Blog posts tagged with 'VALVE TESTING'


Valves were from time to time, returned to either the main Mullard service department or one of the regional service centres.   Reasons for return were varied and included,   defective new product and devices that had failed during their guaranteed life. Today's blog entry will describe what happened to these naughty devices when received at the service centre.

On receipt, the suspect device would be unpacked and listed in a return register attached to a numbered report form on which the results of any tests carried out would be recorded.  At this point, the report contained details of the valve type, manufacturers coding, parcel condition on receipt and result of a subjective visual inspection of the valve.

The device and report would then be passed to a service bench where a range of static tests would  have been made using a test board.  These tests were carried out in a definitive sequence and should a test device fail at any point, the sequence was discontinued.  The test schedule comprised of the following tests: -

The first static test was always a filament check where both continuity and filament current draw were checked.  

The second test was for insulation resistance where the cathode was energised to a positive potential and measurements of leakage current to other electrodes was measured whilst gently tapping the envelope with a rubber hammer to reveal possible intermittent faults.  

The third test was a vacuum test where the valve under normal operating conditions was connected to a microammeter in series with the grid to measure any reverse grid current and hence check for a 'soft' valve.  

The fourth test was an estimation of cathode emission where the valve was effectively configured as a diode to form a composite anode with an AC supply being applied to the anode and the total rectifed current hence being measured and compared to specification.  

The fifth and final static test was a check of the anode current-grid volts characteristic curve  at several different grid voltages.

If all static tests passed specification then the test device would be passed to another test area and operated for a period of 15 minutes in a suitable receiver - typically a Mullard or Philips receiver.  This allowed a direct practical check of device behaviour in an actual circuit and additionally allowed a subjective assessment of any noise and microphony.

If necessary, the final test stage would involve valve disssasembly for further visual and microscopic examination, the results of which were recorded on the test report.

At the conclusion of testing, the report would be submitted to a claims department who decided, based on report contents whether or not the valve could be replaced under the terms of the guarantee 

Today's photo presented below shows Doreen Snailpen operating a valve test board at the Mullard Waddon Service Department:-





Valves as we know, have many different functions and for each application there are certain parameters which ensure the best use in these applications, for example: -

For an audio amplifier - knowledge of the maximal power output and distortion of an output valve for various levels of signal input under variable anode load resistance and grid load resistance is important.

For a radio receiver - knowledge of the cross-modulation factor allows effective choice of frequency changer valves and IF amplifier valves so that effective AGC circuits may be designed.

In order to measure these parameters and others such as equivalent noise resistance, hum level, conversion conductance, power output and harmonic distortion, the Mullard Valve Measurement & Applications Laboratory designed and built in house a range of test rigs which would test these parameters  The design of each of these instruments was particularly elegant in that circuit components and input voltages were infinately variable and easily switchable and that obtained test results could be directly read from meters and indicators which were calibrated in the units required without having to  resort to complex calculations.

The results of these basic dynamic measurements were used at key stages in the history of a valve type.  During development and pre-production trials, samples were tested to ensure the performance of a new valve met the required specification.   In production, similar measurements provided information for publication in valve data manuals and application reports.  When the valve was an established production line, checks were made on sequential valve batches to ensure that consistency of performance was being maintained.

In today's Mullard archive picture shown below, you can see a Mullard Valve Applications Laboratory physicist in the foreground measuring Equivalent Noise Resistance, on his right, a colleague is working the test rig for measuring Power Output and Distortion: -



As well as the production testing we discussed in a recent blog entry, Mullard employed Quality Control (QC) testing where a sample of ten valves were taken at regular intervals  and subjected to a range of tests.  These tests differed from the production tests which were go/no-go type in that the QC tests were quantitative and the results were recorded using a series of Shewart Charts to individually log test characteristics both within a single batch and also inter batch.

The Shewart Chart is a time related plot which allowed trends in measurement to be monitored.  The chart would have on it's y (vertical) axis two sets of bars - the first, the warning limit, the second, the specification limit.    Corrective action would be taken if measurements approached the warning limit before the specification limit was exceeded and the valve product had to be scrapped.

Although some of the QC tests followed the 'Test 1' protocol, additional tests were also carried out.  For indirectly heated valves, there was a test on heating time and for battery valves, a series of tests to indicate performance when HT & LT batteries run down.  Inter-electrode capacitances were measured for all HF valves along with tests for input damping and equivalent noise resistance.  For voltage amplifying valves (ie - audio ) measurement of gain was undertaken.  For output valves, total harmonic distortion for a given output was measured and microphony tests were made when considered appropriate. Frequency changers were given a custom sequence of tests to measure conversion conductance and oscillator performance.

As you can imagine, this barrage of tests required custom built equipment which was designed and constructed in house and below, you can see a photo of a Mullard Blackburn microphony test station giving an EL37 a good seeing to: -



Another QC test was the Life Test where test valves were over - run for a 600 hour test and subjected to the 'Test 1' protocol at intervals throughout the test period.  

