[IamIan]

Opinions and preferences are like A____ , everyone has one.

So I don't encroach on the threads / ideas of others I figured I would post some of mine here , separately.

I see a lot of this as a personal preference and/or usage conditions effected choices kind of thing ... ie ... there is no ... one answer that is equally correct for everyone ... in all conditions ... all the time... everywhere... etc.

aka .... A build in HI will have significantly different considerations than a build in Death Valley ... or Northern Alaska ... A Rally car PHEV will have very different usage considerations than a LB highway cruiser ... etc.

I wanted to share some of the data I've collected and thoughts I've had so far in my snail's pace of a project to PHEV my I1.

My snail's pace is likely to continue ... it's who I am ... and the restrictions of time and $ I have to spend on the project.

At this time I have no intention of marketing a 'kit' to the masses ... if anyone finds any of my data and ideas useful to them for their own projects ... they are welcome to use them ... but don't expect an IamIan Store near you.

The other aspect of my snail's pace to keep in mind in this thread ... is that I enjoy the journey more than the destination ... I run experiments because I enjoy it ... I am doing what I enjoy , not just trying to reach a destination ... as such , I might have spent a sizable amount of time ... fun for me ... but others would skip much of the time I have spent , for their PHEV builds ... kind of like watching Star Wars movies is more fun than someone just skipping to the end and telling you 'good guys win'... sure those 3 words are correct and accurate ... but the journey through the movie is far more entertaining than just jumping to the end result.

After this first post to give and idea of what is to follow ... I will next walk through my past on this project to share some information about dead ends , changes in direction , etc ... that happened over time ... The past will work it's way to the present and what I am currently planning for the future ... then as I make additional small snail pace steps forward I'll post about it here.

 

[rayholan]

Perspectives

Opinions and preferences are like A____ , everyone has one.

So I don't encroach on the threads / ideas of others I figured I would post some of mine here , separately.

I see a lot of this as a personal preference and/or usage conditions effected choices kind of thing ... ie ... there is no ... one answer that is equally correct for everyone ... in all conditions ... all the time... everywhere... etc.

aka .... A build in HI will have significantly different considerations than a build in Death Valley ... or Northern Alaska ... A Rally car PHEV will have very different usage considerations than a LB highway cruiser ... etc.

Hi, Ian. Great meeting you at Mike's in August. I commend your emphasis on different usage scenarios and geographical locations. We often gloss over those key variables when discussing PHEV and other mods here. Look forward to your future posts!

Ray

 

[retepsnikrep]

Looking forward to your detailed and precise project analysis.
The more projects the merrier. We can all learn a lot from each other.

 

[aisbell]

I look forward to reading about your experiences, experimentations, etc., because I am a long-time (11 years) Insight owner who has recently added to his electrified auto fleet a mass-produced conversion EV, a Mitsubishi i-MiEV, the EV model of the small Mitsubishi i kei car series. This has made me acutely aware of the many advantages that the Insight has over a mass-produced conversion, although Mitsubishi's seems to be one of the best conversions.

Few of today's mass-produced EV's are purpose-built, so they aren't light like the Insight and their aerodynamics aren't as good as those of the Insight. So a well-built Insight EV conversion could be a great EV.

However, a concern about an Insight EV conversion is the growing lack of availability of Insight body parts and their relatively high cost where I live. One significant traffic accident could easily total one's expensive Insight EV conversion even at its increased value over an original Insight.

Bus as IamIan said, not all concerns, advantages, conditions, etc., apply to all regions of the world, so please tell us your story at your own pace.

 

[IamIan]

The Past

Thanks for the votes of encouragement.

- - - - - -

As a Base vehicle the Gen-1 Insight is hard to beat for Energy Efficiency carrier of 2 people at highway speeds and such.

Even the last Aptera production intent vehicle ... prior to folding , was only about ~12% less wh of energy per mile @ 50MPH ... and it was to cost over $30,000.

This made the Insight my base vehicle of choice 10 years ago .. when I was on the fence between converting my existing 1990 Chey Cavlier to a PHEV or BEV ... The Insight won out ... and it is still on top of what is available today ... even ~14 years after it was designed by Honda.

- - - - - -

A lot of options have popped up over time ... each with their pros and cons ... before I get to where I am now I'll do a quick over view of some of those other options I am no longer actively pursuing ... and touch quickly on why.

- - - - - - -

#1> OEM NiMH + MIMA + Grid Charger
Although I already have this ... the ~500wh of usable OEM NiMH capacity is too limited for usefulness as a PHEV ... at best that is about ~1/22 of a gallon per charge ... it would be something ... but for me ... I use my grid charger more as a maintenance item ... only doing an occasional Grid Charge 2-3 times a year.

