Jump to content
Sign in to follow this  
Tony S

Mercury Arc Thyratron

Recommended Posts

Tony S

This is heavy going to read. Please spare me a thought, I had to research and write it, with help from Phil D, Glennspark and 80 year old textbooks.


Before we start read the Mercury Arc Rectifiers thread to give you an understanding of the history of the Mercury Arc Thyratron.


That fancy box of frogs you wired up yesterday aint as modern as you thought!


Mercury Arc Thyratron (Inverter):


In the early days of the Mercury Arc Rectifier (MAR) it was soon discovered that as the arc travels around the cathode pool two arcs are active at the same time all be it briefly. This causes a short on the transformer output and can develop into a short circuit. If this backfire occurs the anodes blast away at each other ignoring the cathode. Not good as far as a rectifier is concerned and damages the anodes to the point where the bulb could be ruined.

Fuses protected each individual anode but a method was needed to stop the other anodes conducting. By introducing a grid just below the anodes with a potential just above that of the cathode the arc would stop. The grid current being regulated by the series resistor.


Normal running



Fault condition.





Only one fault relay is shown but a multi armature relay would be used to “kill” all the anodes.


Of course someone had to start playing around with these grids and reverse the action of the MAR to DC in AC out.


In 1913 the Thermionic Thyratron was known about and by 1926 the Mercury Arc Thyratron was in commercial production. Peter Cooper Hewitt (the inventor of the MAR) again being involved in the development.


If we were to introduce a DC supply to a MAR we would just get DC out. Not what we want. It would just act as a diode.


The basic 3 phase waveform:





By using a 6 phase star connected transformer we get this. The original waveform is mirrored on itself.




The exciter introduces a second 6 phase waveform of a lower amplitude. This lower voltage controls the anode grids and can be varied by altering the DC voltage to the exciter rotor field. All though it looks to be about half the output voltage, in reality it would be a small fraction of the cathode voltage. As you can see from the anti backfire drawings where a battery was used. If I did the drawing to scale you’d probably just think the graph base line was a bit wobbly.





Because the mercury arc can only conduct in one direction it gets rid of the negative half cycle.





To simplify things even more I’ll chuck away the bits not pertaining to the red phases, leaving just the actual red phase and the grid voltages that control it.





Because the output of the mercury arc thyratron is fed in to a transformer with 180° opposed windings half the waveform is reversed.




Look familiar ? ? ? ?



The exciter:


The exciter is needed to give a reference frequency for firing the cathode grids.

Depending on the application the exciter can be driven by either a synchronous AC or DC motor. The synchronous AC motor would be used where two feeds of different frequencies need to be connected. The motor being fed from the grid that is being exported to, it also makes synchronizing the supplies simple as it is locked in to one of the systems. If you’ve read the transformers thread I included a bit about the growth if the national grid where places wanted their private network supply linked in to the grid. Some of the collieries around Nottingham where I live had various voltages and frequencies. Voltages and phase reversal are easily sorted with transformers, frequency needs to be converted via a DC link using our friend the MAR. Through this link a 33⅓Hz private network supply could be linked to the 50 Hz grid. To do this would need a 4 pole synchronous motor driving a 6 pole exciter.


12½Hz, 13½Hz, 25Hz and 33⅓Hz were all to be found in various parts of the UK. At least where I started work they generated at 50Hz all be it the rotation was reversed. Just to be different the works network was at 11.4kV. That was solved by swapping the phases at the intake sub and cranking the NWEB 33/11KV tap changers up a bit.


If DC is generated in house and an AC supply is needed then a shunt wound DC motor would drive the exciter. Changes in the loading on the system won’t adversely affect the speed of a shunt motor and as far as I know a flywheel would be used to take out any slight variance.





A transformer can be used to reference the control grids to a grid supply, again the reference has to be shifted to above the mean base line.






Although the exciter appears big in the first drawing it would in fact be quite small, it is after all only supplying the grid voltage.


The AC exciter is connected in a strange way where the star point is connected to the DC negative rail via a dropper resistance that injects a small voltage between the DC exciter rails and the star point. This has the effect of raising the control waveform above the negative. This is to give the cathode grids a positive reference to the anode pool, by altering the output of the dropper resistance the wave form can be moved up and down. The DC voltage injected must equal the control RMS value. If it falls below the cathodes will backfire.





The mercury arc inverter:


The heart of the beast!





Like the MAR drawings only two out of the six arms are shown.

Note the exciter diodes shown would be metal oxide plate diodes. The resistors control the current along with the field excitation inductance.


When the voltage on a grid increases the output of its associated cathode is reduced. So as the exciter waveform varies the output inversely follows. The voltage of the grids is miniscule compared to the voltage it’s controlling.


Output transformer:


The primary windings are 180° opposed windings so the waveform is corrected when it reaches the secondary.



Practical use:


As said if you want a different frequency for what ever reason this was the answer. OK you would also need a MAR to convert you’re supply to DC, followed by the thyratron to convert it to whatever frequency you needed. Even if it’s RF frequency, the thyratron will do it.

Railways used high speed circuit breakers to switch between rectifier and thyratron bulbs to handle regenerative currents.


Valve technology it may be but it will handle power in the MVA range.


New technology, BAH! HumBug!


As a hint, have a look for the first cross channel HVDC link.



OK I couldn’t resist doing a full drawing.

If you get a crick in you’re neck sorry (no I’m not)




Well that didn't work. The drawing should be four times the size. Suppose I was pushing my luck, 1700 X 600 pixels.

View full knowledgebase

Share this post

Link to post
Share on other sites
Sign in to follow this  


Important Information

Terms of UseWe have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.