Friday, June 4, 2010

Life in LiFePo4 - Cycle Life and Attenuation Data

Kind of a late blog entry. Busy week with Rickard Electric Speedster design. The May 28th show rounded up a lot of little parts such as the accelerator - with a little valet function box I refer to as "daughter mode."









The basic concept is that we can use the TBS E-Xpro instrument to monitor state of charge by AH. And we can kind of flexibly define a percentage SOC based on any AH value we like. The unit will close a relay when a defined value is reached. But a relay to what?

I have a bevy of daughters. Not to denigrate them, but they are not precisely EV enthusiasts, although they like cars and like the feel of an electric. But when out and about, if they did start to "run out of gas", I'm trying to visualize something that would convey to them that it is time to go home, and if they don't go home, it will cost Dad about $9000 in batteries.

An LED probably won't cut it. I can picture a daughter saying "Hmmm, a light. I wonder what that's about. I better ask Dad when I get home." She would then continue her travels until it suited her.

With a buzzer or audio alarm - "Hmm, that's certainly annoying. I better ask Dad to fix that next time I see him. "

So what can I do to get her to have a sense of urgency that there is something wrong with the car that needs attention NOW.

Well of course we could easily wire this up to a controller input to disable the controller. But this is a little extreme. What if she is in traffic and needs to get off the road?

The concept we've settled on is in no ways original. It is commonly referred to as limp mode and indeed there are some ways to do it in the controller. Our Curtis determines state of charge by voltage and I'm none too enamored of that. I want to use the AH counter.

So we devised a very simple little box, it has not enough circuitry to even be considered a circuit. Indeed there is no circuit board. It's a couple of resistors wired across the terminals of a relay.

In normal operation, it simply passes the 0-5v throttle signal from the throttle through the normally closed contacts of a 12v automotive relay to the controlller.

We can use a small dash mounted switch OR the TBS amp hour counter relay to pass 12v to the valet box to activate it.

This simply divides down the voltage to about 1/4 of its normal value before connecting it to the controller. I've added a 5K pot to trim this from the 1/4 value down to zero basically.

We also made kind of a discovery with our Vantage Vehicles GreenVan. This miniature van is a very el cheapo little Chinese cargo van that is SO small and SO basic and so ugly it actually starts to look cute. The company is based in Corona California (http://www.vantagevehicle.com).

But we were surprised to learn they used some pretty high quality components in the "conversion" they do to these vehicles. The system is 72v but use some unidentified but physically enormous AGM cells of 198Ah. The charger is a DeltaQ. These are really pretty nice chargers though not adjustable. The only reason we don't talk more about them is they are just focused on the low voltage market. I think the highest voltage they will do in their product line is 96volts. And just about no one is doing 96v in EVland. It's kind of a Neighborhood ELectric Vehicle thing.

The motor is also a little mysterious, but seems to be a 10HP three phase AC induction system with a 35 HP peak rating.

To drive that, we were a little astounded to find a Curtis 1238-6501 three phase controller. This is the 80V version of the controller we are using in the Speedster (Curtis 1238-7501).

The system appears to do regenerative braking only on applying the brakes. I'm increasingly leaning toward this on the Speedster and the Mini is now dialed down to very low values when you take your foot of the accelerator, but quite a bit when applying brakes.

The GreenVan does this a little differently. Instead of simply keying on the brake light signal, they employ a pressure transducer in the brake line to convert brake pressure to a 0-5v signal into the controller. The transducer gets 5v and ground from the controller, and puts out a voltage varying across its 500 psi range.

I thought this transducer might be hard to find. We located the company that makes it, MEAS Specialties, and contacted them. Never even got a reply to an e-mail. Before calling, I looked around a bit on the net and worked out their part number scheme. Turns out, it is VERY commonly available on Digikey at $104 each - Digikey part number MSP3501P2-ND. And you can get them at various pressures, 250 psi, 500 psi, 1000 psi etc. This might be important. I don't know how well our Speedster pressures will match the GreenVan's.

Anyway, this will take some experimentation to get the feel right, and may not be enough improvement to justify the effort. But I like the feel of the GreenVan's brakes.

