Low Cost PCB's Low Cost Components

Author Topic: Modelling an Alkaline Battery in LTspice  (Read 6799 times)

0 Members and 1 Guest are viewing this topic.

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Modelling an Alkaline Battery in LTspice
« on: August 15, 2015, 03:38:29 AM »
Hi,

Recently on the Forum there has been a lot of interest in the discharge characteristics of Alkaline batteries. This thread is dedicated to building a reasonably accurate LTspice model of an Alkaline battery.

This is not new, it is based on a model explained in this document:

http://www.del.ufms.br/appnts.pdf

The battery modelling starts on page 260.

The model presented in this document has been translated into LTspice.

I will compare the performance of the model to actual test results found in this report:

http://www.amstat.org/publications/jse/v21n1/dunn/RE11_021_ALDI_Stores_Primary_battery_testing.pdf

There are test for a electronic game, an electronic flash and a digital camera.

Once the model has been developed it can be used to evaluate battery life extenders.

I will show the development of the model in stages.

Open Circuit Voltage versus State of Charge

The TABLE function is used to generate a voltage as a function of the SOC.
The SOC charge was inverted to give a DoD Depth of discharge to match the tabulated data provided in the reference.



The result is:



The Horizontal axis represents fresh to fully depleted.

Modelling the Battery Capacity


C1 was added to model the battery capacity. The capacitor is discharged at a rate proportional to the load current.



In this test circuit a 0.1Ahr battery is discharged in 3600K seconds or 1 hr.




Stepping the load current

The load was modified to step the load current:



The resulting curves show the battery life is inversely proportional to load current:




ESR modeling

So far the ESR of the battery has been ignored. The ESR is not constant. The model has been set to use constant ESR from 100% to 20% SoC, then increase in linear fashion to 2x at 0% SOC.



The results:




To be continued...

Jay_Diddy_B



« Last Edit: August 15, 2015, 06:39:29 AM by Jay_Diddy_B »
 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modeling an Alkaline Battery in LTspice
« Reply #1 on: August 15, 2015, 03:38:50 AM »
Calculating Discharge Rate 'C'

The discharge rate C = Capacity / output current is need to make an adjustment to the capacity of the battery at high discharge currents.



The RATE signal shows:




Adjusting the Capacity for High Discharge Rates

The model has been modified to reduce the capacity of the battery for high discharge rates. The values were obtained from the reference in the first post.

I have also added an integrator to measure the mAh that can be extracted from the battery.



These are the results:




I have attached a zipfile containing the LTspice model.

To be continued...

Jay_Diddy_B

« Last Edit: August 15, 2015, 06:53:17 AM by Jay_Diddy_B »
 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modeling an Alkaline Battery in LTspice
« Reply #2 on: August 15, 2015, 03:39:09 AM »
Testing the model

Standardized test are defined in an IEC publication:



(No Monkeying around here !!)



Electronic Game

The official test for an electronic game is 250mA for 1 hour (3600 seconds) in a 24 hour (86400 second period):



And the test results show:





About 8 days, 8 hours of playing time.


Electronic Flash

The official test is 1A 10 seconds per minute, 1 hour per day:




The results show:



That about 6 days worth or 360 flashes before the end point of 0.9V is reached.

The results in the test report are:



The model is showing reasonably good results and can be used to explore battery life extenders.

Enjoy !!

Jay_Diddy_B



« Last Edit: August 15, 2015, 07:02:02 AM by Jay_Diddy_B »
 

Offline LabSpokane

  • Super Contributor
  • ***
  • Posts: 1899
  • Country: us
Re: Modeling an Alkaline Battery in LTspice
« Reply #3 on: August 15, 2015, 03:57:51 AM »
Beautiful.   :clap:
 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #4 on: August 15, 2015, 07:48:43 AM »
Hi group,

I am going to introduce the concept of the 'Battery_Booster'. This is a power supply that will attempt to keep the output at 1.4V, providing the input voltage is greater than 0.7V. I have assigned an efficiency of 95% to my 'Battery_Booster'.

Here is the 'Battery_Booster' model:



The properties of the 'Battery_Boost' are:

The output voltage is constant at 1.4V

The output power is constant, if the load is constant.

The input current increases as the input voltage falls.

The input power is 105.5% of the output power.

The 'Battery_Booster' shields the ESR of the battery from the load.



Electronic Game test with the 'Battery_Booster'

I have placed the 'Battery_Booster' between the battery model and the load.



