I want to make my own BMS

[John Frum]

@Watch Me Build has a fet based bms.
It may have failed closed.

JBD bms cell overcharge issue.

Ok so one of my cells in a diy 280ah LifePo4 will hit 3.6v while another is at 3.55 and 2 others are in the upper 3.4s. I realize I need to top balance these cells again, but am living off grid and this is my only power source, so that's not an option at the moment. I am hooked into an...

diysolarforum.com

 

[Factory400]

We're talking about overvoltage disconnects.

Among numerous other failure modes.

 

[Factory400]

@Watch Me Build has a fet based bms.
It may have failed closed.

JBD bms cell overcharge issue.

Ok so one of my cells in a diy 280ah LifePo4 will hit 3.6v while another is at 3.55 and 2 others are in the upper 3.4s. I realize I need to top balance these cells again, but am living off grid and this is my only power source, so that's not an option at the moment. I am hooked into an...

diysolarforum.com

I did not read this, but will take a moment to point out that the circuits driving the FET can kill it really fast.

An easy example is a system where the gate control allows it to oscillate - rapidly turning it on/off and effectively putting the FET's int the linear region. With even a small amount of current flow - those FETs will be damaged.

The FET gate drivers have to be bulletproof.

 

[BiduleOhm]

@Watch Me Build has a fet based bms.
It may have failed closed.

JBD bms cell overcharge issue.

Ok so one of my cells in a diy 280ah LifePo4 will hit 3.6v while another is at 3.55 and 2 others are in the upper 3.4s. I realize I need to top balance these cells again, but am living off grid and this is my only power source, so that's not an option at the moment. I am hooked into an...

diysolarforum.com

That's a cheap chinese BMS and exactly why some of us design our own

Also:

I am hooked into an MPPSOLAR pip1012lv-mk. And normally it would charge until the bms reaches cell overvoltage (set at 3.6). And then it stops the charge and when it drifts back to 3.55 it restarts. It will do this a few times, and then the MPP unit decides it has been charged and stops trying to bulk charge it.

He was using the BMS as the main and only way to stop the charge... Not the proper way to do it, and also why you want redundancy

 

[Cal]

What about the other protections? (low voltage, over current, over temp, low temp, ...)

Also, how do you disconnect the loads which aren't the inverter? because they won't have a control input to turn them off...

I currently have a DIY SSR low voltage protection in one of my two RV's. Just a simple high side fet switch. I like maintaining a common ground to all loads (and power sources) with no ground disconnects. I'm not concerned of the 2500W inverter 160A load as that's only ON when using the microwave. I haven't gotten around to building a second SSR. DIY SSR

Over current protection is a fuse at the battery and additional small fuses at individual loads.

I'm not that concerned with over temperature. Probably more concerned for my own health than the batteries during over temp.
Low temp is handled just like a over voltage condition.

I'm with Bob, if necessary I would disconnect the inverter at the inverter switch. But not necessary as it's only active when microwaving. The rest of the loads are all under 8A (the forced air furnace is 8A). A simple DIY SSR can handle those loads.

 

[BiduleOhm]

The switch of a FETs based BMS is basically a SSR...

 

[Watch Me Build]

I'm lost... What do you mean I have it as a main?

I have a few different breakers. And shutoffs to be able to disconnect the battery as well as the panels from the mppsolar and the 12v fuse box.

What would you suggest otherwise?

 

[Cal]

The switch of a FETs based BMS is basically a SSR...

That’s correct. However, just the load requires a common SSR, not charging sources.

 

[Bob B]

What about the other protections? (low voltage, over current, over temp, low temp, ...)

Also, how do you disconnect the loads which aren't the inverter? because they won't have a control input to turn them off...

In my case, for LVD, the inverter will be turned off if it hasn't already turned itself off ..... I will have a low current SSR to disconnect my DC loads. None of them pull much current. I will set this to happen at a lower voltage than what the inverter shuts itself down.

I will have 2 low temp cutoff points ... One will shut off charging ..... the loads will all shut off at a lower temp. Over temp will shut off everything and over current will depend on whether it is charge or load.

