Blown fuse detection parallel battery bank

[Mrusten]

I can not wrap my head around circuit designing, it's highly annoying!
Could please someone help me to design the circuit I'm looking for?
And please I'm looking for a circuit not other solutions like microcontroller LED across fuse, etc.

So my issue:
I have 6 leaf modules in parallel 8.4v when fully charged. That means that each cell is 4.2 V and I want to fuse each cell, the center pin is connected in parallel across all 6 modules that means I need a circuit for the negative terminal sensing and a circuit for the positive terminal sensing. It's a part of a large bank so the currents is max 25 A for each module (each fuse is 25 A).

I already found part of what I'm looking for here from this post.
but i think a voltage difference sensing circuit could work as well.
the circuit should also latch the outputs for the led and transistors just to make sure there is no intermitted faults passing by, and a reset switch to reset the latching

I have 384 fuses and 32 banks of 6 modules so that's why I'm looking for a circuit I can use for each of the parallel banks and wire it up to the Siemens PLC, no addressing required as the LED will show what fuse is blown.

 

[KlausST]

Hi,

LEDs need at least 2V to operate .... I can't see that this circuit can satisfy this.
LEDs need a current limit to about 20mA .... I can't see how this can be satisfied.
If some state gets latched ... how is it released?
I don't know how the PLC is involved at all.

There is no clear question.
And since you don't want some recommendations ... I 'm not sure what is allowed to recommend .. or to be asked.

Klaus

 

[BradtheRad]

By putting a single led across a good fuse, it stays dark. However a blown fuse automatically subjects the led to system voltage.

 

[KlausST]

Hi,

But the problem here is: you have a bus .... and you have batteries.
So if one fuse blows ... the voltage across the battery does not drop the same way as with a resistive load. The battery voltage may drop some 100mV ...
And the same way ... the bus voltage does not rise. So the voltage across the blown fuse may be some 100mV.

I don't know the idea behind the circuit. But maybe it is some energy backup or storage system.

I don't understand the "center connection" at all. The batteries are never fully combined and never fully separated.
It's very likely that when one fuse in one row (top or bottom) trips ... the other fuse in the same row will follow to trip, because the total current is then taken by the remaining (fewer) paths ... which means the current in each remaining path will rise -> overcurrent --> fuse trip. Avalanche like. Total system fail.

Let's say only one fuse trips. What happens next? I guess the "faulty" battery will be replaced by a new one. If so, it's likely that it is not charged to the same level as all the other ones. And - because of fresh - the internal resistance may be lower than at the old ones. Which means the current (either charging or discharging) may be much higher than with the old ones. --> overcurrent --> fuse trips at the most healthy battery --> total system fail.

In my opinion the idea should be to keep the sytem alive as long as possible.
Usually weak batteries (connected in parallel) draw the lowest current (unless a battery internal short circuit happens. But how likely is this?)
So the fuse at the weak battery will not drop.

There is a good reason for "true health check circuits" on paralleled batteries.

Klaus

 

[Mrusten]

Hi,

LEDs need at least 2V to operate .... I can't see that this circuit can satisfy this.
LEDs need a current limit to about 20mA .... I can't see how this can be satisfied.
If some state gets latched ... how is it released?
I don't know how the PLC is involved at all.

There is no clear question.
And since you don't want some recommendations ... I 'm not sure what is allowed to recommend .. or to be asked.

Klaus

hi thanks for answer
im missing the circuit, thats what im asking for.
transistors, led's, latch and reset switch have to be in the circuit.
im using a plc to monitor the whole system and it wil tell me if a fuse is blown somwhere and i wil physically go and look for the led witch is on.

if you read my post i put information about the system and what is needed, all information is needed to be able to design the circuit or to modify the circuit i found. the question in my post is easy: circuit? help?

im all in for advises but im already setteled for this way of fusing the system so there have to be a citcuit involved

--- Updated Oct 11, 2023 ---

By putting a single led across a good fuse, it stays dark. However a blown fuse automatically subjects the led to system voltage.

View attachment 185391

thanks for answer but it wil not detect anything before segnifisent voltagedifferense and there is no feedback to plc and latching of the led

--- Updated Oct 11, 2023 ---

Hi,

But the problem here is: you have a bus .... and you have batteries.
So if one fuse blows ... the voltage across the battery does not drop the same way as with a resistive load. The battery voltage may drop some 100mV ...
And the same way ... the bus voltage does not rise. So the voltage across the blown fuse may be some 100mV.

Yes the voltage across a blown fuse wil be low, thats why the circuit i found is using current to determend the state of the fuse.