The results show that the electronic game will run for 5 hours and 20 minutes:



This is significantly less playing time than without the 'Battery_Booster'

The reason is the nature of the IEC 60086-2 test program, it calls for a constant current load of 250mA.

With the 'Battery_Booster' in place the power consumption is constant throughout the test. If the battery voltage is allowed to fall, the power consumption falls during the test.

I have attached a zipfile with my LTspice model.

Regards,

Jay_Diddy_B

« Last Edit: August 15, 2015, 07:56:29 AM by Jay_Diddy_B »
 

Offline pickle9000

  • Super Contributor
  • ***
  • Posts: 2131
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #5 on: August 15, 2015, 08:01:19 AM »
 :-+ :-+
 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #6 on: August 15, 2015, 08:16:06 AM »
Hi group,

The behaviour that is predicted by my model is included in the Batter... Patent application.

https://www.google.com/patents/US20120121943

Here is the picture:



They have chosen for some unknown reason to boost the battery voltage to 1.8V

Note: they show a shorter running time with the Battery_Booster connected. :palm: :palm:

Regards,

Jay_Diddy_B





 

Offline pickle9000

  • Super Contributor
  • ***
  • Posts: 2131
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #7 on: August 15, 2015, 08:31:58 AM »
Can you do that with the boost set to 1.6, for a comparison?

Edit (sorry I see you already have sufficient info in an earlier post)
« Last Edit: August 15, 2015, 08:34:47 AM by pickle9000 »
 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #8 on: August 15, 2015, 08:46:29 AM »
Can you do that with the boost set to 1.6, for a comparison?

Edit (sorry I see you already have sufficient info in an earlier post)

I have arrange the 'Battery_Booster' to work at different output voltages:



Here are the results showing a 1.3V and 1.6V boost voltage:





The electronic game test is 1 hour per day. At 1.6V the playing time is about 4.5 hours.

If the electronic game uses:

1) An LDO, it will present a constant current load.

2) A smps it will present a constant power load. In this case the play time should be independent, (1st order) of the Boost voltage.

Regards,

Jay_Diddy_B
 

Offline ElectricGuy

  • Regular Contributor
  • *
  • Posts: 227
  • Country: pt
Re: Modelling an Alkaline Battery in LTspice
« Reply #9 on: August 15, 2015, 09:03:29 AM »
Great Job!!
Thank you!
Regards
ElectricGuy
 

Offline dcac

  • Regular Contributor
  • *
  • Posts: 58
Re: Modelling an Alkaline Battery in LTspice
« Reply #10 on: August 15, 2015, 09:40:05 AM »
Hi group,

The behaviour that is predicted by my model is included in the Batter... Patent application.

https://www.google.com/patents/US20120121943

Here is the picture:



They have chosen for some unknown reason to boost the battery voltage to 1.8V

Note: they show a shorter running time with the Battery_Booster connected. :palm: :palm:

Regards,

Jay_Diddy_B

Here's the patent text 'explaining' fig 7:
Quote
FIG. 7 shows measurements that illustrate the advantages of the various embodiments. Three popular AA battery brands, Panasonic, Duracell and Sony were chosen for the measurements. Active load circuitry that drew a fixed 50 mA current was placed at the output of these batteries and the voltage of each battery was measured over time. The horizontal access shows time and the vertical access shows the battery voltage. The starting voltage for these fresh batteries was 1.6V. The amount of time it takes for the batteries to reach 1.39V, which is where a lot of electronic equipment stop operating, are listed. The Panasonic battery took 6.3 hours to reach that level, while it took 4.5 hours for the Sony battery. The Panasonic battery when used in conjunction with a regulator, according to embodiments of the invention, took 27.9 hours before it stopped providing 1.5V, and the Sony battery when used with a regulator took 32 hours before its stopped providing 1.5V. Thus, with the regulator, it takes 4.5 to 7 times longer before the battery needs to be replaced. Thus, the total number of batteries that need to be manufactured and consequently discarded would be reduced by 4 to 7 times.

I've no idea what the horizontal line at 1.8V is supposed represent, but I believe the vertical lines shows where *their* 'Regulator' stops functioning. The 1.39V which they claim "is where a lot of electronic equipment stop operating" isn't shown, and probably the cause of much confusion. Also 1.35V is mentioned in the table underneath but isn't drawn in fig 7 either.
« Last Edit: August 15, 2015, 09:41:49 AM by dcac »
 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #11 on: August 15, 2015, 10:54:56 AM »
Hi,

The 1.8V is the voltage that they have chosen to boost the battery voltage to.

