DIY ajustable current source : test cell fuse wire

akiraEC

100 W
Joined
Jan 29, 2015
Messages
133
Hello,
I'd like to build my battery pack with cell fuses. I can look at tables to choose the best wire gauge but ultimately, I'd like to test it with real current. The idea would be to flow an increasing current in the fuse wire to test two things :
- it does not get too hot under normal load.
- it fuses quickly enough when reaching a high current.
Do you know of a schematic of a circuit to generate an ajustable (0-40A) current ? I guess I could use a pack of battery as input but I'd need something to limit the current to the required value.
 
When fusing individual cells that are high current, but are assembled into a small pack...the fuse melt amperage and separation time are a critical decision.

When using 10A cells in a very large pack, the acceptable performance criteria are much broader and easier to meet. An internally shorted cell is like replacing the shorted cell with a solid copper wire, and current from that entire paralleled string will flow the dead-short current through the bad cell.

If a 10A cell is in a 7P string, and in a dead short, each cell might provide over 20A per cell, so the fuse will be force-fed over 140A. The fuse criteria is then, barely warm at most during 10A peaks, and quickly pops at (perhaps) 70A. That should be easily done.

The bad cell will continue to vent, and the other six cells in the 7P string will be separated from the short.

High-amp cells in a small pack will require a much more precise fuse. An HG2 is rated at 20A (*which I consider it's temporary peak, in order to enhance a long life). So the fuse should be barely warm at 20A, and if it is in a tiny 2P pack, these cells would likely provide over 40A in a dead short, meaning 2P would force feed over 80A through the fuse of the shorted cell.

I am searching for the real numbers, and will post them when I can
 
Bare fuse wire has a composition that is similar to lead-free solder, tin/zinc. When it blows, there will be a small amount of conductive splatter. At the lower amperages, the fuse wire is very thin, and the splatter is small.

When amperages get high, the fuse-wire becomes thick, and the volume of splatter is more significant. In those situations, a thin strand of copper wire is recommended to reduce the volume of splatter.

In applications where there are high enough volts and amps for an arc to bridge the fuse wire stumps, it is recommended to encase the splatter inside a glass bulb.

In cases where the blown fuse inside the glass bulb might still allow the production of a conductive plasma, there is a type of glass bulb fuse that is filled with a particular type of sand.

I will search for these values and post them as soon as I have a reliable reference
 
Here's a circuit that would work for providing a regulated current:

Current-source-circuit.png


For this, you apply an adjustable voltage to the non-inverting input of the opamp. The opamp will try to keep the voltage on the sense resistor the same, so by adjusting the input voltage, you can set how much current the circuit will allow to flow through the load (fuse wire in your case).

However, if you really want to test up to 40A with this circuit, the mosfet would be getting extremely hot. It would be better to use some kind of adjustable DC/DC converter that can provide the current that you need without burning up a lot of heat. You can add current limiting to many kinds of power supplies using the same method as Fechter's current limiting boards: https://endless-sphere.com/forums/viewtopic.php?t=36958
 
Addy said:
Here's a circuit that would work for providing a regulated current:

For this, you apply an adjustable voltage to the non-inverting input of the opamp. The opamp will try to keep the voltage on the sense resistor the same, so by adjusting the input voltage, you can set how much current the circuit will allow to flow through the load (fuse wire in your case).

That circuit might work OK but the 1k resistor needs to be much lower, like 1mOhm so it can handle 40A. And the FET would get super hot, but if the fuse blows right away, the thing won't have time to heat up much. It needs a fuse too, in case the output gets shorted and the FET blows. You could use a 12v battery to supply the circuit.

A bench power supply with CC CV controls would be much nicer, but 40A bench supplies are stupidly expensive.
 
Hi Addy,
Thank you !! My knowledge in electronics is quite limited, so I am very grateful for you help (not only you but all of you)!

Would you have suggestions for the FET ?
I have found some 200A capable FETs in DirectFET package :
https://www.mouser.fr/ProductDetail/Infineon-IR/IRF7749L1TRPBF?qs=sGAEpiMZZMshyDBzk1%2fWi7r8jvGh7mrdbhMhKEPxsdQ%3d
Maybe another in T0220 would be easier to put on a heatsink. How about this one :
https://www.mouser.fr/ProductDetail/Texas-Instruments/CSD19506KCS?qs=sGAEpiMZZMshyDBzk1%2fWizSW%252bwlCH%2fhMnfHjD3Efj2PhrFxZjTkdpQ%3d%3d

Would it be better to put several in parallel ?

Addy said:
Here's a circuit that would work for providing a regulated current:

For this, you apply an adjustable voltage to the non-inverting input of the opamp. The opamp will try to keep the voltage on the sense resistor the same, so by adjusting the input voltage, you can set how much current the circuit will allow to flow through the load (fuse wire in your case).

