Cold cathode, gas filled regulators for amplifier screen/preamp voltage rail dropping

[trobbins]

MOV's G.R.Woodville presented a KT66 PP audio amp in WW Dec 1948 that used a Cossor neon S130 regulator to drop 500V output stage supply to circa 385V for the KT66 screens and preamp circuitry. https://worldradiohistory.com/UK/Wir...ld-1948-12.pdf

That circuit configuration is failsafe when the regulator stops conducting (given that the screen voltage collapses), and if the regulator ages and requires a higher striking voltage (given the high B+ available).

Has anyone experienced other failure modes of typical gas regulators - and in particular if they could somehow go low resistance?

I'm just preparing to restore a PA amp that uses an OC3 for the same function for a 6L6 PP output stage with 415V B+. This is not a hi-fi amp where screen voltage regulation is important, but rather just a means to easily provide suitable voltage rail levels.

I also enjoyed Alex's recent article on the intricacies of cold-cathode regulator noise generation, and some improved ways to use such regulators - not something I've seen discussed in that depth before:
http://www.valveradio.net/radio/impr...regulator.html

Ciao, Tim

[kalee20]

Cor blimey, I just had a glance at the article, as I have a lot of gas reference tubes, but I never knew about relaxation oscillation if a few mV in Zener diodes! And I thought I knew everything I needed to know about using Zeners.

I'll read the whole thing later. Every day is a school day... thanks for putting it up!

[Ed_Dinning]

Hi Tim, it is a pretty massive valve, so electrode structures could possibly break and go s/c, similar (but highly unlikely) if there is a massive flashover.

Could also be protected with a voltage sensing circuit similar to those used for bias supplies.

Let me know if you need one and I'll bring it out on my next visit

Ref the relaxation osc, I believe there was an article many years back where this size of gas reg was used in an organ as a tone generator

Ed

[trobbins]

Ed, I've got gas regs in my Selmer organ to provide oscillator frequency stability, and in a Williamson pentode variant amp to provide regulated screen voltage, and have a few spare parts floating around, but yes they often are the large octal type.

Kalee20, I rang up Alex (the author) a few weeks ago after reading that article, and another article he wrote up that I was interested in related to marrying opamp driving of valve output stage audio amps, as I hadn't stumbled across his website prior, and google led me there! Alex is in Sydney, so in more 'lockdown' than me. He has a strong electronics background, so the write-ups were a joy to work through. I should get to play with the gas reg in the amp I'm renovating to see what benefits I can include from Alex's insight.

[Radio Wrangler]

Real Zener diodes only go up to a few volts.

'Zener' diodes above that are actually avalanche diodes. And avalanche diodes are regularly used as noise sources.

In the past, current-limited thermionic diodes controlled by starved heater supply were used as shot noise generators. The noise level being calculated from metering te anode current and knowing the charge on an electron. Microwave sources used discharge tubes pierced through waveguide.

Since then, noise diodes (avalanche jobbies) have taken over.

You can use almost any avalanche diode ( 'zener') of several volts to make some noise, as per the linked article. But the junction capacitance is a bit large for high frequency response... these are intended as power dissipating thingies after all.

So amateurs building their own noise sources for noise measurements of RF stuff came up with the idea of using the reverse-biased b-e junction of a small UHF transistor to get low capacitance. Breakdown voltage is about 7-10v and they need to be current controlled to quite low current or else dopant migration will ruin them as noise diodes. (This operation already ruins them as RF transistors)

Purpose-made noise diodes have boundary rings of heavily doped semiconductor around the junction region to discourage dopant in the region from going walkies. Specialist microwave semiconductor firms made these things. HP had their own semiconductor arm do diodes for the 346 noise source family, though they later farmed them out. NOISE/COM inc also got into the noise diode business and rather kindly would sell individual diodes with tested noise data to amateur constructors.

So, gas discharge regulators are noisy. Higher voltage (fake) zener diodes are noisy. Avalanching transistor junctions are noisy.

Ironically, one of the quieter voltage references in general use is the REAL zener diode! The low voltage sort usually made as a 'buried zener' in a 3-terminal regulator chip, ADC or whatever. They seem to be quieter than bandgap references.

For RF noise measurements, if the budget will stand it and if fundamental thermal noise standards are a bit too much (size and £££) then the HP/Agilent/Keysight N4000 series seem to be the most commonly used standards, taking over from the HP346 family.

For voltge reference use, whatever you choose, you need to look out for slope impedance
versus bias current to get the best performance. Some diodes use an inconveniently large current before their promised accuracy manifests. Look out for hysteresis leading to oscillation (VR150/30, OA2 etc) where you have to design circuit capacitances and impedances to tame them.

If using little neon bulbs, some low voltage ones had a bit of additve in their gas just to help them get going and to lower their striking voltage....radioactive additives and with half-lives passing by the striking voltage goes up. Even without the additives, erosion of electrode surface smooths them and up goes the striking voltage. In a number of classic oscilloscopes, little neons used in power supplies are starting to be troublesome.

In regulators for screen supplies for large RF power tetrodes, shunt regulators are the sort to go for. High perveance tetrodes are rather frisky, and the damned things can easily output screen current.

David

[Chris55000]

Hi!

There is a circuit in the "Radio Laboratory Handbook", 1961 edition:–

https://archive.org/details/in.ernet.dli.2015.16346

. . .for a portable mains noise source based on an 85A2 reference tube capacitor–coupled to a simple three–valve aperiodic amplifier – the actual noise spec. or the suggested uses of this circuit wasn't very clear tho!

In regulators for screen supplies for large RF power tetrodes, shunt regulators are the sort to go for. High perveance tetrodes are rather frisky, and the damned things can easily output screen current.

C.A. Quarrington made reference to this on the chapter 6 on "Testing Valves and Transistors" in the third volume of the "Radio and Television" set, third 1963 edition – it can play havoc in valve–testers and be a cause of all sorts of nonsense in high-power amplifiers, and I'm not entirely certain many of the the modern miniature triode–pentode valves, both voltage–amplifier and power types, are completely free of this either!

Chris Williams

[Oldcodger]

From memories ( circa 40 years ago), cold cathodes "strike" with 70 v on the anode when the grid is pulled at a certain voltage above the cathode.
Something used in the older inner island exchanges of the late 60's. At one time in the late 70's in our TRS, we had a cable pressure tester stepping around a few pressure circuits looking for a circuit with a fault condition. Then some bright spark decided to remove the exchange pulse to our TRS . We had to find a way to replicate this pulse. My design was to use a CR network to strike a CC with the relay used to "strike " the CC "killing" the charge.
of course in lower voltage systems, a Zenner could have been used, but in 70's technology ,the power supplies were higher. And we had to work with higher current things like 2000 type relays .