LiFePO4 Battery Packs Small vs Large Cells

[Roger Heuckeroth]

This may be obvious to some of you, but I have been deliberating over  
LiFePO4 battery options for some time now and have come to the  
following conclusions.  I figure I'll put this out there and see if  
anyone disagrees.

The conclusion that I have come to is that building packs from  
multiple smaller cells is better that using the larger prismatic  
cells... at least for EV applications.  For the following reasons:

1.  Smaller cells are typically capable of higher C discharge and  
charge rates.  I don't know why this is, but it seams to be the case  
almost clear across the board.  Higher C rates allow for better peak  
performance (acceleration, climbing hills).  Also the need for  
expensive Ultra-caps becomes a mute point.
2.  The smaller cells are manufactured in a more controlled manner  
sometimes by the tens of thousands per day.  All by machines.  All  
these machines are designed and built in the USA or Europe, even if  
the manufacturing is done in China.  I hope I don't offend anybody's  
nationalism, but keeping unskilled labor out of the process would lead  
to less variation from cell to cell within a given batch. Many of the  
larger cells are still built in part with manual labor.
3.  Cells still vary in performance, however, when you tie a large  
number of them in parallel they have to act as one large cell.  The  
characteristics variations from one cell to another are averaged out.  
Statistically speaking you are building an averaging circuit by  
putting them in parallel. All of them share the same voltage from full  
charge to full discharge.  This is of course figuring that you are  
putting parallel strings in series, not series strings in parallel.
4.  By virtue of the averaging of parallel cell performance I believe  
you increase the performance of the whole pack.  Theoretically, if all  
parallel strings have the same capacity you don't even need a BMS...  
OK some of you are thinking " oh yes you do."  You do need a simple  
charge and discharge limit control, but you do not need an active  
equalization system that shuffles power about while your discharging  
the cells.  With a well balanced (averaged out) pack all the cells  
will hit bottom at about the same time.  Without active charge  
equalization a series pack of larger cells is limited by its weakest  
cell.  It is the first to hit full charge causing wasted charging  
energy as its over-voltage protection circuit shunts current around  
the cell, and its the first to trigger the low voltage shut off  
circuit once it bottoms out.  There could still be several kWhrs of  
energy left in the other cells, but to protect the week one from over  
discharge the whole pack shuts down.  It takes a very sophisticated  
BMS system to actively shuffle power about while the pack is being  
discharged.
5.  Small cells give you more design flexibility to custom design the  
pack to fit what space you have available.  When I was looking at  
prismatic cells I was going to have split the pack up into several  
areas.  With the smaller cell pack it looks like I can fit the whole  
pack 300V, 15kWhr in one spot, and it looks like it will be about 150  
lbs less than the prismatic cells.
6.  Should the pack need active cooling there is space between the  
cells for air flow by nature of the design.
7.  Since smaller cells are more of a commodity, their price should  
come down faster than the larger cells in the future.
8.  Simpler BMS, or did I already say that in #4.  Simple over/under  
voltage protection should work fine.

Ok so there are a few disadvantages:

1.  You need a battery tab welder to construct the pack.  A bit  
pricey, but actually, for me it will be a business deduction, so that  
helps.  Or you have to have someone else construct the pack for you  
that has the right equipment and expertise.
2.  You have to build your own custom battery case.  I'll fabricate my  
own... we do a lot of plastic fabrication in one of my businesses.
3.  If you don't do a good job at welding the tabs, one could come  
loose throwing the pack off.  If the pack is properly built this  
should be easy to diagnose and fix.  Not so easy if all parallel  
strings are welded together like the SSI Racing A123 pack (no  
offense).  However, get your weld right and there should be no problem.
4.  More time labor intensive to put the pack together.  But then  
again I have probably spent more time researching this stuff than it  
will actually take me to build it.

So, who agrees with the above and who disagrees?


_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Jack Riggi]

I don't know big or small witch is best but check this out for a tab welder,
to bad it sounds like you already have bought one. they say for around 100
bucks    http://www.ledhacks.com/power/battery_tab_welder.htm
----- Original Message -----
From: "Roger Heuckeroth" <[hidden email]>
To: "Electric Vehicle Discussion List" <[hidden email]>
Sent: Thursday, July 24, 2008 11:49 PM
Subject: [EVDL] LiFePO4 Battery Packs Small vs Large Cells

> This may be obvious to some of you, but I have been deliberating over
> LiFePO4 battery options for some time now and have come to the
> following conclusions.  I figure I'll put this out there and see if
> anyone disagrees.
>
> The conclusion that I have come to is that building packs from
> multiple smaller cells is better that using the larger prismatic
> cells... at least for EV applications.  For the following reasons:
>
> 1.  Smaller cells are typically capable of higher C discharge and
> charge rates.  I don't know why this is, but it seams to be the case
> almost clear across the board.  Higher C rates allow for better peak
> performance (acceleration, climbing hills).  Also the need for
> expensive Ultra-caps becomes a mute point.
> 2.  The smaller cells are manufactured in a more controlled manner
> sometimes by the tens of thousands per day.  All by machines.  All
> these machines are designed and built in the USA or Europe, even if
> the manufacturing is done in China.  I hope I don't offend anybody's
> nationalism, but keeping unskilled labor out of the process would lead
> to less variation from cell to cell within a given batch. Many of the
> larger cells are still built in part with manual labor.
> 3.  Cells still vary in performance, however, when you tie a large
> number of them in parallel they have to act as one large cell.  The
> characteristics variations from one cell to another are averaged out.
> Statistically speaking you are building an averaging circuit by
> putting them in parallel. All of them share the same voltage from full
> charge to full discharge.  This is of course figuring that you are
> putting parallel strings in series, not series strings in parallel.
> 4.  By virtue of the averaging of parallel cell performance I believe
> you increase the performance of the whole pack.  Theoretically, if all
> parallel strings have the same capacity you don't even need a BMS...
> OK some of you are thinking " oh yes you do."  You do need a simple
> charge and discharge limit control, but you do not need an active
> equalization system that shuffles power about while your discharging
> the cells.  With a well balanced (averaged out) pack all the cells
> will hit bottom at about the same time.  Without active charge
> equalization a series pack of larger cells is limited by its weakest
> cell.  It is the first to hit full charge causing wasted charging
> energy as its over-voltage protection circuit shunts current around
> the cell, and its the first to trigger the low voltage shut off
> circuit once it bottoms out.  There could still be several kWhrs of
> energy left in the other cells, but to protect the week one from over
> discharge the whole pack shuts down.  It takes a very sophisticated
> BMS system to actively shuffle power about while the pack is being
> discharged.
> 5.  Small cells give you more design flexibility to custom design the
> pack to fit what space you have available.  When I was looking at
> prismatic cells I was going to have split the pack up into several
> areas.  With the smaller cell pack it looks like I can fit the whole
> pack 300V, 15kWhr in one spot, and it looks like it will be about 150
> lbs less than the prismatic cells.
> 6.  Should the pack need active cooling there is space between the
> cells for air flow by nature of the design.
> 7.  Since smaller cells are more of a commodity, their price should
> come down faster than the larger cells in the future.
> 8.  Simpler BMS, or did I already say that in #4.  Simple over/under
> voltage protection should work fine.
>
> Ok so there are a few disadvantages:
>
> 1.  You need a battery tab welder to construct the pack.  A bit
> pricey, but actually, for me it will be a business deduction, so that
> helps.  Or you have to have someone else construct the pack for you
> that has the right equipment and expertise.
> 2.  You have to build your own custom battery case.  I'll fabricate my
> own... we do a lot of plastic fabrication in one of my businesses.
> 3.  If you don't do a good job at welding the tabs, one could come
> loose throwing the pack off.  If the pack is properly built this
> should be easy to diagnose and fix.  Not so easy if all parallel
> strings are welded together like the SSI Racing A123 pack (no
> offense).  However, get your weld right and there should be no problem.
> 4.  More time labor intensive to put the pack together.  But then
> again I have probably spent more time researching this stuff than it
> will actually take me to build it.
>
> So, who agrees with the above and who disagrees?
>
>
> _______________________________________________
> For general EVDL support, see http://evdl.org/help/
> For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev 

_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[SteveS-5]

I'll be interested to see what the consensus is on  BMS for small vs.
large cell packs. What you say makes sense, but is that how it really
works? What do they do in the drag bikes that use the A123 cells?

