Hi-Fi 17W Tube Amplifier

If I have a vaccum tube and I'm over powering it at the point the anode gets red hot, what sort of damage it get?

Sooner or later the glass envelope may crack. Also you'll end up stripping the cathode coating prematurely. Usually it takes years to strip it, overloading the tube may wear it out much much faster.

And if I keep on the tube (at the same power level that it would make the anode red) for a limited amount of time so it doesn't heat up that much (and of for long enough so it cools a bit), the cathode coating will strip of the same or a bit less (for each time it's on)?

It's even worse I think. Tubes usually work best when not cycled on/off. This further exacerbates the cathode stripping. I'm not exactly sure of the stripping mechanism, but they work best (and longest) if always kept on, within their operating conditions (not more, not less also), and the HV supply must be introduced, once the filaments have had enough time to start the thermionic emission process. There's a lot of info on the net on extending their lifetime, some of which, I've used to design the automation system.

I see, anyway I can't keep it on constantly even within its operating conditions, that's why i asked. For the filament i read all your writings and I was already taking care of it.

There are many things to consider. Make a project, run it by me, I'll try to help with what I can. Tubes work best at center bias. At low level bias, or when the cathode is heated, but with no current passing, it gets poisoned. Instead of getting stripped, it gets coated with an extra layer of high resistance oxides, which reduces the tube's usability over time. The gain becomes lower, there is more self induced noise etc. The best way to run a tube is as it follows. 1st you preheat the heater filaments. For a pair of EL84's give it a good 15-20s before doing anything else. It's also beneficial to use some sort of soft start, because cold filaments have a very high surge current. I'm currently working on such a system. 2nd After enough preheating time, enable the HV supply. A soft start is good, but not necessary. If you're using a Pi filter (as me), it should already delay the power up to a given extend. 3rd Tubes work best when not cycled. The more time they spend in an ON position, and less time in an ON/OFF cycling position, the better. A little historical fact. The first ENIAC computer, used to cycle the ON/OFF states of the tubes, and they would constantly fail. Keeping them permanently ON, reduced the failure rate from a tube every single day, to a tube or two per month or so. Check out my automation system for the Amplifier, it covers all the things I described above. It enables the filaments first, 20s later the HV supply, and after a considerable time of idling (16min or so), only then it disables the amp. As I said, make a project, run it by me, and I'll try to be helpful as much as I can

Well, thanks for the help. Anyway it would be a power armstrong oscillator (which will drive a tesla coil, what a surprise😂), made with a pl509 (i hope), or a pl504, which will surely get red hot in little time. It will be powered by a microwave oven transformer since it is often used. I wanted to ask you some tips to not destroy the tube in little time since they are quite costly😅.

Contrary to common beliefs, tubes don't go out in a big bang. They usually go out quietly after a long time of abuse. It won't redplate if you drive it within acceptable limits. For instance the power dissipation is 16W and the supply voltage I reckon around 2-3kV. Don't go above 3.5kV, because with an output transformer (the tesla coil itself acts as such) the swing will go twice above it (7kV), which is the maximum for these tubes. Ok, so you have 2kV for instance and 16W. This means you need I=P/V I=16/2000 I=8mA. Don't exceed 8mA! Okay, next you need to calculate the nominal load for such a power rating. R=U/I R=2000/0.008 R=250kOhm. Next, you need to choose a resonance frequency and calculate the coil inductance, that matches a 250kOhm reactance at the given frequency. For instance ~2mH primary inductance matches a reactive load of 250kOhm at a frequency of around 20MHz, which is on the upper end of the tube's capabilities. This is somewhat good because the tube will not be able to amplify any random harmonics. Next choose a valid bias point for the amp to match the 8mA/2kV point. Since the datasheet doesn't go above 300V or so, do it experimentally and very carefully. Set the Vg2 at 200V and start at a Vg1 of 0V. Slowly decrease the Vg1 voltage to to a negative level (use fixed bias), and measure the current at the cathode. When you reach 8mA, fix the Vg1 voltage and continue with your circuit. And now a fair warning ⚠. Keep in mind that at such voltages, the electrons bombarding the anode, have sufficient kinetic strength to start emitting X-ray radiation. There are some types of kenotron tubes, which were exclusively made with a metallic envelope instead of glass, to shield the users from the x ray radiation. This is a veeeery dangerous experiment, one which I'd never ever even consider trying. I suggest you do the same!

I can't go higher than 1MHz, 5-400kHz is actually optimal for spark propagation, also the primary coil cannot be that big, it will almost be as big as the tesla coil itself, which leads to several problems as to high coupling, which leads to high electrical stress in the secondary with the result of a failure on the wire, and a low impedence path for the high voltage to reach the primary and potencially ruin the tube. In general these circuits are supplied with a half wave on 3kV peack value (and very often only for one half cicle per second, ignoring the other 49 or so), because of that i see most of the coilers say that that way you can push more current on the anode without damaging it, another reason why I asked to you about this. For the x-ray thing ¯\_(ツ)_/¯

Then you need some impedance matching. Tubes are high impedance devices, and need to be coupled with high impedance loads. A tesla's primary is typically a low impedance load, as the tesla coil is nothing more than a double tuned step up transformer.

And p.s. Yes you can pump more current through the tube without damaging it, but it mustn't be continuous. The tube must work in class C mode, and the surge currents must be very short in duration. Most datasheets give an absolute maximum pulsed rating.

Indeed most of the papers I read say that the oscillator is a class C Armstrong oscillator were the tube is actually turned on for only 20-30% of the rf cicle

Yes, exactly. For instance a 20% cycle will allow 5 times more current through the tube at peaks (around 40mA) for your tube.

Calculations get a little weird for me for such devices. I have the basics about tubes, but other than that I've only examined tesla coils by proxy

It's more likely to me to use a pl509 or even better a el509(very unlikely).

The E/P prefix is only for the filament voltage. E is common designator for 6.3V filament voltage. P is usually for a 13.3V filament, although PL504 uses around 25V if I recall correctly

Yes, the pl509 needs 40V for the filament while the el509 needs indeed 6.3V, which are more comfortable to use. Another thing, it is better to preheat the filament at a lower voltage before giving the nominal voltage to the filament or this will poison the cathode?

Yes the cathode may become poisoned after time. I've thought about it for my last circuit, but you need to allow some minimal anode current to flow, to keep the cathode from growing an interface resistance. In time the tube will become useless, even though it might have plenty of hours of lifetime left in it. My advice, don't keep the heater partially on, just make a soft start circuit, to limit the initial surge of current.

Yes of course, I was designing the circuit so it would be partially on for a couple of seconds then it would be switched fully on and about 15 to 20 seconds after that the hv power supply will be turned on

No need for such complexities. Just find a thermistor, add it in series to limit the surge current, and it'll slowly bring it up in order. Anyway power tubes rarely fail due to filament burnout. Mostly small preamp tubes fail this way, and again, not very often

Well I still need to supply the hv after the filament warmed up, and for that I only need one opamp and one relay, since opamps usualy cames in pairs and I don't really like losses, I thought to use one opamp for preheating the filament and the other for the delay needed to the filament to be fully hot. I know there are single opamps but I don't have any.

Well I don't know. The circuit is usually more complex than a single op amp. Just drive it manually.