Reducing Inrush Current with NTC

Hi everyone, need some help figuring out what went wrong here:

I am looking for solutions (preferably cheap, simple and not power hungry) to reduce or minimize inrush current of small single phase motors, pumps and compressors of up to 1 or 1.½ HP. I'm pursuing this in order to make it possible to use such equipment with a small off-grid inverter, since almost all pure sine wave inverters (high frequency models) rated at about 1~2 kW have the rated steady state power necessary to run these appliances (small fridges, pumps, etc), but they simply lack the capability to start them due to relatively high inrush currents.

For a reference test, I thought about trying some stuff with my home fridge. My fridge runs at 127Vac single phase from the mains power (power here in Brazil runs at 3-phase 220/127Vac), draws 2A at steady state, power factor of 0.55 and steady power consumption of about 120W. At first glance, you'd think an off-grid pure sine inverter rated at 1.000W would have no problem running this fridge. However, the inverter cannot start the fridge no matter what. I tried different input voltages (12 and 24V, 2 different inverter models), tried a larger battery bank (220Ah and 440Ah, fully charged), tried using larger input cables (tried up to 95mm², which is about 0000AWG), no sucess. The inverter simply cannot handle the output surge caused by the inrush current. The alarm indicated on the inverter was "Output Overload" everytime. I had to buy an expensive clamp meter to measure the inrush current to figure out what was happening. Using the mains power, the inrush current was measured to be 50A. Instantly I thought that was the problem. 50A was too much for the inverter, even if it was for only a few miliseconds.

Many have suggested I simply use a larger inverter with superior power rating or a low-frequency model. We all know this works, that is obvious. However, that is the lazy solution, because you are wasting money and power to run an inverter @5% load at steady state. I don't think it's reasonable to use a 5.000W inverter to run a 120W fridge and low-frequency inverter models here in Brazil have limited options and are ridiculously expensive. The only thing stopping my 1.000W inverter from running my 120W fridge is the inrush current, so I looked for ways to reduce the inrush current.

Searching for solutions, I stumbled upon NTCs. In theory, NTCs are components with significant resistance at ambient temperature and negligible resistance at steady state current flow. Manufacturers (one of them is Ametherm) advertise them as inrush current limiters. So I thought about wiring a NTC in series with my fridge, which in theory would add some resistance at startup reducing the inrush current and at steady state would not interfere with the compressor since NTC resistance drops to negligible values once the temperature rises due to the passage of current.

I bought a pack of NTCs on the internet. I searched for a suitable model, found one that had 10Ω @25°C and max steady state current of 5A, model NTC 10D-15. Since steady state current of my fridge was 2A, the NTC would be able to handle it with ease. Using Ohm's Law the max inrush current I expected was around 127V/10Ω = 12.7A. What I predicted would happen would be a peak current of something around that 12.7A, which would be a reduction of around 5× from the normal 50A measured. I thought the fridge compressor would probably have a slow start due to that inrush reduction, would take more time to achieve steady state, but the initial surge would at least be supressed I thought. Once steady state was achieved, the NTC would be at a higher temperature, thus it's resistance would be around 0.15mΩ as specified in the datasheet, hence not interfering with steady state current of 2A and the fridge would run normally. So I prepared my setup as follows:

• I wired a 6-in-1 wattmeter with the circuit so I could monitor voltage, current and power at the same time. The wattmeter doesn't have refresh rate high enough to perceive the inrush event though. The wattmeter was merely to observe steady state behavior with NTC.
• I placed my clamp meter in the hot wire so I could effectively measure inrush current value.
• Wired the NTC in series with the fridge power cord.
• Took a few steps back to observe what would happen LOL.
• Switched the power on.

To sum it up, the experiment failed. After powering the circuit on, the compressor took a few 15 seconds before kicking in. The NTC literally instantly exploded the moment the compressor started. I saw a little spark as a chunk of the component flew away across my lab haha. What really suprised me is that even after the NTC exploded the circuit achieved steady state normally. After making sure nothing was on fire I approached the circuit and checked the wattmeter. Measuring normally 2A and 120W as if the NTC wasn't even there. So I concluded that at steady state the circuit behavior was as expected, the NTC resistance was indeed negligible and the fridge compressor drawed power normally.

However, when I checked the clamp meter, it registered an inrush current of 48A! That puzzled me a lot. My hypothesis is that the NTC was not able to supress the surge, maybe because of the duration of the surge? Maybe the NTC achieved a higher temperature too fast and the fridge was still demanding current for its startup? If so, then for a brief moment, the NTC had already achieved its final resistance of 0.15mΩ due to temperature increase, but the fridge still wanted to draw inrush current from the mains power, thus the inrush current measured a little lower 48A and that current through the NTC made it explode I guess.

I searched online for more data on the NTC model 10D-15, maybe some specification I overlooked. I couldn't find any. However, on Ametherm's website I found some info on NTC sizing and discovered a calculation I ignored: inrush energy. Any NTC has a total energy in Joules it can absorb before reaching max temp. They use the equivalent capacitance of the circuit to calculate the energy of the surge, but I don't have this capacitance value for my fridge. In this case, Ametherm suggests calculating the worst case scenario, in which the inrush current has a full cycle duration. So we have E = V × i × t = 127 × 50 × (1 ÷ 60) ≈ 106J. So probably the NTC I picked wasn't rated for 100J.

Then I tried something different. I tried 4 of them wired in series with the fridge. I believed that would split the energy between the 4 NTCs, about 25J in the worst case scenario for each. Additionally, the series resistance now would be about 40Ω, which should limit the inrush current even further. The results were almost the same, the main difference none of the NTCs exploded, but I smelled a light burned scent from them. The inrush current reduced a little to 46A, still far from what I expected from using a 40Ω NTC configuration. The steady state operation was normal as expected.

Any ideas on what made the behavior not as expected on inrush current limitation?
I suspect your problem is in the nature of the load - you are dealing with an induction motor and an unspecified and uncontrolled load (i.e. a load that you have NO control over). You are attempting to start an induction motor by, basically, reducing the voltage applied to it. But that unfortunately just increases the amps demanded by the induction motor. This is NOT a recipe for success. I rather suspect that the load is rather intractable - you are trying to start from a 'stalled' condition where the torque requirement is high and the torque requirement means a very large current is required. Which you are denying the device. Ouch. The only (IMHO) solution is: start with no load and then supply ample current to get the motor going. Then you start increasing the load slowly so the current stays within bounds. Of course, you can not do this.

I guess the bottom line for me: you have to start with the nature of the load. Lick that and maybe you will have a prayer.
Hi tio_lennon, I think what you need is a high energy NTC like the Ametherm MS35 series. I think it´s small and it will fit your needs.
One problem will be its availability here in Brazil. I've found it online here. It's not expensive.
PS: Yes, I am in Brazil too.... My son Fabio Jesus once developed a off-grid solar feeder. He had the same issue with pumps and freezers.

One more thing you should notice: the inrush current may last for SECONDS ! So you should use the highest energy capable NTC you can afford.