How does pricing compare?

The tab welding is what scares me off the small packs. From what I've
seen you need a real tab welder to be sure not to damage the cells and
they are expensive.

I do like the idea that small cells give you a lot more freedom in
designing the pack form factor.

- SteveS

Roger Heuckeroth wrote:

> This may be obvious to some of you, but I have been deliberating over  
> LiFePO4 battery options for some time now and have come to the  
> following conclusions.  I figure I'll put this out there and see if  
> anyone disagrees.
>
> The conclusion that I have come to is that building packs from  
> multiple smaller cells is better that using the larger prismatic  
> cells... at least for EV applications.  For the following reasons:
>
> 1.  Smaller cells are typically capable of higher C discharge and  
> charge rates.  I don't know why this is, but it seams to be the case  
> almost clear across the board.  Higher C rates allow for better peak  
> performance (acceleration, climbing hills).  Also the need for  
> expensive Ultra-caps becomes a mute point.
> 2.  The smaller cells are manufactured in a more controlled manner  
> sometimes by the tens of thousands per day.  All by machines.  All  
> these machines are designed and built in the USA or Europe, even if  
> the manufacturing is done in China.  I hope I don't offend anybody's  
> nationalism, but keeping unskilled labor out of the process would lead  
> to less variation from cell to cell within a given batch. Many of the  
> larger cells are still built in part with manual labor.
> 3.  Cells still vary in performance, however, when you tie a large  
> number of them in parallel they have to act as one large cell.  The  
> characteristics variations from one cell to another are averaged out.  
> Statistically speaking you are building an averaging circuit by  
> putting them in parallel. All of them share the same voltage from full  
> charge to full discharge.  This is of course figuring that you are  
> putting parallel strings in series, not series strings in parallel.
> 4.  By virtue of the averaging of parallel cell performance I believe  
> you increase the performance of the whole pack.  Theoretically, if all  
> parallel strings have the same capacity you don't even need a BMS...  
> OK some of you are thinking " oh yes you do."  You do need a simple  
> charge and discharge limit control, but you do not need an active  
> equalization system that shuffles power about while your discharging  
> the cells.  With a well balanced (averaged out) pack all the cells  
> will hit bottom at about the same time.  Without active charge  
> equalization a series pack of larger cells is limited by its weakest  
> cell.  It is the first to hit full charge causing wasted charging  
> energy as its over-voltage protection circuit shunts current around  
> the cell, and its the first to trigger the low voltage shut off  
> circuit once it bottoms out.  There could still be several kWhrs of  
> energy left in the other cells, but to protect the week one from over  
> discharge the whole pack shuts down.  It takes a very sophisticated  
> BMS system to actively shuffle power about while the pack is being  
> discharged.
> 5.  Small cells give you more design flexibility to custom design the  
> pack to fit what space you have available.  When I was looking at  
> prismatic cells I was going to have split the pack up into several  
> areas.  With the smaller cell pack it looks like I can fit the whole  
> pack 300V, 15kWhr in one spot, and it looks like it will be about 150  
> lbs less than the prismatic cells.
> 6.  Should the pack need active cooling there is space between the  
> cells for air flow by nature of the design.
> 7.  Since smaller cells are more of a commodity, their price should  
> come down faster than the larger cells in the future.
> 8.  Simpler BMS, or did I already say that in #4.  Simple over/under  
> voltage protection should work fine.
>
> Ok so there are a few disadvantages:
>
> 1.  You need a battery tab welder to construct the pack.  A bit  
> pricey, but actually, for me it will be a business deduction, so that  
> helps.  Or you have to have someone else construct the pack for you  
> that has the right equipment and expertise.
> 2.  You have to build your own custom battery case.  I'll fabricate my  
> own... we do a lot of plastic fabrication in one of my businesses.
> 3.  If you don't do a good job at welding the tabs, one could come  
> loose throwing the pack off.  If the pack is properly built this  
> should be easy to diagnose and fix.  Not so easy if all parallel  
> strings are welded together like the SSI Racing A123 pack (no  
> offense).  However, get your weld right and there should be no problem.
> 4.  More time labor intensive to put the pack together.  But then  
> again I have probably spent more time researching this stuff than it  
> will actually take me to build it.
>
> So, who agrees with the above and who disagrees?
>
>
> _______________________________________________
> For general EVDL support, see http://evdl.org/help/
> For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev
>
>
> No virus found in this incoming message.
> Checked by AVG - http://www.avg.com 
> Version: 8.0.138 / Virus Database: 270.5.5/1570 - Release Date: 7/24/2008 6:59 AM
>
>
>
>  

_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Roger Heuckeroth]

I came across that on the internet. when doing a search.  There's a  
bunch of homebuilt plans out there.  Its tempting to build my own, but  
good welds are a critical part of the assembly process.  Typically  
there are multiple welds on each battery side.

Years ago I built my own vacuum forming machine for my business.  We  
were making parts 1-10 at a time and no professional shop wanted to  
take my business.  When they set up a machine they want to do 100s or  
1000s of parts.  I also made my own wooden molds.  It actually worked  
out terrific.  I built the whole machine for under $500.  Vs $20,000  
for a used machine.  We still use this vacuum former today.  Its  
definitely slower than the production machines out there, but this  
particular product is still built in small runs with frequent mold  
changes.

However, a professional dual pulse battery tab welder will probably do  
a much better weld than something than something I throw together.  I  
don't really have a surplus of time these days, and my business could  
use some write-offs, so I'll probably go ahead and buy one if I go  
this route.  Also, this is my first conversion, but may not be my  
last.  I may even get into the custom battery pack building  
business... who knows.