This is not to say this couldn't be more used as a small PHEV ... it is just that ... for me ... If I'm in the mind set to treat it as a PHEV the OEM ~500wh is just not enough ... I find I do net better staying in the mind set of using MIMA and such to cover those miles as an efficient HEV ... this is mostly a limitation / failing I've identified in my mind set for me personally... If I know I'll be plugging it in as a PHEV ... I over use MIMA beyond the benefit point for the small ~500Wh capacity... and if I'm not thinking of it as a PHEV I usually don't bother to plug it in.

- - - - - - - -

#2> Multiple OEM NiMH + MIMA + GridCharger
There are those who have tried this approach ... and It was a strong contender for me at one point ... the high power OEM sticks solve the problem of a battery that can take the power loads of assist and more importantly regen ... the NiMH is very forgiving of abuse and treatment ... The OEM IMA battery area has enough dead space ... that I figured I could fit as many as 80 sticks in that space ... which would give me up to 4xOEM as much usable capacity ... Between the 20 sticks in the car ... and the 40 other sticks I already had from previous experiments , this would have been the least expensive option.

But ... if I take those other 40 sticks I have to use for a PHEV project ... I have to give them up as the energy recycling storage system for my regenerative discharge battery tester ... which would mean they aren't just free ... because I would have to buy a replacement battery pack for the regenerative discharge testing ... A battery pack that can do ~4kw or ~126 Amps of charge or discharge rate and also have a decent amount of energy storage ... Lead would have been a fairly cheap alternative for a replacement , and as a stationary testing system the weight wouldn't have been too bad ... but Lead has a shorter cycle life , the weight would be a pain in moving it ... and eventually one of my other projects was to make that regenerative testing system more portable / mobile than it is now ... more weight is the wrong direction.

Although this would give 4xOEM usable capacity ... it would have weighed 4x as much as well ... and at about ~40wh/kg these are not the best energy density batteries for a mobile application where weight is a vehicle efficiency penalty.

- - - - - - - - -

#3> 2nd Booster Battery + OEM NiMH + MIMA + Charger
This was also a strong contender for a while ... it solved the IMA peak power problems by continuing to use the OEM power cells ... and it reduced the amount of additional weight problem as the 2nd battery with much smaller booster current rates could use some of the nearly ~200wh/kg battery options.

However , moving to higher wh/kg energy density batteries meant less forgiving of abuse batteries ... a more robust BMS would be needed ... and some of those higher wh/kg battery options are fairly hazardous for fire and such ... if one was going to redo the BMS , make a better one and fake out the OEM system to use it , and design and build a control electronics between the two packs ... it seemed like one was already doing about ~90% of the work to just replacing the OEM pack with a better battery.

Because this option kept the original OEM power cells it's combined energy density would be better than #2 above ... but not as good as #4 bellow ... for example ... 50kg at 500wh and 50kg at 10,000 wh = 100kg at 10,500 kg = the same as a ~105 wh/kg single battery... ie 100 kg of ~105 wh/kg battery would be functionally the same as the combo.

- - - - - - - - -

#4> Enter the A123 style cells as a OEM replacement.

The highest power A123 Cells would be able to reduce the HEV battery pack weight and still handle the high power rates ... but would not be a total usable energy benefit for a PHEV ... which was the direction I was going.

The 20Ah pouch cells at Roughly the same kg of weight as space as the OEM system would give about ~3x as much usable capacity ... virtually no weight penalty ... improves vehicle net efficiency.

The A123 cells are slightly more cycle energy efficient than the OEM NiMH power Cells were ... not massive ... but it is one more step in the correct direction.

Multiple tests and example builds by people ... including Peter ... have shown that a well built system of these cells can easily take higher loads than the assit and regen loads of the OEM IMA system.

At about ~120wh/kg ... 100kg of these will net more usable energy than 50kg of 40wh/kg and 50kg of 200wh/kg ... which net only ~105wh/kg ... the same 100kg of A123 is about ~20% more.

By having one battery pack instead of a 2nd booster it increases the system net efficiency ... doesn't have the loss of the battery to battery electronics ... doesn't have the loss of the additional battery cycle in the system.

The A123 cells have a lower self discharge rate than the OEM NiMH ... this slightly improves the yearly energy efficiency of any PHEV option ... Although LSD-NiMH has similarly low self discharge rates ... they don't handle the high power rates as well ... they don't have as high energy densities ... and they are slightly less cycle efficient.

It isn't all Pro though ... this option does require some other cons be dealt with ... more fragile cell case , more need for BMS , etc ... etc ... but more on that latter.

So it is with the A123 style system I am now , still actively pursuing... more into the finer details on that , next time.

