Blog posts tagged with 'VALVE MANUFACTURE'


The electron gun we described in a previous blog entry was mounted and sealed into the neck of a tube using an automated machine which comprised of a number of fixtures which moved around the carousel and revolved on their own axis.   At the first position, the tube was mounted top downwards in the top part of the fixture and the electron gun  was mounted base downwards in the lower part of the fixture .    In the following photo, you can see the loading arrangement: -


Operation of a lever moved the gun such that it entered the neck of the tube and the fixture then moved to a sequence of stations at which the bulb was rotated in a flame from carefully postioned gas blowpipes playing onto the join area such that the neck and gun were gradually fused together this leaving the gun seated into the tube with the pumping tube projecting downwards from the end of the neck - "sealing in" was thus completed.  

At this stage, the glass was allowed to cool slowly to provent heat induced stresses being left in the glass which had the potential to weaken the glass and hence pose an implosion risk at pumping - more on pumping in the next blog entry.


In this process, where sealed valves were passed from the last station of the sealing machine, a mechanical tri-claw hand fitted with an asbestos glove (imagine the 'elf n safety implications today!!!) transfered the valve to the first vacant position on the pumping machine.

The pumping machine comprised of another rotating table which had a sequence of rubber bungs around the periphery.  Each bung was connected to a vaccuum manifold supplying a high torr vaccuum.  As the valve abutted a bung, the process of pumping commenced.  As the valve rotated around the table, the valve passed through a heated convection tunnel where the high temperature assisted in driving off air and other gaseous contaminants.  

The valves then passed through loops of copper tube carrying radio - frequency (RF) current.  The RF served three purposes by raising internal temperatures, firstly,  to induce RF eddy currents in the metallic parts of the sealed valve which drove off any interstitially occluded gas; secondly, to sublime the getter material such that an active flashing coated the internal envelope; and finally to convert the filament barium and strontium carbonate coating to form the emissive oxide - all this going on whilst the pumping still continued!

Finally, when a satisfactory vaccuum was pulled, the pumping sprue was cinched and the valve finally sealed off using a fine glass flame played at a point close to where the pumping sprue meets the envelope top, accordingly,  the glass fuses with the characteristic pip at the envelope top being left.   A second tri-claw hand then removed the pumped valve from the pumping table.

In the photo below, we see Roger Beardshall at one of the Blackburn pumping machines: -




Today, we shall look at valve sealing.  First off, the cage-base assembly was inserted into the bottom end of the envelope and then mounted vertically onto a rotating table on the sealing machine.  At the first station on the table, the location of the electrode assembly with respect to the bottom of the valve was automatically adjusted. The aligned assembly then passed through a four stage sealing process where gas blow pipes played upon the the walls of the envelope at a point exactly opposite the rim of the glass base.  Gradually, the glass softened and hence the envelope and base were fused together and with gentle pressure from forming guides, the join was shaped and minor irregularities were smoothed out.

The final product of the sealing process is shown in the photo above as '3'.


Electrode assemblies returned from feeder to parent factory were complete except for the getter and in the case of battery valves, the filament.  In both cases, these components were added at the parent factory, immediately prior to the assemblies being sealed into the envelope and the valve being pumped.   This was to avoid contamination of these labile components which were vacuum stored until useage.

Attachment of the getter is a simple process where a strip of metal ribbon coated with getter material is welded to the stirrup wire at the top of the electrode assembly.   

Insertion of the battery filament however, is a very delicate process, requiring a keen eye and superb hand-eye co-ordination as the filament is intoduced through the top of the assembly.  The filament tab, forms the lower connection between filament and pin.  The spring at the top of the filament is attached to the appropriate conductor to form the second connection,  A complete assembly is shown below as '2' shown with it's attendant envelope having a pumping manifold sprue shown as '1'.



