[dimitri]

As I told you in the private Email, you make very bold assumptions and your statements clearly show lack of basic understanding of what BMS actually needs to do to be effective, while remaining cost effective. I could have added all those bells and whistles too, but I found thru extensive testing and experience that it wasn't really needed.

So, before you go labeling someone else's work, take some time actually testing your product in your EV and get some real EV miles on it. There is plenty of armchair designers on this forum and all others. Just because something can be improved, doesn't mean it needs to be improved.

As for open source, it was all my work and I chose to make it public to the extent that I wanted. I never promised to make every single detail public.

I am all for free market competition, Jerry. Make a better product, just don't bash something you haven't actually used, only to make yourself look smart.

 

[Jerry Liebler]

I certainly have made many assumptions! A big one is: When a battery manufacturer specifies a maximum charge voltage do the mean it in an absolute sense or do they allow tolerance? My assumption is they mean it with the tolerance of readily available measurement instruments. Thus I chose 3.6 volts +/- 1%. If I should have chosen the absolute maximum I can easily "recenter" the design to 3.6volts + 0% -2%. Another assumption I have made is that an interconnect in a battery pack may become loose and no fire should result, hence the 300 volt DC rated fuse. Another "assumption" is that switching an unclamped inductive resistor WILL result in shorted transistors after a while. This is my experience with over 40 years of building electronics. Regarding cost, a build of 1000 cell modules could easily be sold at $12.50 each and result in a small profit even if the rest of all of the minimum order quantities had to be scrapped.

One holds secrets for ONLY 2 reasons, the knowledge would benefit a competitor or it would embarrass oneself.
Regards
Jerry

 

[EVfun]

The "open source BMS" from Dimitri is still available in the archives. From that anybody could build their own working system, but I doubt they could do it for less than buying them from Clean Power Auto.

If you have another BMS system in mind share what you want about the design, and for help and input as you wish, sell us on what your system will do for us, and offer it for sale. Some people here want nothing to do with a BMS, some want cell level management, and some want cell level monitoring. Personally, I'm mostly in the last group because I can manage them myself but would like some watchdog letting me know I should do that *now*.

I did have one other thought on the cell loop circuit. Older PB acid regs generally used a normally open system so you paralleled the optical isolator outputs and if any one turned on the system went into action. The problem with that is any break in the wiring looked like an all is O.K. message from the regs. Most of the new Lithium systems have the optical isolators on when everything is O.K. and use a series connection so any break in wiring shows up as a fault and action will be taken. The downside of the NC loop system is the constant slight drain on the cells, usually between 3 and 15 milliamps depending on the system. My idea is a third way of monitoring the cell modules. I was considering a normally open bus with the cell isolators paralleled but with a termination system to monitor for breaks. So you would put a limited current of about 5 to 10 milliamps on the chain of paralleled cell isolators. At the far end of the chain you would have a terminator made of a couple of LEDs or other method to create a measured voltage on the chained connectors. The BMS monitor would be window comparator circuit. All is O.K. only if the voltage stays right around the terminator voltage. At the cell level the drain should be able to be reduced to microamps with most of the power now coming from the 12 volt system which can be easily unplugged.

 

[DavidDymaxion]

Nah, lots of reasons for secrets! Not saying these apply in this case, but there are lots of reasons to not blab everything about a design:

• You are pushing a part beyond spec, and are afraid it'll encourage others to do so
• You are afraid of emboldening newbies to try something dangerous
• You are respecting a nondisclosure agreement with someone that helped your design
• Your employer constrains you from sharing intellectual property as a general policy
• You are afraid of waking a "sleeping giant" that'll refuse to sell you parts if they see you as a competitor
• You are afraid someone will mess up your design and then wrongly blame you for the poor performance when they can't replicate it
• You might be exposing yourself to legal liability

... One holds secrets for ONLY 2 reasons, the knowledge would benefit a competitor or it would embarrass oneself. Regards, Jerry

 

[Jerry Liebler]

EVfun,
Yes the static drain on each cell with a normally closed loop is a concern! But just having a quad comparator on each cell "costs" up to 2 milliamps. Adding a voltage reference and 2 sets of voltage dividers adds another milliamp or so. So adding 2 milliamps for the opto relay and another 2 milliamps for a green LED is still "noise" for even a 40 AH pack (it would take over 100 days to use half of the packs energy). This loading is quite easily justified by the large gain in reliability and ease of implementation over any normally open architecture.
Regards
Jerry

 

[dimitri]

Another assumption I have made is that an interconnect in a battery pack may become loose and no fire should result, hence the 300 volt DC rated fuse.

