Long Life Batteries

[Fratink]

Hello everyone,

Has anyone had experience with long life primary batteries that can handle cold temperatures and relatively high discharge current?  The best I've found is the L91 AA from energizer which is good but might not have the capacity I need.  Everything else I've been able to find is much more expensive per Wh.

Here's a summary of what I've found:

• Alkaline C/D have the best Wh/$ ratio, into the 30Wh/$ but capacity decreases to 1/10 at -20
• Lithium Manganese Dioxide (Like the L91) is around 20Wh/$ for a AA
• lithium thionyl chloride like the Tadiran TLH-5930 D come to 3Wh/$ or so but have very low max discharge current
• Specialy NiHM cells could be used as primary cells as some have low self discharge

Requirements are to have discharge current up to 1A for short periods (1-5 minutes a day) and then draw is down to uA rest of the time.

If it wasn't for the -40C requirement I'd just toss in alkaline cells as they are by far the most cost effective.  I guess I could rent a battery tester and put some alkaline cells into a temperature chamber and just see how they do.

Any thoughts would be super appreciated, I'm not that familiar with batteries for use in long life and extended temperature applications.

[mjs]

Been there, done that. Check out Saft LSH series.

[Fratink]

Thanks, those things look amazing.  Price point is a bit higher than I'd hoped but I guess you can't be too choosy when you're asking for something that works at -40.

[mjs]

With high volumes you can get better price, but even then they're not exactly cheap. Be sure to read all the application information regarding those batteries and if you're using them in any larger quantity, I warmly recommend contacting Saft sales and going through your application with their support.

There are quite a few things to take into account if you want reliable high performance any time at any temperature. Like chemistry delays, voltage+current capacity/impedance with temperature and consumed energy, passivation and so on.

[jbb]

I can’t say too much, but experience at my work shows wide temperature range is a Problem.

High temperatures cause greater leakage currents in electronics and battery self discharge.

Low temperatures can cause huge increases in cell internal resistance when the cells are partially depleted. The open circuit voltage might look OK, but when you apply load it drops like a rock.

So testing at low temperature is critical. Ideally you would test with the real load too, but that might take too long.

Usually we think about coin cells or cylindrical cells. There are manufacturers out there making pouch type cells with much lower resistance. They look a bit like LiPo cells.

It’s also important to think about moisture. I got a unit back from the field recently which died before it’s time. Turns out there was some moisture under the wrapper on a coin cell; it bridged the anode and cathode and that sucker went flat.

On the electronics side, it’s always worth having a look at your current consumption. Can you:
- turn off some stuff when you’re not using it?
- reduce the peak current?
- extend the minimum operating voltage a little?
- upgrade some parts with a lower current version (note, much re-validation may be required)
- thoroughly test the source code for power consumption? Some sensitive current probe can be very helpful here (I use a 10 Ohm shunt the. Analog Devices instrumentation amplifier for x100 gain and few hundred kHz bandwidth).
- upgrade your power supply, e.g. replace a ‘generic’ buck converter with high efficiency type (especially at light load)?

[Fratink]

Yeah, everything I've been reading matches that temperature problem.  I hadn't thought about the increased leakage at high temperatures so that's really good to know.

We've actually had some moisture problems in the past so I'm hoping to resolve that.  The field solution is apparently to drill holes in the unit so I think that means there was an engineering error somewhere haha.

Those are all great suggestions about reducing power.  My design right now is:
An always on 3.3V bus with 500nA quiescent current.  Total load most of the time is less than the quiescent.  Basically just an RTC with an alarm to wake up the MCU.
A high current 3.3V bus with high efficiency switchmode supply that is connected to the battery through a relay.  It's off 99% of the time now haha
A charge pump for he 5V bus since that is also off 99% of the time.

It's those pesky sensors and wireless communications that have to turn on once in a while that take all the power haha.

[jbb]

Yeah, I know what you mean about pesky sensors.

On always-on 3.3V: this will be a pain if using some Lithium cell. Could you go down to 2.5V?

On switched 3.3V: maybe you could throw away the relay and use a converter with an enable pin? Also see comment above about voltage range.

On switched 5V: try to have as little load as possible up here, because it’s going to load the battery really heavily (in terms of mA) when the battery is low.

Also take a close look at all the MCU GPIO pins. It’s very easy to leave a pin connected to the 3.3V switched domain enabled, which leads to lots of wasted current. Same with weak pull-up resistors.

Little things like pull-up and pulldown resistors can also add up when you’re in sleep state. Did someone just pick 10k and call it a day? Could those be increased a bit?