Linear Tech 12-cell battery monitor/BMS

[Arlo1]

Methods I love you man! The only reason I havent ridden my bmx in the last 4 weeks is I dont have time to get the battery ballancend with my MKSppro!

 

[liveforphysics]

It's good to see you back Arlo1

I'm damn excited about Methy's project too.

 

[Arlo1]

It's good to see you back Arlo1

I'm damn excited about Methy's project too.

IM just in the middle of putting an offer in on my first house! So me quite buisy IM not going anyware though once this is done and i get some renters in I will be building some realy cool sh#t. Three goals for the near future are the 6-71 on the road runner, my brushless motor jet wing!, and a 150 volt 2wd ebike!

 

[asynchrono]

Hello Everybody.

I am sorry for entering the forum with this request. However, desperation has driven me to do so.

I am in need of two or three LTC6802-1 chips very very urgently. There appears to be a shortage of these in Europe and the US.

If any of you guys have some spare ICs, even if they are already poplulated onto an old prototype board, I would be willing to reward you for helping me out in my time of need Just send me a private message.

Thank you very much for your time guys. And again, sorry for the personal request on this board.

All the best.


Nick

 

[Arlo1]

Uhm so....... I have been a bad boy and neglected my bmx with my bm6 moniters pluged in while I was not around and uhm yeh two cells were below 2 volts one was about 1.1 or somthing like that. SO I have been slowly bringing them back up. All I use for the charger is the 2 power suplies set to a total of 100.8 volts. So I chaged the battery up till the highest hit 4.22 and my moniter started beeping and then I set the MKSPPRO to discharge it to 4.0 volts for the whole pack and that took 24 hours! I then charged it again and the same battery was still the highest but I got the others to 3.9 so far so At this rate by the end of the week I should have a ballanced battery pack again. Remember I am running 24s 2p 5000 mah turnigy lipo so there is alot of energy to shunt off in heat from the little mksppro!
Come on methods I need to buy a proper ballancer asap!
We should start a donation fund for methods to get this going faster! :wink:

 

[methods]

I am in need of two or three LTC6802-1 chips very very urgently. There appears to be a shortage of these in Europe and the US.

Nobody ever has these in stock.
You have to buy directly from Linear. They say 2 week lead time but they will ship right away. I just ordered 100 of them a few days ago ($1,300.00 !#@!%!^!)

You can also request a sample - which they definitely will ship right away.
If they refuse a sample - get someone at a tech firm to request the sample.
If you dont know anyone claim to be a student.

Where there is a will there is a way.

-methods

 

[grindz145]

Me like bms porn

Indeed.

I think it is fair to say that this would step up the battery monitoring 10x from what just about anyone in the ebike community has right now. Amazing.

Are you ever going to be able to tell us who the customer is?

It's interesting that you won't have any monitoring capabilities without USB power. This is definitely a departure from the norm in manyyy ways.

 

[asynchrono]

I am in need of two or three LTC6802-1 chips very very urgently. There appears to be a shortage of these in Europe and the US.

Nobody ever has these in stock.
You have to buy directly from Linear. They say 2 week lead time but they will ship right away. I just ordered 100 of them a few days ago ($1,300.00 !#@!%!^!)

You can also request a sample - which they definitely will ship right away.
If they refuse a sample - get someone at a tech firm to request the sample.
If you dont know anyone claim to be a student.

Where there is a will there is a way.

-methods

Methods. Thanks for the advice. I usually buy from Linear Technology direct. Usually a two week wait is OK. But, on this occasion I have been forced to redesign a board to take the LTC6802-2!

You are right, where there is a will there is a way

All the best.

 

[methods]

It's interesting that you won't have any monitoring capabilities without USB power. This is definitely a departure from the norm in manyyy ways.

Lets talk about that - not exactly sure where you are going with that idea.

As far as passive monitoring - of course I have LED indicators on the shunt lines so - just like an astro-blinky - users will be able to see which cells are shunting. When it comes to a power light, LVC light, or any other indicator light... I think that is pretty asinine. It is a continuous drain on the pack for no good reason. LVC indication must be the worst of them all. Hmmm let me think.... Ok - my battery is dead so now I will light up some 15mA LED's to drain it further

So - that leaves the idea of having an LCD screen right on the BMS which is an insane power hog so... that is a big no.

The #1 offense that a BMS can make is to draw power right?
If you store your bike over the winter you dont want to have to tear into the BMS and open up a bunch of balance taps
A BMS should be able to be left connected to the pack indefinitely.