 

[MN Driver]

If you have the A123 cells already, they would be an excellent pack. As far as I can tell you can't get the cells from the usual Chinese sources anymore like you used to. As an alternative, if you are willing to give up the space and about 1.4kg per cell if you go with CALB CA series gray cells for about 70kg pack for 50 cells or about 84kg for 60 cells, you could have 6.4kwh or 7.7kwh nominal. The CALB CA cells have more capacity than they are rated for and there are folks who have been charging them at 3C and discharging at up to about 10-12C and they have little voltage sag. Much better than the HiPower cells that Peter had available to him when he bought them for his first PHEV conversion in 2009/2010. I'm convinced they would make a great pack and that they wouldn't struggle at all if you put 60 into a pack and used voltage and current mods to get max power(if desired). If the capacity is what you want they aren't a bad idea otherwise your plan for A123 cells is a solid one IMHO but I figured I'd put that out there.

To get an idea of the performance of CALB CA cells, here's a video: ..although you may have already come across this from DIYEC.

My personal choice, however was to go full BEV with one Insight and leave the other Insight stock(with clutch switch) for the long roadtrips. I'll have two decent OEM Insight packs in each car to create one strong one and maintain it for the long haul. Based on Peter's comments about efficiency while cruising on the highway with a light engine load and pushing assist, I've decided that PHEV wasn't what I was after. 19.2kwh(100 60Ah cells) pack of CALB CA cells will go into my conversion.

 

[IamIan]

If you have the A123 cells already, they would be an excellent pack.

I do already have 55 cells... more details on those later.

Although my testing of other battery options was not 100% complete at the time ... I rolled the dice a bit ... the A123 pouch cell was a strong contender among the options I was considering ... The Insight Central Group buy was just a little bit before I would have been ready to pull the trigger on my own ... but it was better to be part of the group buy ... cheaper to get the cells themselves ... helped to keep me from slowing my already slow snail pace ... and it was a IC community project I was happy to be part of.

As far as I can tell you can't get the cells from the usual Chinese sources anymore like you used to.

AFAIK ... There are still routes you can get them ... A123 products are still being made ... and sold.

A bit more care needs to be taken now than was needed before ... More important than ever to test all the cells when you get them now ... always was a good idea.

As an alternative, if you are willing to give up the space and about 1.4kg per cell if you go with CALB CA series gray cells for about 70kg pack for 50 cells or about 84kg for 60 cells, you could have 6.4kwh or 7.7kwh nominal.

That is a reasonable alternative ... and there are many BEVs out there with owners happily running on CALB cells.

But 70kg for 6.4kwh = just over 90wh/kg ... which is good ... but a considerable step less than the ~120wh/kg from the A123 pouch cells... The A123 cells would weigh less to give the same kwh of capacity... 100 of the A123 cells (as others have done) for a 2p50s pack would be ~50kg of battery for ~6kwh.

I'd rather not increase the weight that much if I can ... 70+ kg ( ~154 Lbs ) of battery ... is a pretty big battery pack for a PHEV... especially in a light weight insight ... an option definitely ... and there are many happy CALB BEV owners.

For now my target for the PHEV project is a increase but only to about ~3kwh of capacity... about ~3xOEM usable... I might in the future do a increase of this ... but for now that's where I'll be starting.

My personal choice, however was to go full BEV with one Insight and leave the other Insight stock(with clutch switch) for the long roadtrips. I'll have two decent OEM Insight packs in each car to create one strong one and maintain it for the long haul. Based on Peter's comments about efficiency while cruising on the highway with a light engine load and pushing assist, I've decided that PHEV wasn't what I was after. 19.2kwh(100 60Ah cells) pack of CALB CA cells will go into my conversion.

A Full BEV conversion of my insight has been an option I've been considering for many years ... even before I bought the car ... It is a fantastic base vehicle ... but I have been thus far more tempted / attracted by all the other improvement projects I could do to a PHEV.

That should be an excellent BEV ... please do start a thread about it , as the project get's under way.

 

[RedDevil]

Subscribed!
Looking into building a portable 1kWh mini-PHEV kit for my Gen 2... but I'll just watch and learn here.

 

[IamIan]

One other aspect of the CALB I forgot to mention.

Their ridged cell shell removes some of the fabrication issues some people have had with using the more fragile pouch cells which have a thin foil shell.... and that is a good thing for CALB.

For my own two bits ... just a bit of my own personal preference.

I have also seen people .. with a bit more care in the pack outer case and cell assembly ... have been able to compensate for this A123 pouch cell disadvantage ... ie if done right , the more fragile foil pouch cell is not necessarily a con ... but it is one more aspect one has to consider / take into account when choosing a cell.

For me ... I'm hoping to take advantage of what others have learned and ... compensate the A123 pouch build ... so that there is no functional issue about their more fragile foil shell ... which many who use them have also been able to do.

