Soft start for 12V sump pump

[large_ghostman]

Got a slightly irritating problem , Two small sump pumps of 12V each, one is 3A the other is 5A. They are run from the same 12V line everything else runs from, including the regulators for 5V and 3.3V for the control system.

The pumps switch in and out by relay at the moment (I might change this to Mosfet at some point), they are on for around 4-6 mins at a time and can be off from 12mins - 2 hours depending.

When they start up i get some pretty big spikes on the 5V and 3.3V lines. despite have diodes across the relays and pumps, and general filtering on the lower voltage lines. So any ideas how best to soften these spikes? I dont have much room but getting frustrated with this, i am tempted to just stick a big ass cap on the pumps so the few Ms at start up they draw from the caps.

As that is lazy, i am after some better ideas to stop the spikes, I noticed the 3A pump pulls around 8A for a few mS when it starts up, I cant get to the other pump to take a reading.

The board wasnt mine and has had several people modify it. At the moment i cant remove the pumps or do a better board, but next upgrade i will make sure its done properly!

 

[Cicero]

Just some things off the top of my head:

• I don't feel adding some tank caps around the place is necessarily a lazy method, and I'd probably do that straight away. Those caps would need to be fully charged when your relay switches, so upstream of the pumps and relay, otherwise when you switch your relay it may be worse, charging your empty cap and your pumps inrush at the same time.
• Maybe an appropriate NTC in series with the pumps - thats pretty lazy, haha.
• Then I'd look at adding a snubber network across the pumps to suppress the transients.
• Maybe if you're having problems with the spikes affecting the rails and causing real problems, then add further filtering to your main 5V rail.

I'd be interested in how quickly the spikes happen. Also be certain the spikes are actually as bad as you think, and not induced on your scope probes or whatever.

 

[dknguyen]

Things like diodes and snubbers do not deal with startup transients or inrush currents. They only deal with shutdown transients. So if you were wondering why adding diodes seemed to do nothing for the startup spikes or inrush currents, that is why.

Use an NTC. It will start as high resistance when cold and thus limit inrush currents, but as the motor current warms up the NTC, the resistance will decrease and allow for more efficient steady state operation.

A PTC is also sometimes used this.. A PTC does the opposite. It is a low resistance when "cold" but when the inrush current spikes happens (or any excessive current), it will cause the PTC to heat up which increase resistance and thus limit the current spike. So it doesn't limit inrush current specifically, as much as it limits any overcurrent that may occur.

If the inrush spike is too short to allow the PTC to heat up fast enough to do its job then you have to use an NTC and not a PTC.

 

[large_ghostman]

Thanks guys, the pumps are hard to get too, the board isnt much better placed! I used the scope because the 3.3V chip locks up now and then when the pumps kick in. Its a bit odd but the smaller pump tend to lock it up more than the larger one, it took 4 hours to get the pump out to service and then 3 hours to get it back in! So really reluctant to take out until we upgrade next.

Good idea with the NTC, i might have to cut the cables and add another board, i am really not in the mood to try and get the board back out!! The 5V line dosnt lock up but for some reason it throws the clock out and displays some random time on it, so we have to reset the start time on the pumps to fit what the clock says .

The entire thing is an awful design, one of the relays is making that clack noise you sometimes get just before they give up for good! So maybe divine intervention might solve it for me. The 3.3V and 5V boards are filtered about as much as i dare go, whats really annoyed me is we sent them out to a local company to have altered. What came back was a dogs dinner, i didnt have a choice as we just didnt have time to do it ourselves.

The whole thing is a old board thats been updated a couple of times, IF i get time i might design completely new boards and install separate supplies for them, rock and hard place as its not our system. We get to service under contract and they want to put in a newer system, but the newer system isnt ready yet. Not sure when its going to ready at the moment, it wouldnt pass safety checks these days. nice small pumps apparently from a old boat (canal barge), really robust but suck hard when they kick in.

Awww i just know i will end up being bugged with it until i sort it.... so i guess thats my weekend sorted

 

[MikeMl]

A stalled DC motor looks like a dead short until the rotor begins to move and starts making a back emf to reduce its current demand.

The stalled current can be a factor of 5 higher than the running current, so your 5A motor might draw 25A from the power supply for several ten of ms while the rotor starts moving...

25A can pull-down the voltage on the power supply until the motor starts. One method is to put a series diode going to from the 12V supply to a (new) large electrolytic filter capacitor, and put that in the path to the voltage regulators. As the supply is pulled down, the diode blocks.

That way, even if the in-rush to the motor pulls the supply voltage down to 3V, the regulators simply draw off the energy stored in the electrolytic capacitor until the motor comes up to speed, and the supply recovers. See below:

 

[large_ghostman]

A stalled DC motor looks like a dead short until the rotor begins to move and starts making a back emf to reduce its current demand.

The stalled current can be a factor of 5 higher than the running current, so your 5A motor might draw 25A from the power supply for several ten of ms while the rotor starts moving...

25A can pull-down the voltage on the power supply until the motor starts. One method is to put a series diode going to from the 12V supply to a (new) large electrolytic filter capacitor, and put that in the path to the voltage regulators. As the supply is pulled down, the diode blocks.

That way, even if the in-rush to the motor pulls the supply voltage down to 3V, the regulators simply draw off the energy stored in the electrolytic capacitor until the motor comes up to speed, and the supply recovers. See below:

View attachment 112825

Thanks Mike, i will try and get that in on the current setup, but definitely the way to go on the new boards. Irritating as its not every time, just enough times to really bug you but not enough to strip it all down and spend time redoing it all if you see what I mean. That and the fact a bit cheated we paid good money for a half assed job! The main panel we are also replacing, once we decom i will take some pics. I am no neat freak but its appalling.

But we inherited it from whoever did the system to start with, I dread problems with that panel. Every wire is red and not a single identifier, nothing neat and EVERYTHING cable tied into some kind of modern art doo dah.

Its a rail system so i am a bit puzzled why they didnt simply put separate rails in for the different voltages, if they had really wanted to its three phase power board, could have easily gone extreme and stuck the transformers on different phases.

Maybe just 10 years of constantly adding things without much thought given to break downs.

 

[DGM]

You might also want to take into consideration that attempts to restrict load current during startup might tend to make the motor have more trouble breaking static friction -- which does tend to be an issue if the pump has moving seals. Isolation seems like a safer route.

 

[BobW]

It also depends on the kind of pump you have. If it's a positive displacement pump, then it's going to need full torque from standstill all the way up to full RPM, which means that if you try to limit the inrush current, the motor probably won't start. On the other hand, if it's a centrifugal pump, then the torque requirement at low RPM is very low, and increases with speed. So, for a centrifugal type pump, it should be no problem to put some kind of current limiter (NTC or fixed resistor & timer) in series with the motor. A centrifugal pump will just run at lower speed until the current limiting disappears.