# Limiting 1000 µF capacitor inrush current to protect fuse and power source

I have a WS2812b (NeoPixel) strip controlled by ATmega and powered by 5 V 2 A cell-phone style charger (from Digi-Key). The strip itself is located about 1.3 m away from the power source and microcontroller. The whole circuit would not consume more than 1 A (theoretical limit 4-6 A at full brightness).

Design guidelines on Adafruit and other sources suggest adding a 1000 µF capacitor at the beginning of the strip to help with power surges and stability, which I have in my existing circuit and there are no issues running this set up whatsoever. However I would like to add some safety components to the circuit, primarily in a form of a fuse straight out of the power supply. I am thinking adding 1.5 A fuse, but I am concerned that sudden inrush current on a large capacitor can blow the fuse upon initial connection to power. Would a 1.5 A slow blow (or even fast blow) fuse be OK for this application or do I need to design a dedicated slow-start circuit?

Schematics is below:

Update: I added the theoretical maximum of the circuit (4-6 A as there are close to 100 pixels current limited by software - thus the need for a fuse)

• Sep 19 at 5:07
• ~3.5 Amps/m max Did you use this ? Sep 19 at 6:28
• the power supply will most likely shutdown before the fuse, so assuming your wiring is suitable for 2A, the fuse is superfluous. If it were a battery supplying your power, then, yes, adding a fuse is wise. Remember - the fuse is to protect the wire, not the load. Sep 19 at 6:50
• I don't know whether I should post this as an answer or leave as comment. The whole concept makes little sense. Since this device has USB, USB devices are allowed to have a maximum of 10uF of capacitance, the 1000uF capacitor is 100x more than the USB charger expects to have as load and it might have problems starting up, due to the inrush current period taking so long they can go into overcurrent or undervoltage shutdown. And as they only provide say 2A rated and usually have overcurrent protection, if the fuse survives the inrush, it is likely it never blows. Sep 19 at 8:56
• You could charge the capacitor through an inductor with a low DC resistance. Just make sure you don't set up any unfortunate resonance frequencies. Sep 20 at 13:54

If you are interested in my ideas: Here you go in order.

1. Read the datasheet for the charger carefully. Search for terms like 'short circuit protection' or 'current limit'. Most devices will have some sort of hiccup implementation. My guess is that sufficient safety measures are already included and qualified - so no need to complicate your design. Please also consider: These devices are fully qualified and tested. You know us Germans: These are big words.

2. If you plan on integrating a fuse: have a look at the I2T rating in the datasheets. This value will determine the 'snappiness', or in other words: The tolerance to inrush currents. You could also consider using a so called PTC-Fuse. If you are interested in further education regarding the topic, please see: https://www.littelfuse.com/~/media/automotive/catalogs/littelfuse_fuseology.pdf If not: Use a 3A Slow Blow (5x20mm) - will work just fine.

3. If you want to overkill your problem in a simple way: Add a resistor (1 ohm, for example) in series with the fuse. This will limit the inrush current as desired, but will waste a lot of power over time. Also the voltage drop must be considered. You could also use an NTC design which reduces the power losses. Please see: https://www.tdk-electronics.tdk.com.cn/download/541612/b1b77484fb39733c7d16858074bb9490/pdf-applicationnotes.pdf

4. If you want to go the extra mile: There are active circuits, so called softstarters, out there. Please have a look at https://www.ti.com/lit/an/slva670a/slva670a.pdf This is most likely, almost definitly, an absolute overkill. To make matters worse: You could use a device implementing a so called E-Fuse and a softstarter - Most likely the IC will have some sort of I2C interface. In this way, the controller could retrieve current power consumption and stuff like that - but what for?