If I modify my model so I can do tests with and without the 'Battery_Booster'





If the product that is being powered has an undervoltage detector set at 1.35V the product will run for 13.3 ks (kilo seconds) 3.7 hours.

If the 'Battery Booster' is used it will run for 84 ks (23 hours). This why they are claiming a 6x improvement.

The issue here is that the 1.35V undervoltage threshold is really high. The IEC 60086-2 test use thresholds of 1.05 or even 0.9V for some devices.

A device with a 1.35V (per cell) under voltage threshold is poorly designed in my opinion.

If I look at an undervoltage threshold of 1.1V:



There is no improvement from the  'Battery_Booster', below 1.1V the devices will last longer without the 'Battery_Booster'.

Regards,

Jay_Diddy_B

 
« Last Edit: August 15, 2015, 10:56:36 AM by Jay_Diddy_B »
 

Offline pickle9000

  • Super Contributor
  • ***
  • Posts: 2131
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #12 on: August 15, 2015, 11:05:58 AM »
Is Fig7 in the patent the batterizer or what they claim to be a typical use curve?
 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #13 on: August 15, 2015, 11:32:22 AM »
Is Fig7 in the patent the batterizer or what they claim to be a typical use curve?

I believe it is a typical use case, to support their claim that the batteries last 4-7 times longer.

They test three name brand batteries with a 50mA constant current load, with and without the Boost circuit.

They choose a very high 1.35V cut off voltage to support their claims.

It is a Patent Application. I am not sure if the patent has been granted.

Jay_Diddy_B
 

Online edavid

  • Super Contributor
  • ***
  • Posts: 2371
  • Country: us
Re: Modelling an Alkaline Battery in LTspice
« Reply #14 on: August 16, 2015, 01:04:07 AM »
Jay, amazing work!

Have you looked at adding a bypass mode to your booster?

(Does anyone know if the Batteriser has a bypass FET?)
 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #15 on: August 16, 2015, 01:55:34 AM »
Jay, amazing work!

Have you looked at adding a bypass mode to your booster?

(Does anyone know if the Batteriser has a bypass FET?)


To keep it simple I have not added a bypass mode. It would be fairly easy to do I would just change the efficiency of the 'Battery_Booster' as function of input to output voltage.

In fact, in practice, you get the bypass mode for free in a synchronous boost converter. If the input is equal to the desired output voltage the boost regulator will stop switching and keep the top FET on 100% of the time.

You have the opposite problem that you have a path that is only limited by the ESR of the battery and the voltage drop of the body diode if you short the output.

Jay_Diddy_B


 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #16 on: August 16, 2015, 02:28:03 AM »
Hi group,

Here the model for a 'Battery_Booster' with a bypass:



And the results show:




Horizontal axis is input voltage.

The rules are:

Vout = Vin if Vin > Vboost (Bypass Mode)

else Vout = Vboost if Vin >0.7V

else if Vin < 0.7V then Vout = 0

and

if Vin > Vboost then efficiency = 99% (Bypasss mode)

else

efficiency = 95%

The change in efficiency will have very little impact on the battery life.

I have attached a zipfile with the Battery Booster model for those playing along at home.


Regards,

Jay_Diddy_B


« Last Edit: August 16, 2015, 02:31:50 AM by Jay_Diddy_B »
 

Offline Stephan_T

  • Contributor
  • Posts: 35
  • Country: de
Re: Modelling an Alkaline Battery in LTspice
« Reply #17 on: August 20, 2015, 09:06:21 AM »
Hi Jay,

great thread!

Hi dcac,


[...]
Here's the patent text 'explaining' fig 7:
Quote
FIG. 7 shows measurements that illustrate the advantages of the various embodiments. Three popular AA battery brands, Panasonic, Duracell and Sony were chosen for the measurements. Active load circuitry that drew a fixed 50 mA current was placed at the output of these batteries and the voltage of each battery was measured over time. The horizontal access shows time and the vertical access shows the battery voltage. The starting voltage for these fresh batteries was 1.6V. The amount of time it takes for the batteries to reach 1.39V, which is where a lot of electronic equipment stop operating, are listed. [(A)] The Panasonic battery took 6.3 hours to reach that level, while it took 4.5 hours for the Sony battery. [(B)] The Panasonic battery when used in conjunction with a regulator, according to embodiments of the invention, took 27.9 hours before it stopped providing 1.5V, and the Sony battery when used with a regulator took 32 hours before its stopped providing 1.5V. Thus, with the regulator, it takes 4.5 to 7 times longer before the battery needs to be replaced. Thus, the total number of batteries that need to be manufactured and consequently discarded would be reduced by 4 to 7 times.

