Mullard 5-20 PSU

[PJL]

Originally Posted by G8HQP Dave

Ion damage to the cathode tends to happen when the valve is running in temperature-limited condition: cathode cooler than normal, but still hot enough to emit electrons. Under this condition there is no protective space charge around the cathode, as emitted electrons get immediately grabbed by the anode.

You can generally reckon that modern valves will be no better than decent NOS samples, and perhaps worse in some areas such as heater-cathode leakage, grid current, service life and sample variation.

This is getting a bit OT but I would like to understand more about cathode poisoning. If this was the case then there would be current flowing wihtout an anode voltage.

Electrons escape the cathode from thermal excitation. When there is +ve anode potential they move to the anode and bombard it making it hot due to loss ot the acquired kinetic energy. If there is no anode potential, the expelled electons leave a positive charge on the cathode and so fall back to the cathode. This action alone may result in excessive heat. I might do a little test and monitor the heater current with anode current and without.

[GrimJosef]

Originally Posted by PJL

Electrons escape the cathode from thermal excitation ... If there is no anode potential, the expelled electons leave a positive charge on the cathode and so fall back to the cathode. This action alone may result in excessive heat.

I think 'conservation of energy' means this isn't the case. The electrons take some energy away from the cathode when they leave and they bring it back with them when they return. So the cathode temperature will be unaffected by this process whether the electron flow (current) is weak or strong.

Cheers,

GJ

[G8HQP Dave]

Originally Posted by PJL

This is getting a bit OT but I would like to understand more about cathode poisoning. If this was the case then there would be current flowing wihtout an anode voltage.

I don't understand your point.

Cathode poisoning due to ion bombardment occurs when the the cathode is hot enough to emit, and the anode is positive enough to attract the electrons and give them some energy but the cathode is not hot enough to build up an appreciable space charge around it. The electrons shooting off to the anode hit the odd air molecule and ionise it. A positive ion shoots back to the unprotected cathode and hits it, damaging the surface. Under normal valve operation the ions still get generated but most of them can't reach the cathode because the negative space charge is in the way.

So, running a valve with full heater and no HT can cause cathode interface. Running a valve with reduced heater and most or full HT can cause cathode poisoning. Valves last longest when either running normally, completely off, or maybe with reduced heater and significantly reduced HT too (so current is not temperature-limited). Very few 'standby' switches do the correct thing.

[PJL]

Originally Posted by GrimJosef

I think 'conservation of energy' means this isn't the case. The electrons take some energy away from the cathode when they leave and they bring it back with them when they return. So the cathode temperature will be unaffected by this process whether the electron flow (current) is weak or strong.
GJ

Agreed, but my theory is a conducting valve is drawing some energy away from the cathode. The cathode temperature should fall with increasing anode current.

[G8HQP Dave]

You may be right, but the change will probably be too small to measure. The thermal energy of an electron as it leaves the cathode will be somewhere around 0.1-0.2eV. The energy as it hits the anode will be somewhere around 100-200eV - a factor of 1000 greater. Assume both cathode and anode have around 2W heating - the cathode from a 6.3V 0.3A heater, the anode from 10mA current at 200V. So there are enough electrons to deposit 2W at the anode - they will carry 2mW away from the cathode. You would need to start by regulating heater voltage to better than 0.1% before you even begin.

So ion bombardment happens when the cathode is emmiting a small amount but not enough to give a space charge, I have two (and three quarters) questions...

Q1 How long is the period (assuming full HT from time zero) that ion bombardment happens.

Q2 How much damage to the valve will this cause.

Q2 3/8 And how good are getters in minimising the stray atoms that give rise to ions.

Q2 3/4 People say using a valve rectifier solves the problem, but surley not to said rectifier.

MM

[G8HQP Dave]

A1. as soon as the cathode is hot enough to emit, but not hot enough to maintain a space charge
A2. depends on how long it stays in this condition, and how gassy the valve is
A2 3/8. very good, but not perfect
A2 3/4. yes, the rectifier can suffer this problem for a few seconds every time you switch on

As you will appreciate, you were asking me about the length of a piece of string. The valve doesn't fail immediately, it might last a long time but not as long as it might without this problem.

Given I am using new valves and not reducing the life of irreplceble ones, I am sticking with the full on at start semiconductor PSU. Thank you all for the input, better the devil you know etc..

The postie delivers a CD today (Dvorak, New world symphony) I will wait until I have finished the amp to give it a whirl.

[julie_m]

It's only ever going to be a concern during warm-up (during cool-down, the reservoir capacitor will discharge long before the cathodes cool sufficiently), and one rectifier is cheaper to replace than two output pentodes. With a true directly-heated rectifier (aot an indirectly-heated rectifier with one heater lead commoned with the cathode), the warm-up is even quicker (but this might then stress the other valves).

I would expect that if the vacuum was poor enough for there to be enough gas to make ionisation a real problem, the valve would be noticeably noisy; as the ionisation process would surely disrupt the electron stream. Force = mass * acceleration, and the heavy negative ion formed from an electron slamming into a gas atom will travel to the anode much more slowly than a naked electron - it's still only carrying one unit of charge, so the force on it is the same, meaning the acceleration will be less.

Edit: G8HQP Dave pipped me to the post.

It works, and for your delight(?) a picture of the underneath. The design differs a bit from the standard 5-20, based on this http://www.gtlab.net/gtlab2/filemana...alve_amp_1.pdf with Lundahl LL1663/PP transformers for the output and a Canterbury Windings TM019A for the supplies. The three 'sticks' at the top of each output stage are panel mount multiturn pots for bias and AC balance cunningly using the mounting bushes as monitoring points (wander plugs fit).

I won't do a seperate build thread, it will consist of 'I drilled a hole' etc. but I will post a final pic when the pretty front has been made and fitted.

[maninashed]

Well done, interesting use of diecast ally boxes too. Did you use a matched set of 4 EL34s and what have you set the bias at? Are you planning to post your final cct diagram ? Should sound nice !

maninashed

I didn't get matched EL34s because unmatched ones are cheaper, what are they matched to? and as I have seperate bias and an AC balance pot I can match them in situ. Bias set at 30mA for now, it's plenty loud enough to the point where it goes class AB (i.e. the bias monitor volts go up a bit). The circuit is the same as the one in post 30 with seperate bias pots rather than one and balance. Also the pole in the anode of the EF86 is 15k and 47p, suits the LL1663s better.

[julie_m]

Matched pairs of valves are matched against each other; so that not only do they have the same gain but if they share a common cathode resistor, they will pass the same anode current.

Negative feedback makes a circuit tolerant of a wider range of gain. A separate bias supply allows per-valve adjustment of the operating point. So you don't really need matched pairs for your amplifier anyway.

As promised a picture of it looking pretty and two wander plugs monitoring one of the cathode currents.

[Whaam68]

The Amp looks great!

Have you thought of putting a CL90 Thermistor in series between the + on your IEC & the fuseholder....cheap....easy....slow turn on. I've used this on my last two amps with no problems. Farnell have them in stock.

Mike

[qualityten]

Originally Posted by merlinmaxwell

It works, and for your delight(?) a picture of the underneath. ...

I won't do a seperate build thread, it will consist of 'I drilled a hole' etc. but I will post a final pic when the pretty front has been made and fitted.

Congratulations on this successful and attractive build. The main value of a build thread is to help learners (like me) learn what is obvious to the more experienced. So, if you still feel any desire to offer a build thread, I think it would be valued. Even the decisions about layout and hole drilling are actually important.