Add a test for microphony to your valve tester

[radiomobile]

From time to time one meets a valve which tests fine but proves to be excessively microphonic or generates crackles or fizzing noises etc.
A test for this can be added to a valve tester if it uses smoothed dc for testing a valve. Testers such as the Sussex or other home-brew types typically but sadly not AVO or Taylor or other makes which feed the valve on test with ac

To add this facility simply connect a small valve output transformer, wired "back to front" ie 3 ohm secondary in series with the anode of the valve under test and connect the primary to a small transistor amp (typically around £3 from eBay) and speaker. The amp should have a volume control. The small output transformer should be fitted well away from any mains transformers to avoid induced hum. Tapping and wiggling the valve while on test can show up any potential problems. This has proved very effective in my own valve tester.

Lester Moon

[G6Tanuki]

A fascinating idea and implementation.

In times-past I have had microphony-issues in the local-oscillators of military comms-receivers that used the likes of the 1T4/1R5 1.4V-heater battery-valves. Specially when there's a Vibrator-PSU sitting only a couple of inches of mechanical-coupling away. a 100-200Hz wobble on the local-oscillator is really annoying when listening to SSB/CW transmissions.

I've never been able to source any sane documents on the resistance-to-microphony or intrinsic self-mechanical-resonant frequency of these little valves. I solved the problem by replacing the vibrator-PSU with a modern SMPS.

[radiomobile]

A friend who worked for Mullard told me that when the B7G DK91 series valves were first introduced the microphony proved so troublesome that radio shops demanded radios with the old octal DK32 range. Mullard obviously sorted the problem out quickly.

[GrimJosef]

Microphony is an important issue but I can, perhaps, understand why valve tester manufacturers didn't include it as one of the standard tests in their machines (besides the fact that, as you say, most traditional testers put AC through the valves).

The basic problem must be "How can a meaningful number be assigned to the degree of microphony in a valve ?". Without a number it wouldn't be possible to say whether a given valve passed or failed the test.

Your anecdote about the DK valves is interesting. I'm not clear, though, whether the problem with the DK91 was that all of them were more microphonic than the DK32s, so the basic design needed changing, or was it that a fraction of the DK91s were fine but the rest were too microphonic to be usable ? If the latter then your tester would be handy for picking out quiet ones and using only those in sets where low-microphony was needed.

I've heard it said that one of the reasons for Mullard changing the design of the ECC83 from the early 'long plate' metalwork to the later types with shorter anodes was to reduce the microphony. This sounds plausible as, in general, larger structures are 'floppier' than smaller ones and their resonant frequencies are lower, bringing the problem into a part of the audio spectrum which it's more difficult to suppress with vibration isolators. Sadly some members of the guitar-playing community actually liked the microphonic nature of the earlier '83s and the long-plate ones can now fetch a premium price if only the seller can locate a buyer who likes their 'sound'.

Cheers,

GJ

[radiomobile]

The test is not just for microphony but for other spurious noises, crackling etc. which do not necessarily show up on normal valve testers. A badly soldered lead-out wire on octal (and earlier) valve bases for instance is more readily picked up by this addition to a valve tester.

[Craig Sawyers]

Mullard papers on studying microphony in valves, in 1962

https://www.pearl-hifi.com/06_Lit_Ar...llard_Rsch.pdf

Craig

[radiomobile]

A very comprehensive study indeed.

On a lighter note I distinctly remember a picture in a magazine, Wireless World I think, in the late 1950's of a manufacturers valve test rig which showed a small rubber hanging from a thin chain which was used to test microphony!

[Robert Gribnau]

Like it says in a footnote on page 1 of the Mullard paper, Part 1 of this paper is an abridgement of "Microphony in Electron Tubes" as published in "Philips Technical Review", 1960.

See page 71 and further:

https://worldradiohistory.com/Archiv...eview-1960.pdf

[Craig Sawyers]

Well found!

Craig

[FrankB]

G6Tanuki- The 100 cycles is what most of the U.S. vibrators operate at. Supposedly that is filtered out. Not always though.

[trobbins]

It is fairly easy to use modern day spectrum analyser software and a soundcard to make a shock assessment of microphony output from a valve. Perhaps the most difficult aspect is using a repeatable 'clanger', as the other test conditions can be made repeatably robust. Software, like REW, can 'capture' the peak of the spectrum amplitude for a clang event from any practical circuit that operates the valve as a typical audio gain stage. A specific valve type will typically have a spectrum signature of peaks.

As the Philip's authors found, even samples from the same manufactured batch of a valve type can exhibit a wide 20dB level variation in microphony levels, so testing and sorting could be a benefit to some.

Ciao, Tim

[chriswood1900]

It is worth noting several of the US testers like Hickock and Jackson included a socket to listen for noisy tubes usually connected to Aerial of an AM radio, tap the tube and you should hear extra hash or so they claim..

[CambridgeWorks]

Valves that I have here worthy of such microphony tests include the sometimes notorious EF86 and the ECC83. I have a 5-10 amplifier to test them in.
However, anyone considering selling such valves need to be aware that microphony within a valve is a difficult thing to establish just how good or bad it is for a prospective purchaser to judge.
Even when you give exacting figures obtained by using digital meters for mA/V, some buyers will disregard such descriptions as they much prefer terms like Strong, or Very Strong, which are totally meaningless.

Unless wishing to sell-on valves, testing for microphony is not really required, as for personal use, a flick of a finger nail or tap with a biro will usually identify any troublesome valves.
A true microphony test that is repeatable by a buyer would need calibrated equipment and repeatable methods, far beyond the capabilities of enthusiasts such as us.
Rob

[radiomobile]

My own valve tester works on mains or from a built in 12 volt 17vA battery, and proved very useful at the last NVCF meeting (some members of the forum may remember seeing it) when I sold a couple of valves and bought some as well. It saved me from a potentially expensive mistake and reassured buyers too.

[G6Tanuki]

Originally Posted by FrankB

G6Tanuki- The 100 cycles is what most of the U.S. vibrators operate at. Supposedly that is filtered out. Not always though.

This issue was mechanically conducted vibration wobbulating the local oscillator. Which meant listening to SSB or CW was a horrible experience.

I looked at the LO on a speccy, it was producing a 'carrier' about 400Hz wide on 14MHz.

Testing for microphony is frequency dependent!!

[Craig Sawyers]

Tektronix used to age valves on an on/off schedule for typically 70 hours. Then check for grid current, and carry out a vibration test.

https://w140.com/tekwiki/images/1/14..._procedure.pdf

From https://w140.com/tekwiki/wiki/Vacuum_tubes

Has all this sort of data.

Craig