0
\$\begingroup\$

I'm designing a LiPo battery and I'm struggling with the protection, or more specifically, the fuse selection. Can I please have some help? This is a really convoluted problem with a lot of factors, and I’m struggling to write this post in a clear way so please bear with me.

Context: The BMS can protect the series cells and pack, but given that I have parallel cells, I need fuses to protect the individual cells. If 1 cell is damaged, it’s parallel buddy could suffer a short circuit, which is just 1 of many reasons why parallel cells would need a fuse each.

The BMS is also only good for continuous discharge or pulses with a duration greater than 10 ms. However, it’s likely that we will have shorter pulses at roughly 1.2 ms. These pulses are most likely under the cells limitations, but protection would still be needed.

Problem: I’ve looked at a lot of fuses and not found any that are suitable. If I pick a fuse that will blow somewhere between the products expected continuous discharge and the cells limitation for continuous discharge, that same fuse will not protect the cell from its rated pulsed/burst discharge because the fuse will blow at a current far greater than the cell can handle. This is true for all the fuses I've looked at. Does the right fuse exist? What can I do?

Key points:

  • The battery (3S2P) has parallel cells (and series but let’s not get into that), which means I need balancing, cell matching, and individual cell fuses, as well as the BMS that I would have needed anyway. Everything except fuse selection is done.
  • BMS can protect the series cells, but not the individual parallel cells. It also can't do anything to protect against pulse loads with a duration less than 10 ms. It has short circuit detection which is measured in µs but that won’t help with pulsed loads or with parallel cells having a short circuit.
  • A PCM on each cell is not an option because of self discharge. We're relying on cells with a very good self discharge rating, and losing that would change our whole product architecture.
  • The cell I’m using is custom, and will not have a complete datasheet written until I proceed with manufacture.
    • Cell max continuous discharge: 20 C (90 Ah)
    • Cell max pulsed (3 ms) discharge: 30 C (135 Ah)
  • Pack limitations are cell limits x 2 and reduced by 20 % to account for imperfect matching and spot welding. 20 % was made up and I’m happy to be told that I should reduce or increase it.
    • Battery max continuous discharge: 16 C (144 Ah)
    • Battery max pulsed (3 ms) discharge: 24 C (216 Ah)
  • Product (still using 20 % for imperfect matching).
    • Product continuous discharge: BMS will step in at 71A for the pack, so the cells could be exposed to 42.6 A each.
    • Product pulsed (1.2 ms) discharge: The BMS would have been set to 155 A for the pack, which means the cells could be exposed to 93A each. But the BMS can’t do much about a 1.2 ms duration.
  • I can fit a fuse that has a footprint of 9.315 x 5.24 mm, but those dimensions are for the copper, not the fuse body.

Example:

  • 0685P9500-01 (50A) will protect the cell during continuous discharge since it’s rated for 50 A . However, for a 3 ms pulse it won't blow until it sees ≈ 500 A
\$\endgroup\$
3
  • \$\begingroup\$ "20C (90Ah)" A C rate would translate to a current, not a capacity. Do you mean 90 A? \$\endgroup\$
    – winny
    Sep 1, 2022 at 9:20
  • \$\begingroup\$ Thanks Winny. Yes, I meant 90A. Sorry. \$\endgroup\$
    – Dan
    Sep 5, 2022 at 7:08
  • \$\begingroup\$ Please edit accordingly. \$\endgroup\$
    – winny
    Sep 5, 2022 at 8:02

1 Answer 1

1
\$\begingroup\$

Fuses blow after an amount of time at a current value.

The specs for them are very vague. Often a 5A fuse will have a description which could be paraphrased to “this will blow after something like 2 minutes at around 5A, about that, more or less”. Note that this does not say “this will carry 4.9A for 5 hours”, nor “this will blow once current is over 5A”. There is a current and a time factor, and then some HUGE tolerance bands. Famously, when testing official fuses from reputable suppliers against some very dodgy back street supplied ones, there was equal spread from all suppliers.

The value they blow at depends on lots of factors, including, but not limited to: ambient temperature, recent current flows, age of the fuse etc. They are not accurate devices. The usual rule of thumb for physical fuses is you want to use one rated for 1.5 to 2 times the maximum normal current to reduce false blows. If you take things apart, you’ll often find a 5A fuse protecting something which should never draw more than 2.5A. This is so that the fuse never blows under “extreme normal usage”, which could be 3.5A.

You talk about “normal” current of about 43A per cell, so that’s a starting point. You then have pulsed 93A, but that is only for 1.2ms, suggesting that a slow blow fuse should be able to ignore that kind of spike. Using the rule I mentioned above, that would mean you’d want a slow blow 90A fuse. But then, that 1.2ms spike might mean you want something slightly higher, say around 120A to make sure you don’t’ get too many false blows. If you check datasheets you can review the charts showing how a fuse blows of current against time. There is no easy way to solve this, just check every datasheet of something close to what you want, as slow blow fuses can vary a lot. But I think you’ll struggle to find anything that’ll blow as quickly at as low currents as you seem to be hinting at in your question.

As this fuse is going to be built into the cell, they are a last line of defence. They are not replaceable, once they blow you throw them out. This means you want to completely remove the chances of false blows. So the example fuse you mentioned is probably rated too low.

Personally though, I think you are over thinking this. Fuses are very blunt tools. They are not brilliant at fine tuned set ups like you seem to be asking for. I’d say a standard 150A slow blow fuse would be suitable, it won’t catch a fast pulse, but fuses don’t do that. If you want to do that, you need to design some complicated circuitry, or risk blowing (and throwing out) lots of cells needlessly.

TL:DR fuses are too vague to do what you are asking.

\$\endgroup\$
7
  • \$\begingroup\$ Thanks Puffafish. That was very detailed. Sorry about my delayed response. The best fuse I can find that will fit is the 0685P9500-01 which is a 50A fuse in a 1206 footprint. There are others in a 1206 package, and that leaves a heaps of space left. But most others seem to be physically larger and won't fit. \$\endgroup\$
    – Dan
    Sep 5, 2022 at 7:07
  • \$\begingroup\$ It also sounds like 50A per cell isn't enough of a margin when I'm expecting the BMS to step in at 71A on the pack. I'm not sure what to do about avoiding false blows since I can't seem to get anything greater than 50A to fit. \$\endgroup\$
    – Dan
    Sep 5, 2022 at 7:08
  • \$\begingroup\$ In the case where 1 parallel cell is damaged and becomes a short circuit, the smallest current the other cell can be exposed to is 375A (± some large tolerance) which will trip this fuse in roughly 7.5ms. That assumes the highest internal resistance of 8mΩ for the dead cell and the lowest voltage of 3V for the live cell. For UN38.3 certification, the cells are tested with a 100mΩ short circuit for an hour, and since we're also measuring cell temperature, we might be ok. \$\endgroup\$
    – Dan
    Sep 5, 2022 at 7:08
  • \$\begingroup\$ For the 1.2ms pulses that exceed the cells capabilities, I'm not sure what to do about that either. \$\endgroup\$
    – Dan
    Sep 5, 2022 at 7:08
  • \$\begingroup\$ @Dan Fuses are dumb. Your BMS has brains, used those brains to monitor current and to turn off power FETs at high currents. Fuses can just be lengths of wire, so getting something to fit in a small space shouldn't be too hard, talk to your cell manufacturer, they'll be able to get the correct fuse in there. \$\endgroup\$
    – Puffafish
    Sep 5, 2022 at 7:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.