LiFeP04 Batteries for Solar & BMS

[Living Large]

Excellent stuff and great write up. For an EV application.

My whole mantra is to differentiate between EV and solar, and treat the batteries according to usage.{snip}

Interesting observations.

I know little about this subject, other than appreciating what you are saying. So I am hoping to see some discussion by the people who are knowledgeable. I do have a few questions about what this means practically for a PV application with LFP, but I will reserve them for the moment.

I did anticipate that now that we have identified a possible tool to implement LVD and HVD, there would be a science to determining the actual thresholds as well as a timer interval to account for sag. I didn't think or know about the implication of a slow charge rate - I only had considered charging by genny.

[PNjunction]

Excellent stuff and great write up. For an EV application.

My whole mantra is to differentiate between EV and solar, and treat the batteries according to usage.

For instance, in a solar bank where each cell may be realistically charged and discharged at 0.05 to maybe 0.1C, voltages need to be set differently.

Consider - if you charge your cells at 0.05C to maybe 0.1C (which could be the case with a regular solar setup), you are *already* charging at what could be considered the absorb current. Then once you reach about 3.45 to 3.5v, you are done. If you do a capacity test at this point, you'll find you have reached full capacity. I have tested it.

Going further - what happens when you set your charge limit for 3.7v, but tickle your cells with 0.05 to 0.1C? TWO THINGS happen, and it is slow enough to actually witness.

Once the cells reach full capacity at about 3.45v, it more or less stalls as the charge current takes a bit of time to actually drop to zero amps. So it sits there around 3.45v with no current flowing. BUT, you are still set for 3.7v. If you watch it long enough, the voltage will *suddenly* ramp up very quickly trying to reach 3.7v, yet there is still no current flowing. So what is making this voltage increase? Electrolyte heating and parasitic reactions. Not good.

In addition, because we are not in an EV application, we have time to do pack-level monitoring (assuming you start out sanely) and run with conservative values. Ie, no more than 3.5v or so at the top. And if you wisely engineer your capacity size so you never reach 90% DOD on a regular basis, your LVD can also be conservative at say 3.2v. There is enough time for the LVD to trip, and no cell will be doing a fast cliff-dive at the currents we are drawing.

This is the point that often gets missed with these batteries coming primarily from EV sources - from a practical standpoint, we WANT to do top balance (a misnomer surely as you have pointed out), and bottom balancing means that you have not sized your bank appropriately for solar.

So the warning here is that when doing solar, essentially tickling our cells, what is good for EV is not necessarily good or may be a case of overkill (depending on your risk factor, which can be mitigated by being conservative.)

But do check out that electrolyte heating phenomenon with low current and high-ish voltage. It was fascinating to watch. Put a cell on the bench and try it. Like a dolt, I actually sat there and watched it. Much like Jack does on his bench.

[Sunking]

Please see the edit and screen shot in post #31. The Orion BMS software is on the Ewert website.

Great you found it. That sub menu is where you will set up the LVD and over current trip points.

[Living Large]

Oops my bad. That was a PM I sent Bob as a list of questions for him to prepare when you reached out to him. Not sure how it got here for you to see. Well at least it lets you know I am trying to help you in the background.

Please see the edit and screen shot in post #31. The Orion BMS software is on the Ewert website.

[Sunking]

I don't know that yet. Worst case, the output would have to be filtered. It's close though, I think. I need to go do some research - the Orion Jr PC software is available on the Ewert site, as are various manuals that may answer the question.

Oops my bad. That was a PM I sent Bob as a list of questions for him to prepare when you reached out to him. Not sure how it got here for you to see. Well at least it lets you know I am trying to help you in the background.

[Living Large]

Can LVD be done on time delay? Say 2.6 volts if greater than 15 seconds? Check your email and PM

I don't know that yet. Worst case, the output would have to be filtered. It's close though, I think. I need to go do some research - the Orion Jr PC software is available on the Ewert site, as are various manuals that may answer the question.

For what it is worth, the TI datasheet for that 6 channel A/D reflected that is has a timer of up to 3.2 seconds, I think. Not saying the Orion has that chip - just an observation.

EDIT The software looks extremely user friendly, and I believe I found a timer that will do what you (I) want.
Here is part of a screen shot that applies to both the charge limits and discharge limits tabs:

[Sunking]

Dereck, I received an email back from Chris Ewert. Good news:

"Yes, absolutely. That is a core feature of the Orion Jr BMS. There is
an output called "Discharge Enable" and one called "Charge Enable".
Those outputs turn on if everything is OK, and turn off if any one cell
drops below the min. cell voltage, or goes above the max. cell voltage
respectively."

I'll email you my address, so I can ask further questions. Thanks a lot.