 

[Bob B]

I did not read this, but will take a moment to point out that the circuits driving the FET can kill it really fast.

An easy example is a system where the gate control allows it to oscillate - rapidly turning it on/off and effectively putting the FET's int the linear region. With even a small amount of current flow - those FETs will be damaged.

The FET gate drivers have to be bulletproof.

All the more reason to avoid using them for switching high power circuits.

If I were going to automate a high power circuit, I would use one of the high powered relays that has very low hold current.

 

[Pidjey]

Mosfets fail closed when voltage in Vgs is exceded or a high frequency current spike in Vds is induced in Vgs. They do not fail spontaniously.

An easy example is a system where the gate control allows it to oscillate - rapidly turning it on/off and effectively putting the FET's int the linear region. With even a small amount of current flow - those FETs will be damaged.

The FET gate drivers have to be bulletproof.

Another reason of failing is due to oscillations in the gate as you said. A common source of this are MOSFETs in paralel due to the parasitic inductange in the gate, parasitic capacitance of drain to gate and parasitic capacitance of gate to source. If not properly design you can create an oscillator in the gate of the mosfets. This can be prevented with different techniques but one of them is adding ferrite beads (or resistors). If you want I can post here some documentaiton.

I am using Fets with maximum Vgs of 20V and I am swithing them at 11V. Also mosfets are in paralel with capacitors and to filter high frequency noise on Vds. I am not very worried of mosfets faling on. Mosfets are used to switch thousands of Amps. I am not worried of 100A here. Inverters and chargers are also done with mosfets. It would be funny to see a switched mode power supply using relays. I am more concerned on enough mosfets and proper thermal design. (I also cannot guarantee that I can fail here of course)

Sometimes I think that problems with cheap chinese equipment comes from hiring cheap unexperienced designers or designers copying other designs and adapting them without the propper knoledge. I hope I can get a propper desing.

@Watch Me Build has a fet based bms.
It may have failed closed.

JBD bms cell overcharge issue.

Ok so one of my cells in a diy 280ah LifePo4 will hit 3.6v while another is at 3.55 and 2 others are in the upper 3.4s. I realize I need to top balance these cells again, but am living off grid and this is my only power source, so that's not an option at the moment. I am hooked into an...

diysolarforum.com

Did @Watch Me Build check physically if the mosfets really failed? it can be logic or something else. I am just curious.

 

[John Frum]

Did @Watch Me Build check physically if the mosfets really failed? it can be logic or something else. I am just curious.

How does one physically test the mosfets?

 

[Factory400]

How does one physically test the mosfets?

Scorch marks are the most obvious.

Basic testing to see if the MOSFET is totally blown can be mainly accomplished with a DMM - it can be tricky and deceiving though. In circuit testing can create challenges, taking the device off the PCB is best but impracticle in many situations. Depending on the other circuits connected - the various meaurements require some intuition to make any sense. That is especially true if it is only partially damaged. I don't think you can fully analyze the MOSFET in circuit, but a totally dead one should be reasonably obvious with some measurements.

The gate can hold a charge for a while and give you the impression that the FET is failed in an ON state. The DMM can be used to partially charge/discharge the gate to see if the DS and SD resistance changes. With the gate discharged - you should see high resistance in one direction and some lower resistance in the other direction (indicating the body diode is conducting and blocking). Some gate drive schemes will have a drain path for the gate to prevent it from ever floating, that will prevent you from using the DMM to manipulate the state of the device - you would need another voltage source for that.

The GS and GD resistance should never be very low - on an out of circuit FET the GS and GS impedance is very high. If the gate was blasted, it is possible you can see a low resistance here which is not good. The only thing you should see is a path created by the gate driver circuitry. I low resistance seen here is not necessarily MOSFET damage either - it could be a failed driver.

This is only the most basic of guidance.......but gives you something to chew on.

 

[Bob B]

An option to consider ..... Using a contactor instead of FETs ..

 

[Factory400]

An option to consider ..... Using a contactor instead of FETs ..

If you don't know how to use/design with FET's - contactors are the next best option.