However, if you really want to test up to 40A with this circuit, the mosfet would be getting extremely hot. It would be better to use some kind of adjustable DC/DC converter that can provide the current that you need without burning up a lot of heat. You can add current limiting to many kinds of power supplies using the same method as Fechter's current limiting boards: https://endless-sphere.com/forums/viewtopic.php?t=36958
 
fechter said:
That circuit might work OK but the 1k resistor needs to be much lower, like 1mOhm so it can handle 40A. And the FET would get super hot, but if the fuse blows right away, the thing won't have time to heat up much. It needs a fuse too, in case the output gets shorted and the FET blows. You could use a 12v battery to supply the circuit.

Thank you for your help !

1mOhm would result in a 2W power dissipation under 40A, right ? This is not even much for a resistor.

Something like this might do the job :
https://www.mouser.fr/ProductDetail/KOA-Speer/PSJ2NTEB1L00F?qs=sGAEpiMZZMtlleCFQhR%2fzfOI4xHs7IDUutBqgYbi4Il52bsccUJZvA%3d%3d
 
akiraEC said:
Hi Addy,
Thank you !! My knowledge in electronics is quite limited, so I am very grateful for you help (not only you but all of you)!

Would you have suggestions for the FET ?
I have found some 200A capable FETs in DirectFET package :
https://www.mouser.fr/ProductDetail/Infineon-IR/IRF7749L1TRPBF?qs=sGAEpiMZZMshyDBzk1%2fWi7r8jvGh7mrdbhMhKEPxsdQ%3d
Maybe another in T0220 would be easier to put on a heatsink. How about this one :
https://www.mouser.fr/ProductDetail/Texas-Instruments/CSD19506KCS?qs=sGAEpiMZZMshyDBzk1%2fWizSW%252bwlCH%2fhMnfHjD3Efj2PhrFxZjTkdpQ%3d%3d

Would it be better to put several in parallel ?

You're welcome. For the FET you don't need an expensive one. In this circuit the FET is used in it's linear region - it's not switched on fully, only enough to get the current flow that you set.

Try to find a FET that:
-is able to handle the maximum current you plan to use
-has a package that is easy to heatsink (TO-220, TO-247 etc)
-has a reasonable threshold voltage

To turn the FET on, the opamp's output voltage has to be greater than the FET threshold voltage, plus whatever voltage drop is across the sense resistor. Many FETs have a threshold around 4V, so if you use a 12V supply and choose a sense resistor that doesn't develop a large voltage drop, you should be totally fine.

It would be better to have several FETs in parallel on a heatsink if you plan to be testing at high currents for extended periods of time. If your fuse wire is going to blow fast at 40A, you don't need as much heatsinking and multiple FETs.

akiraEC said:
1mOhm would result in a 2W power dissipation under 40A, right ? This is not even much for a resistor.

Something like this might do the job :
https://www.mouser.fr/ProductDetail/KOA-Speer/PSJ2NTEB1L00F?qs=sGAEpiMZZMtlleCFQhR%2fzfOI4xHs7IDUutBqgYbi4Il52bsccUJZvA%3d%3d

P = I^2 * R, so for a 1mOhm resistor at 40A:
P = 40^2 * 0.001 = 1.6W

This would be totally fine for the resistor you're looking at. Another thing to think about is the voltage across the sense resistor at these currents.
At 40A you would get only 40mV. This means you need to make a reference voltage between 0-40mV for the input of the opamp. You might want to have a higher sense resistance so you have a bigger voltage range to work with. You also have to make sure your opamp can work with this voltage range.
 
Addy said:
P = I^2 * R, so for a 1mOhm resistor at 40A:
P = 40^2 * 0.001 = 1.6W

This would be totally fine for the resistor you're looking at. Another thing to think about is the voltage across the sense resistor at these currents.
At 40A you would get only 40mV. This means you need to make a reference voltage between 0-40mV for the input of the opamp. You might want to have a higher sense resistance so you have a bigger voltage range to work with. You also have to make sure your opamp can work with this voltage range.

Good point. A higher resistance would work as long as the total dissipation is reasonable. You might go as high as 5mOhm. A decent op amp will still work OK with a 1mOhm resistor (most bike controllers are in this range). A LM358 is pretty commonly used to amplify the shunt voltage.
 
I apologize if it sounded as if I felt a 40A "blow" spec is ideal, I really don't know what is ideal, but I do strongly feel this is a very timely subject to discuss. The 30Q cell has become very popular over the last year, and the higher-spec HG2 is rare due to a higher price, and also because it is most popular for vapes as a single-cell. It's original design spec for the HG2 was for cordless tools with a small and light 1P pack.

Another spec I am curious to research is the amount of time that causes a fuse-wire to blow. When I suggested that cells provide double their rated amps in a dead short, the amount is actually much higher, but I don't know the actual figures. Few are willing to fry these expensive cells just to gather data.

A pack that is always cool may be leaving a lot of performance on the table, and a little warmth can indicate that a certain cell was properly spec'd for the job. My research indicates that 140F is a hard ceiling that no cell should ever be allowed to reach, and 100F is an optimal high-mark for heat. (I am open to considering alternative specs and data). Of course, a battery running cooler is not a problem.

In a large pack, the cells on the outer shell will passively shed some heat to the ambient air. The cells at the core of the pack will run slightly hotter, so...if the handful of cells at the center of a pack die an early death, the whole pack is discarded. The cells at the physical core determine how long the pack will be usable.