The results of all of these various QC tests were produced at a weekly 'round table' where representatives from the Mullard departments of Production, Laboratories, Development and Technical Service could discuss the findings and as necessary recommend and implement minor modificatons in construction and manufacturing methodology to improve performance and hence correct undesirable trends.


The Mullard valve factories had numerous test schedules for valves produced but today, we are going to look at the production test that EVERY valve that went into stock received which was known colloquially throughout the Mullard organisation at ' Test I'.     

It is interesting to note that, with the exception of the knock test, the test schedule is identical to that used by Mullard Magic before any of our valves go on sale.

The Test I schedule was conducted by two girls, however, by mid 1953, for some popular valve types, Test I was automated meaning that approximately 1500 valves per hour could be tested in this way by a single girl and this is what was done: -

KNOCK TEST - in this test,  a valve was supplied with appropriate voltages whilst seated in a standard valve base and then  treated to several sharp raps with a rubber hammer by the operator.  A milli-ammeter in the anode circuit would indicate faults such as low or no emission, open circuits, short circuit or intermittent operation.  An IF component abstracted from the anode current was amplified, rectified and passed to a loudspeaker and neon network where variations in anode current would cause characteristic sounds or indication flashes.

Having survived the 'Ben Nock' test, the valves under test were passed to another test bench where they were seated in banks of valve bases and preheated before being subjected to additional quantitative tests. which included:.................  

INSULATION RESISTANCE  - in this test,  with the cathode maintained at a positive potential, measurements of leakage between the cathode and other electrodes were made under specified voltage conditions.  For indirectly heated valves, the heater to cathode insulation was also checked.

EMISSION - in this test, all electrodes excepting the cathode were strapped to form essentially a diode and an AC supply at a specified voltage was applied and the total rectified current measured.

MUTUAL CONDUCTANCE - in this test, the anode voltage was maintained at a specified value and three different values of grid bias were aplied to the control grid and the corresponding anode current values were noted and checked against the slope of the theoretical V- Ia curve.

VACUUM - in this test, the valve was operated at nominal working paraeters and the vacuum was tested by measuring the reverse grid current or 'gas' flow using a series connected micro-ammeter in the grid circuit.

VISUAL INSPECTION - in this test, each valve was inspected under a 4x bench magnifier in daylight balanced (3200K) light for general cleanliness and internal appearance.

In the following photograph , you can see a valve test station at Mullard Backburn, on the left station is the 'knock test' board, the centre station, the characteristics test board and at the right station, the microphony test board: -



After passing the knock test, valves were passed to the main test bench and plugged into a test board where a specially designed  eight position switch  applied successively, appropriate voltages such that values were obtained for the following parameters: -

Heater current

Cathode current

Control grid current

Screen current

Anode current

Total emission

Mutual conductance ( at 4 - 7 points on the characteristic curve dependant on valve type)

Inter electrode insulation for all electrode combinations

Only those valves whose performance fell within specified limits were then passed into "Factory Stock."

 In the photograph below we see Enid Barribal testing valves using one of the Mullard Blackburn 'Test 1' main test boards: -


Mullard Blackburn also had a semi-automated final test board where valves were plugged into sockets mounted on a revolving platter.  As each valve passed from station to station, the various tests were automatically applied but the indicating instruments on the test board were replaced by sensitive relays which would energise if a faulty valve was detected hence activating an automatic claw to pull and dump the faulty valve in a reject tray.


After ageing, the valves were ready for a series of intensive tests during which any valve which failed specification was rejected.  The first of these tests and the first hurdle at which the valve could be failed was known as the "Knock Test."

To conduct a Knock Test, the valve was plugged into a test board and the appropriate voltages were applied to all electrodes. The valve was then sharply tapped several times with a rubber hammer and any hidden faults such as intermittent short circuits or loose connections were indicated by readings on a series of meters or the lighting of neon lamp indicators.  The object of this test was to weed out any valves which were hopelessly defective and which had the potential to damage the very delicate instruments utilised for the final tests which we will be discussing in my next blog entry.



And so the story continues.......... a couple of blog entries ago, we left the production process at the sealing stage at which point we had a sealed and functioning valve, however, further processing was required before it was suitable for use for it was required to undergo a process called ageing.

Ageing consisted of operating the valve under load under carefully controlled conditions.  Large numbers of valves were plugged into a socket rack like the one in the photo below: -

Once the rack was fully loaded, the valve heaters were activated at a 20% higher than nominal rating, so for a 6.3V valve, a Vf of 7.5V would be applied.  The control grid would have a positive potential applied and both these effects served to 'boil off' electrons from the emissive cathode which in turn caused a small proportion of the barium and strontium oxides to be reduced to metallic barium and strontium.  This was a desirable characteristic as the presence of these transition metals served to increase the cathode's emissive properties.   After a prescribed time, the grid potential was reduced to zero, cutting off electron flow and the ageing was completed by running the valve at a low cathode current and with the Vf reduced to it's nominal rating.

Across the water, at Philips Herleen, they also aged produced valves but utilised a slightly more modern and certainly more ergonomic valve ageing rack as you can see from the following photo Bart Fokkink is tending an active rack: -