sticky relay fix?

[BANE]

Hello all,
I have a 24v 70amp relay serving as a E-stop for my large robot.  The relay is placed between the 24v 66ah lead acid batteries and a 2x25 sabertooth.  The relay 12v coil are controlled with an N channel mosfet which is connect to the axon II.  My problem is the relay seems to get stuck in "ON" position %40 of the time  (aka not good for an e-stop).  I've opened up one and mechanically everything is good.  Also, i have an identical one and the same thing happens to is as well.

My guess first is that, being it is high current, the hammer is getting welded to the anvil a little.  generally i can tap the outer casing and it will brake loose. possible solution: use graphite or some lubricant??

Second guess: that the relays mosfet circuit isnt powering off the relay all the way.  Ive checked with multimeter and there is 0v in the coil when off.

anybody have any input?  So far my idea of a remote e-stop isnt working out so well  

thnx in advance

[BANE]

never mind. doing away with the relays and going oldschool with a switch

[Soeren]

Hi,

My guess first is that, being it is high current, the hammer is getting welded to the anvil a little.  generally i can tap the outer casing and it will brake loose. possible solution: use graphite or some lubricant??

Yes I'm 99.9% sure that it's due to contact welding, which is a common problem with high current contacts (whether in relays or regular switches).

Some relays (and switches) counter it by letting one contact slide over the contact when touching ( a sort of "L" shaped movement), but the best is getting relays with wolfram (tungsten) contacts, as they're very hard to affect (I have seen slight pitting in wolfram contacts, but never experienced them welding).

A snubber network over each contact pair is yet a method (as long as the contacts are rated for the current you're switching), as this kills the high voltage inductive spark that makes the welding happen.
If used unipolar, a fast diode could be used.

[BANE]

thanks for the response Soeren,

(as long as the contacts are rated for the current you're switching)

These relays actually came from two salvaged wheelchair motor controllers so they should be rated high enough.

A snubber network over each contact pair is yet a method

i googled "snubber network" and found this http://www.electronics-tutorials.ws/io/io_5.html this looks like a good solution

Just out of curiosity about he snubber network, the sabertooths can draw 50A for a couple ms which is pretty nice when demanding torque.  Will adding this have a insignificant performance impact?

[Soeren]

Hi,

Just out of curiosity about he snubber network, the sabertooths can draw 50A for a couple ms which is pretty nice when demanding torque.  Will adding this have a insignificant performance impact?

No, the snubber is only in action when switching the contacts. When the contacts are driving an inductive load, the voltage spikes over the contacts will go sky high, the surrounding air is ionized and sparks can fly. A bouncing contact only worsen this, as it happens when turned on. The only thing you'll feel from a snubber is a slightly slower off (you'd need a 'scope to see the small difference), but you gain the clean switching.
The snubber should be calculated for the load.

[BANE]

thanks an ton; i will defiantly keep this in mind for future projects.

Im currently working on redesigning the whole power setup and probably going to drop the relay idea and just have an e-stop button