May 15th marked the 1 year anniversary of our first video post. As many of you know I was in publishing before (print magazine) and just love the business. I've tried other businesses, but after publishing, nothing really makes any sense to me. I've seen them and they didn't make sense. So I would try them myself, and was surprised to learn I had nothing to learn there. Most of them just don't make much sense. It's hard to move heavy objects around for a 30% markup.

One of the things about publishing that I understand viscerally is about the flow of information. You give it away. To anybody who will listen. And you can't give it away FAST enough to avoid becoming successful at it. The only way to fail at publishiing really is through a lack of generosity of spirit. If you start viewing information as proprietary and valuable, you generally sell off the office furniture at auction a few months later. If you strive to pick up information gems and IMMEDIATELY give them way, it seems it rains money and other good things.

This is a bit counterintuitive. But one of the keys is that the information is not one way, and you are not really the producer. If people find the info useful, and gradually come to trust that it isn't a trick, they start to add to it. It's kind of a stone soup process. You of course have to initiiate it with the stone. But soon people are showing up with cartots, vegetables, onions, potatoes, etc.

Basically, you become a goto focal point where people can usefully promulgate useful information. As these contacts improve in quantity and quality, your information flow improves dramatically, and becomes something quite beyond what you are actually capapable of. Your role kind of devolves to sorting and categorizing, trying to filter through to find the gems that your readership, in this can viewership, might find useful. The more useful, the more people tune in, and the information sources grow and improve.

The good news is that you wind up with a lot of information. The bad news is, there are some real world problems publishing all of it. This gets a little hard to explain. But what made us very successful at Boardwatch was we got very good at sorting out the bullshit from the infogems - initially through hard work and experimentation. But over time we just developed a feel for it. But the other skill that is kind of hard define, goes hand in hand with the generosity of spirit.

It's actually hard to describe. If I use the word "tact" the people that know me in the forums I've just left are going to laugh outloud. Simillarly "appropriate judgement" rather fails. How about a "sensitivity to info issues."

Most everyone loves secrets. And the greatest joy is in telling them. Even greater is seeing them told widely. And often the value of the infornation is kind of tied to how secret it is. Scarcity being a key element in the value of anything.

Unfortunately, in the real world, we face certain ethical dilemmas. We may not quite approve of what our employer is doing and have a need to get some information out, but on the other hand, you don't want to lose your job in today's economy - with a couple of young children to feed and an upside down house payment.

So the GM Volt engineer loves to call and chat about some very interesting things. But if her employer knew she was putting this info out, there would be a rather immediate career change. And often, just the nature of the information points to the likely source.

So one of the frustrations is that I wind up with some inflows that are terribly interesting, but I don't know precisely how to deal with them without wrecking lives and careers. At one time, I was pretty brutal with this. "If you have a secret that needs to be a secret, why in the hell are you calling a magazine editor?"

Unfortunately, our actions have consequences, and often the consequences are someone ELSE's consequence. And if you see this go down a couple of times, you just can't help but become a bit more discrete. Time heals all. Often a month or so in the "aging cellar" changes the whole situation and no harm no foul then.

Sometimes, I can just generate this info out of thin air without attributing a source. That makes me look very smart. That's not precisely the goal. If it came to me in a dream or I'm a little vague about it, occasionally there is a reason. But it appears it would damage someone or something if I get too detailed on where it came from.

That's all a rather long winded tour of the joys of journalism. I really like the position and I'm frightfully good at it. But like Chess or Bridge, you never really master it. Actually, I always assume some portion of our viewership/readership is actually a lot more knowledgeable and a lot more intelligent than I am. It's not ALWAYS the case, but it's a good assumption to start from and particularly appropos in technical fields. Everyone is a specialist. And by definition I have to be a generalist. So, it's pretty real. Those in that ca
tegory will hopefully pick up from this why some elements of EVTV are always going to look a little odd.

For the rest, here's a little example. You may or may not know that I'm particularly enamored of these Chinese prismatic cells. Of course the reason is obviously weight and volume to power - they make electric cars viable in a way that lead acid, NiMH, etc simply cannot in a very ratiometric fashion. The particulars of the lead acid battery are immaterial. At that size and that weight and that power, it just doesn't work except for toy demonstrators.