So back to monitoring -

The only other time/way to monitor would be serially - so why not steal the power from the USB?
I suppose I could have gotten clever and tried to tap charger power but that just adds complications IMHO. I would prefer to hold the charge throttle at 0% until the user turns on monitoring - then allow charging. The irony is that 5V USB power actually comes from an isolated 12DC-DC that comes from the battery anyway so.... I am actually drawing my power from the pack - but only when the user wants me to.

In that sense - why would you want a BMS to draw power from the pack any time other than when the user requests it?
Why would you ever want to charge and *not* datalog the charge profile?

Those are rhetorical questions - I am just trying to share some of my thinking.

Did I miss something?


So looking at all the situations:

Discharge:
* Due to self discharge
* Due to BMS
* Due to Controller
* Due to Accessories

Charge
* Due to charger
* Due to regen

Lets look at those. Self discharge and BMS discharge need to be minimized - right? I did that. Any time the Controller or Accessories are powered the USB is also powered - so that is safe. Seems I have minimized leakage and protected from all "normal" discharge paths. If any cell reaches 2.7V (Critical LVC) an opto output can trigger the main contactor to "self destruct" and cut power. Otherwise, it is pretty easy to set up a system that will gracefully shut down once it has reached standard LVC (think cell phone, laptop, etc)

For Charge - In both cases the USB power will be active. I think it is reasonable to deny charge without datalogging being on - if for nothing else - than to get a record if a failure occurs. Imagine having dataloggs for every single time you charged or discharged your bike over its entire lifetime

I hope I dont sound argumentative - you got me thinking and now I want to know what you were thinking about
Care to share any ideas??

-methods

 

[methods]

EDIT:

I just realized that you are looking at this from an Ebike perspective.
Now your comment makes a lot more sense - like who is going to duct-tape a laptop on their bike right
I have been thinking about larger size EV's that have user interfaces etc. A UI with USB is just assumed I guess.

As far as a version for Ebikes - they would end up having to have a UI of some sort controlled by a power switch.
Of course a simple version could be made - but why?
Gary can sell people a BMS that just protects you from LVC, shunts HVC, and throttles current.

This toy is for people who want to have an "experience". Battery geeksters - ya know?
Could be as simple as hijacking the CA.
More likely it would be an android type device - right? Mine has USB

-methods

 

[liveforphysics]

More likely it would be an android type device - right? Mine has USB

-methods

I love android.

However, our Android devices don't output any power from the USB port, only milk it from whatever they connect to.
Dedicated screens and drivers are cheap. PSP screens can be had for $35 shipped, and are way better in every way to our phone's screens.

 

[methods]

Yea - I was just trying to illustrate a point. I had not really thought it out that far.

Whatever device you would choose to use would be "plugged in" so power would be available.
I am not to worried about the source - I am sure that we can find a switched 5V power source.

One thing I am sure of is that a BMS should be as power efficient as possible so that it can be "forgotten" about.
My average current is about 150uA for 36 cells. Round that up to 200 for margin.

200uA/hour
4.8mAh per day
335mAh per month
4Ah per year

So you could hook my BMS up to a 24S 5Ah Hobby City pack, leave it sitting all winter, and you would probably be alright.
More realistically it would be 12S - so divide those numbers by 3

1.33Ah / year BMS discharge

Now: If I were to try to power the monitoring directly from the pack those power numbers would skyrocket - even if I did something clever like slept 99% of the time. Sure I could get the LTC power down pretty low but what about all the external devices. There would be bias resistors, individual part quiescent currents, and all sorts of "new" current draws on the system. How about the uC power? Lets say we go a step further and I carefully select parts that have sleep modes - or parts that can gracefully wake up from power down.... sigh... I have done a fair deal of low power design and I would say the ultimate solution to a low power design is to "plug in" Then I dont have to worry about that crap.

* LTC Power x 3
* uC power
* USB chip driver power
* Converter efficiency
* Temperatures probe power
* Current sensor power
* Opto isolator power
* Bias resistor power (pull-downs on com lines and gates)
* Power across temperature

I think people are going to start realizing why BMS's are not built directly into batteries when we have people coming home to find dead batteries.
Especially when they took the time to discharge their pack to "storage voltage" LOL -> I discharged my cells to 3.7V for storage to protect them - then the BMS ran them dry

-methods

P.S. Before anyone points out ultra low power parts like the MSP430, 1uA op-amps, etc - yea, I know - that is what i do at my day job. :|

 

[methods]

I just tested the balancing section.
All 36 channels work right out of the box.

I wrote a script to blink the lights twice per second.

