Defective polyfuse?

[mrmp17]

Hi,

I'm developing a battery powered circuit that powers a heating element (2.9 Ohm PCB heater) through a USB type-C connector which is also used to connect a charger. The bidirectional mosfet switch that connects battery voltage to the connector (when powering a heater) also has a polyfuse in series to protect it from shorts. I've had two polyfuses die on two of the circuits I have assembled (under normal operation, no shorts) - when heater is connected, I get only around 0.5V on my output, that should otherwise be full Li-Ion battery voltage. When not under any load, the fuse measures around 50Ohm (after plenty of time to let it cool down). One weird thing is, after desoldering one of the failed polyfuses, it started to function as expected again. This failure happened after testing the circuit for many hours (around 20h of operation). One thing I did a few days before failure is, I have coated the PCBs with a silicone compound (Kafuter K-705), to protect it from moisture. I have also used the circuit outside (in a case) in cold weather, but I don't have a clue how any of those two things could affect anything.

Is this a bad batch of polyfuses, or am I missing something? Schematic of this part of the circuit attached below.

The fuse is TECHFUSE SL1206125 bought from LCSC (part number C70164).

[Zero999]

How much current does the fuse normally pass?

I've had the same problem before. It was caused by using too lower polyfuse rating. The silicone coating probably isn't helping either.

[TheMG]

According to the datasheet, that polyfuse has a holding current of 1.25A. Assuming a USB voltage of 5V, your 2.9 ohm heater draws 1.72A. This is greater than the holding current.

Polyfuses are not a precise device by any means. That is why there is a "hold current" and a "trip current". If operated at a current between the hold and trip current, there is no guarantee that the polyfuse won't trip unexpectedly after some time.

The hold and trip currents are also temperature dependent, and the datasheet specifies those values at 25C temperature. At a higher temperature, it will trip at a lower current, and at a lower temperature will trip at higher current. Conformal coatings or potting can indeed affect the trip current by holding in heat.

https://datasheet.lcsc.com/szlcsc/1809211118_TECHFUSE-SL1206125_C70164.pdf

The main issue if that you have simply selected a polyfuse too small for your load. Look through the datasheet closely, they also have a table and graphs that show the effect of ambient temperature on the hold current. Select the polyfuse which has a hold current greater than your worst case load, for the worst case ambient temperature that your device will be subjected to.

[Zero999]

Thanks for taking more time to do calculations and write a more detailed response than I did.

Polyfuses only have a limited number of trip cycles. After each cycle it takes a certain amount of time to recover and the resistance will be permanely increased, by a small amount. This has the effect of lowering the trip current after every time it trips. Eventually there comes a point when it will either trip at the load's normal operating current, or doesn't recover afterwards.

Use a polyfuse with at least double the tripping current, of the nominal load current. If you're concerned that it won't provide sufficient protection, then you need to design your circuit to handle the extra current, or provide an extra layer of protection. The polyfuse should be there as a means of last resort protection, rather than something which is repeatedly tripped, such as the current limit on a bench power supply.

[mrmp17]

Thank you both!
Sorry for not including all the load data in my question, so here is my worst case calculation. The heater is powered from battery voltage, so worst case 4.2V, which makes for around 1.5A. Heater is also PWM-ed (frequncy of about 0.5Hz), with a duty cycle of around 50% in worst case. This calculates to a RMS current through the fuse at 1A, which is below the fuse hold rating of 1.25A. Even at 1.5A (during the high time of PWM) I would not expect the fuse to trip for at least 10s, even at higher ambient temperaturs. Is this still considered a underrated polyfuse? Would these operation conditions be expected to damage the polyfuse during the 20 hours of operation to the point of complete "open circuit" (50ohm when given time to cool down) failure?

[abyrvalg]

Coating the polyfuse contributes to the problem too. One popular home appliance by Xiaomi had this problem - a polyfuse with adequate current parameters was failing indeed because of being painted and having heat dissipation problems. In the next device revision they’ve changed the coating mask to exclude the fuse and the problem have disappeared.
If you need it to be coated, you need to scale the fuse current rating somehow to take the slower heat dissipation into account.

[uski]

I would like to add that polyfuses are not that great. I mean, they are great - but not always.
Specifically, they are useful where there is a reasonable chance of intermittent defect due to external causes, which are likely to be removed by the user. For instance, a transient short circuit on an output connector.

But it seems like in your application, there is no fault that could be transient. If there is a fault, it would be internal, and it would require a repair of your product anyway.

In these conditions, I would use a normal fuse. Yes it would not auto reset, but there is no need for an auto-reset anyway. And polyfuses are not ideal because they have an internal resistance and also when they fail, they still allow some current to go through which can cause issues too.

I'm really not a big fan of polyfuses frankly. They have their applications, but I am not a fan of people just using polyfuses everywhere where fuses would be appropriate.

[mrmp17]

The reasoning behind the polyfuse is exactly that - intermittent external faults. The heater cable and the heater itself are subject to mechanical stress, so a torn and thus shorted cable is a very real possibility so a normal fuse blown in the battery pack because of this is not really a great design of output protection in my opinion. Is there a better implementation of such protection, that has a reasonable component count?

One other question: There is significant capacitance (10uF + 4.7uF) on Vbus rail (needed by the charger) but when the circuit is powering a heater, this capacitance is charged every rising edge of the before mentioned PWM (charger IC is of course disabled during heating operation). Could this be a factor in polyfuses dying so quickly (besides maybe being a bit underrated)? Are such repeated current spikes likely to kill a polyfuse?

[ejeffrey]

Second not being a big fan of polyfuses.  You need to oversize them pretty drastically and then they can let through a lot more current for a lot longer than you would like.  It might still be a reasonable approach here.

PWM could be a problem.  The thermal time constant of the polyfuse is much slower, but due to I^2R, the RMS current seen by the fuse well be higher than the average current and this causes more heating.

[Siwastaja]

Always derate holding current and expected life cycles.

For example, if you are expecting 1A operating current and 10 cycles, pick a polyfuse with at least 2A hold current and 100 cycles. Yes, this derating will cause the trip current to go even further from your normal current, making it even less precise. Polyfuses aren't that great.

It's more design, but consider doing an active e-fuse (with traditional single-use fuse in series) circuit. Seeing you have "PWM", I'm assuming you have a controller, probably a micro? Why not add current sense and actively stop if the output is shorted, this would be cheaper than a polyfuse? Just make sure the switching transistor SOA can handle the delay of the current sense + control delay (analog comparator delay + interrupt latency, for example).