But for now, the price is high. So high, that the only way this works is if the cells last a very very long time compared to the other battery technologies. If you have to replace them every three or four years, you have simply moved the operating expense from the gas pump to the battery store.

But at some point in longevity, rather keyed to the life of the car, the expense moves out of the recurring operating category and into the capital expense of the vehicle itself. We no longer talk about miles per engine. In fact, miles per tire used to be an EXTREMELY important factor in tire sales. It just isn't anymore. True, we go through a few sets in the life of the car. But somewhere past 50,000 miles, we ceased caring very much.

The heart of all this is that a good Trojan lead acid cell is rated at 350 cycles. The LiFePo4 cells are spec'd at 2000 cycles if we limit discharge to 80%, and 3000 cycles at 70%. Thundersky recently went to 3000 and 5000.

Those deeply experienced in the lead acid pool, have pooh-poohed this forever as being hopelessly optimistic. I feel their pain, but I've known for some time that it is off base. Part of this is cultural and part technical. The good news is that your LiFePo4 cells will last a VERY long time unless you actually destroy them, and of course a BMS is a quick way to accomplish this.

The cultural part is pretty broad. You have to understand the Chinese a little bit. "Managers and Scientists" are very highly regarded there. "Salesmen" are almost viewed in the same light as street hawkers selling roast peanuts. And it is not seemly to brag on your children. Indeed there are some serious karmic issues with mentioning your children in a positive light. And finally, if you ask a question of a superior in China, you are basically challenging him. If it is something for you to know, he would have already told you.

This all sets us up for some serious disconnects with America. First, in the good ol' USA, if you will do 180AH sometimes, even for a second, label it 180AH. Better, let's call it 190AH. Who measures? If it's a problem, we can always add a four point disclaimer at the bottom of the back page.

In most American corporations, the sales and marketing guys not only pretty much rule the game, that's almost always where the CEO's come from.

What you will find in dealing with the Chinese is that if they tout it as a feature or specific value, it will ALWAYS exceed that value - or at least they really did believe it would. They actually pad things the OTHER way. We're right around 190AH, - label it 180AH and we will not appear unseemly.

Second, the sales person you might be dealing with is essentially POWERLESS in the organization. They have to get permission to use the restrooms. And if you press them for technical information necessary to make the sale, you put them in a very awkward position. They either have to go appear to challenge their boss, a very painful prospect, or risk losing the sale. They do NOT work on commission as is common here, but failure is its own reward in China and you do NOT want to fail as a salesmen. There is no lower place to fall to. So you might note a little stress from a simple question about how they come up with cycle life data.

In China QUALITY is the golden horizon. The executives there actually do not understand why we want all this info. Don't we know they are claiming to have the HIGHEST QUALITY - GUARANTEED. Who would want more than that?

Finally, they're just a bit secretive about everything germane. I myself don't quite undetstand this, but all Chinese drama centers on a secret. The companies who make these batteries are not scientific in nature. They discover nothing and they invent nothing. They do develop manufacturing processes to ensure quality and output. The cathode material in the China Avaition Lithium Battery Company cell comes from a company titled ALEES in Taiwan. They buy it by the barrel. Similarly, you can buy aluminum foil and copper foil in China in a wondrous array of widths, lengths, and weights and some if it even perforated in various ways.

What you know of as a battery company in China develops clean rooms and workshops with some automation and they order these materials and produce a lot of cells - 1000 cells per day say. They know about as much about the chemistry as my daughter does.

They DO do testing, for quality purposes. And of course to develop better batteries with higher outputs etc. But the data is for internal use only and indeed, the salesmen aren't normally very privy to it.

I did stumble upon a pretty bright sales guy with Sky Energy. And he was pretty good at hunting down information although this was a painful process taking weeks and dozens of e-mails to get even the most basic info. He appears to have not made the cut in the transition to CALB - probably too inquiring and uppity.

One of the most fascinting things I got from him was an actual .XLS spreadsheet listing the AH values from 500 cycles of testing for what they term "attenuation" or the decrease in capacity over time from cycling the cells. I got one of these for the 180AH cells and one for the 100AH cells, with two cells tested and the actual values. I've prepared a couple of simple graphs from this data.


There are a couple if interesting things about this test. First, the 500 cycle length is impressive. This is 500 actual cycles, in a world where lead acid cells are ESTIMATED to have a 350 cycle life. They show a pretty uniform 5% degrade from their initial value - and it appears to be quite linear. That implies 2000 cycles easily to 80% of initial capacity.

More encouraging, the initial value is ABOVE the rated capacity. And indeed we are just starting to break below advertised specs AT 500 CYCLES.

The tests are of course at a very moderate 0.3C. But the thing that just blows me away is that all 500 cycles are to 100% discharge - not 80% or 70%. They run until the voltage at 0.3C breaks 2.0v which is their definition on the spec sheets of 100% discharge.

So what the data appears to show is a minimum 2000 cycle life at 100% discharge. They are fudging this to 2000 cycles limited to 80% and 3000 cycles limited to 70%.

In America, we would test at 50% DOD, claim the 3000 cycles, and not mention this silly depth of discharge thing at all. Or else it would appear at the bottom of the back cover of the report with a tiny little cross and in 2 point type.


So I am very taken with the prospects for long life from these cells. They just appear to be bullet proof to me.

Another interesting aspect that explains some of my own observations. These cells appear to INCREASE in capacity over the first 50-75 cycles. Further, the curve made looks just a little bit like the discharge curve. It appears steeper on the left than it does on thte right where i appears to flatten. This might imply even LONGER lifetimes if it si flatter at 1500 cycles than at 400 for example.



Jack Rickard

16 comments:

  1. Nice data. Must go back to "Watch Paint Dry" ;-)

    For what is is worth I am putting my inital charge/discharge/charge cycles on the web: http://www.citystromer.dk/LiFePO4.html

    I am about to define where to go from here. I had done the first 10 out og 68 LFP90AHA in pretty much a single cell cycle and I am trying to figure out how to speed thing up, without lossing to much data.

    I think I will do the bulk charge (0.3C) of a series string of 10 cells and then do the rest of the charge one-by-one. The discharge could be done in the same manner - I have 3 different batches, where the two of them are very similar, and third is just holding spec (92 - 94 Ah). A discharge could be done until the first cell reaches 2.8V and the estimate the rest of the cells in the string from their curve shape.
    This would cut the time spent with a factor of 5.

    I have 4 cells in excess which will be used for temperature dependency measurement.

    Thanks for your posts and videos,
    Martin.

    ReplyDelete
  2. Jack,
    Did they happen to provide any data for a given charge cycle (I and V vs. Time curve)?
    Thanks.
    Martin

    ReplyDelete
  3. Any idea of what causes the high and lows on the CALB graphs?

    Martin.

    ReplyDelete
  4. A plot of current and voltage during charge and disharge at 1C we have posted here. No, they did not give me one for the 1/3C but I've posted those before as well.

    The 500 cycles is something that takes some time and automation.

    Yes, there are anomalies and questions and I don't have the answer.
    There are discharge cycles quite beyond the curve. Then therre are a series of cycles that vary signficiantly, followed by a series that are dead linear, and then a more random set that look a bit suspicious.

    What was most interesting was the propensity to reverse and for 20 or 30 cycles the capacity seems to grow, then reverses again. What is THAT about.?

    It has been my experience that all good answers contain the seeds of further questions.

    Jack Rickard

    ReplyDelete
  5. Hi Jack,

    Is your 3C charge/discharge testing coming soon to EVTV, you have been teasing us for weeks now!

    We love watching paint dry.

    Keep up the good work, good on you for pulling out of the forums.....

    ReplyDelete
  6. You just don't know what your missing in the EVDL. EV Components drama is in full swing. I suspect that several of these people will not be on the list as well, but not voluntarily.

    A wise move leaving. I've always been skeptical of the folks there with possible hidden agendas, and your prodding has only proven my fears correct. Bravo Mr. Rickard.