Now we come to the bit where the valve base and the electrode cage are married together.  It was vitally important that the cage and base were in perfect alignment otherwise the assembly would not fit centrally within the glass envelope and hence a perfect seal betwen base and envelope would be difficult if not impossible to achieve.

Correct alignment was ensured at Mullard by making three key welded connections whilst the base and cage were held in a precision jig as shown in the picture below: - 

As you can see, these jigs were beautiful precision pieces wrought by extremely skillful toolmakers.   Taking the jig, an operator would locate the cage in the perspex cradle which slides on two cylindrical guides.  When the base and cage were mounted, the cradle carrying the cage was moved such that an anode and two screen tags were correctly aligned with the corresponding base support wires.  The two were joined using a spot welding machine which used two pointed copper alloy electrodes and a heavy electrical current to effect the weld.

Once these three key welds were made, sufficient structural rigidity was provided to remove the partially welded assembly from the jig in order to access and make the remaining welded connections to the control grid, screen grid and remaining anode tag.  

I think you are by now getting the message - these valves truly were hand made by a very labour intensive process using specialised custom made equipment expertly wielded by very skilled operators having supreme dexterity.   So, I would exhort you all to revel in the enjoyment that these difficult to make, enigmatic, thermionic devices make to your audiophillic pleasure now you're finding out how much of a kerfuffle making them entailed!

Once all welded junctions were made, the assemblies were placed in metal dust proof caddies to be returned to the parent factory for the next stage in valve assembly...... to be continued..............................


With the electrode cage sorted, the next stage was to prepare the valve base.  In the valve build described - the DF91 - this has a B7G glass base, the manufacture of which was the subject of an earlier blog entry where we left the valve base as a glass button having seven composite wires, each comprising of a valve pin, a seal and an electrode support wire, all very nice, but an electrode cage cannot be mounted on the valve base as it stands.

The valve base prep team had to first prepare the base using a special machine that both cuts the electrode support wires to correct length and then bends them to the exact position to enable mating with the corresponding connection from the electrode cage.

In the photo below, on the left, you can see a valve base as supplied and then to the right after electrode trimming and bending prior to mounting the electrode cage onto the base: - 




We last left valve assembly where we had formed the electrode cage, today, we will describe what happened next in the valve assembly process.  Before the electrode cage could be mounted on the valve base, four additional components needed to be added and welded into position.  The first two components were the top and bottom screen plates and projecting tags which are welded through onto tags on the outer screen as per the photo below: - 

The third extra component is an L shaped piece of metal which is first welded to the bottom screen plate and then bent over and welded to one of the suppressor grid support rods, finally, the last piece, the getter frame is welded to the top screen plate as shown in the photo below: -

Whew, and now you can see that after all of this dexterous handiwork, the elctrode cage is finally assmbled, ready for the next stage in valve assembly.... and my next blog entry.


Previously, we looked at wire production and although this is used for numerous valve componentry preparation, the biggest use by length has to be filaments so, today, we are going to look at filament preparation.

Tungsten wire is the base substrate for filament wire with some wires being just 8 microns thick or 3/10000 of an inch or 1/10 the thickness of a human hair.  The wire is then coated with a admixture of strontium and barium carbonate using the process of cathetophoresis where an electrostatic field induced into the wire attracts colloidal carbonates suspended in a polar carrier solvent.  The 'wet' wire is then dried in a tunnel furnace leaving a carbonate deposit.

Once the wire is coated it is 'filament tabbed' where the filament is supported at one end whilst having a nickel tab attached at one end and then a shorter tab and spring wire then fitted to the first end.  In the next two photos taken from Mulalrd standard operating procedures you see the process flow and finished tabbed filament: -

In the filament tabber, the coated wire is passed through the machine where after a set length has unravelled, a small hammer breaks the coating after which a compressed air jet blows away the unwanted portion and a pressure pad then does a final clean.  At the next machine station, an 8mm wide but 8 micron thick nickel tape is wound around the cleaned area of the wire then spot welded onto it before finally being clipped to the required length.

