haven't stop to think about this yet. Is very likely I'll be making most of the details about the project available but need first to focus on bringing it to life.

Cheers,
Mike M.

Good stuff! You manage to cover everything... maybe just a PIR sensor to control the lights?!

Don't forget to replace the breadboard with some protoboard and proper soldering! I hate when my prototypes start to misbehaving, and after 2 hours debugging, I discover that was a bad contact from the breadboard...

Also an Arduino Nano/Pro might help you saving some space and will look great if you attach everything into a protoboard and connect your sensors with those 2.54mm white board-to-wire connectors (like those ones: http://uk.farnell.com/productimages/standard/en_GB/CN09580-40.jpg).

Cheers,
Mike M.

Hi @Reza , there's no need to increase the digital output voltage . You just need to have a 5V source somewhere and you can use a N-MOSFET to drive the relay.

It would look like that:

Just ignore the "INPUT" it should be an OUTPUT on the diagram above. Also the MOSFET needs to have a lower VGs and support the current required by your Relay. Something like that should do the trick: http://au.element14.com/on-semiconductor/ntr4501nt1g/n-channel-mosfet-20v-3-2a-sot/dp/1431305.

Just avoid the common mosfets like BS170 and BS250 as they need some high voltage on the Gate to actually let the current flow. The datasheet will always have a few charts showing how much current the mosfet let flow for different VGSs.

Cheers,
Mike M.

@micah
Just need to be careful with those Gas sensors, they normally have a heater which needs to be ON well before you take any measurement, and it might draw some current. There are other solutions for Carbon Monoxide, like those ones: http://www.figarosensor.com/feature/tgs5042_tgs5342.html, but they are very expensive. Maybe there are other manufactures using the same electrochemical, but never stop to searched for it.

All other sensors you should be able to put it to sleep and only read very quickly.

About the battery supply, I've posted some comments on another topic: http://forum.mysensors.org/topic/3460/cheap-ldo-for-battery-usage/17 which might interest you.

Cheers!

Hi @alexsh1 , yes I recently find out those 1.5V rechargable lithium batteries. I haven't bought one myself but I'm wondering if they have an internal step-down circuit there from 3.7V there

Most of the rechargeable batteries will have a self-discharge much greater than non-rechargable Alkaline/Lithiun. The rechargeable lithium-ion have not so much... maybe 2%/month, but still adds up if you wish to run your project for 1 or 2 years.

Anyway, my point here is if you just want something to stay running for over one year in your basement for example, any rechargeable battery can add a significant cost and some trouble as you'll need to use a very specific rechargeable cell. It's not very practical IMO. On the other hand cheap non-rechargeable Alkaline and Lithium are common, offering very low self-discharge and are more resilient to temperature variations.

In the other hand if you project is a toy or a hand held device which need new batteries every second week, well, rechargeable lithium seems to be the way to go. In this case maybe just stick a LiPo or a big 18650 battery... just need to pay attention for any possible regulation as those ones the voltage will vary from 4.2V (fully charged) to 3V (discharged)... and 4.2V might fry some 3.3V component

At the end it all comes down to the project design and how it's supposed to run. If you have access to energy harvesting, like a sensor in your roof, even more fun to design...

Cheers!

Hi @alexsh1 , yes I recently find out those 1.5V rechargable lithium batteries. I haven't bought one myself but I'm wondering if they have an internal step-down circuit there from 3.7V there

Most of the rechargeable batteries will have a self-discharge much greater than non-rechargable Alkaline/Lithiun. The rechargeable lithium-ion have not so much... maybe 2%/month, but still adds up if you wish to run your project for 1 or 2 years.

Anyway, my point here is if you just want something to stay running for over one year in your basement for example, any rechargeable battery can add a significant cost and some trouble as you'll need to use a very specific rechargeable cell. It's not very practical IMO. On the other hand cheap non-rechargeable Alkaline and Lithium are common, offering very low self-discharge and are more resilient to temperature variations.

In the other hand if you project is a toy or a hand held device which need new batteries every second week, well, rechargeable lithium seems to be the way to go. In this case maybe just stick a LiPo or a big 18650 battery... just need to pay attention for any possible regulation as those ones the voltage will vary from 4.2V (fully charged) to 3V (discharged)... and 4.2V might fry some 3.3V component

At the end it all comes down to the project design and how it's supposed to run. If you have access to energy harvesting, like a sensor in your roof, even more fun to design...

Cheers!

@micah
Just need to be careful with those Gas sensors, they normally have a heater which needs to be ON well before you take any measurement, and it might draw some current. There are other solutions for Carbon Monoxide, like those ones: http://www.figarosensor.com/feature/tgs5042_tgs5342.html, but they are very expensive. Maybe there are other manufactures using the same electrochemical, but never stop to searched for it.