On Jul 25, 2008, at 8:26 AM, Jack Riggi wrote:

> I don't know big or small witch is best but check this out for a tab  
> welder,
> to bad it sounds like you already have bought one. they say for  
> around 100
> bucks    http://www.ledhacks.com/power/battery_tab_welder.htm
> ----- Original Message -----
> From: "Roger Heuckeroth" <[hidden email]>
> To: "Electric Vehicle Discussion List" <[hidden email]>
> Sent: Thursday, July 24, 2008 11:49 PM
> Subject: [EVDL] LiFePO4 Battery Packs Small vs Large Cells
>
>
>> This may be obvious to some of you, but I have been deliberating over
>> LiFePO4 battery options for some time now and have come to the
>> following conclusions.  I figure I'll put this out there and see if
>> anyone disagrees.
>>
>> The conclusion that I have come to is that building packs from
>> multiple smaller cells is better that using the larger prismatic
>> cells... at least for EV applications.  For the following reasons:
>>
>> 1.  Smaller cells are typically capable of higher C discharge and
>> charge rates.  I don't know why this is, but it seams to be the case
>> almost clear across the board.  Higher C rates allow for better peak
>> performance (acceleration, climbing hills).  Also the need for
>> expensive Ultra-caps becomes a mute point.
>> 2.  The smaller cells are manufactured in a more controlled manner
>> sometimes by the tens of thousands per day.  All by machines.  All
>> these machines are designed and built in the USA or Europe, even if
>> the manufacturing is done in China.  I hope I don't offend anybody's
>> nationalism, but keeping unskilled labor out of the process would  
>> lead
>> to less variation from cell to cell within a given batch. Many of the
>> larger cells are still built in part with manual labor.
>> 3.  Cells still vary in performance, however, when you tie a large
>> number of them in parallel they have to act as one large cell.  The
>> characteristics variations from one cell to another are averaged out.
>> Statistically speaking you are building an averaging circuit by
>> putting them in parallel. All of them share the same voltage from  
>> full
>> charge to full discharge.  This is of course figuring that you are
>> putting parallel strings in series, not series strings in parallel.
>> 4.  By virtue of the averaging of parallel cell performance I believe
>> you increase the performance of the whole pack.  Theoretically, if  
>> all
>> parallel strings have the same capacity you don't even need a BMS...
>> OK some of you are thinking " oh yes you do."  You do need a simple
>> charge and discharge limit control, but you do not need an active
>> equalization system that shuffles power about while your discharging
>> the cells.  With a well balanced (averaged out) pack all the cells
>> will hit bottom at about the same time.  Without active charge
>> equalization a series pack of larger cells is limited by its weakest
>> cell.  It is the first to hit full charge causing wasted charging
>> energy as its over-voltage protection circuit shunts current around
>> the cell, and its the first to trigger the low voltage shut off
>> circuit once it bottoms out.  There could still be several kWhrs of
>> energy left in the other cells, but to protect the week one from over
>> discharge the whole pack shuts down.  It takes a very sophisticated
>> BMS system to actively shuffle power about while the pack is being
>> discharged.
>> 5.  Small cells give you more design flexibility to custom design the
>> pack to fit what space you have available.  When I was looking at
>> prismatic cells I was going to have split the pack up into several
>> areas.  With the smaller cell pack it looks like I can fit the whole
>> pack 300V, 15kWhr in one spot, and it looks like it will be about 150
>> lbs less than the prismatic cells.
>> 6.  Should the pack need active cooling there is space between the
>> cells for air flow by nature of the design.
>> 7.  Since smaller cells are more of a commodity, their price should
>> come down faster than the larger cells in the future.
>> 8.  Simpler BMS, or did I already say that in #4.  Simple over/under
>> voltage protection should work fine.
>>
>> Ok so there are a few disadvantages:
>>
>> 1.  You need a battery tab welder to construct the pack.  A bit
>> pricey, but actually, for me it will be a business deduction, so that
>> helps.  Or you have to have someone else construct the pack for you
>> that has the right equipment and expertise.
>> 2.  You have to build your own custom battery case.  I'll fabricate  
>> my
>> own... we do a lot of plastic fabrication in one of my businesses.
>> 3.  If you don't do a good job at welding the tabs, one could come
>> loose throwing the pack off.  If the pack is properly built this
>> should be easy to diagnose and fix.  Not so easy if all parallel
>> strings are welded together like the SSI Racing A123 pack (no
>> offense).  However, get your weld right and there should be no  
>> problem.
>> 4.  More time labor intensive to put the pack together.  But then
>> again I have probably spent more time researching this stuff than it
>> will actually take me to build it.
>>
>> So, who agrees with the above and who disagrees?
>>
>>
>> _______________________________________________
>> For general EVDL support, see http://evdl.org/help/
>> For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev
>
>
> _______________________________________________
> For general EVDL support, see http://evdl.org/help/
> For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ 
> ev

_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Roger Heuckeroth]

 From what I have seen, joining cells in parallel, and then in series  
is how all the pros do it.  Parallel strings are forced to run at the  
same voltage, so they hit 3.65 V all at the same time, and 2.5 volts  
all at the same time, so you only have to monitor and control one  
voltage for each parallel string.  So BMS is the same architecture as  
with the large cells.

A123s are still super expensive, however, you can get cells that will  
be comparable to TS cells in $/Ahr, and are a hell of a lot more  
potent... 10X current with minimal voltage sag compared to TS blocks  
of the same Ahr.

Yes, welds are critical.  Good equipment that is specific to the job  
is key.

On Jul 25, 2008, at 8:50 AM, SteveS wrote:

> I'll be interested to see what the consensus is on  BMS for small vs.
> large cell packs. What you say makes sense, but is that how it really
> works? What do they do in the drag bikes that use the A123 cells?
>
> How does pricing compare?
>
> The tab welding is what scares me off the small packs. From what I've
> seen you need a real tab welder to be sure not to damage the cells and
> they are expensive.
>
> I do like the idea that small cells give you a lot more freedom in
> designing the pack form factor.
>
> - SteveS
>
> Roger Heuckeroth wrote:
>> This may be obvious to some of you, but I have been deliberating over
>> LiFePO4 battery options for some time now and have come to the
>> following conclusions.  I figure I'll put this out there and see if
>> anyone disagrees.
>>
>> The conclusion that I have come to is that building packs from
>> multiple smaller cells is better that using the larger prismatic
>> cells... at least for EV applications.  For the following reasons:
>>
>> 1.  Smaller cells are typically capable of higher C discharge and
>> charge rates.  I don't know why this is, but it seams to be the case
>> almost clear across the board.  Higher C rates allow for better peak
>> performance (acceleration, climbing hills).  Also the need for
>> expensive Ultra-caps becomes a mute point.
>> 2.  The smaller cells are manufactured in a more controlled manner
>> sometimes by the tens of thousands per day.  All by machines.  All
>> these machines are designed and built in the USA or Europe, even if
>> the manufacturing is done in China.  I hope I don't offend anybody's
>> nationalism, but keeping unskilled labor out of the process would  
>> lead
>> to less variation from cell to cell within a given batch. Many of the
>> larger cells are still built in part with manual labor.
>> 3.  Cells still vary in performance, however, when you tie a large
>> number of them in parallel they have to act as one large cell.  The
>> characteristics variations from one cell to another are averaged out.
>> Statistically speaking you are building an averaging circuit by
>> putting them in parallel. All of them share the same voltage from  
>> full
>> charge to full discharge.  This is of course figuring that you are
>> putting parallel strings in series, not series strings in parallel.
>> 4.  By virtue of the averaging of parallel cell performance I believe
>> you increase the performance of the whole pack.  Theoretically, if  
>> all
>> parallel strings have the same capacity you don't even need a BMS...
>> OK some of you are thinking " oh yes you do."  You do need a simple
>> charge and discharge limit control, but you do not need an active
>> equalization system that shuffles power about while your discharging
>> the cells.  With a well balanced (averaged out) pack all the cells
>> will hit bottom at about the same time.  Without active charge
>> equalization a series pack of larger cells is limited by its weakest
>> cell.  It is the first to hit full charge causing wasted charging
>> energy as its over-voltage protection circuit shunts current around
>> the cell, and its the first to trigger the low voltage shut off
>> circuit once it bottoms out.  There could still be several kWhrs of
>> energy left in the other cells, but to protect the week one from over
>> discharge the whole pack shuts down.  It takes a very sophisticated
>> BMS system to actively shuffle power about while the pack is being
>> discharged.
>> 5.  Small cells give you more design flexibility to custom design the
>> pack to fit what space you have available.  When I was looking at
>> prismatic cells I was going to have split the pack up into several
>> areas.  With the smaller cell pack it looks like I can fit the whole
>> pack 300V, 15kWhr in one spot, and it looks like it will be about 150
>> lbs less than the prismatic cells.
>> 6.  Should the pack need active cooling there is space between the
>> cells for air flow by nature of the design.
>> 7.  Since smaller cells are more of a commodity, their price should
>> come down faster than the larger cells in the future.
>> 8.  Simpler BMS, or did I already say that in #4.  Simple over/under
>> voltage protection should work fine.
>>
>> Ok so there are a few disadvantages:
>>
>> 1.  You need a battery tab welder to construct the pack.  A bit
>> pricey, but actually, for me it will be a business deduction, so that
>> helps.  Or you have to have someone else construct the pack for you
>> that has the right equipment and expertise.
>> 2.  You have to build your own custom battery case.  I'll fabricate  
>> my
>> own... we do a lot of plastic fabrication in one of my businesses.
>> 3.  If you don't do a good job at welding the tabs, one could come
>> loose throwing the pack off.  If the pack is properly built this
>> should be easy to diagnose and fix.  Not so easy if all parallel
>> strings are welded together like the SSI Racing A123 pack (no
>> offense).  However, get your weld right and there should be no  
>> problem.
>> 4.  More time labor intensive to put the pack together.  But then
>> again I have probably spent more time researching this stuff than it
>> will actually take me to build it.
>>
>> So, who agrees with the above and who disagrees?
>>
>>
>> _______________________________________________
>> For general EVDL support, see http://evdl.org/help/
>> For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev
>>
>>
>> No virus found in this incoming message.
>> Checked by AVG - http://www.avg.com
>> Version: 8.0.138 / Virus Database: 270.5.5/1570 - Release Date:  
>> 7/24/2008 6:59 AM
>>
>>
>>
>>
>
>
> _______________________________________________
> For general EVDL support, see http://evdl.org/help/
> For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ 
> ev

_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Janet Plato-2]

> The conclusion that I have come to is that building packs from
> multiple smaller cells is better that using the larger prismatic
> cells... at least for EV applications.  For the following reasons:
>
> 1.  Smaller cells are typically capable of higher C discharge and
> charge rates.  I don't know why this is, but it seams to be the case
> almost clear across the board.  Higher C rates allow for better peak
> performance (acceleration, climbing hills).  Also the need for
> expensive Ultra-caps becomes a mute point.

  I am not sure, but I seem to recall the prismatic cells had
higher internal resistance owing to the way the material was
packed into the corners of the square space.  The material is
further away and has a longer path to egress the cell, hence
higher resistance.

Perhaps someone on the list can confirm that.

Janet


     


_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Lee Hart]

Roger Heuckeroth wrote:
>  From what I have seen, joining cells in parallel, and then in series  
> is how all the pros do it.

Except that nobody is really a "pro" at this, except in the sense that
they are paid for their labor. The whole concept of paralleling many
small cells is an experiment. No one really knows how well it will work
long-term, or the risks as the number of paralleled cells increases.

Lots of people are trying it, because they can't get large cells. It's
probably a solution born out of expediency and necessity, rather than
because it's a good idea.

There are some pretty serious risks involved. If the cells aren't very
consistent, they won't share the load well. You have to test every cell,
as bad ones are virtually impossible to find or replace. Paralleled
lithiums don't self-equalize at the same state of charge well.
Inconsistencies in the welds will cause more problems. Cooling becomes
an issue. If a cell shorts, the ones in parallel dump *all* their energy
  into it, creating a fire hazard.
--
Ring the bells that still can ring
Forget the perfect offering
There is a crack in everything
That's how the light gets in    --    Leonard Cohen
--
Lee A. Hart, 814 8th Ave N, Sartell MN 56377, leeahart_at_earthlink.net


_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Roger Heuckeroth]

Lee,

Thanks for you feedback.

Its not because we can't get larger cells.  Its because the smaller  
cells have a higher C rating with less voltage sag.  It would be more  
expedient just to buy TS cells and wire them up, but the performance  
is not there.

I think it would be smart to test every cell first... just a simple  
test to measure discharge / charge capability and to make sure the  
cell isn't a dud would suffice.

I don't understand you theory about paralleled cells not  self  
equalizing.  Can you expand on that?  The way I see it they would all  
act together at the best of their ability, but they are all tied  
together, so they can not work at different voltages.   If 3.65 V is  
100% SOC and 2.5V is say 5% SOC then if after you load them for a  
while they will all balance out at a certain voltage and SOC.  No  
single cell will hit 2.5V before another.  Sure the stronger ones will  
do a higher proportion of the work, but the all are forced to work as  
a team.  Help me out here, I don't understand how they can not be  
equalized?

Large cells an develop an internal short also.  This normally comes  
from lithium plating if you over discharge a cell.  How would a short  
in a parallel pack be any different than a short in a large cell?  
Also if a short did happen because of a failed vow voltage cut off  
circuit for example, it would not accept a charge, and your BMS would  
notice this upon the next charge.  We are talking LiFePO4 cells here  
that can be completely shorted out and not cause a fire.

Roger

On Jul 25, 2008, at 11:21 AM, Lee Hart wrote:

> Roger Heuckeroth wrote:
>> From what I have seen, joining cells in parallel, and then in series
>> is how all the pros do it.
>
> Except that nobody is really a "pro" at this, except in the sense that
> they are paid for their labor. The whole concept of paralleling many
> small cells is an experiment. No one really knows how well it will  
> work
> long-term, or the risks as the number of paralleled cells increases.
>
> Lots of people are trying it, because they can't get large cells. It's
> probably a solution born out of expediency and necessity, rather than
> because it's a good idea.
>
> There are some pretty serious risks involved. If the cells aren't very
> consistent, they won't share the load well. You have to test every  
> cell,
> as bad ones are virtually impossible to find or replace. Paralleled
> lithiums don't self-equalize at the same state of charge well.
> Inconsistencies in the welds will cause more problems. Cooling becomes
> an issue. If a cell shorts, the ones in parallel dump *all* their  
> energy
>  into it, creating a fire hazard.
> --
> Ring the bells that still can ring
> Forget the perfect offering
> There is a crack in everything
> That's how the light gets in    --    Leonard Cohen
> --
> Lee A. Hart, 814 8th Ave N, Sartell MN 56377,  
> leeahart_at_earthlink.net
>
>
> _______________________________________________
> For general EVDL support, see http://evdl.org/help/
> For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ 
> ev

_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Christopher Frost]

Roger,

Actually I think it's more so that no one wants to pay the premium for the
larger cells. There are larger cells with higher C ratings. TS cells however
do not make that list. :-)

Yes, testing every cell would be smart, though it wouldn't answer everything.