    ReplyDelete
  7. Check this out...from a company who sells EV parts. They designed or are designing a BMS and don't even have any cells to test them. Assuming somebody sends them a dead cell and they build this BMS, would they actually sell a BMS having never tested it on a working battery cell?

    "Anyone have a dead Thundersky/Sky Energy cell? Need a unit to use as
    a BMS mounting mockup."


    What a scary wild west industry we are in. It will be nice when it matures, but there are plenty of growing pains to come.

    ReplyDelete
  8. Your characterization of the Chinese practice of delivering more than promised with battery capacity does not seem to have carried over into their other industries. I've seen a number of overrated Kelly controllers fail, and a couple of recent failures of the new Chennic controllers as well, (which look like reverse engineered Curtis's). My second Kelly throttle is also failing after about a month as did the first. Didn't you also have a failure in your Kelly hall effect throttle?
    JRP3

    ReplyDelete
  9. JRP:

    The failure of Kelly controllers seems quite traceable to input rush currents by people who don't understand such things. We've run a 1400 amp Kelly controller for 18 months now without ANY kind of incident.

    We did have a failed kelly hall effect pedal. Worse, it failed ON ala Prius fashion, though not very dramatically. We are currently using a very similar Chennic throttle with a different hall effect sensor in it. They're aware of the problem.


    How were the Kelly controllers "overrated." We have achieved everything spec'd on the Kelly controller we have, although we normally do not run it at that level. The throttle is rather a life cycle item with no specific life specified.

    It is my experience that it is quite carried over into all products we've looked at. Does that mean that products do not fail in use? Errr.. no. Not on either side of the sea. But I find their specifications broadly much more conservative than U.S. products in the field.

    Jack Rickard

    ReplyDelete
  10. "But I find their specifications broadly much more conservative than U.S. products in the field."
    I have only found that to be true regarding their batteries, not their electronics. I'm sorry but a hall effect throttle that fails after a few months of use, and two pot throttles that fail after two months of use each, or less, falls far short of any specified or implied specifications. The two throttles I've received from Kelly, that failed in less than two months each, are obviously crap. Throttle Position Sensor pots last for thousands of miles, even in American built vehicles.
    PS. Any progress on your forums, so we can discuss such topics properly?
    JRP3

    ReplyDelete
  11. "We've run a 1400 amp Kelly controller for 18 months now without ANY kind of incident."
    And how often have you pulled 1400 amps from it? I'm guessing never.
    JRP3

    ReplyDelete
  12. JRP. You online forum guys never give up. Just negative rant and nattering from the sidelines with nothing really useful to say at all. Yes, you would be guessing of course. And you have NOTHING to base the guess on at all except your ongoing logic from your previous post. Nothing useful to contribute. No insight. Nothing positive. Just a GUESS at the most negative thing you can dream up.

    Keep typing. If you would drop that EV work and just type full time, you could be a genius.

    It matters nothing that we drive a car daily for 18 months and 10,000 kilometers with no issues from the controller at all. We haven't LOOKED at the setup since month ONE. It just works. But in your mind they fail.

    Having one, I have followed up and inquired on the failures I've seen. Every one has been somebody turning on the maintenance switch with the ignition on and killing the input capacitors. But the story never dies because you don't WANT it to die. It must be inferior because it is inexpensive and it is Chinese.

    Hey pard. It is what it is. The main asset in dealing with new and emerging technologies and products is to look at whats in front of you and what you see and accept that it is what it is. If you filter it with what you want it to be and color it with all your biases, you start down the road of myth and misinformation that led me to such frustration that I left your little fantasy world on DIYelectric.

    Try it. Observe it. Measure it. Your theories and my theories of what ought to be and how it all ought to go just simply don't matter even in the SMALL scheme of things.

    WHY would you guess? Get a 1400 amp Kelly. Theyr'e not that expensive. And then set it up and test it. At any current level you want. Please do report the results - even if embarassingly they do not match your preconceived notions of how they ought to work or NOT work.