I've no idea what the horizontal line at 1.8V is supposed represent, but I believe the vertical lines shows where *their* 'Regulator' stops functioning. The 1.39V which they claim "is where a lot of electronic equipment stop operating" isn't shown, and probably the cause of much confusion. Also 1.35V is mentioned in the table underneath but isn't drawn in fig 7 either.

I have added some colored annotations to the graphic that make things a bit more obvious. The original patent image only shows the dropout values of the 'regulated batteries' (B). What they don't present is the ridiculous dropout value (A) "where a lot of electronic equipment stop operating" which they use for comparison.
I have also added a column (C) to the table for a dropout voltage of 1V, a conservative interpretation of the IEC 60086-2 test threshold.
This demonstrates how much less performance the regulator will provide, than (IEC-) standard test procedure expects.
Just compare (A) to (C) and everyone can see how stupid the claim of the 800% really is.
53.4 / 6.3:  Standard IEC tests of the Panasonic battery gets almost 8.5 times the runtime than what they compare their product with. :palm:
49.4 / 4.5: And for the Sony battery, the IEC test gets even 11 times of their benchmark.  :palm::palm:

PS. Has anyone any idea, what the pink area on the right side of the diagram represents? Constant 50mA at 0.1V for as long as I want? My bullshit detector smells a potential free energy source of 5mW of power.
 

Offline timofonic

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: es
  • Eternal Wannabe Geek
Re: Modelling an Alkaline Battery in LTspice
« Reply #18 on: August 20, 2015, 11:42:37 AM »
Hello.

I'm a total n00b to simulators.

Anyone knows if I can import your model to QUCS?

Call me an Open Source zealot, but I prefer to use FOSS whenever possible.

I'll try it this week, but I'm totally dumb with simulators and I even need to improve a lot at electronics theory.
 

Online Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 1454
  • Country: ca
Re: Modelling an Alkaline Battery in LTspice
« Reply #19 on: August 20, 2015, 01:11:18 PM »
Hello.

I'm a total n00b to simulators.

Anyone knows if I can import your model to QUCS?

Call me an Open Source zealot, but I prefer to use FOSS whenever possible.

I'll try it this week, but I'm totally dumb with simulators and I even need to improve a lot at electronics theory.

It may be possible to translate the  model to QUCS. I am not familiar with QUCS and did not find a reference manual.

I translated the model from PSPICE.

You would need to check if QUCS supports behavioral Current & voltage sources and the table function.

LTspice is not open source, but it is free and very widely used.

I would try downloading LTspice from here:

http://www.linear.com/designtools/software/

Regards,

Jay_Diddy_B
 

Offline CustomEngineerer

  • Frequent Contributor
  • **
  • Posts: 337
  • Country: us
Re: Modelling an Alkaline Battery in LTspice
« Reply #20 on: November 17, 2015, 04:28:29 PM »
we agree with you that voltage run time under load would be different from no load condition, which was the whole point of the Monkey video to demonstrate that batteries are different than power supplies LTspices. The use of a bench power supply LTspice (with no current limit setting) to determine device cut off voltage run times is inaccurate. Prior to our video release, the most important battery parameter, its Internal Resistance , was being ignored, because no one discovered Internal Resistance until our Engineerers came along.
Pls stay tuned for our next video explaining our complete lack of understanding of technical details.
 

Offline timofonic

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: es
  • Eternal Wannabe Geek
Re: Modelling an Alkaline Battery in LTspice
« Reply #21 on: November 18, 2015, 12:46:28 AM »
we agree with you that voltage run time under load would be different from no load condition, which was the whole point of the Monkey video to demonstrate that batteries are different than power supplies LTspices. The use of a bench power supply LTspice (with no current limit setting) to determine device cut off voltage run times is inaccurate. Prior to our video release, the most important battery parameter, its Internal Resistance , was being ignored, because no one discovered Internal Resistance until our Engineerers came along.
Pls stay tuned for our next video explaining our complete lack of understanding of technical details.
Who are you? :)
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf