This little project was inspired by the accidental discovery of some unusual looking components inside a discarded high power lighting controller.  The controller was used to switch power to a bank of flourescent lighting fixtures for a large lecture theatre complex. Each switch circuit might have been switching over one hundred 100W flourescent lighting fixtures.  These nasty inductive loads would have made short work of a standard relay and probably even a conservatively rated large TRIAC would not have withstood many on off cycles.  This particular controller used Mercury filled relays of a type I had never encountered before.


Each relay consists of a reservoir of Mercury metal ( nearly 40cc !), an iron float, an electromagnet and a hermetically sealed chamber containing the Mercury, electrodes and iron plunger.  When the coil is energised, this pulls the iron plunger into the Mercury pool, which displaced rises up the chamber to make contact with the top electrode.  This works because Iron floats on Mercury and is also one of the few metals that Mercury will not dissolve.

This makes for almost the "perfect" relay.  There is no possiblity of contact chatter, a true metal to metal contact is made of miniscule contact resistance and no possibility of the contacts welding to each other as a result of overload or arc flash during opening.  In fact the arc flash will help to break the contact as vapouring Mercury thrusts the liquid away from the flashing electrode. Compared with convential relays and contactors, the size per equivalent rating is very small.  There is no prospect of the contact becoming RF rectifying.

There are downsides to these relays, a few that come to mind are that it depends on the action of gravity, so orientation is critical. Mercury is a potent poison, poses  significant disposal problems and in extreme cold weather can freeze. (Not a problem for me here in Oz)  . The other problem is that it is not possible to make a set of changeover contacts, unlike a conventional relay.

The design of RF relays poses some problems, the main one being the minimization of stray inductance. The Contact material should be Gold or Platinum or Rhodium.  Silver is the next compromise !  None of these are real options in the commercial world for obvious reasons.  Silver is not such a good choice, despite the minimum bulk resistivity contact corrosion is a problem. Not so bad for DC or power AC applications but no good for RF. Commercial power contactors use proprietary alloys consisting of a sintered tablet of silver oxide and other metals. The appeal of a true metallic contact for RF is compelling  and a mercury contactor is the answere.


In this design the changeover is implemented, as per the circuit, by a conventional relay which alternatively activates either relay.  A good feature ( or mis-feature depending on point of view ) of this design is when totally deactivated, the antenna is disconnected by default, a good state of affairs for radio hams.  

This simple design  features both PL259 and BNC conntectors for all input output connections, heavy RG8 braid used for all internal RF current paths. A true copper groundplane for the IO connectors ensures minimum insertion loss. There is a toggle switch for local control.  The little red things are metal oxide varistors, I prefer to use these across relay coils because they are insensitve to polarity and permit more rapid coil  deactivation. These MOVs fire at about 32V so I put two in series to ensure compatibility with a 24VDC system. The little PCB was made with the dremel method.   When soldering to the relay terminals it is essential to put a heat sink where the terminal meets the hermetic seal. You really dont want the Mercury to leak out!

Allthough I am much enamoured of my TR relay, I would discourage reproduction due to the extreme cost of the relays. Mine were free, thats the only and true reason for their use!

The reason for bringing all the contacts to one side, rather than a symetrical layout  was cable dressing issues in my shack. This relay has to be bolted to the wall  and RG8  is a pain !  otherwise, in an ideal world, the layout should be symmetrical.