Magnetic connector arcing prevention

[OwenH]

Hello,

I am designing a custom magnetic connector, it's rated for up to 15A, featuring 8 spring pins (each rated for 10amp)  (4 for + and 4 for –). On the device side, I have a TI LM74900 ideal diode controller with back-to-back MOSFETs to provide reverse current protection, as well as under-voltage and over-voltage protection.

My original plan included a dedicated “enable” pin that would disconnect the power before the main contacts separate, helping to prevent any arcing during a hot disconnect (like Apple MagSafe). However, due to mechanical constraints, implementing such a pin is challenging.

Before finalizing the mechanical design, I would like to explore additional strategies to prevent arcing under load. Specifically:

the existing Ideal Diode:
Can the LM74900 respond quickly enough to effectively prevent arcing if the connector is disconnected under load? This controller has OV/UV and overcurrent protection.  Although I imagine none would be fast enough to protect from arcing?

Transient Suppression:
Would incorporating an RC snubber, TVS diode (already included in the design), or other transient suppression methods reduce or eliminate arcing in this scenario?

Other options:
Are there other strategies or best practices for ensuring a clean, arc-free disconnect without relying on a dedicated enable pin?

Any insight into  is greatly appreciated!

Owen

[spostma]

I would use coaxial cables, as that is inductance free by nature.

The inrush current can be ramped up simply using a mosfet in series and an R/C fiter on the gate to slowly turn on the MOSFET.

R/C filters on both sides of the connector, with R being about equal to the cable impedance to dampen high frequent ringing or spikes.

[D Straney]

Don't have direct experience doing the exact same thing, but from what I know:
Protection features on an ideal diode chip definitely won't be able to eliminate all arcing during de-mating (esp. since the current sense etc is usually filtered pretty heavily), but it could limit it to a shorter time (us instead of 10s or 100s of ms, maybe). If I were you, I'd build a simple mockup and do some measurements to get a feel for the scale. I don't think there's any way you can remove arcing completely with active control.

I can't think of any situation where a snubber on the LOAD side would help, as with mechanical contacts opening, that necessarily has to be across the contacts to absorb the inductance's stored current that would otherwise cause a "flyback" voltage spike and arcing. So to use a snubber effectively, you'd either have to have pins that are guaranteed to disconnect later (back to the mechanical issue), or put a snubber inside the SOURCE side connector.

(For what it's worth, coax isn't "inductance free", just lower-inductance than two side by side wires, and harder to terminate/get heat out/get large cross section areas. You can do some inductance-per-meter lookups/simulations/measurements with different candidate cable geometries, like coax vs twisted pair, to see which one does best while meeting the 15A requirement)

[inse]

Arcing depends on the applied voltage, how much do you supply?
Did you do any experiments how much arcing is to be expected in your application?
From the magnetic connectors I know, I guess the connecting process would be probably more problematic than disconnecting.
Do you have any protection measures on the supply side or is the voltage always present on the contacts?

[Georgy.Moshkin]

Is it possible to monitor the current using fast ADC and cut the power as if it was done manually after trigger from an oscilloscope? Some comparator/fpga/mcu combination may provide adequate cost I think.

[amyk]

Look at the magnetic mains connectors used on kitchen appliances.

[mtwieg]

I am designing a custom magnetic connector, it's rated for up to 15A, featuring 8 spring pins (each rated for 10amp)  (4 for + and 4 for –). On the device side, I have a TI LM74900 ideal diode controller with back-to-back MOSFETs to provide reverse current protection, as well as under-voltage and over-voltage protection.

Based on the mention of the LM74900, I'm guessing this cable is carrying DC at 60V or less.

My original plan included a dedicated “enable” pin that would disconnect the power before the main contacts separate, helping to prevent any arcing during a hot disconnect

If you have multiple power contacts in parallel, then if you were able to sense current on each contact, then you could get similar functionality by detecting sudden imbalance in current between the pins as they separate. This would require more electronics, but so long as the time between when the first and last contact breaks is long enough (a millisecond should be plenty) then you should be able to disable the LM74900 before the last contact breaks, preventing any arcing.

(like Apple MagSafe)

AFAIK, MagSafe is not a connector at all. It's just a way of aligning Qi charging coils.

[OwenH]

Thank you for all the responses!

I’m leaning towards incorporating an enable spring pin connector within the cable assembly, as it seems to offer the most reliable operation. The cable design already includes a MOSFET-based load switch to ensure the connector pins remain unpowered when not in use.

Initially, the plan was to integrate the power switching circuitry directly into the connector housing, with all the connections at the terminal end. However, space constraints have made this difficult. My current idea is to relocate the switching circuit to the source end of the cable, housed in a small enclosure integrated into the cable assembly the opposite end, where the additional size is less of a concern.

Owen

[Berni]

Yep the best way to do this is having an enable pin.

Just make sure the pin is designed in such a way that it always gets interrupted before any other power pins do.

This is not only used in Apple MagSafe (Not the wireless kind, Apple has dumb confusing ways of naming their products, this is regarding a 5 pin connector design used on laptops that they call MagSafe and predates wireless charging features in apple products) but this is also used in things like USB-C where the high power 20V 5A charging modes also prevent a arcing problem.

The problem of doing this without a detection pin is that you can only detect it once the arc has already formed, at what point it is already too late. It doesn't matter if the load or source does it, as long as the current drops sufficiently before the power pins finally loose contact.