Can LVD be done on time delay? Say 2.6 volts if greater than 15 seconds? Check your email and PM

[Living Large]

LL I am going to put you in contact with a sales rep for Orion products. He is a avid DIY EV guy. Not a Bottom Balancer, but he knows all about it and how to make the unit work This takes me out of the middle. Send me a PM with your e-mail address please.

Dereck, I received an email back from Chris Ewert. Good news:

"Yes, absolutely. That is a core feature of the Orion Jr BMS. There is
an output called "Discharge Enable" and one called "Charge Enable".
Those outputs turn on if everything is OK, and turn off if any one cell
drops below the min. cell voltage, or goes above the max. cell voltage
respectively."

I'll email you my address, so I can ask further questions. Thanks a lot.

[Sunking]

Back to the starting gate. Your cart has individual boards, and they are passive. The Orion Jr advertises that it has "integrated passive balancing," but there are no external cell boards. For some reason I was going to guess the Orion would be "active".

Seems like this product would do what I would want for the PV system we have discussed, and your bottom balancing scenario. It is programmed via RS-232 and a PC, and I notice it draws 1.2W typical (nice). Has a battery backup. My expectation would be the PC can be disconnected and this runs autonomously.

Here is the advertised list of inputs and outputs:

-2 digital signal outputs for controlling charge and discharge limiting mechanisms
-1 digital signal output for controlling a battery charger
-1 CANBUS 2.0B interface [optional] (both standard and extended IDs supported)
-1 digital RS-232 interface for programming and diagnostics
-2 multi-purpose outputs with software assignable functions
-1 multi-purpose input with software assignable functions
-3 analog 0-5v outputs that represent the following signals: Charge Current Limit (CCL), Discharge Current Limit (DCL), State of Charge (SOC).
-2 thermistor inputs (additional monitoring possible with thermistor expansion module)

These "digital" outputs - do you think this means logic high and logic low? The voltage swing appears to be the battery bank voltage, which is a little surprising. EDIT I found the wiring manual, which says the outputs are open drain, which is why they go to the PS voltage.

Between the digital and the two multi-purpose outputs, do you think LVD and HVD for a PV system can be implemented? EDIT I sent Orion's manufacturer a question of if voltage above/below a threshold can be detected and control outputs.

Based on the description of Columb counting, would you expect that the SOC function on this is accurate, compared with the "meters" built into some equipment? Orion has its own optional meter, but with the analog output you could make your own or just use a multimeter.

Would the CCL and DCL functions be used, in your opinion? We've been focusing on voltage.

EDIT One thing I don't see is if there are timers to allow for voltage sags to avoid false tripping a LVD.

At a glance, this looks like a good candidate. All that would be left is determining how to interface it to the actual system.

LL I am going to put you in contact with a sales rep for Orion products. He is a avid DIY EV guy. Not a Bottom Balancer, but he knows all about it and how to make the unit work This takes me out of the middle. Send me a PM with your e-mail address please.

[Living Large]

You are not alone. Myself I am just going to go with the Orion Jr as I know it will do what I want. I just will not use the Top Balance feature it has built-in which eliminates balance boards. That is the difference between active and passive Top Balance. Tight now my cart has passive Balance Boards. 16 independent boards that do not communicate with anything. They are in their own little world oblivious to anything around them. They turn on at 3.65 volts, and turn off at 3.55 volts. That is all they know.

Back to the starting gate. Your cart has individual boards, and they are passive. The Orion Jr advertises that it has "integrated passive balancing," but there are no external cell boards. For some reason I was going to guess the Orion would be "active".

Seems like this product would do what I would want for the PV system we have discussed, and your bottom balancing scenario. It is programmed via RS-232 and a PC, and I notice it draws 1.2W typical (nice). Has a battery backup. My expectation would be the PC can be disconnected and this runs autonomously.

Here is the advertised list of inputs and outputs:

-2 digital signal outputs for controlling charge and discharge limiting mechanisms
-1 digital signal output for controlling a battery charger
-1 CANBUS 2.0B interface [optional] (both standard and extended IDs supported)
-1 digital RS-232 interface for programming and diagnostics
-2 multi-purpose outputs with software assignable functions
-1 multi-purpose input with software assignable functions
-3 analog 0-5v outputs that represent the following signals: Charge Current Limit (CCL), Discharge Current Limit (DCL), State of Charge (SOC).
-2 thermistor inputs (additional monitoring possible with thermistor expansion module)

These "digital" outputs - do you think this means logic high and logic low? The voltage swing appears to be the battery bank voltage, which is a little surprising. EDIT I found the wiring manual, which says the outputs are open drain, which is why they go to the PS voltage.

Between the digital and the two multi-purpose outputs, do you think LVD and HVD for a PV system can be implemented? EDIT I sent Orion's manufacturer a question of if voltage above/below a threshold can be detected and control outputs.

Based on the description of Columb counting, would you expect that the SOC function on this is accurate, compared with the "meters" built into some equipment? Orion has its own optional meter, but with the analog output you could make your own or just use a multimeter.