>From what I have read, individual Lithium based cells do not equalize well at
all in strings. This is why Victor (I believe with the Honda CRX?) attached a
BMS to each cell in his pack. The cells don't all charge at the same rate
even if they are the exactly the same down to tested specifications.
Temperature and individual cell wear also effect charging over time.

If stronger cells are strung together with weaker cells the weaker cells with
take a higher hit on the discharge. These cells would not charge to full in
the same timeframe and overcharging lithium based batteries (I am told) isn't
the wisest.

With Lead or Nickel based battery technologies this isn't as much of a problem
due to overcharge abilities.

To do this properly you really should have a decent battery management system
worked out. I haven't personally done a lot of testing on these batteries,
I've just been doing several years of reading. I could be 100% wrong as
always. :-)

Christopher

On Friday 25 July 2008 01:14:41 pm Roger Heuckeroth wrote:

> Lee,
>
> Thanks for you feedback.
>
> Its not because we can't get larger cells.  Its because the smaller
> cells have a higher C rating with less voltage sag.  It would be more
> expedient just to buy TS cells and wire them up, but the performance
> is not there.
>
> I think it would be smart to test every cell first... just a simple
> test to measure discharge / charge capability and to make sure the
> cell isn't a dud would suffice.
>
> I don't understand you theory about paralleled cells not  self
> equalizing.  Can you expand on that?  The way I see it they would all
> act together at the best of their ability, but they are all tied
> together, so they can not work at different voltages.   If 3.65 V is
> 100% SOC and 2.5V is say 5% SOC then if after you load them for a
> while they will all balance out at a certain voltage and SOC.  No
> single cell will hit 2.5V before another.  Sure the stronger ones will
> do a higher proportion of the work, but the all are forced to work as
> a team.  Help me out here, I don't understand how they can not be
> equalized?
>
> Large cells an develop an internal short also.  This normally comes
> from lithium plating if you over discharge a cell.  How would a short
> in a parallel pack be any different than a short in a large cell?
> Also if a short did happen because of a failed vow voltage cut off
> circuit for example, it would not accept a charge, and your BMS would
> notice this upon the next charge.  We are talking LiFePO4 cells here
> that can be completely shorted out and not cause a fire.
>
> Roger

_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Roger Heuckeroth]

Christopher,

Thanks for your take on it.

What larger cells with higher C-ratings are you referring to?  I have  
not come across any large LiFePO4 cells, 50 Ahr or above, that have a  
10 C continuous rating.  The best I have seen is 5C, but that was with  
voltage sagging down to 2.6V.

Victor used a series string, no parallel cells in his pack as far as  
I'm aware of.

We use BMS as a catch all term that can mean a lot of things.  You  
definitely should have charge regulators to prevent over voltage on  
any one cell. The circuit simply clamps the voltage at some level  
(usually 3.65V) you could leave a cell on this voltage for an extended  
amount of time.  It won't over-charge, it will simply not accept any  
more current.  If you clamp a parallel string at this voltage the  
current will decrease until there is hardly any current flowing.  Each  
parallel string needs one over and one under voltage protection  
circuit.  That's all.

Roger

On Jul 25, 2008, at 1:49 PM, Christopher Frost wrote:

> Roger,
>
> Actually I think it's more so that no one wants to pay the premium  
> for the
> larger cells. There are larger cells with higher C ratings. TS cells  
> however
> do not make that list. :-)
>
> Yes, testing every cell would be smart, though it wouldn't answer  
> everything.
>
>> From what I have read, individual Lithium based cells do not  
>> equalize well at
> all in strings. This is why Victor (I believe with the Honda CRX?)  
> attached a
> BMS to each cell in his pack. The cells don't all charge at the same  
> rate
> even if they are the exactly the same down to tested specifications.
> Temperature and individual cell wear also effect charging over time.
>
> If stronger cells are strung together with weaker cells the weaker  
> cells with
> take a higher hit on the discharge. These cells would not charge to  
> full in
> the same timeframe and overcharging lithium based batteries (I am  
> told) isn't
> the wisest.
>
> With Lead or Nickel based battery technologies this isn't as much of  
> a problem
> due to overcharge abilities.
>
> To do this properly you really should have a decent battery  
> management system
> worked out. I haven't personally done a lot of testing on these  
> batteries,
> I've just been doing several years of reading. I could be 100% wrong  
> as
> always. :-)
>
> Christopher
>
> On Friday 25 July 2008 01:14:41 pm Roger Heuckeroth wrote:
>> Lee,
>>
>> Thanks for you feedback.
>>
>> Its not because we can't get larger cells.  Its because the smaller
>> cells have a higher C rating with less voltage sag.  It would be more
>> expedient just to buy TS cells and wire them up, but the performance
>> is not there.
>>
>> I think it would be smart to test every cell first... just a simple
>> test to measure discharge / charge capability and to make sure the
>> cell isn't a dud would suffice.
>>
>> I don't understand you theory about paralleled cells not  self
>> equalizing.  Can you expand on that?  The way I see it they would all
>> act together at the best of their ability, but they are all tied
>> together, so they can not work at different voltages.   If 3.65 V is
>> 100% SOC and 2.5V is say 5% SOC then if after you load them for a
>> while they will all balance out at a certain voltage and SOC.  No
>> single cell will hit 2.5V before another.  Sure the stronger ones  
>> will
>> do a higher proportion of the work, but the all are forced to work as
>> a team.  Help me out here, I don't understand how they can not be
>> equalized?
>>
>> Large cells an develop an internal short also.  This normally comes
>> from lithium plating if you over discharge a cell.  How would a short
>> in a parallel pack be any different than a short in a large cell?
>> Also if a short did happen because of a failed vow voltage cut off
>> circuit for example, it would not accept a charge, and your BMS would
>> notice this upon the next charge.  We are talking LiFePO4 cells here
>> that can be completely shorted out and not cause a fire.
>>
>> Roger
>
>
> _______________________________________________
> For general EVDL support, see http://evdl.org/help/
> For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ 
> ev

_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Lee Hart]

Roger Heuckeroth wrote:
> Lee,
>
> Thanks for you feedback.
>
> Its not because we can't get larger cells.  Its because the smaller  
> cells have a higher C rating with less voltage sag.  It would be more  
> expedient just to buy TS cells and wire them up, but the performance  
> is not there.

It's not really there with the small cells, either. One cell by itself
may have a high C rating; but when you bundle large groups of them
together, the composite cell has a lower C rating. This is because you
have added resistance in connecting them in parallel, and because they
heat each other.

You can think of a large cell as many small cells in parallel, sharing
the same package and having one pair of terminals instead of dozens.
Only in this case, the manufacturer has already taken the self-heating
and extra interconnection resistance into account in their ratings.

The advantage of larger cells is that their extremely close proximity
means they all have to be at the same temperature, same age, same
voltage, etc. So, there is a much higher probability that they will all
match.

> I think it would be smart to test every cell first... just a simple  
> test to measure discharge / charge capability and to make sure the  
> cell isn't a dud would suffice.

That's a good start. It will weed out the obviously defective ones. But
if you're going to go to a lot of trouble to package them in a way that
makes in nearly impossible to test or replace individual cells later,
then I think more thorough testing is called for.

> I don't understand you theory about paralleled cells not  self  
> equalizing. Can you expand on that?