I don't know the idea behind the circuit. But maybe it is some energy backup or storage system.

idea is the same as mine, to monitor fuses in a paralell bank under any situation

I don't understand the "center connection" at all. The batteries are never fully combined and never fully separated.

senter is connected to make sure all cells is balanced, a bms is connected to the center pin as well

It's very likely that when one fuse in one row (top or bottom) trips ... the other fuse in the same row will follow to trip, because the total current is then taken by the remaining (fewer) paths ... which means the current in each remaining path will rise -> overcurrent --> fuse trip. Avalanche like. Total system fail.

in a small system yes but in my case its a large bank with small loads, purpose of the bank is to provide electricity several days, when drawing max inverter capacity the current is 22A from each cell, typical use of the inverter is 10 - 20% so the other fuses could trip but its not likely in normal operation of the system

Let's say only one fuse trips. What happens next? I guess the "faulty" battery will be replaced by a new one. If so, it's likely that it is not charged to the same level as all the other ones. And - because of fresh - the internal resistance may be lower than at the old ones. Which means the current (either charging or discharging) may be much higher than with the old ones. --> overcurrent --> fuse trips at the most healthy battery --> total system fail.

all modules is from used cars, sertanly not new so its again a posibility for that to happen but not likely, the leaf modules can provide 500A and im using 25 at apsolutely max eaven if a cell is new and other old at normal low power usage i do not think its enough of a difference to trip the fuse

In my opinion the idea should be to keep the sytem alive as long as possible.
Usually weak batteries (connected in parallel) draw the lowest current (unless a battery internal short circuit happens. But how likely is this?) So the fuse at the weak battery will not drop.

yes the system is overdimmentioned just to make it last long and keep it allive for longer, if a fuse trip i could remove one fuse from the other paralell banks as well to rebalance the system with less batteries at work.
my main consern is the fact that old leaf modules could chort circuit internally and when 5 other cells is connected in paralell that means 2500A discharge in to the shorted cell, so thats my reason of fuses

There is a good reason for "true health check circuits" on paralleled batteries.

thats a good point

 

[BradtheRad]

inverter

The frequent advice here is it's more practical to string batteries in series rather than parallel.

Since you mention an inverter it implies you wish to create high voltage. Consider adding the small battery voltages (in series) so that you have high voltage DC. Then build a switching circuit to chop it into bipolar AC. Experimenters at this board use such an arrangement about as often as a power inverter containing a step-up transformer.

The battery banks can be assembled positive and negative, creating 0V in between. Then a half-bridge (2 transistors) is sufficient rather than a full H-bridge.

It's easier to work with reduced current at high voltage, than vice-versa.

 

[KlausST]

Hi

The frequent advice here is it's more practical to string batteries in series rather than parallel.

I agree.

Series connection.
* benefit: low current
* problem: the weakest battery gets stressed most, because of permanent under_discharge and over_charge. And to worsen this: the weakest battery determins the system usable capacity and livetime. It becomes unusable very early.

Parallel connection:
* benefit: it uses all the battery capacity sumed up. No special stress of one particular battery.
* problem: high current

Klaus

 

[Mrusten]

the system is already 240vdc so its high enough, i could have 6 series strings but i went for 6 in paralell module banks to save on bms cost. all this have already been up to consideration thats why im asking about a circuit to sens the fuse state

 

[BradtheRad]

An RS flip-flop can respond to a momentary voltage change, and latch. It's a memory cell. You can make it from a logic gate with hysteresis feedback if you adjust resistor values properly. Connect all inputs of a gate having more than one input. If you use invert-gates, assemble two invert-gates to create one memory cell.

 

[Mrusten]

An RS flip-flop can respond to a momentary voltage change, and latch. It's a memory cell. You can make it from a logic gate with hysteresis feedback if you adjust resistor values properly. Connect all inputs of a gate having more than one input. If you use invert-gates, assemble two invert-gates to create one memory cell.

View attachment 185423

thank you, interesting, that could work fine, is it some how possible to set treshold value / adjust sensitivity?

 

[BradtheRad]

possible to set treshold value / adjust sensitivity?

A potentiometer (or resistor divider) might do the job. Or zener diode and potentiometer in series. Substitute a string of diodes/led's for the zener diode.

Experiment to find ohm values that create volt levels to allow the hysteresis to perform latching.

 

[D.A.(Tony)Stewart]

Your biggest problem is to define all the assumptions for this design to work.
This has not been done yet. Once this is done, the design is easy.

