Whats most efficient step-up or step-down?

[tintin]

Hi

I'm working on a temperature logger project with a microcontroller and IC's that run on 3.3V which I want to power with 3V coin cell batteries. I'm want to get the device as power efficient as possible.

I have two ideas to drive this circuit.
1. Two coin cells in series (3+3=6V) and a step-down circuit to 3.3V.
2. Two coin cells in parallell (3V) and a step-up circuit to 3.3V

Does anyone have any idea which is the most power efficient way?

[David_AVD]

I gather that the 3V is not enough for the rest of the circuit to operate correctly?

[duskglow]

I removed my comment.  I didn't notice that the cells were 3V. I'm an idiot. 

[mariush]

I'm working on a 9v battery replacement using supercapacitors and one IC that I found nice specs wise was LT1613 : http://www.farnell.com/datasheets/1605123.pdf

It's up to 90% efficiency, it can do 3.3v @ about 30mA (which should be enough for your needs) from a 1.5v battery and obviously much more with higher input voltage.

Considering you plan on using 3v batteries, which can usually do about 3mA of current, I guess your circuit doesn't use much power so other than being expensive, this IC would work just fine.
The datasheet also shows the regulator working in SEPIC configuration, so it could work with wide input voltage range, below and above 3.3v with no problems.

Just consider A CR2032 (3v battery) usually has about 200mAh while even a small 1.5v AAA alkaline battery has about 1200mAh. It would also take about the same space on pcb as two 3v batteries in battery holders.

Anyway, the only problem is that it's a VERY small chip.. I put it aside because i practically have to etch a pcb (and i don't have the chemicals at the moment) or order one just to make a prototype.

So I had a second look and found LT1307: http://uk.farnell.com/linear-technology/lt1307cn8-pbf/ic-dc-dc-up-converter-1307-dip8/dp/9560394
Here's the datasheet: http://www.farnell.com/datasheets/25129.pdf

This is also available in DIP format, and works as low as 1v, but it's only about 70% efficient with low input voltage compared to the previous one... and it's still quite expensive.
But it would still be much better to do 1.5v aaa to 3.3v instead of 3v cr2032 to 3.3v

You might also want to check ISL9111 in fixed 3.3v output: http://www.intersil.com/content/dam/Intersil/documents/fn76/fn7602.pdf
Should do 3.3v with up to 80-95% efficiency with input voltages between 1.2 to 3.3v  (this one's only boost, you don't want to give it more than 3.3v)

[mikeselectricstuff]

I'd say series and step down - lower current means battery internal resistance is less of an issue. You will also probably get a lower cut-off voltage, though this depends on the dropout of the chosen converter.
Be very wary of current-draw specs of low voltage step-ups, in particular no-load current, as this is often supplied  from the output,  and so multiplied by the step-up ratio. 
I recently saw some data on a new low-voltage step-up part from Touchstone - TS3300 - which has quiescent about a tenth of the best I could find when I last looked a while ago. http://touchstonesemi.com/products/ts3300

Also consider using a bigger battery if possible - a 2450 can supply a lot more current than a 2032

[nctnico]

For many (low power) battery powered devices a linear regulator is more efficient than a step-down. IMHO it would be better to use parts which can work from 3V (or less). In several of my recent designs I had to make do with 1.8V. Having 3V from a coin cell is pure luxury 

[Legit-Design]

I'm working on a temperature logger project with a microcontroller and IC's that run on 3.3V which I want to power with 3V coin cell batteries. I'm want to get the device as power efficient as possible.

Maybe try to change your circuit around to allow more flexibility with operating voltage? What micro and IC's are you using currently? I'm sure there might be some good options for parts which are designed to have wider operating voltage range for battery applications.

[duskglow]

That's a good point, I'm sure there are microcontrollers out there that are designed to run extremely low power and on a battery voltage.  I just can't remember the name of one offhand.

[nessatse]

I can't really tell you which is more efficient, but for reference, I have been working on something similar, and decided to use a AA alkaline cell and boost converter.  I used the Microchip MCP1640 (http://ww1.microchip.com/downloads/en/DeviceDoc/22234B.pdf).  It has been running on my desk for the last 4 months or so and the battery still looks perfect.  This while sampling every 10 seconds and transmitting the results using one of those cheap 2.4GHz RF (nRF24l01) modules.  My initial objective was for at least a 12 month battery life when sampling every 5 minutes, but extrapolating the results above, should give significantly more.  The controller is a ATTiny2313 btw.

[fcb]

Step-down tends to be more efficient as you can (easily) do things like synchronous rectification - there's the added bonus that you can shut them off compared to a typical boost converter and there are buckets of them to choose from (certainly compared to boost converters). Size-wise: getting a 1MHz buck is alot easier than a 1MHz boost.

The most efficient thing would be to use the 3v directly.  As far as which micro? I would look to Microchip (mainly cause you'll always be able to get them, there well documented and I've been using them for 20+yrs), but the Texas MSP430 seems to have won tonnes of design-ins in low power applications - it's got some nice converters too.  But the choice of micro is not as important as the sampling and wake-up regime.

[nctnico]

That's a good point, I'm sure there are microcontrollers out there that are designed to run extremely low power and on a battery voltage.  I just can't remember the name of one offhand.

The MSP430 series from TI rule when it comes to ultra low power. Most ARM Cortex-M0 controllers from NXP will also work reliably at 1.8V which can't be said about other vendors (like Atmel) who claim a device can work at 1.8V but in reality it can't. The thing to look for is the absolute minimum operating voltage which should be a bit lower than the actual operating voltage.

[Psi]

Even with a single a 3V lithium coin cell you have plenty of options MCU wise.
Those 1.8V atmega/tiny will be quite happy with a 3v coil cell voltage range ( 2V -> 3.2V)
As said above 1.8V is really their absolute min, but above that they're fine

But if you run a MCU off a unregulated battery you need to consider ADC sampling.
With a moving VCC you have to use a spare ADC channel or use the internal 1.1V reference to monitor and calculate what VCC is so you can calculate other ADC readings correctly.

VCC = (ADC_max_value / ADC_reference_sample) * Ref_voltage

It does introduce more errors though, so depending on your application it may not be suitable.

[tintin]

Thank you for all reply!

To clarify more what I'm trying to achieve with my circuit.
The logger is for high-temperature logging with thermocouple. I have found a IC for thermocouple to digital conversion, MAX31855. It need power supply 3,3V+-0,3V. Maybe there is another I can use with lower supply voltage?

The whole circuit will be in a pretty hot environment ~60-70°C. So I need to use components and batteries that can work in such temeprature. I have found these batteries, BR2477A lithium coin cell 1000mAh, temeprature range up to ~125°C.
For microcontroller I will use a ATMega Automotive rated,  can work down to ~1.8V.

I can get 6 of those batteries fitted in my box, but question is what will the best setup be? Series ? parallell?  combination of both series and parallell? Step-up or step-down regulator?

The most power efficient setup would probably be to get the circuit working directly from battery supply without voltage regulator, but I don't know how I can achieve that?

Thanks!

[madires]

I'm using the MCP1702 for low power stuff. The low input quiescent current is very nice if I'm able to send the MCU into sleep mode. Since the MAX31855 allows +/-0.3V you could switch its Vcc via a MOSFET by the MCU.