I also tested the WDT. Works great - if I unplug the USB (which powers down the uC but leaves the LTC's on) after a couple of seconds all the shunts turn off and the LTC's go into low power mode. When I plug the USB back in - everything comes back on line.

SHWEEET!

Now I need to solder up 1,000,000 more shunt resistors and do some hard-core burn-in tests

-methods

 

[liveforphysics]

I just tested the balancing section.
All 36 channels work right out of the box.

-methods

Not bad for a twink.

I'm super proud of you bro.

-Luke

 

[methods]

Looks like my fans are not enough.
I populated all 216 shunt resistors and ran them at 50% duty cycle
With the 3 fans I intended to use I was seeing 70C surface temps.
The resistors are rated up to 155C

When I ran the duty cycle up to 100% the temp crept up to 150C. Granted this was not in the enclosure - which will focus the air flow better - but for a rough test I can tell that the 3 fans I selected are not going to cut it. I am going to just stop screwing around and bolt up one of my Mega 12V @ 2.5A insane computer fans. These blow hard enough that they float on the table. I am positive - beyond a doubt - that these will do the trick. After I see how cool I can get the resistors to run I will try to select a quality 3" fan that is priced reasonably.

I made a little video but it is pretty Ghetto. I will probably wait until I get things in the case to post up any more pictures.

-methods

 

[liveforphysics]

Looks like my fans are not enough.
I populated all 216 shunt resistors and ran them at 50% duty cycle
With the 3 fans I intended to use I was seeing 70C surface temps.
The resistors are rated up to 155C

When I ran the duty cycle up to 100% the temp crept up to 150C. Granted this was not in the enclosure - which will focus the air flow better - but for a rough test I can tell that the 3 fans I selected are not going to cut it. I am going to just stop screwing around and bolt up one of my Mega 12V @ 2.5A insane computer fans. These blow hard enough that they float on the table. I am positive - beyond a doubt - that these will do the trick. After I see how cool I can get the resistors to run I will try to select a quality 3" fan that is priced reasonably.

I made a little video but it is pretty Ghetto. I will probably wait until I get things in the case to post up any more pictures.

-methods

Gimme a call fatty. I've got some ideas.

 

[fechter]

Now: If I were to try to power the monitoring directly from the pack those power numbers would skyrocket - even if I did something clever like slept 99% of the time. Sure I could get the LTC power down pretty low but what about all the external devices. There would be bias resistors, individual part quiescent currents, and all sorts of "new" current draws on the system. How about the uC power? Lets say we go a step further and I carefully select parts that have sleep modes - or parts that can gracefully wake up from power down.... sigh... I have done a fair deal of low power design and I would say the ultimate solution to a low power design is to "plug in" Then I dont have to worry about that crap.

That's one of the stickier parts of a BMS design. Nobody really wants to have to plug in two plugs every time you charge, but that is a simple and reliable solution to the problem. The LVC function has to work all the time except when charging too.

My old ver. 2.x boards used a jumper at the charger plug to do the switching. Another way would be to use a magnetic reed switch on the battery side connector and a small magnet on the charger connector that would turn on the control circuit whenever the charger was plugged in. This would allow for a single connection if there was an on-board 5v supply that runs off the charger voltage.

Stanby drain of the cell circuits is another unavoidable feature of any BMS. That's why it's a good idea to have a connector between the cells and the BMS so you can unplug it for long term standby. With large format cells, it's probably not really an issue as long as you charge at least every 6 months or so. Just for reference, my analog cell circuits drain about 200uA at full cell voltage and the drain tapers down to about 10uA at 2.5v.

 

[grindz145]

For Charge - In both cases the USB power will be active. I think it is reasonable to deny charge without datalogging being on - if for nothing else - than to get a record if a failure occurs. Imagine having dataloggs for every single time you charged or discharged your bike over its entire lifetime

Sorry I haven't had a chance to check the forum in a few days.

Charging is the most important thing I can think of. In general a BMS is a redundant protection even while off, so that any detrimental charge attempts can be protected against. I didn't mean to say that you're BMS is flawed, just a completely different approach, which makes sense considering what you're doing.

Alot of the fundamental BMS features do not make any sense here. For instance short circuit protection ... yeah right.

I do like the idea of having a circuit monitoring the cells all the time though. Generally the safety measure you perform is to isolate the pack using a fet or a fuse. In this case perhaps a fire suppression device is more realistic Even in a BMS which is always on, it will eventually enter a sleep mode until some type of interrupt wakes it up so as not to enter deep discharge.