Current exceeding fuse rating ( excess voltage) may not be a reliable indicator since fuses can take a long time when operating near the threshold. What you want to know is, I think, is the battery drawing current from an internal short when it should be supplying current. Another possible feature is detecting a failed cell that is degrading the pack from low current due to high ESR and possible low C. The most critical test for repacking cells in a used pack is matching the ESR when all are charged to the same voltage. Having a cell ESR, Capacity tester is a key tool to have which pulses current and measures dV/dt for charge/discharge.

Fuses have thermal runaway above a certain temperature at a constant current. They are rated in I^2*t and Rs for room temp or simply the holding current. This translates to a voltage drop near 100 mV but is bipolar with charging and discharging. In order to detect both conditions a precision rectifier is needed. Precision comparators are required to detect above this threshold in order to reach a latched threshold. After the fuse is blown its voltage drop depends on how much the no-load battery voltage returns compared to the parallel string voltage. One can model a battery with a Cap, 10kF for 16850 and some ESR then another RC pair for the long-term soak charge values.

In parallel cells, the battery with the lowest ESR draws or supplies the most current unless there is an internal partial short which is undetectable by voltage but perhaps an array of thermal sensors or measuring each fuse voltage drop for relative changes.

Dendrites can form when the cell is with excess current and may be bigger than the fuse, which at least blows it from the pack. Dendrites can also form at very low lower SOC but tend to be smaller than the fuse rating so heating occurs and can age a cell more rapidly when V drops well below the others in series and takes out the other paired cell too. Pulse charging can mitigate this.

Although current sensors tend to be unipolar only unless there is a differential output.

I think a smart monitor with a uC would be best for BMS and fuse detection.

Recommendation.
One reliable way to detect an open circuit is to drive a constant current and measure voltage like any RLC meter. It could be DC or AC or a multiplexed pulse to scan all fuses.

These fuses being on the outside of every series rail may cause a design requirement to create an extra supply voltage outside the V+,V- rails to power for those designs that are simply voltage sensors with latches.

If anyone wants to try a design on a simple battery pack , I created a model here. I had previously chosen 14A fuses which will eventually blow from a Reset charge voltage and CC setting just more than 15A.

https://tinyurl.com/yvjfnq36 Falstad Simulation

Matching the ESR's and battery voltages is critical as these are simple indicators of battery capacity. In parallel packs the strongest cell gets the biggest stress, while in series strings the weakest cells get the most stress. This means a low C battery changes voltage the fastest I=CdV/dt and high ESR are also weak batteries or low SoC.

--- Updated Oct 15, 2023 ---

The problem with driving the fuse to sense it is that its battery is in parallel with the fuse nodes in the loop so that these nodes are never high impedance.

 

[Mrusten]

Your biggest problem is to define all the assumptions for this design to work.
This has not been done yet. Once this is done, the design is easy.

Current exceeding fuse rating ( excess voltage) may not be a reliable indicator since fuses can take a long time when operating near the threshold. What you want to know is, I think, is the battery drawing current from an internal short when it should be supplying current. Another possible feature is detecting a failed cell that is degrading the pack from low current due to high ESR and possible low C. The most critical test for repacking cells in a used pack is matching the ESR when all are charged to the same voltage. Having a cell ESR, Capacity tester is a key tool to have which pulses current and measures dV/dt for charge/discharge.

Fuses have thermal runaway above a certain temperature at a constant current. They are rated in I^2*t and Rs for room temp or simply the holding current. This translates to a voltage drop near 100 mV but is bipolar with charging and discharging. In order to detect both conditions a precision rectifier is needed. Precision comparators are required to detect above this threshold in order to reach a latched threshold. After the fuse is blown its voltage drop depends on how much the no-load battery voltage returns compared to the parallel string voltage. One can model a battery with a Cap, 10kF for 16850 and some ESR then another RC pair for the long-term soak charge values.

In parallel cells, the battery with the lowest ESR draws or supplies the most current unless there is an internal partial short which is undetectable by voltage but perhaps an array of thermal sensors or measuring each fuse voltage drop for relative changes.

Dendrites can form when the cell is with excess current and may be bigger than the fuse, which at least blows it from the pack. Dendrites can also form at very low lower SOC but tend to be smaller than the fuse rating so heating occurs and can age a cell more rapidly when V drops well below the others in series and takes out the other paired cell too. Pulse charging can mitigate this.

Although current sensors tend to be unipolar only unless there is a differential output.

I think a smart monitor with a uC would be best for BMS and fuse detection.

Recommendation.
One reliable way to detect an open circuit is to drive a constant current and measure voltage like any RLC meter. It could be DC or AC or a multiplexed pulse to scan all fuses.

These fuses being on the outside of every series rail may cause a design requirement to create an extra supply voltage outside the V+,V- rails to power for those designs that are simply voltage sensors with latches.