Sure. Start by looking at lead-acids:

Lead acid cells wired in parallel tend to self-equalize at the same
state of charge. That's because voltage is a reasonable indicator of
SOC; it changes linearly at about 0.1v change per 10% SOC change for a
12v battery.

Also, the internal resistance of lead acid cells increases as they
discharge. Under load, this means the least-charged cell has the highest
resistance, and so supplies less current. This coupled with the drop in
voltage means that paralleled cells all tend to discharge to the same SOC.

On charge, lead acids can be mildly overcharged without serious harm. If
one parallel cell reaches full first, it simply gets slightly
overcharged while the others come up to full.

Now for lithiums:

With lithium cells, the voltage is virtually flat over a wide range,
from 20%-80% state of charge. Simply wiring cells in parallel does not
mean they are at the same state of charge. Connect a 25% and a 75% SOC
lithium cells in parallel -- the current that flows between them is very
low, so it would take a very long time for them to equalize.

On discharge, the series resistance of each cell does not change. This
means that when you parallel cells, other factors determine how much
current each one supplies. If one happens to have a lower internal
resistance, or better external connections, it supplies the most
current. It gets discharged deeper on each cycle. The higher-resistance
parallel cells don't pick up more of the load until the best cell is
nearly dead. This is a good way to cause some cells to fail sooner than
others.

On charge, you can't overcharge lithiums to equalize them. You have to
depend on the slight difference in voltage between 90% and 100% SOC to
get a little more current into the less charged cells.

Ultimately, you need to TEST your planned setup. Assuming it will work
based on hearsay or wishful thinking could be a very costly mistake! Get
some cells. Put current sensors in series with each one. Charge and
discharge them, and *measure* the current, voltage, and amphours going
in/of of each one. *Prove* that the specific cells and specific
connection method you plan will actually work, before you commit to
using it for thousands of dollars worth of cells.

> Large cells can develop an internal short also. How would a short  
> in a parallel pack be any different than a short in a large cell?  

The sheer mass of the larger cell provides more cooling capacity in case
of a short. The manufacturer will also have designed in "fuses" into the
interconnections inside the cell so that a truely bad internal short is
less likely to spread to become a dangerous failure.

You can do this with paralleled cells by putting a fuse in series with
each (or a thin spot in the interconnects designed to act as a fuse). If
a cell shorts, the total current from the paralleled cells will then
blow this fuse to prevent the problem from spreading.

> Also if a short did happen because of a failed low voltage cut off  
> circuit for example, it would not accept a charge, and your BMS would  
> notice this upon the next charge.  We are talking LiFePO4 cells here  
> that can be completely shorted out and not cause a fire.

A single failed cell may not cause a fire; but that doesn't mean a large
*group* of cells won't. I find the difference between LiCr and LiFe
cells is like the difference between matches and toothpicks. A box of
matches lights easily, and once started, spreads ferociously fast! A box
of toothpicks is harder to light and spreads slower; but is still highly
combustible.

--
Ring the bells that still can ring
Forget the perfect offering
There is a crack in everything
That's how the light gets in    --    Leonard Cohen
--
Lee A. Hart, 814 8th Ave N, Sartell MN 56377, leeahart_at_earthlink.net


_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Roger Heuckeroth]

Lee,

I'm trying to understand all this.

How are we adding resistance by putting cells in parallel.  Doesn't  
resistance follow the rule of

1/Rp = I/R1 + 1/R2 ... + 1/Rn

If each cell had an internal resistance of say 8 mohm, the equivalent  
parallel resistance of say 10 cells in parallel should be 0.8 mohm.  
Am I missing something?

Are you are referring to workmanship issues like sloppy spot welds, or  
not enough weld surface area?

Roger

Lee Hart wrote:
On Jul 25, 2008, at 3:39 PM, Lee Hart wrote:

> It's not really there with the small cells, either. One cell by itself
> may have a high C rating; but when you bundle large groups of them
> together, the composite cell has a lower C rating. This is because you
> have added resistance in connecting them in parallel, and because they
> heat each other.


_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Metric Mind]

Roger Heuckeroth wrote:
 > Christopher,
 >
 > Thanks for your take on it.
 >
 > What larger cells with higher C-ratings are you referring to?  I have
 > not come across any large LiFePO4 cells, 50 Ahr or above, that have a
 > 10 C continuous rating.  The best I have seen is 5C, but that was with
 > voltage sagging down to 2.6V.
 >
 > Victor used a series string, no parallel cells in his pack as far as
 > I'm aware of.


In my opinion, and it based on limited experiments I did, externally
paralleling Li cells is lousy idea unless you can guarantee quality
of such connection to be no worse than internal connections inside
large prismatic cells. As Lee pointed out, in essence OEM manufacturers take
large collection of small cells (in form of internal plates) and do such
packaging for you in one enclosure to get large cells, but there are
critical differences - some ill effect they can minimize with right
process, but you can't if you do it externally.

As far as BMSing and equal SOC if paralleling cells, Lee is right
on the spot. Everyone naturally wants to get away cheap, substituting
voltage for SOC (Ah amount) and it doesn't work very well for LiIons,
worse for LiP, and doesn't work *at all* for LiFePO everyone these days
seems to be locked on. Voltage is easy to measure compared to Ah, so
people pretend that 3.3V = 75% SOC (insert you own numbers here).
Not so.

A "BMS" which maintains voltages equal on each cell at all times
is certainly far better than nothing, but it is NOT really a BMS.
Such BMS has no idea what the state of charge is for each cell. Thus you
cannot put more or less Ah in certain cells to equalize the pack so you
can utilize its full capacity.

You can have a series string of 100Ah pack of LiFePO4 cells and
hardware to ensure each cell has exactly 3.215V on it. Yet, all the
cells could be at 70% SOC and one or few at 20%, so your *useable* pack
capacity is really near 10Ah before lowest cell gets ruined by further
discharge. Worse - you have no way of knowing it even if you suspect
this is the case unless you know actual capacity of each cell and
continuously track Ah in and out of the pack which of course takes more
than voltmeter per cell (in a form of A/D and central computer
collecting data or whatever implementation; these are irrelevant details).

If you want to get away cheap (without tracking Ah, e.g. having
balanced pack at *any* SOC level in the middle where all voltages are
the same and are no indication whatsoever what is going on), you can only
rely on the tail of charging curve where voltage increases sharply -
then the voltage differences up there will tell you if cells are close to
100% or not. If voltages are lower than knee, all you can tell that it
is not
about 85...90% yet, but how far away you are, you still don't know.
You must keep charging individual cells until they are all at the
same voltage *and near voltage indicating 100% of SOC*. THEN
it "resets" the pack and you can be reasonably sure all the cell are at full
capacity. This is very workable solution, in fact I will implement it
as very next pack in my ACRX. The only "inconvenience" (or condition
for this to work if you will) is you must often charge completely,
to near 100% SOC - this is where equalization using this method
will take place. Sort of like Rudman regulators clamping on top, but
otherwise doing nothing in the middle. If you only cycle between 30% and
70% SOC many times,
e.g. do partial charges, it is pretty much guaranteeing you'll drift out
of real [Ah] balance without any reliable ways to measure how far.
Amount of drift of course depends on how alike the cells came from
the manufaturer, their temps differences, etc. and properly profiling
them is large effort for a hobbyist, and besides makes sense only if
you can take advantage of this data and it is half science and half of
empirical art.