But for those who get concerned about it ... a more rigid cell shell like the CALB is an option ... just not an option I will be taking at this time.

 

[uhtrinity]

A variation on #2:

Using battery packs from other car manufactures like Toyota (Prius) or Ford (Escape). Packs could either be stock size or some greater multiple. I might be building a Prius sourced pack for a friend who has an Insight. My thought is to build it in a series configuration, but go with max voltage short of throwing an error code. He already has a dual charger that can be converted to a dual stage charger similar to what I run on my Escape pack.

 

[IamIan]

A variation on #2:

Using battery packs from other car manufactures like Toyota (Prius) or Ford (Escape). Packs could either be stock size or some greater multiple. I might be building a Prius sourced pack for a friend who has an Insight. My thought is to build it in a series configuration, but go with max voltage short of throwing an error code. He already has a dual charger that can be converted to a dual stage charger similar to what I run on my Escape pack.

That is a 100% Valid option ... and if the batteries used can be had at a good price ... it might be hard to beat financially... and sticking with a very forgiving of abuse Battery type like NiMH makes the entire system easier / less complex.

For my own PHEV path ... I eventually left #2 mainly because I can get the same wh of energy in 1/2 to 1/3 the weight ... which means a more net efficient vehicle... that's the elephant in the room for me , and why I am currently pursuing Non-NiMH battery options for my PHEV.

 

[IamIan]

One of the considerations with going to a different than OEM chemistry is the number of cells one chooses to put in series.

The OEM electronics operate in the range of ~185v to ~120v.

As I recall ... Peter Started with 45 cells ... although I'm not psychic .. my guess is because that would have a similar nominal voltage to the 144v nominal OEM NiMH pack.

45 cells at ~120v has the average cell down to about ~2.67v ... and at ~185v the average cell goes as high as ~4.11v.

Although there is some amount of debate about the fine details ... more on that later ... The Guidelines Section #1 from A123 ( Attached) site 2.5v as a minimum low ... and 3.6v as a high ... a max of 3.7v as a peak while being charged.

If we take the OEM A123's recommendations at face value ... this means 45 cells in series has some room left at the bottom ... but can be potentially pushed too high at the ~4.11v of the IMA system.

By the end Peter ended up with 50 cells ... not sure of all of the logic ... but my guess is ... that is what could be made to fit reasonably.

@50 cells ... 185v high is ~3.7v average per cell ... and 120v low is average of ~2.4v per cell... this is much closer to A123 Recommendations... just a bit low at the bottom end.

But if one shifts toward the bottom to get it closer in line with the 2.5 that would simultaneously shift the top up higher ... and it is already at the max of the A123 recommendation voltage.

48 cells in series @120v would give us the average low of ~2.5v ... but a high @185v of ~3.85v.

50 cells lines up at the top ... but goes a bit low ... 48v lines up at the bottom but goes too high.

49cells in the middle would be .... ~120v low of ~2.45v ... and a ~185v High of ~3.78v.

- - - - - - -

Weighing Potential usage.

In the Car the system is less likely to get fully charged from gasoline ... the whole point of a PHEV after all is to do most of the charging at home ... where the charger controls the top end... which means the higher upper end point may be less likely to hit the Car's ~185v ... and we can leave the pack's top end to the charger... which can be anything we design / want it to be for voltage.

If all other things were equal this wall charger aspect of the PHEV design intent would lead us more toward protecting the bottom end ... with something like a 48 cell pack.

- - - - - - -

Energy difference:

50 @ ~60wh per cell = ~3,000 wh pack.
49 @ ~60wh per cell = ~2,940 wh pack... ~98% of 50 cell.
48 @ ~60wh per cell = ~2,880 wh pack... ~96% of 50 cell.

Although there is a energy difference ... at most ~4% is not a large margin.

- - - - - - - -

Handling ... the other consideration for cell count is packaging and handling.

an odd number like 49 makes dividing the pack in half with a breaker and or fuse not even halfs ... which might not matter a great deal ... but if we wanted to preserve that half way middle aspect a even number is needed.

One issue with the OEM NiMH pack is that is awkward to service ... or get access to ... and no matter how well designed eventually needing service is something to consider in the design ... 50 cells in two modular 25 cell subpacks ( ~28 lbs each just batteries ) would make that servicing easier... the 48 cell pack has the most potential for the modular ease of service ... it could be [email protected] ( ~27 Lbs each just batteries ) cells each ... [email protected] each ( ~18 Lbs each just batteries ) ... [email protected] each ( ~14 Lbs each just batteries ) ... [email protected] ( ~9 Lbs each just batteries ).

- - - - - - - -

I bought 55 cells ... so I would have some (5) extra to sort out the weakest ... and still have 50 left if I chose to go with 50 cells... 4 could be used for a ~12v battery for other projects.

- - - - - - -

I like the idea of the even split pack and the ability to do modular ... which leads me to 48 or 50 .... at the moment I am leaning more to 48... even with the minor ~4% wh hit compared to 50 cells.

The low end system voltage is nice ... more modular ability is nice for fabrication and such ... and by happy coincidence the three PL8's I already have can each do 8 cells in series for charge or discharge tests ... which would mean , if needed in the future I could do pack check ups down to the cell level fairly quickly ... up to ~4kw of PL8 test on half a pack at a time... with built in 3Amp per cell bypass for maintaining balance... all 48 cells could be tested in just one day ... like a weekend.... which might be nice to see how the pack ages over time in the long run.

The other benefit of slightly reducing the cell count ... is that is opens up a little bit of space in the pack for cooling and or heating flow ... about ~14mm of it ... this is enough for one bigger space toward the middle ... or two smaller spaces splitting the pack into 3rds.... more on that latter.

- - - - - -

I'll include some of the other A123 files uploaded in this and future posts ... so people who don't already have them ... will be able to know / see some of what I am talking about.

 

[retepsnikrep]

As I recall ... Peter Started with 45 cells ... although I'm not psychic .. my guess is because that would have a similar nominal voltage to the 144v nominal OEM NiMH pack.

Ian it was 50 HIPOWER cells, same nominal voltage etc as A123.
I'm using the same 5 year old packs in my HCH1 Civic now so that should help to address the Lithium longevity concerns.

The reason was, I was using the BCM taps fed from groups of 5 of the cells in the first prototype, rather than the BCM fooler resistor matrix. This enabled the BCM to work as in the OEM setup monitoring blocks of 5 cells (Each 5 cells equalled one pair of sticks) 10 taps = 50 cells. A cell group more than a volt off would trigger a neg recal just like the OEM system and stop further assist. Ditto pos recals etc.

However I found that restricting later and went for the BCM fooler to mimic a perfectly balanced pack, and later added the BCM interceptor as well to fix the soc at 19 bars.
Assuming you have a BMS of some sort you will almost certainly want the fooler and interceptor to make it work without those restrictions.

In use I think you will unlikely push the cells right until empty and the 2.5V figure is quite conservative.
My Abs Min cell Voltage setting under load is 2.3V in my BMS. I've never seen that in anger.
The pack is exhausted by then and I simply stop using assist or the car won't let me use it anyway due to low pack V.
If the cells are all down to say an average resting voltage ~2.8V (140V) then you lose IMA start and assist etc anyway irrespective of the indicated or faked Soc.
They start getting quite warm at low soc which you will notice.

If you intend to operate at normal IMA power levels then the higher the system voltage the lower the current will be for that same output power. This means the 100A fuse is less stressed with 50 cells than with 48 or 45

If you use current hacks the 48 cells will not be able to push as much current through the system as 50 cells with the extra 6 volts to help overcome the motor back emf. If I was building my setup again I would go for 60 cells for absolute maximum IMA power potential. 50 cells is a good even number for a road car with some power mods though. A bit higher V than nominal but not high enough to cause issues.

Feel free to tell me to shut up at any time.

PS Once you get going and taste unlimited assist and/or increased IMA power/current you will want more capacity/voltage not less

 

[IamIan]

Feel free to tell me to shut up at any time.

Not at all ... all feedback , experiences , correction of any of my errors , insights , suggestions , opinions , etc ... are all welcome , and appreciated.

However I found that restricting later and went for the BCM fooler to mimic a perfectly balanced pack, and later added the BCM interceptor as well to fix the soc at 19 bars.
Assuming you have a BMS of some sort you will almost certainly want the fooler and interceptor to make it work without those restrictions.

I have one of your BCM Interceptors.

Which is definitely a needed piece in order to use any significant amount of additional capacity ... as others discovered ... the OEM system will only go a little bit above the OEM ~500wh ... the BCM Interceptor makes a pack of virtually any size possible.

I was planning on a BCM Fooler as well ... I am planning to build one up based on the threads ... hopefully with no major issues ... but there will always be a few... haven't built it yet ... but it is on the list.

In use I think you will unlikely push the cells right until empty and the 2.5V figure is quite conservative.

I agree... but I am leaning more toward the conservative direction.

There is also very little left to be had down that low anyway ... more details latter ... but I'm seeing the cells are at~6% AhSoC or ~2%WhSoE at ~3.2v(rested) ... there will be some dV that could go lower due to resistance of the batteries and wires and such ... and there is less voltage sage under load with these cells than the OEM NiMH.