Are you implying that fire can be started with my current miniBMS product installed, due to lose connection? Please explain this hypothetical scenario in more details. It should be amusing, especially coming from someone who doesn't own an EV AFAIK and only knows LiFePo4 cells from datasheets.

You are posting same theoretical issues here as you do on TS Yahoo list, which should be renamed to "blind leading the blind" list, as most of active posters on that list have not held a LiFePo4 cell in their hands, let alone have an EV.

Oh, and BTW, every miniBMS module is fused, always been, always will be.

 

[EVfun]

EVfun,
Yes the static drain on each cell with a normally closed loop is a concern! But just having a quad comparator on each cell "costs" up to 2 milliamps. Adding a voltage reference and 2 sets of voltage dividers adds another milliamp or so. So adding 2 milliamps for the opto relay and another 2 milliamps for a green LED is still "noise" for even a 40 AH pack (it would take over 100 days to use half of the packs energy). This loading is quite easily justified by the large gain in reliability and ease of implementation over any normally open architecture.
Regards
Jerry

But why use comparators on the cells? It seems like a voltage reference for the high end and a low voltage detector for the low end could turn an optical isolator on outside of mid range. The 2 together shouldn't demand one milliamp when off. Something crudely like this:

        R3         
+ ----^^^------    
    |     |   |    
    |     >  \/=   
    |  R4 >  -- O1 
    |     >   |    
    |     |---|    
    |     |  --    
    |-----)--|| V1 
    >     |  __    
 R2 >     |   |    
    >    --   |    
    |----/\   |    
    >   D1|   |    
 R1 >     |   |    
    >     |   |    
    |     |   |    
- -------------    

R1, 6k8
R2, 15k
R3, 680
R4, 10k
D1, TLV431 (1.24v ref)
V1, low voltage detector
O1, opto-isolator

Might want to throw a low power red LED on right on top of D1 so the current through O1 isn't much higher for over voltage compared to under voltage.

 

[Jerry Liebler]

Dimitri,
My EV will have 52 100 AH CALB cells in series. What will happen if a bolt loosens and 182 volts DC from a freshly charged pack is put across a "MiniBMS cell module? There is a reason fuses have voltage ratings!
Yes I wouldn't risk my car to the current MiniBMS!
Regards
Jerry

 

[dimitri]

Dimitri,
My EV will have 52 100 AH CALB cells in series. What will happen if a bolt loosens and 182 volts DC from a freshly charged pack is put across a "MiniBMS cell module? There is a reason fuses have voltage ratings!
Yes I wouldn't risk my car to the current MiniBMS!
Regards
Jerry

I can tell you what happens, since I tested it myself on my own EV and few customers tested it as well and reported same results. When high voltage is applied to single module, fuse blows immediately, which results in open cell loop and generates BMS signal. No fire, no drama. Total cost, $12 for module replacement, which is a good lesson to keep those bolts tight and follow install instructions.

Not risking your car to miniBMS is your natural right, since I am not twisting your arm to buy it. But, I will not stand here listening to some self proclaimed expert bashing my hard work. Unless you have proven a specific failure of miniBMS that ocurred in real EV under specific real conditions, you have no right to bash it.

Create something that works and offer it to people, competition is good. But stop referencing my work in negative light unless you have some real data to prove it.