You got to know your cells better than manufacturer does to predict
anything ahead of time, and no one does. By the time you collect enough
statistical data and come up with predictive cell model that is close to
reality, manufacturer will likely discontinue this type of cell and your
big effort will now worth very little. Coming up with adaptive learning
algorithms to combat this and do it right is possible but no trivial
task. So you just define what is "good enough" for you, and as with
everything compromise between how BMSing should really be done and
time/patience/knowledge you got + size of your wallet.


--
Victor
'91 ACRX - something different
'01 in-AUDI-ble - handsome EV with 0.4 MW AC drive (work in progress)



_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[txhokie4life]

Roger Heuckeroth wrote:

Lee,

I'm trying to understand all this.

How are we adding resistance by putting cells in parallel.  Doesn't  
resistance follow the rule of

1/Rp = I/R1 + 1/R2 ... + 1/Rn

If each cell had an internal resistance of say 8 mohm, the equivalent  
parallel resistance of say 10 cells in parallel should be 0.8 mohm.  
Am I missing something?

I agree -- however this does explicity support the concept of hammering
the better cells (until they are not :-).  The cell with least resistance will
supply the most current.

Are you are referring to workmanship issues like sloppy spot welds, or  
not enough weld surface area?

Roger

Lee Hart wrote:
On Jul 25, 2008, at 3:39 PM, Lee Hart wrote:

> It's not really there with the small cells, either. One cell by itself
> may have a high C rating; but when you bundle large groups of them
> together, the composite cell has a lower C rating. This is because you
> have added resistance in connecting them in parallel, and because they
> heat each other.


_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[txhokie4life]

So Victor what is your recommendation for those looking for
higher Ahrs, high C and low ESR on a (reasonable) budget :-)

Sounds like we are trying to defy the laws of physics (and our wallets :-)
or should that be :-(

mike

Victor Tikhonov wrote:

Roger Heuckeroth wrote:
 > Christopher,
 >
 > Thanks for your take on it.
 >
 > What larger cells with higher C-ratings are you referring to?  I have
 > not come across any large LiFePO4 cells, 50 Ahr or above, that have a
 > 10 C continuous rating.  The best I have seen is 5C, but that was with
 > voltage sagging down to 2.6V.
 >
 > Victor used a series string, no parallel cells in his pack as far as
 > I'm aware of.


In my opinion, and it based on limited experiments I did, externally
paralleling Li cells is lousy idea unless you can guarantee quality
of such connection to be no worse than internal connections inside
large prismatic cells. As Lee pointed out, in essence OEM manufacturers take
large collection of small cells (in form of internal plates) and do such
packaging for you in one enclosure to get large cells, but there are
critical differences - some ill effect they can minimize with right
process, but you can't if you do it externally.

As far as BMSing and equal SOC if paralleling cells, Lee is right
on the spot. Everyone naturally wants to get away cheap, substituting
voltage for SOC (Ah amount) and it doesn't work very well for LiIons,
worse for LiP, and doesn't work *at all* for LiFePO everyone these days
seems to be locked on. Voltage is easy to measure compared to Ah, so
people pretend that 3.3V = 75% SOC (insert you own numbers here).
Not so.

A "BMS" which maintains voltages equal on each cell at all times
is certainly far better than nothing, but it is NOT really a BMS.
Such BMS has no idea what the state of charge is for each cell. Thus you
cannot put more or less Ah in certain cells to equalize the pack so you
can utilize its full capacity.

You can have a series string of 100Ah pack of LiFePO4 cells and
hardware to ensure each cell has exactly 3.215V on it. Yet, all the
cells could be at 70% SOC and one or few at 20%, so your *useable* pack
capacity is really near 10Ah before lowest cell gets ruined by further
discharge. Worse - you have no way of knowing it even if you suspect
this is the case unless you know actual capacity of each cell and
continuously track Ah in and out of the pack which of course takes more
than voltmeter per cell (in a form of A/D and central computer
collecting data or whatever implementation; these are irrelevant details).

If you want to get away cheap (without tracking Ah, e.g. having
balanced pack at *any* SOC level in the middle where all voltages are
the same and are no indication whatsoever what is going on), you can only
rely on the tail of charging curve where voltage increases sharply -
then the voltage differences up there will tell you if cells are close to
100% or not. If voltages are lower than knee, all you can tell that it
is not
about 85...90% yet, but how far away you are, you still don't know.
You must keep charging individual cells until they are all at the
same voltage *and near voltage indicating 100% of SOC*. THEN
it "resets" the pack and you can be reasonably sure all the cell are at full
capacity. This is very workable solution, in fact I will implement it
as very next pack in my ACRX. The only "inconvenience" (or condition
for this to work if you will) is you must often charge completely,
to near 100% SOC - this is where equalization using this method
will take place. Sort of like Rudman regulators clamping on top, but
otherwise doing nothing in the middle. If you only cycle between 30% and
70% SOC many times,
e.g. do partial charges, it is pretty much guaranteeing you'll drift out
of real [Ah] balance without any reliable ways to measure how far.
Amount of drift of course depends on how alike the cells came from
the manufaturer, their temps differences, etc. and properly profiling
them is large effort for a hobbyist, and besides makes sense only if
you can take advantage of this data and it is half science and half of
empirical art.

You got to know your cells better than manufacturer does to predict
anything ahead of time, and no one does. By the time you collect enough
statistical data and come up with predictive cell model that is close to
reality, manufacturer will likely discontinue this type of cell and your
big effort will now worth very little. Coming up with adaptive learning
algorithms to combat this and do it right is possible but no trivial
task. So you just define what is "good enough" for you, and as with
everything compromise between how BMSing should really be done and
time/patience/knowledge you got + size of your wallet.


--
Victor
'91 ACRX - something different
'01 in-AUDI-ble - handsome EV with 0.4 MW AC drive (work in progress)



_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Steve West-6]

On 26/07/08 10:03 AM, "txhokie4life" <[hidden email]> wrote:
> higher Ahrs, high C and low ESR on a (reasonable) budget :-)

High vs Low is a matter of opinion. What size pack are you after? How much
power does it need to deliver continuously and peak? Given the requirements
we can work out what solutions are most economic.

Steve West



_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Christopher Frost]

Roger,

Ah, you are right. I generally group chemistry's together in my head. On this
occasion I was grouping Lithium together. I was thinking of Enax, Kokam, and
Saft.