It makes sense that once you enter your car and turn it on it would give you the warning (LV, OT, ETC) but I would almost like it immediately, before I even enter the car. This is completely overkill I know, because the cells I'm used to are MUCH more dangerous than anything you would (in your right mind) use large scale. Still, if somehow you get something draining your battery (some type of fault) it would be nice to initiate some type of safety device, even if not physically in the car/motorcycle or whatever. I've seen a few burnt garages even from EV guys with PbA. It is really a whole new ballgame with this mammoth of a pack, and discharge rates as they are.

But to be clear I understand that the main function of this BMS is balance and monitoring, and not radio alerting fire department Any addition control will happen downwind of this.

P.S. I think this thing will absolutely be a killer way to monitor the pack, and I can see building the coulomb counter software on top of this GUI with CA interface. Would be intense...

 

[methods]

I am getting really excited...

The cases are at the machine shop right now being set up for a pair of 48V fans. I tested the fans to 68V each and they did not fail - almost 50% margin. The good news is that i wont need a power resistor or anything silly like that - the fans can hook directly to the charge input so they come on with the charger.

I have wired up the charge throttle circuit and I am starting some 100V 20A starter tests.
I am trying to think of a metric for measuring margin without actually smoking parts off the board. I suppose mosfet case temperature is my best bet.

Wait till you see this thing.... totally enclosed and totally bad-ass.
I had a realization yesterday that makes the design about 2.7 times as bad-ass (in terms of ease of use).
I will wait until it is all together to show pictures.

-methods

 

[methods]

WORKING 100% !!!!!!!!!!!!!!!!!!!!!!

I got the charge throttling working 100% right out of the box

At first I had the PWM frequency set around 300hz -> this would sometimes trip the protection on my power supply.
I then bumped the frequency up to around 3khz - this sounded like a swarm of locus but worked perfectly.
I made a final bump up to 30khz and it works like a frigging dream.

- dead silent
- smooth as silk
- no interaction with the charger
- 1024 steps of resolution (1/10th percent or basically 100mV)

I wrote a quick script to sweep the current from 0A up to 15A and back to zero over and over.

oh man..... SWEET!

and guess what? Fets are hardly even getting warm. Who knew? If you populate your board with 400A worth of 150V mosfets it stays cool at 20A
(Credit to Jamie who asked me why the hell I would spend $2 to heat sink 1 fet when I could just run 2 fets for the same price and have no heat)
(so of course I added 4 - because more is better )

Eat your heart out analog boys Papa methods whupped up on this! Nothing like being able to change the resonant frequency of your circuit with a few keystrokes :|

I am going to go do my victory dance.

Methods LTC BMS V2.0 -> 36 channel, 1.2A balance current, 150V 50A charge , ... worked right out of the box.


* A big thank you to everyone who encouraged me along the way...... and a big F-U to everyone who though I couldn't do it

-methods

P.S. Movies soon - since I know Doc does not believe anything he does not see with his own eyes :wink:

 

[liveforphysics]

Hoorah!!!! Good man Methy!

I'm so proud of you bro!

 

[mdd0127]

Subscribed for V3.0. I need one asap and hopefully a bunch more soon. Wow!

 

[voicecoils]

Awesome Meth, hard work paying off

 

[Hyena]

Nice work Methy!
Aren't you glad it's mainly guys on this forum, if there were more girls around you'd be swamped with requests to have your babies :lol:

 

[Doctorbass]

WORKING 100% !!!!!!!!!!!!!!!!!!!!!!


P.S. Movies soon - since I know Doc does not believe anything he does not see with his own eyes :wink:

Methy I am so exited too.. I just reed all that thread i just.... discovered... :lol:

Your work seem very clean and profesional! I still appreciate alot the Analog version of Fechter and Gary, but that digital programmable version is very interesting !

Will it be availlable for us who ask for

I agree with a versio that could be operated with a pocket PC or other smaller PC than a laptop... i just think about the TTXGP guys that could very appreciate device like that!!!

Remember 5 month ago when i was searching for a 144V bms solution?..I finally assembled a 40s version of the v2.6 gary fechter BMS and it work well with a Zivan charger but now you just did it!!!.. that digital compact BMS i searched for!

A great solution to help cooling these shunt resistor at 100% duty wold be to cut hoed between each resistor channel on the PCB to let the air to circulate fron teh back to the top of the PCB.

man... 144V at 1.2A is... IS....IS.............IS.... more than 170W of heat to sissipate!!!! .. I guess your airflow will need to be optimized!!!!

Doc