 

[Jerry Liebler]

Dimitri,
I have NO intention of "bashing" the MiniBMS or your work, I only cited is a starting point and offer, to one or all, a design that is complete and has the attributes I've stated. I believe all components should be used within their ratings. When a battery manufacturer says 3.6 volts they mean 3.6 +/-1% because they used 2 significant figures and if they say 3.85 volts they mean a tolerance of +/-0.1% since they used 3 significant digits. In my view operating any component outside it's ratings, even under fault conditions is courting disaster. Your experience operating outside manufacturer's ratings and surviving is not an endorsement of such practices that I'll accept.
Regards
Jerry

 

[dimitri]

Dimitri,
When a battery manufacturer says 3.6 volts they mean 3.6 +/-1% because they used 2 significant figures and if they say 3.85 volts they mean a tolerance of +/-0.1% since they used 3 significant digits.

Clearly you have much to learn about LiFePO4 chemistry and how to interpret what manufacturers say and why they say what they say. I respect your intentions to improve the design and your faith in datasheets. I only ask not to refer to my name or my trademark unless you have solid proven data to support your references.

Good luck!

 

[Jerry Liebler]

EVFUN,
It took a while but I kind of like your idea of a "monitored" normally open loop. This concept is used in virtually all alarm systems. A big advantage is it greatly expands the number of "suitable" available optolsolators allowing a likely cost saving. I'd still use a single voltage reference and a Quad comparator to accomodate a "shunting" function to maintain top balance. so the curent saving would be on the order of 2 milliamps but every bit helps.
Regards
Jerry

 

[Jerry Liebler]

A disadvantage of the "monitored" normally open loop is it roughly doubles the number if interconnnections, is that really worth 2 milliamps?

 

[EVfun]

The extra connections are all part of the vehicle 12 volt system. There are likely thousands of such connections already spread about the vehicle. Some in important things like lighting or ABS. I think another 2 per cell isn't a significant risk, especially since the monitored loop will detect any open connection.

 

[Jerry Liebler]

With the NC architecture every modules health and operation is in fact monitored this is not the case with a monitored NO architecture. The extra connections make a monitored NO system cost more, both in materials and installation labor, than an NC system. The available systems hold per cell module drain under 10 milliamps. By throwing some money and parts at the problem the per cell drain could be reduced to 5 milliamps and still have the "fail safe" benefits of the NC architecture.
Regards
Jerry

 

[EVfun]

Actually, I think the NC system has the potential short coming here. Transistors usually fail on. It looks like Dimitri's system has plenty of margin because I haven't heard of any such failures with it.

 

[Jerry Liebler]

Only Dimitri can comment on his system, at least I won't. The design I've done has a green LED and opto relay driven by one output of a quad comparator which has a 10 milliamp rating. Resitors limit the currents through the LED and relay to a total of under 5 milliamps at HVC.

 

[EVfun]

I'm using 60 amp hour cells in the buggy in my avatar. If I walk away for 6 months I can easily kill the 12v parasitic loads -- pull the fuse to the DC>DC and disconnect the 12v battery + so 12v parasitic loads don't kill it. I've even set it up so the controllers small parasitic load is pulled with that fuse too. The cell regs would be a pain to pull though. 10 milliamps is about 44 amp hours if I walk away for six months. Since I should leave the pack at about 50% SOC (LiFe cells age from being full even if not in use) that would be a problem if for some reason I didn't add amp hours every so often. I can live with 3ma, a one year margin of safety.

I've mostly been looking at some type of battery bridge system, only with isolated output and about a 1/2 to 3/4 volt balance window. I sketched the monitor I posted but never followed up with it or drew up a head board (an good job for a dual comparator.) With a bridge it would be easy to pull the parasitic load, or even set up so its only connected when the pack is being charged or driven.

 

[Jerry Liebler]

As I said with the addition of 2 transistors and 2 resistors (a part cost of less than a dime) I can assure adequate drive to the opto relay yet have a max draw of 6milliamps with the typical slightly over 5 milliamps. The green led may be pretty dim at lvc. The 6 milliamps Would get you 416 days on a fully charged pack so if when you left on your 6 month safai they were full they would be right at 1/2 when you returned.