    Finally, the proper design of the car would imply that you intentionally buy more capacity than you would use. If you want to do 1400 amps regularly, you need a 2800 amp controller.

    As in all semiconductor devices, heat management is the sole issue within your control. If I DID have a problem with a Kelly controller, after reviewing the manual I would assume I need more heat sink. As it happens, I have a really simple system in the first speedster. The Kelly is mounted on an aluminum plate that simply extends quite far down into the air stream. I had plans to mount a fan on it, but it proved UNNECESSARY as the controller operated flawlessly with this heat sink without it.

    I GUESSED I might need a fan. Through observation and experimentation I learned I did not.

    On Speedster Duh, I GUESSED a small heat sink would be sufficient. By experimentation and observation, I can already see that it will not be. I know for a fact your American made Curtis controllers DO fail as a heat function. I don't know that the 1238 does, but the company has a history. So I'm going to address heat from the get go.

    Do I mind being "wrong" either time? Not at all. All part of the process.

    Further, Steven Lui of Kelly has been amazing at responding to e-mail questions and offering advice - often with a 2 hour turnaround. Curtis won't give us the time of day and indeed act like they resent EV people using their forklift controllers.

    Jack RIckard

    ReplyDelete
  13. Yes we "forum guys" are a special breed. We have theories, questions, and look for real world results. Sound familiar? Most of the time I seem to reach the same conclusions that you do, so I guess all the typing is making me smarter. I have no doubt that Kelly controllers can work, and I'm glad they give good customer support. I hope I experience that with my second failed pot. I know the Chinese can make good products as well as poor ones. I know that people can damage a product through misuse. I also know that Kelly lowered the ratings on many of their controllers. I'm not saying that's an improper response, merely that it's in conflict with your premise that the Chinese always under promise and over deliver.
    Personally I don't have much interest in series DC at the moment so I won't be testing a Kelly. I'm only really arguing the point because I'm feeling a little slow this morning and needed some typing to up my IQ points. I am be smarterer already.

    JRP3

    ReplyDelete
  14. The problem with the "forum guys" is actually their total disconnect from real world results. They have theories, questions, more theories, more questions, more theories, more questions, and the entire basis of all of it is previous whacko theories and bizarre questions. It never has any connection to real world results at all, which causes it to wander off into the weeds with ever growing sense of the surreal. That's because it is surreal.

    At any point where it is DISRUPTED by reality, I'll be happy to help where I can.

    I've never said the Chinese ALWAYS do anything. They have a tendency to do some things that are culturally quite different, and it is important to understand why. I'm generally delighted with the results. And I eschew the stereotype anti-Chinese bias that permeates the forums where "good old American" generally translates to buying some overengineered undertested piece of utter garbage at an exhorbitant price from one of the forum participants.

    And yes, with enough sage advice from the wiser and smarter professors at DIYelectric, you can pretty much screw up anything - right up to and including vast clouds of smoke.

    Keep typing. You'll get there. I'm seeing some head swelling even now and even from here.

    Jack RIckard

    ReplyDelete
  15. To Jack: I wholly agree with your sentiments re those forum guys, some of which, who I am guessing, are paid agents of disinformation.

    The EVDL forum posters provided us incorrect suggestions to us in 2006 or so, in respect to issues with the Zivan battery charger and Curtis controller.

    We (Vancouver Gadgeteers and my students in schools for the gifted) finally made the 10 breakthroughs on the Revived Battery Electric Pickup Truck by doing basic experiments, i.e. starting from zero information.

    Jack, you are one of the very, very, few people who is putting out correct information regarding electric vehicles.

    The biggest threshold issue with EV's is (as you've already stated) how to get the battery replacement cost to a point where it becomes a capital expenditure.

    The public will get onboard once EV's per-mile cost isn't 2X to 6X higher than burning gasoline.

    Disclaimer: I have an interest in battery revival technology.

    Rob Matthies
    Vancouver BC

    ReplyDelete
  16. Rob:

    That's odd. I spend most of my time trying to learn how to kill them. If you can do what you do well, I must not be doing what I do well.

    Death to all voltaic cells!

    Jack Rickard

    ReplyDelete