All other sensors you should be able to put it to sleep and only read very quickly.

About the battery supply, I've posted some comments on another topic: http://forum.mysensors.org/topic/3460/cheap-ldo-for-battery-usage/17 which might interest you.

Cheers!

Hi @Reza , there's no need to increase the digital output voltage . You just need to have a 5V source somewhere and you can use a N-MOSFET to drive the relay.

It would look like that:

Just ignore the "INPUT" it should be an OUTPUT on the diagram above. Also the MOSFET needs to have a lower VGs and support the current required by your Relay. Something like that should do the trick: http://au.element14.com/on-semiconductor/ntr4501nt1g/n-channel-mosfet-20v-3-2a-sot/dp/1431305.

Just avoid the common mosfets like BS170 and BS250 as they need some high voltage on the Gate to actually let the current flow. The datasheet will always have a few charts showing how much current the mosfet let flow for different VGSs.

Cheers,
Mike M.

Good stuff! You manage to cover everything... maybe just a PIR sensor to control the lights?!

Don't forget to replace the breadboard with some protoboard and proper soldering! I hate when my prototypes start to misbehaving, and after 2 hours debugging, I discover that was a bad contact from the breadboard...

Also an Arduino Nano/Pro might help you saving some space and will look great if you attach everything into a protoboard and connect your sensors with those 2.54mm white board-to-wire connectors (like those ones: http://uk.farnell.com/productimages/standard/en_GB/CN09580-40.jpg).

Cheers,
Mike M.

Hi @jmd

I've never used the Pololu ones but those are the only interesting ones I found so far... I basically use Dev boards like this one: http://www.microchip.com/DevelopmentTools/ProductDetails.aspx?PartNO=adm00458 for testing. I'm normally building the step-up circuit myself so I have better control of everything and reduce costs. But for one off or if you can't build the PCB your self modules are the best option!

Back to the Pololu, they seems to be pretty well built and most models are based on this chip: http://www.ti.com/lit/ds/symlink/tps61200.pdf, which seems to be very nice! All info you need will be in the PDF!

@rollercontainer , I totally agree that any regulation will have some lost, but at the same time not all components work at low voltage range, like the ultrasonic sensor US-100 on this project, it requires at least 2.4V. One alternative is to run everything directly from the battery as you suggested and have a separated rail with the step-up output at 3.3V and enable it only when necessary... or just run everything at 3.3V and simplify the design

Cheers!
MIke M.

Hey @jmd, the Arduino Pro is a much better option than a stardard one. Now regarding the 5V, if you only need it for the ultrasonic sensor, you might be better using the US-100 instead. The potentiometer is to adjust the voltage divider as all those regulators you normally need a feedback Voltage, you should be able to replace it with a single resitor, just need to look on the datasheet to get the correct value.

About the transistor, that's a great trick to use with those devices that don't have a sleep mode. Sometimes you need a N and P channel mosfet combination and place it as a high side switch.

Anyway, if you still need a 5v for something else, have a look on those modules: https://www.pololu.com/category/132/step-up-voltage-regulators. They're a bit more expensive but the quality of the components and the reduced size might pay off. You get what you pay for... BTW, switching regulators are not generaly cheap... On my project power regulator circuits cost more then the MCU and Flash memory together!

Also the sleep current seems to be right as you have the LDO there. Now the awake consumption is too little... That might be because you're waking up and going back to sleep too quickly and the multimeter won't have enough time to get a real reading. One suggesting is to make you code to keep transmitting in a 10 seconds looping, so you can get a more realistic readings. After that change back to the original but also calculate how many millis you code is being awake. With those two figures you'll be able to estimate better the real consumption.

Now, be aware that 3 AA batteries will deliver something over 4.5V only when brand new... which is ok for the LDO. But as soon the juice start to drain the voltage will drop as well. An Alkaline battery will deliver only 0.8V at its end of life, or 2.4V for three in series: http://data.energizer.com/PDFs/E91.pdf

If your Ldo needs, for example, at leas 3.6V to run, that's 1.2V per battery, which is around 50% the Alkaline life.... Just wasting the other 50% of energy.

My suggestion would be to take off the LDO regulator from your Arduino Pro and put a step-up to 3.3V... Power it with two Alkalines and you'll be fine with the US-100 sensor.

Hope it helps...

Cheers,
Mike M.

@Frank-Herrmann looks great! interested to know what driver are you using. I've being using it on RPI but with a MCU interfacing it via Canbus. Cheers

Hello @hek I haven't stop to think about this yet. Is very likely I'll be making most of the details about the project available but need first to focus on bringing it to life.

Cheers,
Mike M.