The continuous length of tabbed wire is sliced into separate filaments by a tool which cuts the tabs at a point 1mm from the end at which the spring wire is welded after which a mechanical grab places the completed filaments on a strip of belted cardboard from which they can be taken and inserted into the electrode cage build.

The long tab at one end of the filament is the connection that will be welded to the bottom filament support with the shorter tab being the connection that will be welded onto the upper filament support thus keeping the filament rigid and aligned.


Yesterday, we looked at what mica actually is, today we will chat in detail about the mica discs used in valve manufacture.     As you look at the electrode cage of most valves, you will see that the individual support rods and elements of the cage are supported in their relative positions by two or more discs of mica.  These discs are shaped and pierced by a series of precise holes which act as location points for the components to be supported.    The function of a mica disc is threefold; to insulate the electrodes from each other; to maintain correct electrode spacing; to hold the electrodes rigid within the envelope.  Below you can see examples of a typical mica in this quaint Mullard photograph:-

At Mullard, the mica washers as supplied from India or Canada are stamped to shape and punched with the necessary holes to produce the mica disc.  Very close tolerences of these holes is a must and hence samples of each batch of mica discs were inspected using a Shadowgraph - a magnifying projector whose screen is fitted either with a comparator overlay or with a measuring vernier.  The magnification used was 30 times and the tolerence demanded was +/- 0.05mm.  The acceptance criteria was >95% pass.   If this specification was not met then the whole batch was rejected.   The next stage of inspection was a 100% visual inspection where operators would reject any broken, bent or mis-shaped mica discs.

Post inspection, the discs were coated with an aqueous solution of Magnesium Oxide - once dried this forms a white powder layer on the mica disc which has very high insulating properties and served to break up the smooth mica surface increasing potential leakage path lengths and repel getter deposition as a valve ages.

Finally, the mica discs are placed in glass flasks which are evacuated and sealed until they were required for use by the valve assembly department.   Here we see Dilys Dyke studiously sealing mica discs into evacuated bulbs for storage at Mullard Blackburn: -




In the preceding two blog entries, I have discussed how tungsten wire was produced by Mullard's - incidentally, exactly the same process was used to make molybdenum wire too.   Both of these wire types were used to manufacture the grids we see in multi electrode cages.   As you can no doubt imagine, the performance of a valve is dependant upon extreme accuracy in the winding and positioning of these grids so let's have a look at what's involved in doing so.

Let's consider the Mullard EF86, a B9A pentode used for many applications from TV IF strips to a floating paraphrase phase splitter in say a Quad II amplifier.  The control grid in an EF86 comprises of two copper wires 20mm in length and 0.75mm diameter spaced 5mm apart which serve to form a rigid support for the grid spiral.  Onto this support, molybdenum wire of 0.05mm diameter is wound over the copper wires with each turn spaced 0.125mm apart.   That's what I call precision, so, how was it done.....

First, here's a nice picture of a Mullard grid winding machine: -

As you can see, the grid winder is very similar in appearance to a typical engineering lathe.  This machine produced grids in continuous lengths of 120cm which were then cut into individual pieces.    At the outer end of the machine was a bracket carrying two reels of support wire which were parallel drawn into the machine via a travelling tailstock.  The rotating machine head was fitted with a reel of grid wire and two wheels - sharp edged and flat rimmed -  that rotate with the head.

In operation, the sharp edged wheel cuts a series of notches in each grid support wire, the grid wire is wound into these notches and finally, the flat edged wheel presses over the edge of the cut support wire thus cinching the grid wires in place. When each continuous length of grid had been wound, it was chemically cleaned then cut into individual grids and 100% visually inspected.  In the picture below, you can see Agnes Shufflebottom diligently inspecting grids for winding defects: -

If any grids are found to be mis-spaced, they can be delicately adjusted with a special gauge which can correct overall dimensional defects, here you see the hand of Agnes' cousin Euphemia, correcting a wayward grid: -