Coin cell Current pump CR2032

[Adi, Robowild]

checking most of the reliable coin cell CR2032 such AS Panasonic declare on very very low current transfer far lower then 5 mA

if i set my Ble in 5mA IO OUT ....when my ble in on and full power it consume 5mA this is far from the battery manufacturer to supply

1.what should be te hardware and software configuration for charge pump ....

thanks

yaniv

[AmbystomaLabs]

If I have interpreted your question correctly, a charge pump really won’t help on on CR battery designs. Yes TX/RX current can easily be 10mA. This is vastly outside the range of a CR battery. 

The solution is to provide a lot of low ESR capacitance and minimize the BLE duty cycle. In this manner the CR battery sees less than 1mA average during high current consumption periods.

Generally, 22uF of chip caps is a good place to start for local capacitance. Never use tantalum since it has static parasitic losses.

TI published a nice white paper on CR powered BLE designs and there is a lot of info in the devzone.

[Adi, Robowild]

THANKS!  ESR is a nice , but not reliable solution  for mass production, i'm looking to a solution like have buffer energy  between the chip and the battery  , how ever i do believe that Nordic Eng already have a prof working solution (they did wait for me to come with a working one  ....)

[haakonsh]

Like AmbystomaLabs have already said, what you're looking for is an around 22uF MLC capacitor. I can also suggest to use the nRf52832 instead, as is has a lot lower peak currents, and it's DCDC converter can source the entire SoC and not just the radio and CPU as with the nRF51822. 

[AmbystomaLabs]

Haakonsh, since this question is coming up a lot more you guys should do a new whitepaper/tutorial on it.  Most of the people doing these designs don't seem to have any hardware experience and make big mistakes.

Since it would be a new whitepaper, I recommend you do a slightly improved version from MLC caps. Digikey shows a 1210 packaged 11mF 3.3V supercap.  Only $0.67 each in a reel of 2,000. https://www.digikey.com/product-detail/en/seiko-instruments/CPH3225A-2K/728-1067-2-ND/4747399

The design could have a big series inductance to better isolate the CR battery from the peak load.  With 11mF you can get 11mA of current for 0.5seconds with only a 0.5volt drop in the supply rail.  That would allow for some real data handling with the BLE core without destroying the CR battery too quickly.

I suppose a second thought is to put the supercap on the low voltage rail of the nRF(ie, DEC4). But you might have to play with the DCDC software to get the whole thing to work right.  Downside of this approach is gpio will drop with the CR battery voltage.

It's just a thought.  Though if your schedule is like mine there is probably zero free time to do fun stuff like this.

[Adi, Robowild]

low ESR it well old finger trick ....

we already check it few weeks a ago and it fail in the lab and in the simulation , when temp goes under -4C it stops after 5 minutes ..in 25 C it is working for shout time 10-15 then it stops

... if the soc will be wake up for few seconds or even 500 ms maybe it will works !!!

 but we  need  the Soc work to 2-3 minutes and then go to sleep ... we used the SOC as a real mcu to do the all the  job...so...   those caps will do not good since the coin cell battery can not  transfer the required energy  , also you would need few ESR low ESR are very expensive  for 4 usd total product. the only solution i came up is build a buffer of energy when soc is sleep and when soc wake up this is the energy buffer ,this circuits is not cheap , and the mosfet and the coil and the caps also lost energy if system is sleep for few days ....SInce some of ble should be tiny and extrem low power i hoped Nordic already had 100% working circuits for this , still i find Nordic much better then ST or TI