Self oscillating mercury switch

[johnwa]

Recently I was thinking about how to make an inverter circuit without any power semiconductors. Mechanical switching is one possibility, however contact wear is always an issue. However, it turns out it is possible to arrange a mercury switch in a self-oscillating configuration by applying an external magnetic field. The image shows the basic idea: current flows at right angles to the magnetic field, which forces the mercury drop away from the contacts, breaking the circuit. The drop rolls back under gravity, and the cycle repeats.

Using this effect, I was able to build a rather temperamental inverter circuit for running a fluorescent lamp from 12VDC: http://loopgain.net/hgswosc/ I wonder if anyone else can get this to work any better?

[Alex Eisenhut]

https://en.wikipedia.org/wiki/Vibrator_(electronic)

[Seekonk]

I used to work with ignitrons used to discharge a capacitor bank into a coil.  This was used to make plastic magnets. Every couple months the mercury would plate to the glass surface and short it out.  Then I would have to turn the tube upside down and shake it to collect the mercury.  Be aware that may happen to your device if currents are high enough over time.

[johnwa]

https://en.wikipedia.org/wiki/Vibrator_(electronic)

Yeah, I had heard of these being used in old portable radios and things. I did try a few experiments with self-oscillating relays at one point, but they always seemed to end with the contacts welded together. Though perhaps a purpose-made device, with a tuned reed, might be more reliable.

I used to work with ignitrons used to discharge a capacitor bank into a coil.  This was used to make plastic magnets. Every couple months the mercury would plate to the glass surface and short it out.  Then I would have to turn the tube upside down and shake it to collect the mercury.  Be aware that may happen to your device if currents are high enough over time.

I did notice at one point, when the mercury switch got quite warm, the mercury vapour starting to condense on the inside of the glass, I suppose it would be possible to completely short it out eventually. Though I think this circuit is far too unreliable for long term operation anyway...

Plastic magnets sound a bit odd! (if by plastic you mean polymer) Were these just fridge-style things containing magnetic particles, or something else?

[ealex]

I've seen such a thing at the university some time ago.
It was used as a time-base for some relay-based railroad safety "computer", something like this: https://en.wikipedia.org/wiki/Interlocking.
If I remember properly it was only used to generate timing for blinking lights, etc, as it had a 0.75Hz frequency.

I can't find any picture but if I remember it had a spring loaded moving part that had the mercury switches and some kind of latching relays and electromagnets.
The main switch was an U shaped tube that had mercury( quite a lot of it ) and a electrodes that connected to the latching relays and electromagnets.

Also, for reliability all interlocking relays had graphite / silver contact pads -> they could never weld together (and no springs to wear down).

[johnwa]

I can't find any picture but if I remember it had a spring loaded moving part that had the mercury switches and some kind of latching relays and electromagnets.
The main switch was an U shaped tube that had mercury( quite a lot of it ) and a electrodes that connected to the latching relays and electromagnets.

Interesting, looks like this idea has seen some practical application. Your description sounds a bit like a mercury contactor design I have heard of, with a steel core floating in the mercury that was displaced by an electromagnet.

The interlocking systems are impressive bits of equipment too, especially the old mechanical systems. I guess they would have been one of the first examples of a large scale fail-safe system.

[Cyberdragon]

I think you need one of these

http://www.fripro.com/other%20FAIRBAIRN%20inventions.htm

Look for this

[johnwa]

I think you need one of these

http://www.fripro.com/other%20FAIRBAIRN%20inventions.htm

Look for this

I suppose the oscillations have a certain warmth that semiconductor oscillators cannot achieve? 

Nice avatar there, Cyberdragon!

[DTJ]

https://en.wikipedia.org/wiki/Vibrator_(electronic)

Yeah, I had heard of these being used in old portable radios and things. I did try a few experiments with self-oscillating relays at one point, but they always seemed to end with the contacts welded together. Though perhaps a purpose-made device, with a tuned reed, might be more reliable.

I recall pulling a valve (tube) car radio apart as a kid in the 70's. It had a vibrator in it to generate "AC" which was stepped up for the valve HT.

[Cyberdragon]

I melted the contacts in a vibrator connected to a reversed mains transformer while trying to make an arc lamp from pencil leads.

[Siwastaja]

Properly designed motor brush designs have thousands and thousands of serviceless runtime, industry proven. You could try to modify an existing brush&commutator assembly from a quality DC motor to build an inverter.

[Tibby]

This device was used as a timing source in old, relay based, railway signalling systems. The mercury oscillates in the U tube. There is a long contact and few shorter ones. Some are used to set the oscillator in motion, when the mercury breaks the contact it swings back, but others re used to provide various timing signals and flash lamps. Electrodes are at different lengths to provide different timing signals, but of course all in phase as they stem from the same source

[Lukas]

I found some (unfortunately german) documents explaining the operation of the mercury oscillators used in railway signalling sytems:
http://www.drehscheibe-online.de/foren/read.php?17,6363754,6364082#msg-6364082
https://depatisnet.dpma.de/DepatisNet/depatisnet/DE000000536467A_all_pages.pdf?window=1&space=menu&content=download_doc_verify&action=download_doc&docid=DE000000536467A&so=asc&sf=vn&firstdoc=1&struct=&Cl=2&Bi=1&Ab=&De=1&Dr=3&Pts=&Pa=&We=&Sr=&Eam=&Cor=&Aa=&NrFaxPages=3&pdfpage=1&pdfmatrix=1x1&origin=pdf_window&verify_str=null