Todd Dore's LiFePO4 batteries

[robert.mcpheron]

Can you tell what companies supply these batts. The one I know of is elite power

[ted.lowe]

I got the call from Manzanita today - repair bill was about $500.  Yikes!  New capacitors (ultra caps) and a new control board - Rich wasn't happy with the old board (this charger is about 8 years old!).  So, I should get it back in about a week.

I'll keep everyone posted to let you know how the batts continue to perform.

Thanks to everyone who helped out with good advice - Rich Carroll, Jim Dawson, Ted Lowe, and Ken Simmermon.

Long live EVs and take care of those Lithiums!!!!

Yes, yikes!  This isn't a repair, it's an upgrade!  i can see from his point of view... he'd rather not support his older designs.    Ultracaps in an SMPS ?  Perhaps to use fewer capacitors....

[ted.lowe]

The regs on the other hand have small ceramic heat sinks and some of them were very hot.  They were apparently disapating as much energy as possible.  I didn't even know the regs could do this. 

i looked at the miniBMS from the link you sent (http://www.cleanpowerauto.com/MiniBMS.html).  The ceramic items are power resistors, not heat sinks.  Each miniBMS can only shunt up to 1 amp. i think their basic concept is to handle an overcharging event for short periods of time until the auto-shutoff or (human responding to the ERROR buzzer) can react.  If 18amps of charging current is going in, a 1 amp diversion won't help much for long.  It's a bit like a front bumper in a slow speed accident.  The auto-shutoff (as you mentioned above) is the airbag though!   Best wishes and thanks for keeping us posted on your bleeding-edge leadership!

[ted.lowe]

More on MiniBMS shunt regulator.  Here's a picture of it:



This is a very simple circuit.  When the battery voltage is too high, it diverts power past the battery.  It's basic parts include: a power transistor (black rectangle lower middle), power resistor (big cream colored ceramic rectangle on left), voltage comparator (black rectangle in middle center), a shunt-regulator (small black rectangle on the top middle), and a diode (can't tell from too small of picture).   i suspect the schematic/theory is very similar to this:

http://www.solorb.com/elect/solarcirc/shuntreg2/

i'm curious how much power the regs consume while in normal charging mode?  Do any of the components get warm in "green light" mode ?  Thankx!

[todd.dore]

Rob -

Elite Power is where I got the LiFePo batteries from.

Ted - during charging, the CMS modules are drawing about 0.7 amps.  At 3.3-3.4V during most of the charging, that's 2.3-2.4 watts of power draw.  Very minimal.

I do not notice any components getting warm during the 'green' mode - but I haven't checked very carefully.

[todd.dore]

So I charged the batts up tonight while manually watching them.  I got the relay (will install tomorrow) so the fail-safe will be implemented.

On my 56.1 mi. run yesterday, I put 17kWh back into the pack for a full charge.  Total efficiency = 303Wh/mi.  So - 2 good things: [1] the efficiency hasn't changed; and [2] the pack charged up again and seems to be holding a charge, so no immediate damage to the batts from the potential over-charge the other day.

I was manually watching the charger and it went into CV mode (wind-down) at 170V.  Then I checked the current meter, and it went down to about 7 amps.  I went inside for a few moments - when I came out again - Walla!  Another buzzer warning!  WTF??  So soon?

It must have just gone off - I saw the total pack voltage - 181V - how did it climb from 170V so quickly?  Anyway, the charger was in the 'yellow' phase and the current wasn't very much (about 5-6 amps), so I unplugged it.  The buzzer went off < 1 min., which is a good sign.  All the red lights disappeared in about 3-4 min. (good sign).  None of the resistors were warm (good sign).  So I've learned more about the charger - even with the lower voltage setting, the shut-down feature will pump these batts up so that at least some of them will hit 3.8V (triggers the buzzer).  That is why the sure-fire way to avoid overcharging with the Zivan is to get the relay - any batt hits 3.8V and Bammo! Off goes the charger.

I believe Ted - the relay is more like an airbag.

One other thing - some of my batt clusters shift around a bit - and the bolts were a bit loose!  I inspected and tightened all of them - this is routine maintenance that I will have to do.

That's all for tonight!

[todd.dore]

Today I installed the solid-state AC relay.  I tested it and it works well.  Now my Zivan backup charger is complete.  There are a few things about this that make it less than ideal (compared to the Manzanita):

[1] The BMS kicks off the Zivan (via the relay) when the first batt reaches about 3.86V.  That's a little too high for me - I'll have to see if it can be dialed-down.
[2] About 2/3 of the batts are throwing red lights by this time (over 3.65V).
[3] The current in the CV phase is about 4 amps - a bit too high for my taste.
[4] I have heard that if the solid-state relay fails, it does so in the open position - that's a headache you don't want (although there is another semi-safety feature that the CV phase starts at about 174V so about 3.6V for each batt; but it can creep up to 185 or 190V - hence the problem I had earlier).

For the Manzanita, no batt reaches 3.8V - the BMS signals the regbus and the charger goes into shutdown mode sooner - only a handful of regs are throwing the red lights.  Also, the CV phase in the Manzanita pumps out 1-2 amps - much more palatable if you want to preserve the batteries.

I drove the car about 18 miles tonight to Sonic Drive In for Sundaes.  No problems, and I think I'll let the batts rest - so no charging tonight.  I would like to start doing more test runs to see how far my true range is.

That's all for tonight!

[ted.lowe]

Hey Todd,  Is the Zivan specially programmed for using with LifePo bats?  If not, it won't treat them well at all.  For my Zivan NG3 configured for flooded lead batteries (120v):

Solid red LED = bulk = constant current until voltage hits a high mark (current here is about 18a)
Blinking red = consolidation = constant voltage at high mark while current decays to a lower value
Solid yellow = equalization = raise voltage very high with very little current (this is only needed for flooded bats, and when they emit their gases/bubble).

After a fairly long drive, bulk runs for 2-4 hours, consolidation for 10's of minutes and equalization for an 1-2 hours.

Hope your charger isn't trying to equalize your LifePo's!

i didn't quite get your meaning of the current consumed by the regs. Do you have a reg on each battery and a single CMS ?  i'm curious about the efficiency of the whole charge using the shunt reg method.

The shunt reg method is simple and cheap, but i see two issues that are non-starters for my adoption: 1) diverted power is wasted as heat, and 2) provides only limited balancing. 
However, i may be missing the function of your CMS.

Issue 2 is certainly the biggest concern as the bats age and some reach the 'auto shutdown mode' much sooner than others.  The preferred (balancing) approach is to continue charging until the last battery hits the shutdown point (need to send less current to the already 'full' batteries in the string).  This either requires regs that can handle much higher current or separate charge wires to each battery. 

To learn more about balancing, i'm running a small experiment with a 4 series pack of A123 cells that power my portable ipod dock boombox.  i bought a DeWalt 36V pack with 10 A123 cells.  i disassembled it and extracted the cells for other uses (like this experiment).  i reassembled a 4 series pack and added a "balancing harness" connected to each intermediate connection point (5 wires for 4 cells).  i bought the Thunder AC6 Dual Power charger (used by R/C hobbyists) which has LifePo and balancing algorithms.  Once you plug in the "balancing harness" into the AC6, you can see the voltage of each cell through-out the charging process.  If you configure the balancing option, it will back off on the more fully charged cells while the others come up.  When it's done, each cell is typically within 2/100's of a volt of each other (3.59-3.61).  i've only lightly cycled the experimental pack about 5 times so far, but look forward to more and deeper cycles to see how the balancing is affected over time.

[todd.dore]

Ted -

From what you write, I can tell that you are a lot more knowledgable about this.  However, I'll try to answer some of your questions as best I can:
[1] The Zivan is not specifically programmed for LiFePo.  However, it is my understanding that the creator of BMS uses the same charger.
[2] The charger shuts off when the first batt hits 3.85V (not the last).
[3] By that time, about 2/3 of the regs are showing red lights (at least 3.65V, but they are all under 3.85V).  I think the charger does CC/CV (I've measured this).
It is for reasons #2 and #3 above that I think the manzanita is better - it will shut down when the first cell hits 3.65V - thereby offering more protection against overcharge.  However, 1-1/2 weeks with the Zivan (probably less than 6 charges) hopefully won't kill my batts.

As best I can tell, the regs take the heat when the batts are about 3.8V, so when the first batt hits 3.85V and shuts down the charger, the majority of the regs don't even get warm.
During blinking red phase, the charger is putting out 5-7 amps.  During yellow phase, around 5 amps.  The relay turns off the charger during the yellow phase, after about 10-15 mins. or so.  The total pack voltage gets up to 181V (for 48 batts).  I've also been watching this manually, like today, so I did not let the charger go into yellow phase - when the pack voltage got up to about 171V, I unplugged it.

Hope that helps!

[ted.lowe]

[1] The Zivan is not specifically programmed for LiFePo.  However, it is my understanding that the creator of BMS uses the same charger.

Hmmmm...  As long as the current and voltage thresholds are appropriately set for your battery pack, i think the CC/CV phases of the NG3 will be adequate as a backup charger, but i wouldn't use it all the time.  The yellow phase should be avoided (LifePo charging specs specify the charge end at end of CV phase).  i also remember reading that the yellow phase can be disabled (only set by factory technician unfortunately).

This leaves the issue of balancing unhandled though (by both the NG3 and MM charger too).

For our tried and true flooded lead acid bats, the yellow phase raises the voltage very high and that is when all the batteries in series are 'balanced'.  We have much experience with this approach working.

For sealed lead acid (AGM), you can't use the yellow stage and we have a lot of anecdotal evidence that AGM packs fail prematurely due to uneven charging over time. 

i'm afraid that without proper balancing, the same problem will occur with the LifePo cells.   Much effort is put into preventing cells from overcharging, but no effort is put into undercharging.  Eventually, the differences in the cells will become larger and ouch!   

Does my argument seem right ? 

Todd/Folks: please challenge my understanding !!  Thanks!