Would the CCL and DCL functions be used, in your opinion? We've been focusing on voltage.

EDIT One thing I don't see is if there are timers to allow for voltage sags to avoid false tripping a LVD.

At a glance, this looks like a good candidate. All that would be left is determining how to interface it to the actual system.

[Sunking]

I'm trying to run with the right shoe on the left foot and visa versa.

You are not alone. Myself I am just going to go with the Orion Jr as I know it will do what I want. I just will not use the Top Balance feature it has built-in which eliminates balance boards. That is the difference between active and passive Top Balance. Tight now my cart has passive Balance Boards. 16 independent boards that do not communicate with anything. They are in their own little world oblivious to anything around them. They turn on at 3.65 volts, and turn off at 3.55 volts. That is all they know.

[Living Large]

I think the reason they use the TI chip is for the A/D converters and its precision 5 volt supply. If you were to use the Arduino you would have to use A/D converters of some sort for the Arduino to work with

I need to do more looking at the various links you gave. Basically, I was envisioning an A/D interface, monitored by the Arduino. For some reason, as I looked at the TI datasheet, my mind lept past the custom development boards they are using (because I didn't like the thought of dealing with SMD) to thinking that one of these off the shelf BMSs would have the TI or similar part in it, and would perform the function of the custom board. And I could link to the Arduino via SPI. Maybe that is not possible - because hey, why would they be making their own A/D interface? I've never worked with a SPI, and it's very possible they are not meant to go from discrete unit to discrete unit, but rather on the same circuit board or boards mounted close together in the same box. You can see I've just scratched the surface, not familiar with pretty much any of the building blocks. I'm trying to run with the right shoe on the left foot and visa versa.

[Sunking]

I spent some time looking at the Arduino BMS link. For the BMS, he's using a TI BQ76PL536A chip, which handles 6 cells. Can be stacked using more chips up to 192. Surface mount device with 64 pins... maybe more work than I would be willing to invest at my old age. It does LVD and HVD for each cell, has a "Pack" level alert as well, 2 temperature sensors, and optionally does balancing if you add external FETs. Communicates via a SPI interface, which the Arduino has.

This is an example of the building block needed to do just what you suggest. One or more of the devices in the list at the link you provided might have 3 of these or similar chips in it and be ready to go. Wouldn't it be great if you could connect an off the shelf device with the front end to an Arduino, to minimize the amount of work? I just scratched the surface, but this looks very promising.

I think the reason they use the TI chip is for the A/D converters and its precision 5 volt supply. If you were to use the Arduino you would have to use A/D converters of some sort for the Arduino to work with

[Living Large]

Accuracy as I understand it is completely dependent on the 5 volt supply. There is a Forum called Endless-Sphere a E-Bike enthusiast group. Over there there are two threads running. One is an Arduino BMS that sound slike a commercial product development, but anyone can get into the action and buy a test unit. If you can help with development they will sale it to you at cost. Sounds like a Chi-Com manufacture ripped them off already and beat them to market. Since it is open source not squat they can do about it. The have all the code and schematics right there to be downloaded so anyone can copy. Might be a bit more work that I am willing to put into it.

I spent some time looking at the Arduino BMS link. For the BMS, he's using a TI BQ76PL536A chip, which handles 6 cells. Can be stacked using more chips up to 192. Surface mount device with 64 pins... maybe more work than I would be willing to invest at my old age. It does LVD and HVD for each cell, has a "Pack" level alert as well, 2 temperature sensors, and optionally does balancing if you add external FETs. Communicates via a SPI interface, which the Arduino has.

This is an example of the building block needed to do just what you suggest. One or more of the devices in the list at the link you provided might have 3 of these or similar chips in it and be ready to go. Wouldn't it be great if you could connect an off the shelf device with the front end to an Arduino, to minimize the amount of work? I just scratched the surface, but this looks very promising.

[Living Large]

Once the weakest (lowest capacity cell) has been located, is it possible / reliable to use it as a pilot cell to monitor the battery pack, and have it control LVD and 90% full charging ? Spending a couple hours for the initial set up, and a monthly verification that the chosen cell is still the weakest, would eliminate a rats nest of wires and fuses and a couple handfuls of questionable quality BMS sensor board, thereby increasing the overall system reliability ?

This is a very interesting suggestion. But what happens if one of the other cells kicks the bucket at some point, and you aren't monitoring it?

As for the questionable quality BMS boards, with the Arduino you wouldn't have it - you would have 17 wires. Having done digital circuit design and construction, I'd willingly go the hobbyist route. Dereck won me over with bottom balancing, so I would seek to not use shunts. I am *guessing* that some of the BMS units listed at the site above do not have individual boards, but I could be wrong. I agree with you that eliminating boards across your expensive cells would be desirable. If it is just a sensor and not a shunt I would feel better about failure modes.