The pack is exhausted by then and I simply stop using assist or the car won't let me use it anyway due to low pack V.
If the cells are all down to say an average resting voltage ~2.8V (140V) then you lose IMA start and assist etc anyway irrespective of the indicated or faked Soc.
They start getting quite warm at low soc which you will notice.

Agreed ... I intend to have a low end cap I avoid ... I have not yet made a final decision about the operating range window ... be it 10-90 or 20-80 or other ... but I will impose some of that on myself voluntarily... it might be overly conservative ... but it has been shown that a smaller operating window will extend the overall service and cycle life.

The ~140v @ 48 cells is ~2.9v per cell ... there is very little left I'd be giving up in that last 0.1v ... bellow 2.9 down to 2.8 , anyway.

If you intend to operate at normal IMA power levels then the higher the system voltage the lower the current will be for that same output power. This means the 100A fuse is less stressed with 50 cells than with 48 or 45

I am intending on being close to OEM power levels.

I agree the fuse would be less stressed ... but I also do not recall anyone ... at OEM power levels having any regular fuse issues ... I know I haven't in 10+ years ... and with 48 cells at near OEM power levels I should see about ~7% less current and a bit less stress on that fuse anyway.

If you use current hacks the 48 cells will not be able to push as much current through the system as 50 cells with the extra 6 volts to help overcome the motor back emf. If I was building my setup again I would go for 60 cells for absolute maximum IMA power potential.

PS Once you get going and taste unlimited assist and/or increased IMA power/current you will want more capacity/voltage not less

I understand.
And that is possible.

But am not going after more power ... I have some doubts if I would use it ... even when I've run in simulated PHEV mode with the OEM ~500wh ... I almost never push to max assist... it hasn't been my usage ... that could change ... but I don't expect it at this point.

And the lower resistance and higher voltage of 48 cells will give some increased system power anyway... even though I have my doubts about the frequency I will use it.

But that is 100% just a personal preference thing... and maybe latter I will kick myself ... but for now ... I'm thinking the bit more with 48 cells is still the direction I am leaning toward.

 

[IamIan]

An aspect touched a bit on previously by Peter ... but I think deserves a bit more depth... is how conservative to be with the cells ... what the pros and cons are ... etc.

But I think a bit more ground work first.

- - - - - - - - - -

So I was testing various batteries some A123 cells among them.

Then a bit over 1 year ago ( 5-9-12 ) there was a large group buy organized:
http://www.insightcentral.net/forums/modifications-technical-issues/22520-large-group-purchase-a123-20ah-cells.html

In the end as I recall with shipping and all other fees ... it worked out to about ~$1,500 for the 55 cells... just over ~$27 each ... not bad ... around ~$427 per kwh.

So I got on board that ... and then eventually I had 55 A123 20ah pouch cells... picture of 4 happy boxes attached.

Of course that was just the raw batteries ... the Project itself will include other costs for electronics , and such.

- - - - - - - - - -

Although I could have just dived in and starting building it up following the work Peter did back in 2010:

http://www.insightcentral.net/forums/modifications-technical-issues/16928-a123-20ah-lithium-install.html

but ... that's not me.

I wanted to test them ... partially to know more accurately what I was working with... partially to sort out the best cells from the worst cells ... and partially just out of curiosity for my entertainment.

- - - - - - - - - - -

So Begins the Testing.

Attached is a picture of my setup.

I used 2 old OEM IMA NiMH battery packs from junk yard Gen-1 HEV Civics ... I had tested previously ... configured into a ~120Ah ~14.4V ( ~1.7kwh ) battery pack.

3 Power Lab 8 ( PL8 ) battery testers ... although there are a wide variety of other devices that could have been used ... those are what I used... what is good to know is the usable Ah and/or Wh from a full discharge of each cell ... and if you want to go further ... cycle efficiency and self discharge rate might be nice.

The PL8 is fairly versatile with several options for testing ... input to a PC being a nice one .. and PC logging , was a must for me... the other feature I really like about the PL8 , is the regenerative discharge mode ... this feature lets me discharge a test battery ... in this case one of the A123 test cells ... and feed a % of that energy being discharged into to charge a 2nd battery ... in this case the ~1.7kwh NiMh battery pack ... then latter when charging one of the A123 test cells I can take some of that energy back out of the !1.7kwh pack ... the circle is not 100% efficient ... but it is more than the 0% returned from the resistance / heat based discharge methods used by many other battery testers ... I get to recycle at least some of the energy from the battery tests ... which is nice in itself ... but it also slightly reduces my electrical costs of doing the tests.

I used a piece of white melamine clad particle board ( scrap from work cut to size ) as a spacer ... so that my clamps were not pulling up or down on the tabs while in the test rig ... another such spacer above it ... used to distribute a small weight put on the cells.