Victor made a very good reply, since the others are more experienced I won't
argue with their answers. :-)

I agree on the "BMS" term. It would be educational to learn what the best way
to manage and monitor batteries would be though. I might start looking into
that. For someone like me, I have a lot of patience and time. For money..
well not quite. 2 out of 3 though. ;-)

On Friday 25 July 2008 03:32:35 pm Roger Heuckeroth wrote:

> Christopher,
>
> Thanks for your take on it.
>
> What larger cells with higher C-ratings are you referring to?  I have
> not come across any large LiFePO4 cells, 50 Ahr or above, that have a
> 10 C continuous rating.  The best I have seen is 5C, but that was with
> voltage sagging down to 2.6V.
>
> Victor used a series string, no parallel cells in his pack as far as
> I'm aware of.
>
> We use BMS as a catch all term that can mean a lot of things.  You
> definitely should have charge regulators to prevent over voltage on
> any one cell. The circuit simply clamps the voltage at some level
> (usually 3.65V) you could leave a cell on this voltage for an extended
> amount of time.  It won't over-charge, it will simply not accept any
> more current.  If you clamp a parallel string at this voltage the
> current will decrease until there is hardly any current flowing.  Each
> parallel string needs one over and one under voltage protection
> circuit.  That's all.
>
> Roger

_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Lee Hart]

Roger Heuckeroth wrote:

> How are we adding resistance by putting cells in parallel.  Doesn't  
> resistance follow the rule of
>
> 1/Rp = I/R1 + 1/R2 ... + 1/Rn
>
> If each cell had an internal resistance of say 8 mohm, the equivalent  
> parallel resistance of say 10 cells in parallel should be 0.8 mohm.  
> Am I missing something?
>
> Are you are referring to workmanship issues like sloppy spot welds, or  
> not enough weld surface area?

I didn't explain it very well, so I can see where there's room for
confusion.

Suppose the manufacturer specs an internal resistance of 8 mohm typical.
But there is a tolerance on that; and it's usually large. Let's say you
actually measure them at 6-12 mohm.

Second, the cheap spot welder you use to put your tabs on also has a
tolerance. Let's say their resistance adds another 2-4 mohms per cell.
The series resistance of any given cell plus its welds can range from
(6+2) to (12+4), or 8-16 mohm.

This 2:1 range of resistance means the current division between cells in
parallel will also vary 2:1. If one cell is good for 10 amps peak, then
a best-case and a worst-case cell in parallel are only good for 10+5=15
amps peak. The worst-case situation with 10 cells in parallel is
10a+(9x5a) = 55 amps; only half of what you might expect.

I didn't mean that cells in parallel have an absolutely lower C rating;
just that the composite C rating is not simply the sum of all the
individual cells.

--
Ring the bells that still can ring
Forget the perfect offering
There is a crack in everything
That's how the light gets in    --    Leonard Cohen
--
Lee A. Hart, 814 8th Ave N, Sartell MN 56377, leeahart_at_earthlink.net


_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Roger Heuckeroth]

Victor,

Thanks for the explanation.  I think the simplest is often the best  
approach.  Full equalization charges every night with shunt regulators  
is the way I sill go.


Roger

On Jul 25, 2008, at 5:39 PM, Metric Mind wrote:

> Roger Heuckeroth wrote:
>> Christopher,
>>
>> Thanks for your take on it.
>>
>> What larger cells with higher C-ratings are you referring to?  I have
>> not come across any large LiFePO4 cells, 50 Ahr or above, that have a
>> 10 C continuous rating.  The best I have seen is 5C, but that was  
>> with
>> voltage sagging down to 2.6V.
>>
>> Victor used a series string, no parallel cells in his pack as far as
>> I'm aware of.
>
>
> In my opinion, and it based on limited experiments I did, externally
> paralleling Li cells is lousy idea unless you can guarantee quality
> of such connection to be no worse than internal connections inside
> large prismatic cells. As Lee pointed out, in essence OEM  
> manufacturers take
> large collection of small cells (in form of internal plates) and do  
> such
> packaging for you in one enclosure to get large cells, but there are
> critical differences - some ill effect they can minimize with right
> process, but you can't if you do it externally.
>
> As far as BMSing and equal SOC if paralleling cells, Lee is right
> on the spot. Everyone naturally wants to get away cheap, substituting
> voltage for SOC (Ah amount) and it doesn't work very well for LiIons,
> worse for LiP, and doesn't work *at all* for LiFePO everyone these  
> days
> seems to be locked on. Voltage is easy to measure compared to Ah, so
> people pretend that 3.3V = 75% SOC (insert you own numbers here).
> Not so.
>
> A "BMS" which maintains voltages equal on each cell at all times
> is certainly far better than nothing, but it is NOT really a BMS.
> Such BMS has no idea what the state of charge is for each cell. Thus  
> you
> cannot put more or less Ah in certain cells to equalize the pack so  
> you
> can utilize its full capacity.
>
> You can have a series string of 100Ah pack of LiFePO4 cells and
> hardware to ensure each cell has exactly 3.215V on it. Yet, all the
> cells could be at 70% SOC and one or few at 20%, so your *useable*  
> pack
> capacity is really near 10Ah before lowest cell gets ruined by further
> discharge. Worse - you have no way of knowing it even if you suspect
> this is the case unless you know actual capacity of each cell and
> continuously track Ah in and out of the pack which of course takes  
> more
> than voltmeter per cell (in a form of A/D and central computer
> collecting data or whatever implementation; these are irrelevant  
> details).
>
> If you want to get away cheap (without tracking Ah, e.g. having
> balanced pack at *any* SOC level in the middle where all voltages are
> the same and are no indication whatsoever what is going on), you can  
> only
> rely on the tail of charging curve where voltage increases sharply -
> then the voltage differences up there will tell you if cells are  
> close to
> 100% or not. If voltages are lower than knee, all you can tell that it
> is not
> about 85...90% yet, but how far away you are, you still don't know.
> You must keep charging individual cells until they are all at the
> same voltage *and near voltage indicating 100% of SOC*. THEN
> it "resets" the pack and you can be reasonably sure all the cell are  
> at full
> capacity. This is very workable solution, in fact I will implement it
> as very next pack in my ACRX. The only "inconvenience" (or condition
> for this to work if you will) is you must often charge completely,
> to near 100% SOC - this is where equalization using this method
> will take place. Sort of like Rudman regulators clamping on top, but
> otherwise doing nothing in the middle. If you only cycle between 30%  
> and
> 70% SOC many times,
> e.g. do partial charges, it is pretty much guaranteeing you'll drift  
> out
> of real [Ah] balance without any reliable ways to measure how far.
> Amount of drift of course depends on how alike the cells came from
> the manufaturer, their temps differences, etc. and properly profiling
> them is large effort for a hobbyist, and besides makes sense only if
> you can take advantage of this data and it is half science and half of
> empirical art.
>
> You got to know your cells better than manufacturer does to predict
> anything ahead of time, and no one does. By the time you collect  
> enough
> statistical data and come up with predictive cell model that is  
> close to
> reality, manufacturer will likely discontinue this type of cell and  
> your
> big effort will now worth very little. Coming up with adaptive  
> learning
> algorithms to combat this and do it right is possible but no trivial
> task. So you just define what is "good enough" for you, and as with
> everything compromise between how BMSing should really be done and
> time/patience/knowledge you got + size of your wallet.
>
>
> --
> Victor
> '91 ACRX - something different
> '01 in-AUDI-ble - handsome EV with 0.4 MW AC drive (work in progress)
>
>
>
> _______________________________________________
> For general EVDL support, see http://evdl.org/help/
> For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ 
> ev

_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev

[Metric Mind]

Yes, I believe this is very reasonable compromise and small penalty to
pay (having to wait to charge *completely* preferably every time)
for greatly simplifying the "BMS".

Victor


Roger Heuckeroth wrote:
> Victor,
>
> Thanks for the explanation.  I think the simplest is often the best  
> approach.  Full equalization charges every night with shunt regulators  
> is the way I sill go.
>
>
> Roger


_______________________________________________
For general EVDL support, see http://evdl.org/help/
For subscription options, see http://lists.sjsu.edu/mailman/listinfo/ev