Sensor board w/ liPo charger and fuel gauge +BMP180 +HTU21

[Anticimex]

@bjornhallberg Ok, well that sounds much more reasonable to me then :)

[tbowmo]

@ceech

Have you seen si7021? It's pin, and to a large extent software, compatible with htu21, but slightly cheaper at mouser

[Anticimex]

@bjornhallberg Have you tried to compare regulated and non-regulated supply? It would be reasonable to assume you can get more out of the batteries if you can suck them down to 0.5-0.6V. But the step-up regulators are quite "leaky" so will that really translate to a longer runtime in the end? The regulator will be on even if the node is sleeping (and efficiency drops with current drop). So perhaps (depending on usage of course) a regulated supply will actually drain the batteries faster and the end result is that it causes shorter runtime even if more juice is pulled from the cells.
I have not yet set up a proper test environment for this myself.
I was considering having a regulator you could switch off. So that the Arduino itself runs unregulated but the sensors uses regulated power. Then you could turn off the regulator when sleeping. Like one of these.
A cool variant would be to have a regulator that turns itself on when battery voltage drops below a known safe level. The trick is to implement a power rail that can switch from unregulated to regulated supply. The switching can just be done with a comparator. But feedback between regulator output and input is a bad idea I suppose...

[bjornhallberg]

@Anticimex Sorry, I don't have any long time data to offer. So I also don't know if bothering with the SHDN / EN pins (where applicable) is actually worth it. Quiescent current is usually pretty low on some of the better regulators (like TPS61221, LTC3525 etc) so I wonder if it is worth tampering with?

I still think you will gain a few months of run-time using a regulator. Still, no big savings there. The main reason (for me at least) to explore regulators is to enable sensors that would otherwise malfunction as the voltage drops. I.e. most of the common sensor we use (DHT22, DS18B20, Motion). Particularly the DS18B20 has been spotty for me.

Another point is to be able to build really compact sensors that use only one AA/AAA. Not even the nrf24 / atmega would work at 1.5V (and dropping) after all.

Also, according to my latest calculations, a separate pcb with the TPS61221 will cost about $1.75 in materials. So, it wont break the bank.

I wish we could fast-track the entire project a bit and come up with a standard form factor like LowPowerLabs or Harizanov where we could make shields that just plug. Btw, did you see this on the topic of LiPo batteries:
http://lowpowerlab.com/blog/2015/02/03/chinese-lithium-cells-freezing/

[AWI]

@Anticimex Just thinking out loud. Looking at the schematics of the predecessor of this board. There is a mosfet circuit connected to D4. Couldn't you use this to power up an external. regulator or step-up?

[Anticimex]

@AWI yea, but the problem is not activating the regulator. That is a simple IO operation. The problem I think is the output of the regulator, if you want it to power the Arduino itself. And you probably do, since the Arduino packs up probably before your sensors. I need to study some more before I got a plan for that, but I also have a LOT of other things to do so don't expect me to provide the One Solution to it in the coming weeks ;)

[AWI]

@Anticimex Switching the Arduino is probably not a good idea :) but powering up the voltage sensitive sensors with a mosfet switched step-up converter is an option?

[Anticimex]

@AWI yes, but why bother with a mosfet for enabling the regulator (if it already have an enable signal)?

[ceech]

The main reason why I was dragged to the LTC4067 is the fact that it has so called Power path technology. It only uses the battery if there is no other available source of power. The benefit is much longer battery lifetime. It also has a proper 2A Lithium charger. Last but not least is the current monitoring, which can be translated into battery state of charge, which is another thing that interests me.
I chose the voltage regulator for the fact that is fairly efficient and powerful even for ESP8266 modules and as simple as possible to implement. It only uses 35uA, which is as low as I ever saw. And the LTC4067 has Suspend mode that only uses a couple of uA as well.
@tbowmo Prices for the HTU21 are lower, for me at least. And since the pinout is the same it is all for the better.
I'll put some thought into the separate power options. The first thing that can be fairly simply done is to power the Atmega328 from the battery, and the sensors from the regulator.

[ceech]

Which is more useful - an EEPROM or Flash memory chip?

[AWI]

My vote is for flash memory, but no deal breaker

[ceech]

Finally managed to put all things together and made the first two test boards. They look like this:

Main new features are LTC4067 lithium battery charger and XC6210, a low consumption voltage regulator.

The board comes with a Torex XC6210 3.3V voltage regulator . It has low power consumption of 35μA, while delivering at least 700mA. Voltage drop is 50mV @ 100mA.

LTC4067 battery charger with Automatic Battery Charging/Load Switchover

It provides power for the circuit and charges the backup single-cell lithium battery while greatly extends battery life. You can monitor the voltages and currents. It has suspend mode, which reduces current consumption to around 40μA. The power source is a small, 5V solar cell. Connections:

analog input A1 on ATmega 328 is FAULT signal from LTC4067
analog input A0 on ATmega328 is battery voltage
analog input A2 is solar cell voltage
analog input A6 is input current ( I=V/R x 1000 )
analog input A7 is battery charge current ( I=V/R x 1000 )
digital output A9 - drive it high to put LTC4067 in SUSPEND mode

The two trimmer potentiometers are used to determine the current for both the input side - to better match the internal resistance of the solar cell - and for the battery charge current. At shipping they are both set to about 2.5kOhm, which set both currents to about 75mA. Please refer to technical data sheet of LTC4067 for more information. It is available here:
Official web page for LTC4067

Or, ask me.

This is the back side of the board with place for BMP180, HTU21 and EEPROM chip:

[lafleur]

It need an RFM69 radio on the board....??

[hawk_2050]

@ceech , I recently bought one of the predecessor boards to this new design. I was wondering if you had a better feel yet on likely end user pricing for this new one? Do you plan on populating all active parts, or providing them for users to place themselves? Really looking forward to this board becoming available.

[Sparkman]

@ceech Looks great! When do you expect to have them available for sale?

Cheers
Al

[sj44k]

@ceech Some impressive work you are pulling off here! I am also interested so keep us posted when you have it available !

[ceech]

@lafleur The board is meant to be used with either NRF24l01+, or ESP8266 radio modules. @hawk_2050 The price will be around 14EUR for the version without sensors, and 19EUR for the fully populated one. @Sparkman and @sj44k some boards will be available as soon as next week. Those are the test ones and since everything seem to work excellent, I'll make them available for purchase.

[ceech]

For those interested, here is a link with more detailed description of this board:
http://www.ebay.com/itm/331641400414?ssPageName=STRK:MESELX:IT&_trksid=p3984.m1586.l2649
We will probably offer a bit less complex version here, on mysensors.org.
Enjoy.

[MikeF]

I've just built a version of this, using the solar cell from the Solar Motion Light identified in this project: Solar Powered Mini-Weather Station. I'm using Domoticz as the controller.

I'm trying to understand what the solar and battery currents represent (I had to scale these by 1000, as Domoticz will only display currents in amps). Also, at what battery voltage level will the battery start to charge? and what level of solar voltage is required? I'm also concerned that the solar cell voltage can exceed 6V in bright sunlight - is there a need for overvoltage protection on the LTC4067?

(I'll post details and pics in My Projects shortly.)

[ceech]

Here:

float readVcc() 
{
  signed long resultVcc;
  float resultVccFloat;
  // Read 1.1V reference against AVcc
  ADMUX = _BV(REFS0) | _BV(MUX3) | _BV(MUX2) | _BV(MUX1);
  delay(10);                           // Wait for Vref to settle
  ADCSRA |= _BV(ADSC);                 // Convert
  while (bit_is_set(ADCSRA,ADSC));
  resultVcc = ADCL;
  resultVcc |= ADCH<<8;
  resultVcc = 1126400L / resultVcc;    // Back-calculate AVcc in mV
  resultVccFloat = (float) resultVcc / 1000.0; // Convert to Float

  return resultVccFloat;
}

int current = A6;
int chrg = A1;
int cell = A2;
int lipo = A0;
int batterycurrent = A7;

void setup() 
{
  Serial.begin(9600);
}

void loop() 
{
  float napetost = readVcc();

  float tok = ((analogRead(current) * napetost / 1024 ) / 2500) * 1000000; // convert the ADC value to miliamps
  float tokbaterija = ((analogRead(batterycurrent) * napetost / 1024 ) / 2500) * 1000000; // convert the ADC value to miliamps
  float panel = ( analogRead(cell) * napetost / 1024 ) * 2; // measuring input voltage
  float baterija = ( analogRead(lipo) * napetost / 1024 ) * 2; // measuring battery voltage
  int polnjenje = analogRead(chrg);
  
  Serial.print("Vcc = ");
  Serial.print(napetost);
  Serial.println("V");
  delay(400);
  Serial.print("Input current = ");
  Serial.print(tok);
  Serial.println("mA");
  delay(400);
  Serial.print("Charge current = ");
  Serial.print(tokbaterija);
  Serial.println("mA");
  delay(400);
  Serial.print("Solar cell voltage = ");
  Serial.print(panel);
  Serial.println("V");
  delay(400);
  Serial.print("Battery voltage = ");
  Serial.print(baterija);
  Serial.println("V");
  delay(400);
  Serial.print("CHRG = ");
  Serial.println(polnjenje);
  Serial.println("----------------------------");
  delay(2000);
}
/*
Improving accuracy:
To do so, simply measure your Vcc with a voltmeter and with our readVcc() function. Then, replace the constant 1107035L with a new constant:
scale_constant = internal1.1Ref * 1024 * 1000
where
internal1.1Ref = 1,1 * Vcc1 (per voltmeter) / Vcc2 (per readVcc() function)
Example:
For instance, I measure 3,43V from my FTDI, the calculated value of Vref is 1,081V.
So (1,081 x 1000 x 1024) = 1107034,95 or 1107035L rounded up.
Source: http://provideyourown.com/2012/secret-arduino-voltmeter-measure-battery-voltage/
and
https://code.google.com/p/tinkerit/wiki/SecretVoltmeter
*/```