The weight was because much of what I have read ( previous attached ) had talked about it being good to keep the cells under some compression during charge and discharge ... So to play it safe I included some evenly distributed pressure.

Although the PL8s can go up to 40A each ... and can be chained together to act as a single 120A tester ... Initial tests were at a more conservative 10A charge and discharge rate... for a high power test ... it might be a good idea to have the weight added for a little compression also being a heat sink / spreader to prevent the cells from over heating , if one does a high power test.

Even though it has been over a year ... I've been enjoying the testing so far , which I have found to be informative and entertaining ... and I am not done yet ... Still thinking of things I want to test.

Initial / Early testing results I'll go over in the next post ... to prevent this post from being monstrously large.

- - - - - - - - - -

Another good , but dry & technical paper about these kind of batteries is a 150 page pdf from JensGroot Link

If I marked that correctly in my Google Drive you should be able to down load it that way.

 

[IamIan]

So I had my 55 cells.
I had my test rig setup and prepared.

So I did Full Cycle Capacity Check of all 55 of the Cells.
Charge 10A CC to 3.65v CV to ~500mA
Discharge 10A CC to 2.8v

Different setting would have given different results... for example if I had continued the charge CV phase to a lower mA more energy would have gone in ... if I had discharged bellow 2.8v I would have got more out at the bottom ... But I was not ready to be pushing the cells hard ... so softer conservative testing to start.

The Purpose of the testing was not an absolute number ... but for comparative purposes ... does cell #12 have more less discharge capacity from full than cell #37 ... etc ... as long as the testing method is consistent on all cells ... The results will give the type of data desired.

About 1 Month Later ... I had the comparative discharge test done for all 55 cells. See Attached bellow.

The Average was ~58.5 wh ... The Highest was 60.07 wh ... the lowest was 52.1 wh.

Although this variation in discharge wh of capacity is not large ... just ~8wh ... it points out the value in buying a few extra cells ... in this case I had 55 ... if I chose to go with Peter's 50 cell design ... I would be able to remove the 5 weakest from the pack... If I had only bought 50 cells and wanted to use all 50 cells ... I would not be able to match the cells that were closer to each other.

If I had a whole battery pack with that much variation from the strongest cell to the weakest cell ... I would have to design in an intended smaller % SoC/DoD window in order to not over discharge that weakest cell ... ie the bottom ~8wh of the best cell can't be used , because the weakest cell already hit bottom... to try and force out that last wh of the top cell would do damage to all the cells bellow it.

60 vs 52 wh means about a ~13% of the bottom wh cell capacity should not be used from a top balanced pack to avoid over discharging that weakest cell... it would be less than 13% of the whole pack's capacity.

This is one aspect hinting a bit at the ... 'how hard to push them' ... mentioned before.

If the cells are closer to each other is far safer to push them all closer to the end points ... than if they are further out of balance with each other in capacity.

 

[IamIan]

So now that I had a comparison test completed for all 55 cells ... I moved on to Cycle Efficiency per cell.

Like other cell characteristics ... they are not all identical ... and there can be a 'runt' cell... and small variations.

So , by starting from a discharged state ... I measured the amount of energy that went into the cell ... then measured the amount of energy that came back out of the cell ... in order to compare them for cell round trip cycle wh efficiency.

Same Charge and Discharge setting used previously for the wh of capacity tests.

Note ... the cells on the chart attached bellow are in order for this test ... cell#10 on this test is not the same cell#10 from other tests.

I do keep the cell identities organized by the serial numbers printed on the cells ... but those identifiers are too long for a chart like this.

Also ... by putting them in sequence of highest to lowest for this test criteria ... it makes seeing the variation / progression much easier as well.

The Lowest wh cycle efficiency cell came in as 88.95% ... The Highest came in as 94.53% ... The 55 cell average was 93.58%

These are very efficient cells.

Ah Cycle Efficiency will always be higher than Wh cycle efficiency ... which is why some DIY BEV people use AH counting as a method of SoC estimation... but for my purposes I wanted the energy ( wh ) efficiency.

From a ~3kwh PHEV battery pack ~93% cycle efficiency would mean I could potentially see as much as ~200 wh of heat generated over the duration of one cycle... not an enormous amount of heat ... but it helps to quantify it.

The other useful piece of information gained from the Cycle efficiency comparison test ... is that it shows the potential for a gap to develop between the most efficient cell and the least efficient cell ... in this test up to a max of about ~3.35 wh of SoE gap could develop per cycle between the highest and lowest cell ... which is about ~5%.

By doing this test to match up the best cells and exclude the weekest cells ... we can reduce this gap.

otherwise ... we might need an additional % added to our window in our non-used section of the SoC/DoD band ... so the less efficient cell doesn't bottom out between balancing.... up to a max of ~5% per cycle.

