LTC4150 Coulomb Counter giving lower than expected readings.

Hi.

I'm using the LTC4150 on an embedded IOT design.  

The design is running from a 7.2V Lithium Primary Cell  (2 x 3.6V) , and the battery is not chargable, only discharge.

The LTC4150 has been connected between the 7.2V Supply and the device power input , which feeds into a TPS62175 step-down converter.

I'm using the Linear Tech Dev boards, and replaced the Shunt Resistor with a 0.2R (0.5%) resistor. (this equates to 1pulse =43uAh)

The devices spends most of its time asleep taking approx 25 ~ 30 micro amps. 

The input where the batteries normally connect is shunted with a 10,000uF capacitor. As the WiFi connection can take large pulses of current for short periods.

The devices wakes every 60 seconds to record some data (this takes about 300mS), and then every 60 minutes uploads to the cloud. (this can take 16 seconds)

From some basic calculations I estimate the battery life of around 1.2years. 

When I connect the LTC4150 and run for approx 6 hours, I get a life of around 114 days.

Because the LTC4150 is feeding in to TPS62175 (input shunted with 10,000uF cap), I was thinking there may be some noise causing the issue. 

I have a 6u8F Polypropylene cap laying around, and shunted the 0.2R resistor with it. This changed things some what. 

Before I was only getting dischage pulses, but now I get charge and discharge pulses.

But the battery life based on the net number of discharge pulses, has improved the battery life to around 305 days. But this not a good as I expected.

Another engineer I know, suggested moving the LTC4150 to the output of the TPS62175 instead of the input, so the noise would be mostly removed.

Please comment.



Did some grammer corrections. oops
[edited by: nickmercury at 9:04 AM (GMT 0) on 18 Apr 2019]
  • 0
    •  Analog Employees 
    on May 27, 2019 11:54 AM

    Hello,

    I'm not aware of problems with LTC4150 regarding noise. For sure we have newer parts using different technologies / measurement schemes, which you could try instead. E.g. have a look at 

    LTC2943

    or

    LTC2943-1

    the latter has even an integrated 0.05R sense resistor. If that resistance is to small use LTC2943 instead that allows an external shunt. Both operate 3.6-20V, not sure, what is the minimum voltage of your two series cells.

    Patrick