I cannot imagine a scenario where I would choose that contactor over my own MOSFET design. It may, however, be better than some of the super dubious BMS's on the market.

 

[toms]

Love the effort you are all putting in, but i’m not sure what problem you are trying to solve.

In systems i’ve installed or helped install using the REC BMS, i’ve seen well over 100 years of run-time without even having a cell go out of balance.

I use a stand-alone SSR based high voltage disconnect as a 3rd defence between the SCC and the battery, and the LVD on the inverter as the primary load disconnect with the BMS as backup.

There is absolutely no way you could design and build a single BMS for what the REC cost.

I like most of the ideas here, but have yet to see a feature i need that i don’t already have.

What am i missing?

 

[Factory400]

Love the effort you are all putting in, but i’m not sure what problem you are trying to solve.

What am i missing?

I asked the same thing early on....."For fun" was the answer.

Fair enough to me even though personally it would not be fun unless I was getting something unique out of it. This project has nothing unique that I can tell. Still, some interesting conversations.

In systems i’ve installed or helped install using the REC BMS, i’ve seen well over 100 years of run-time without even having a cell go out of balance.

I am going to guess that those systems have well sourced cells that are nicely matched. I would like to see a BMS that can cope with some of the random / low-cost cells that are not well matched. A BMS that can keep them matched at the top and bottom of SOC would be interesting. @Pidjey has indicated he may have some unique active balancing idea once the fundamental monitoring is complete.

 

[Bob B]

Love the effort you are all putting in, but i’m not sure what problem you are trying to solve.

In systems i’ve installed or helped install using the REC BMS, i’ve seen well over 100 years of run-time without even having a cell go out of balance.

I use a stand-alone SSR based high voltage disconnect as a 3rd defence between the SCC and the battery, and the LVD on the inverter as the primary load disconnect with the BMS as backup.

There is absolutely no way you could design and build a single BMS for what the REC cost.

I like most of the ideas here, but have yet to see a feature i need that i don’t already have.

What am i missing?

I just looked at the REC and like what they are doing .... Does it have an optional display or any way to connect wirelessly? Looked like it only had low temp disconnect for charge .... seems like their should be one for load at a lower temp.

 

[toms]

It has an optional remote display, that’s all i routinely use. It has full monitoring and settings adjustment via laptop (wireless if required).

There is no temperature limit for discharge. Not an issue with a house bank that doesn’t see less than -7’C and greater than 1C discharge.

 

[Pidjey]

Love the effort you are all putting in, but i’m not sure what problem you are trying to solve.

In systems i’ve installed or helped install using the REC BMS, i’ve seen well over 100 years of run-time without even having a cell go out of balance.

I use a stand-alone SSR based high voltage disconnect as a 3rd defence between the SCC and the battery, and the LVD on the inverter as the primary load disconnect with the BMS as backup.

There is absolutely no way you could design and build a single BMS for what the REC cost.

I like most of the ideas here, but have yet to see a feature i need that i don’t already have.

What am i missing?

This is like making a handmade sweatshirt, going out for jogging or gardening. You do not really need to be competitive in your freetime activities to have fun. Even if I get something worse than what it is already in the market I will be happy as I have used my freetime doing something I like. Others watch TV or pilot RC airplanes. I am not making a business with this. As an example when I was a kid I made my own chess clock when I already had one. Electronics it is not only my way of making money, it is also my hobby.

Features are something related with each person. Some people only want high voltage protection and other people want 10 type of protections, multiple temperature probes, MQTT output, home assistant integration, diagnosis capabilities, algortihm with power consumption recomendations taking into account the current SoC and weather forecast, integration with the rest of the equipments using their proprietary protocol.....

I am going to guess that those systems have well sourced cells that are nicely matched. I would like to see a BMS that can cope with some of the random / low-cost cells that are not well matched. A BMS that can keep them matched at the top and bottom of SOC would be interesting. @Pidjey has indicated he may have some unique active balancing idea once the fundamental monitoring is complete.

I have already ordered 8x20ah second hand cell pack to start with validating the concept (also for the protections).