[todd.dore]

Ted:

I'm in general agreement with you on the balancing issue.  I've also seen firsthand what the yellow sequence does.

When the Zivan enters the CV phase, blinking red, it cuts the power down from 16amps to about 5-7amps.  What happens, however, is that the power isn't cut down enough, and after about 20 min. or so (I don't know exactly), the pack voltage starts to rise.  The Zivan is set to go into blinking red at about 171V which works ok for my pack.  To reiterate, the current during the CC phase (16a) and voltage that kicks in the CV phase (171V) are ok for my pack.

But, ultimately it climbs to 181 for yellow, which is why I prefer to manually watch this and shut off the charger before it gets to yellow.

The MM cuts down power a lot more ==> 1-2amps at 171V.  So the 'finishing' charge, which you can set how long this is for, does a much better job at maintaining CV.

In terms of balancing, I think the MM does some kind of balancing, although I only have some vague anecdotal evidence of this.  After maybe one or two dozen charge cycles, my batts are still very balanced at most stages of the charge/discharge cycle.  Time will tell in the long run to see how this works after hundreds of cycles.

By the way, I have about 800 miles or so on the LiFePos.  I hope to have 1,000 by the end of this week, and would ideally like to put 1,000 miles each month on the car.

I think this is a good discussion - perhaps we can do a presentation during one of our meetings on this subject?

[todd.dore]

More fun at the EV OK Coral today -

On our way to another Doc visit via the expressway, the warning buzzer went off for a couple seconds.  WTF?  So I pulled over at Medieval Times (parking lot), and within 10 mins. checked all the BMS loop wires (none were loose) and the battery voltage (all was ok - 3.33V each).  So, scratching my head, I thought this may be a false alarm and we continued on to our appointment.

On the way home, I took I-90 East to 294-S.  Just before the Irving Park Rd. iToll plaza, the car dies out - first the ABS light comes on, then the ABS brakes pump on their own very briefly, then walla, a dead car.  Fortunately I was in the far right lane and pulled over.  A Chicago cop was behind me and pulled over as well, asking for help.  He offered to call IDOT, but the IDOT tow trucks are not flat-beds (you need to only use a flat-bed with my VW New Beetle).  I thanked the cop anyway, and he went off to deal with another emergency.  I was grateful that he was there.

So I called a tow truck, the same company from Brookfield, same tow truck driver that I had about 5 weeks ago, and he remembered us.  He towed us back to my house in N. Riverside.  $130 poorer but back home safe.  I was late for work so I had to deal with it later.

After work looked at everything up and down (limited time), spoke with John Emde (thanks John!), and diagnosed the problem - a failed IOTA DC-DC converter!  I have the 55amp one.  I mean, nothing was coming out of this and my 12V was dead on arrival!  No wonder I died on the highway!

So, I need to look at the IOTA to see if there are any obvious problems.  If not, I may need to get a bigger one - the 75amp model.  With the motor blower on all the time, I may be pushing the IOTA to its limits (esp. w/ the power steering, which I use infrequently).  Anyway, the car is out of action yet again (this is getting old and expensive!).  I'll keep everyone posted.

I did get my refurbished Manzanita PFC-20 back today, which was the only good news I've had all week.  It's been one unbelievably bad week for me.  Thankfully I'm healthy, employed, and have my lovely wife, so I try to keep my head up.

I'll write as I make progress...

[todd.dore]

On Saturday I opened up the Iota DLS-55 DC-DC converter-it was totally shot.  Black carbon everywhere!  All the fuses were ok, so the capacitors must have given out.  Thanks to Rich (Pioneer), they had one in stock, and I bought it and installed it.  He kept the old one - he may send it back to Iota, but if not, I wouldn't mind using it for 'show-and-tell' at the meeting.  The new one has a relay installed which kicks the output to 14.4V once the ignition is on.  I don't think that's what happened to my old one, but it helps all the same.

Due to the placement under the hood, the DC-DC replacement was a major pain.  It took several hours!

Ken suggests that a 'pre-charge' circuit needs to be devised and installed before you turn on the main breaker (Anderson connector, in my case).  A way around this, of course, is to leave the main break on all the time, which I will do for the time being.

I installed the Manzanita PFC-20 and adjusted the BMS to match.  I had some minor wiring work, and the RegBus was working again.  The PFC-20 is a far superior charger (CC/CV algorithim) than the Zivan, so I'll be using this at all times.  One curious event - I was able to get the Manzanita up to 24 amps - and the current dial wasn't up all the way!  Perhaps Rich R. threw in a Buck enhancer for 30 amps?  I'll check with him - if so, that would be sweet!

My mileage remains very good.  On the day that the car broke down (Friday) on I-294, I did 49.4 miles (all highway), and it took about 14.5kWh charge - for 293Wh/mile.  Excellent consistent efficiency!

The batteries are holding up fine.

Back later!

[ken.simmermon]

Hi Todd, I think the IOTA my need a pre-charge resistor if it's being disconnected on a regular basis, my 2 cents

Ken

[ken.simmermon]

oops