If anyone wants to try a design on a simple battery pack , I created a model here. I had previously chosen 14A fuses which will eventually blow from a Reset charge voltage and CC setting just more than 15A.

https://tinyurl.com/yvjfnq36 Falstad Simulation

View attachment 185476

Matching the ESR's and battery voltages is critical as these are simple indicators of battery capacity. In parallel packs the strongest cell gets the biggest stress, while in series strings the weakest cells get the most stress. This means a low C battery changes voltage the fastest I=CdV/dt and high ESR are also weak batteries or low SoC.

--- Updated Oct 15, 2023 ---

The problem with driving the fuse to sense it is that its battery is in parallel with the fuse nodes in the loop so that these nodes are never high impedance.

tanks for your reply. isnt this overcomplicated? in my case the fuse is mainly protecting internat short circuit it wil newer blow from over carge/discharge current as there are more fuses down the line

im looking for a circuit that can tell the state of the fuse, it seem like there is 2 ways
1. a circuit is drawing and supplying a small currentdraw across the fuse (in paralell with the fuse) and sense if that currentflow is breached
2. a circuit based on RS-flipflop that wil compare the voltage on each side of the fuse with reference to the battery module center pin so 4,2v at max soc, when any voltage difference is detected the circuit latch

 

[KlausST]

Hi,

when any voltage difference is detected the circuit latch

And here I exactly see the problem.

On the one hand: a fuse will drop voltage. I mean a standard fuse works as a resistor, with current and voltage drop. Multiply both to get the dissipated heat in the fuse wire. It needs to get hot enough to melt for the fuse to trip.
Without voltage drop (during normal operation) --> no dissipated power --> no heat --> not fuse trip

On the other hand: the system - as already mentioned - does not guarantee a meaningful voltage drop to detect a broken fuse. At leat you gave no numbers.

So we need to guess
* the used fuse with all it´s specifications
* the allowed currents on each path
* the "allowed drop" of a non tripped fuse
* the guaranteed drop on a broken fuse.
* and so on
We don´t know the specifications and functions of your system. If you don´t provide them, how can we know?

You even did never answer how you decide to "reset" the latch.

*******
So for me a useful concept could be:
* monitor the voltage across the fuse
* and monitor the current through the fuse
* and get three states:
1) fuse good
2) fuse state not determinable (on too low voltage drop)
3) broken fuse

And you - yes you - need to give the values (numbers and units) of voltage and current thresholds for
* good state
* broken state
And you need to give informations abotu the expected function.

Diagrams, tables, sketches, flow charts ... may be more helpful than a lot of vague text.

Klaus

 

[Mrusten]

Hi,

And here I exactly see the problem.

On the one hand: a fuse will drop voltage. I mean a standard fuse works as a resistor, with current and voltage drop. Multiply both to get the dissipated heat in the fuse wire. It needs to get hot enough to melt for the fuse to trip.
Without voltage drop (during normal operation) --> no dissipated power --> no heat --> not fuse trip

I plan to use regular car 12v fuses Link

On the other hand: the system - as already mentioned - does not guarantee a meaningful voltage drop to detect a broken fuse. At leat you gave no numbers.

So we need to guess
* the used fuse with all it´s specifications
* the allowed currents on each path
* the "allowed drop" of a non tripped fuse
* the guaranteed drop on a broken fuse.
* and so on
We don´t know the specifications and functions of your system. If you don´t provide them, how can we know?

im more of a "make it adjustable" type of guy but i could connect a scope across the fuse and have a look when i pull 25A and disconnect the fuse,
i wil return to this when i get time to do it

You even did never answer how you decide to "reset" the latch.

a reset push button on the circuitboard itself would be great

*******
So for me a useful concept could be:
* monitor the voltage across the fuse
* and monitor the current through the fuse
* and get three states:
1) fuse good
2) fuse state not determinable (on too low voltage drop)
3) broken fuse

And you - yes you - need to give the values (numbers and units) of voltage and current thresholds for
* good state
* broken state
And you need to give informations abotu the expected function.

yes i wil return to it when i test with scope

Diagrams, tables, sketches, flow charts ... may be more helpful than a lot of vague text.

Klaus

 

[D.A.(Tony)Stewart]

The simplest solution would be to use a fuse that can also act as a visual indicator of being blown by discoloration.

One could add that smaller indicator-type fuse in parallel with the actual fuse. ( I assume both will blow at the same time if much (>10x) smaller rating.)

Defining the HMI requirements affects the battery wiring, so define the minimum acceptable indicator location and wiring length.