As long as our Charger can re-balance the pack ... than just ~5% is all that would be needed ... but combined with the ~13% from the wh difference between cells test earlier ... we are up to ~18% safety buffer ... or a ~82% SoC/DoD usage window.

So far the worst case mis-matched cells still would have been fine with a balancing charger and able to safely use ~80% of the ~3kwh every cycle.

The other thing I like about have data like this about the cells ... is that not only can I use it in trying to make the best match of cells for the final project ... but I can also choose where to put certain cells.

Sense the least efficient cells will produce the most heat ... those are the cells that I want closest to the outside edge ... where they can cool the most ... and the most efficient cells are the cells it is safest to put toward the middle of the pack , because they will generate the least amount of heat.

It isn't a big gap ... and not a big impact on final net performance / results ... and I recognize many people would not bother because of the small potential for gains to be had ... but I enjoyed the testing in itself ... the benefits to the battery pack build will just be a bonus.

 

[IamIan]

So the Last and probably least commonly tested criteria for the 55 cells is the self discharge rate.

For good reason ... These cells have a very low self discharge rate... and if the balance is being corrected with every charge ... this would only potentially matter if the cells went months between balance correction charging.

The Self Discharge Rate is higher at higher SoC ... and it is higher at higher cell temperatures.

Unlike the drift from cycle to cycle that could happen from the variation in cell efficiency ... a variation in cell self discharge rate would potentially present itself over time , not cycles.

From the previously posted pages A123 recommends a middle ~50% SoC around ~3.3v ... and at a temperature of about 25 degrees C... at this condition they claim the cells will see about ~3% Self Discharge per Year... at higher temperatures and higher SoC ... they claim as much as ~3% Self Discharge per Month is possible.

This would mean under those storage conditions the ~3kwh pack could see up to 90wh of self discharge per year ... and at higher SoC and Higher Temperatures A123 would expect up to as high as 90 wh of self discharge per month.

Having previously tested what each cell could give as an output from a full discharge ... I end the last round of efficiency testing by fully charging each cell... then wait ... in my case 2 months ... then do a discharge test to see what discharge capacity is left ... in order to create a self discharge comparison for the 55 cells.

The attached Chart is ordered from low to high of this test criteria ... Cell #7 is not the same Cell #7 from other tests ... this ordering helps to show / demonstrate the variation ... I do have the data for each cell's own tests identified by the cell's Serial Number.

My lowest rate ( best ) cell was a 2 month average of 13.81 mWh per day ... or ~829 mWh Self Discharged in 2 Months ... which is about ~0.7% per month ( ~8.6% per Year ) ... the Worst ( highest rate ) cell was a 2 month average of 690.13 mWh per day ... or ~41.4 Wh of Self Discharge in 2 Months ... which is about ~39.7% per Month... The 55 Cell Average was ~53.89 mWh/day ... about ~3.17 Wh Self Discharged in 2 Months ... or about ~2.6% per Month... for Top Upper SoC ... other than that one exception ... these cells are averaging less / better than the ~3% per month upper end A123 warns about.

The worst cell at ~690 mWh/day is off the chart compared to all the others ... the 2nd worst was only 73.58 mWh per day.

To me this indicates that something is potentially wrong with that cell with the off the chart Self Discharge rate ... it just so happens to also be the lowest wh capacity cell at 52.1Wh ... at 88.95% cycle wh efficiency... that same cell is basically the worst in all 3 criteria.... although it is still functional ... being the clearly worst performer in all 3 criteria ... to me , means I would not be putting it in the final pack.

 

[RedDevil]

I read a story today on endless-sphere.com about someone who built an electrified bicycle with a pack that had one bad cell (wanted to link it, but can't find it anymore).
He had a BMS system, but it failed after being charged 10 times.
So he went on charging his cells in packs of 4 with a charger fit for LiFePO4's and had no problems with that.

Having no BMS he stuck to monitoring just that bad cell, knowing that as long as that held its charge the others were certainly still be good too.

He killed the bad cell eventually, by running it low despite having had a warning on the cell. He now cut it out and awaits delivery of a new cell.

But it would be a nice and cheap approach; just test the bad cell to save the pack instead of all of them.
This approach can only be used if we can trust the relative properties of these cells to stay the same; e.g. the cells just deteriorate over time and cycle in a likewise manner, so the worst cell will remain that.
He advises checking each cell before and after charge frequently, at 20 cycles max. Can do that

 

[Cobb]

Just the bad cell? With my enginer kit we had a cell that always had a higher state of charge like cell 8 and 16 of the 16, then a cell with the lowest soc like cell 1 or 7.