Using the 30Q as a benchmark (with others to follow):

At a 15A continuous draw, how warm does the cell get after one full minute?

What continuous amp-draw causes the cell to stabilize at 100F? 110F? 120F?

At a 15A draw, what is the smallest-diameter Zinc/Tin fuse-wire that has an acceptable voltage drop (barely gets warm)?

At a 15A draw, what is the smallest-diameter strand of copper wire that barely gets warm, has low voltage drop?

If we want a fuse-wire to blow in one second (or some other length of time), what diameter blows in one second at 10A, 15A, 20A, 25A, 30A, 35A etc

Any additional questions that should be considered?
 
40A fuse is also somewhere I was aiming for.
I am using HE4 LG cells which have 20A continuous discharge rating. I think that it is more than double when shorted. A 40A fuse would be nice even for a 3p pack (better for a 4p of course).

I have ordered a LM4040 for reference voltage (with resistors for get 0 to 200mV voltage source). I have ordered IRF 1404Z MOSFETs (rated for more than 70A). Also some 20, 22 and 24 gauge copper wire. Also 4 0.02Ohm resistors to put in parallel. They are rated 4W each, I will get 0.005 mOhm and 16W . That should be OK for a short excursion in the 40A region.

I will also put a heavy car fuse in case I get a short in the power circuit.

Let's melt some copper !!!
 
I use dia 0,3mm copper wire as fuse, for my packs, built from Panasonic PF cells. My design criteria was, individual cells should be able to blow their own fuse any case. My boundary conditions were:
1. At 10A (max current of PF) the wire should be warm, but still touchable by naked fingers.
2. At room temperature, a completly discharged cell should blow it in a few secs.
I made tests with different diameters, to determine the required size.

On my electric kick scooter, I regularly pull 40A (from a 6S4P battery) on accelerations, or steep hills without any problem.

T.
 
takyka said:
I use dia 0,3mm copper wire as fuse, for my packs, built from Panasonic PF cells. My design criteria was, individual cells should be able to blow their own fuse any case. My boundary conditions were:
1. At 10A (max current of PF) the wire should be warm, but still touchable by naked fingers.
2. At room temperature, a completly discharged cell should blow it in a few secs.
I made tests with different diameters, to determine the required size.

Hi Takyka,
Thanks for your input. Is it 0.3mm diameter of 0.3mm2 section ?
0.3 diameter sounds freaking thin !! It s something like awg 28 !
 
Some example data:

Samsung 25R: rated for 100A @ 1 sec. DCIR = 22 milli-ohms
Samsung 30Q: No pulse rating on datasheet but DCIR is 20 milli-ohms and short-circuit current is listed as 130A.

You can probably infer burst/short current for other cells by comparing the DCIR to those above.
 
100A. That'a also the range of max current I got for usual 18650 used in eskateboards. I am using HE4 which are very similar to 25R.
It'll be complicated for me to generate such a strong current in order to test my fuses ... I designed the parts for 40$ (continuous). Maybe I can go higher for shorter time.

On the other side, it's a good think that these cells are capable of such a high current for short time. It means the fuses can be made a bit larger to handle normal operation and still blow rapidely in case of short.
 
Seeing how large you can make the fuse wire and still get it to blow with a mostly discharged single cell sounds like a good test method.

This combines two tests into one that is easy to DIY. We need to know how many amps the cell can do into a short circuit and we need to know how many amps it takes to blow a given size fuse wire. Fuse wires should be pretty consistent. Batteries, not so consistent.

As the cells age, their internal resistance will rise and the short circuit current will drop. Once the size is determined for a new cell, it could be reduced by some factor (2x?) to account for aging.

It would also be good to test a number of the same cells to get some statistics on variability.
 
Hi all,
I discussed a bit the proposed circuit on electronics.stackexchange.

https://electronics.stackexchange.com/questions/400514/adjustable-high-current-source-40a
https://electronics.stackexchange.com/questions/403152/parallel-mosfets-to-get-more-current-40a-adjustable-current-source
https://electronics.stackexchange.com/questions/403409/mosfets-in-parallel-for-dc-two-mosfet-altenatively-on-off-to-simulate-dc

Please have mercy and don't laugh to much at me reading my silly questions :)

People there warned me that MOSFET do not like very much to be operated in the linear region. The Current rating does not reflect what they are capable of in this region. Furthermore, when operated in DC, the get hot. And the get hot not homogeneously, which created hot spots and there is a high risk of thermal runaway.
Looking at the SOA (safe operationg area), at 12V, the max current of reasonnably priced MOS in DC is usually below 10A. So I'll have to put many of them in parallel and prepare a hugh heatsink to avoid blowing them.

Not so easy, but I have not given up ... yet. I could also simply sacrifice a cell for science and short it with various wires (22, 24, 28) and see how it blows (I would also add a car fuse in serie of the short to be sure the short would not make the cell explode).
 
That's probably a more practical approach. You could use a long piece of heavy wire in series with the cell to increase the resistance and lower the current. It would be nice to have an accurate measurement of the current during the test but things will happen fast. Something with a max reading capture would work.
 
Back
Top