Valves Seen On Graham Slee's Desk!

Thank you Bruce, but it is really only a case of following procedure - working to the valve manufacturer's data, and the laws of physics.

However, the design is not out of the woods yet...

Many articles on valves warn that cathode followers easily oscillate, but don't explain why. The problem I have now, which the AP didn't make me aware of, is oscillation. A high pitched low-level tone is present on both channels. It is therefore important to use the test gear called ears!

This noise might have been acceptable (being very low level) in the days of tubes, but it isn't acceptable to me.

It is not only cathode followers which can easily oscillate - transistor 'emitter' followers can do exactly the same, but possibly in another frequency band.

Why can this be? Simple! A basic 'follower' is a parasitic Colpitts oscillator. You see so many designs incorporating 'followers' without one jot of consideration for the possibility that it could be oscillating, but the physics says bluntly, it is doing so. It might not be heard in most cases, but that doesn't mean it isn't happening.

Because it can happen, it will happen, and it must be put right - it can't be swept under the carpet.

Being fed by some inductance because all circuits have inductance, that gives you the Colpitts oscillator input inductance. The capacitors can't be seen but they are there and they are real. They are the parasitic capacitances of the grid-cathode junction (base-emitter junction if using a transistor), and the cable capacitance which connects the output to its load completes the picture.

Normally the addition of series resistance prevents oscillation, and this works well with bipolar transistors (sold-state), but the valve insists in carrying on its oscillation. This must be partly due to its extremely high impedance - and also its mass of metal and hence its own internal inductance.

One needs to ask if the effort of tackling all these problems is really worth it??

Sensible level-headed people who by chance could be reading this, would think that valves aren't worth it - but market research suggests the majority of hi-fi buyers are missing some part of that statement (sensible level-headed...).

I also admit to a fascination with the silly old beasties (valves), but mine is based on an admiration of the tenacity (to use Bruce's word) of the engineers of yesteryear - they had no other choice - and as an engineer reading between the lines I think it's safe to say they often used 'suck it and see' techniques to get tubes to kind-of behave.

Perhaps the problem can be cured using that other 'ugly' hi-fi phrase; negative feedback? We will have to see. All the regular cures have so far failed.

If anyone has been looking at Rod Elliot's valve preamp stage like I have, it will be noticed it is DC coupled. You can read about that here: http://www.valvewizard.co.uk/dccf.html.

It might be due to the fact that mine is AC coupled and using grid-leak bias. However, I was of the impression that the DC coupled variety was responsible for a lot more harmonics (distortion), but Rod Elliott's discussion tends to suggest otherwise.

And the question is bound to be asked; is it because I'm using such a low HT voltage? I would like to think not seeing as the 12AU7 data gives curves for it. However, it could very well be the case.

I suppose the story must continue. Seems such a waste of effort to abandon it now - although I do feel like it.

No Bob, I'm a Yorkshireman not a Yorkie

The AP still cannot resolve the residual noise I'm hearing on headphones in the background, however, after applying GOBAL NEGATIVE FEEDBACK (saying it loud for all who detest the very mention), the level has been vastly attenuated. Having said the AP cannot resolve the noise in any of its measurements, it's audio monitor reproduces it.

I have a suspicion it is noise from the 48 volt switched mode power supply and not the oscillation I thought it could be. Apart from the NFB all I did otherwise was to decouple the anode circuit of the cathode follower with a 100nF ceramic capacitor which I suppose will have the effect of attenuating power supply noise. I do a lot of supposing simply because electronics isn't that simple (although in my experience, it is, to some simpletons).

I think a full PSU input filter is in order.

Thanks Barry, but I may already have the preamp circuits. Is one EF86 pentodes and ECC83 twin triode?

I found the oscillation noise and it isn't the cathode follower - it's the 48 volt switching power supply. With some ad-hoc filtering it was greatly attenuated.

With global negative feedback THD distortion was just measured at 0.004% and IMD at 0.012% for the rated output of 775mV rms (0dBu). The voltage gain is 8.12dB (2.5). The load being the 22 kilohm DIN specification.

Out at 10kHz distortion is 0.01% but the residual switching power supply noise is confusing the issue and once suitable filtering is in place it should improve.

S/N is better than 84dB. Upper end of frequency response is somewhere in the region of 120 - 160 kHz (-3dB) at rated 775mV out.

With distortion and noise this low it might not sound like valves at all. It simply doesn't do what valves usually do - it is at solid-state levels.

So, why bother with valves... ? But I just might bother.

This isn't the finished circuit by far, it is an earlier version. I won't be publishing the final circuit or the values for obvious reasons.