5. My two cents: Confirm the safety-features of your wall adapter, and if sufficient, stick to it. Most certainly these devices are not designed to work with highly capacitive loads - but they will just fine. Bear in mind: If you are unsure on how to implement safety features correctly, the probability of the solution not working in case it is required can reach an unacceptable threshold - and then: where is the point of implementing it at first? Don't get me wrong here: Implementing safety features according to datasheet or app note is one part. Testing and confirming them is the other. These are important topics, even for hobby projects. If these struggles have already been taken by the wall adapter guys - why take them again?

• ElectronicsStudent - Welcome :-) Please do not get "chatty" in comments & do not write several short comments. Stack Exchange is different from typical forums. For "chatty" comments, or too many comments in total, we will either delete them or move them to a chatroom (which I have now done). See here and here for rules about comments. Also please see the tour. Thanks. Sep 19 at 18:48
• All - The long comment chain has exceeded what is reasonable for comments. Therefore it has been moved to chat and should be continued there (link below). --- As this bulk moving of comments to chat can only be done once per answer, any further comments posted here might be deleted without notice. Keep it in chat, please! Any updates to the answer which are decided during the chat, can still be made via an edit to the answer. Thanks. Sep 19 at 18:50
• Comments are not for extended discussion; this conversation has been moved to chat. Sep 19 at 18:50

Many people use P-MOS in the linear region to avoid the inrush current during the initial capacitor charging. It is a working approach of course, but it is hard to hold a MOSFET in the linear zone for a long time (all charging process).

I prefer charging through a shunt (like 47 ohm) and once the capacitor reaches a certain voltage (like 3 V) when the inrush current can be tolerable, connect the capacitor directly to the source - in another words, shorts the shunt.

Regarding the fuse, you would have to use a fuse with a higher current rating (or a slower one) to not blow it during the initial charging than is needed during normal operation, so you will lose the fast protecting function.

Schematic:

Design Specs:

5V charger 2A max with OCP shutdown. 2.5 Ohm load min
StripLED 3.5A/m all white User spec: 1.5A max
WS2812b 3.5 to 5.3V for all IC's on strip

e-cap options: 1mF 10 mOhm to 350 mOhm ESR
T= 10 us (low ESR) to 350 us (G.P.)
choose ESR = T/C

Problem: turn-on surge current I = V/ESR >> 2A

Safety with inrush prevented is not a problem: All SMPS of these type now have OCP.

Solution: Choose ICL derated 30 to 50% for max current handling but >= 3 Ohms for 1A @ 3V or 1.7A @ 5V but rated >= 3A

• Inrush Current Limiters (ICL) are ceramic resistor discs with negative temperature coefficient (NTC)

There will be voltage drop in the traces of stripleds and Rmin @ 50% of max current might be 1/0.5^2= 4x Rmin from I^2R=P resistance regulation at high temp. Yet WS2812b can work down to 3.5 V so resistance of stripled must be measured for ground and V+ added together such that I * R max.< 1 V with 0.5V margin.

For Rmax= R(ICL) + R(strip Vcc + 0V) at desired Imax

Recommendation 4A rated ICL 3 Ohms

• lowest cost , safe operation but gets hot.

Desired Outcome:

• Charger stress is reduced on startup without shutdown
• ICL and C choice will attenuate supply ripple >= 30dB by RC filter. e.g. if f = 50 kHz or 10us per pulse then RC=> 30x 10us = 300 us =T using ICL R at Imax and C with low ESR for RC=T

Fuses have a "Nominal Melting" rating (in $$\A^2s\$$) in addition to the maximum steady-state current. If you know this rating and the overall resistance (supply + fuse + ESR of the capacitor), you can calculate exactly how much far you are from blowing the fuse while the capacitor charges, and decide whether you need a bigger fuse / pre-charge circuit or not.

If you do need a pre-charge circuit, here's a simple one with a single FET:

simulate this circuit – Schematic created using CircuitLab

The time constant responsible for switching the FET (R1xC1) is 10 times larger than the precharge time constant (R3xC2), so by the time M1 starts conducing, C2 is already charged via R3.