Davy's KEL84 Grid Voltages

[Richard]

Hi Davy, All,

This amp is built fine, works to spec and sounds great. Before switch-on I checked round and re-did chassis earth which was stood off 2 ohms by paint and tidied a few signal wires; in partic the unscreened end of the R input which was lying alongside the fb wire at the pcb end. Davy's not sure if it's always been like that but the amp hasn't misbehaved at all in many hours testing and that could certainly account for positive feedback. So, nothing exciting but it is often something simple isn't it.

He reminded me about the EL84 grid voltages and I said I'd check them out. I'd checked all voltages at switch on, they were all close to spec, and tested the valves which were all fine and paired up.

Sure enough after 1 min all grids are 0V but,

after 5 mins they varied .01V to .48V
after 20mins they were .01V to .83V

Thereafter they stabilised and even dropped back a little when the mains dropped. Cathode voltages stayed the same at 11V even on the 2 worse valves showing .47V and .83V on their grids.

So, removed the EL84s and checked gridstopper and gridleaks were correct values and they were. Switched on with the valves out and checked for DC on the grids but there was none, just flickering mV which would account for nothing so the coupling caps (plastic) are fine.

Replaced the valves back in the the same pairings but opposite channels of the amp. Lo, and those exact same grid voltages followed the valves.

Researched gridleak values and found KEL84 uses same 470K as Leak used. Mullard used 820K in their 10W amp. EL84 data sheet says max 1M with cathode bias so gridleak value doesn't seem to be the issue.

Google search found one ref to a tired EL84 showing 1.5V on it's grid. Conclusion then is that this might be an ageing thing with these valves.

It does not affect the bias voltage at the cathode but I'm not sure about the actual bias. The voltage showing on the grid must be created across the gridleak and so will be a minute current through the grid of .83/470K = .0017mA. If it is the ageing routine for these valves and that positive grid voltage continues to rise then the valve will gradually pass more current till it self-destructs.

I don't know how many hours are on these valves (Davy?) but I'd check grid voltages in a few weeks time. (Valve marked 4 shows .83V and 6 shows .47V but interestingly these 2 don't draw the most current on test.)

Does anyone have exp or further info?

Rich

[David Mccallum]

Hi Richard,
I see their's no takers on the grid voltages yet,i'ts probably run for about 350hrs or so on the el84's and much less on the ecf80's maybe around 50-60 hrs as i;ve been trying a couple of alternatives to no avail.Hope this is of some help,thank's again.

[Richard]

Hi Davy

EL84 is a powerful valve for its size and that power is used pretty much to the full in the KEL84 design. 2000 hours would be a good life span here I think. Your small signal valves may last 10 times the output valves. Small pentodes tend to go noisey but small triodes can last even longer.

As to the EL84 grid current, I'm really not sure if it is an indication of ageing valves or whether some flow is acceptable with otherwise good valves. It would be helpful if you were to let us know the voltages again after a few hundred more hours use.

Get in the habit of checking the plates in a darkish room occasionally, a few minutes after switch on, for any redness and you won't get caught out by one about to self-destruct.

Can we extend their life I wonder by swapping the gridstoppers to 220K and the coupling caps to .47uF? I suppose we need to confirm the ageing process first.

Rich

[NickG]

Can we extend their life I wonder by swapping the gridstoppers to 220K and the coupling caps to .47uF? I suppose we need to confirm the ageing process first.

I would think that would just hide the problem, but not reduce the actual amount of grid current. I think that if the valve gets to the point where the grid current is enough to case runaway, its going to happen quickly, and would happen with any sensible value of gridleak. I think Mullard would have specified the gridleak value for a valve running in specification, not to cover one that was outside spec.

I can't remember for certain, but I think that one of the causes of grid current is a soft valve.

I guess as you have said, all that can be done is to monitor the valve (or play safe and get another set).

[Richard]

I would think that would just hide the problem, but not reduce the actual amount of grid current.

Hiding the problem may also solve or slow the problem as lowering the bias voltage created would cause the valve to pass less current and increase the margin to runaway.

What we don't know are;

a) if this is indeed a sign of ageing and if grid current will increase at all and

b) what level of grid current the valve can sustain.

If a lower value gridstopper saves the grid from destructing then its life may be prolonged till the thing fades away "normally" with low emmission.

The valves themselves look new with no signs of overheating and full getter silvering. They are decent brand JJ test fine, and the amp works perfectly to spec.

Rich

[NickG]

Hiding the problem may also solve or slow the problem as lowering the bias voltage created would cause the valve to pass less current and increase the margin to runaway.

Maybe, but I can't help thinking (I don't know for cerstain with any of this) that the cathode degeneration will be able to cope with the little extra grid voltage, and reducing the resistor is only going to half it, not that big a difference.

My thought is that either 1. The valve is fine, they do this, and we haven't noticed it before, or 2. its a problem with the valve, and it will progress to the point where it gets fatal. I just think that in both cases a small change will not have any effect. If its ok, it will continue to be ok, if its faulty, that will eventually fail, whatever we do.

I did a bit of digging, and the only reference I ca find is a set of numbers for the valve, where the quoted distortion is a 0.3uv of g1 current, I assume this is caused by nearing the 0v line.

if the valves are to remain, I would just think that a check every couple of months for cathode voltage, and replace the valves if there is any sign of a drift.

But what do I know, I build amps where the anodes glow anyway :-)