Stereo Amplifier
 

Hi All,
I am getting close to getting my long awaited valve amplifier up and running.
I just had to show it off, as I am so looking forward to hearing it going.
I have used 4 6L6EH as output valves, one ECC83 as a phase splitter and an ECC83 as a pre-amp, per side.
I have just got to finish building the chasse, then wire her up and test......
Ever since I can remember, I have wanted to build a valve amplifier. Now it is looking possible.

Re: Stereo Amplifier
 

That chassis must be huge! I hope it goes well.

Just one thing to bear in mind: an ECC83 might struggle to feed 4x6L6 because beam tetrodes need smaller grid leaks than pentode outputs (I'm not sure why), whereas the '83 prefers a high impedance load for low distortion - 2xEL34 is nearer its limit. ECC81 might be a better choice for the phase splitter. See how it goes.

Re: Stereo Amplifier
 

I quote, as I am also a newbie:-"Because of the effective suppressor action provided by space charge and because of the low current drawn by the screen grid, the beam power valve has the advantages of high power output, high power sensitivity, and high efficiency." I think the ECC83 could work, just check the plate current. Are you using DC on the filaments or AC? I once cured a stubborn hum problem by makeing a coil detector, almost a choke, and then changing the alignment of a transformer. It was a -copy, almost, of the Leak Point One amp.

Re: Stereo Amplifier
 

I'm not familiar with the circuit of this. However, assuming that the output valves are not running into grid current why shouldn't the drivers (ECC83) not be capable of driving them? When I was a lad grids were voltage driven.
Alan

Re: Stereo Amplifier
 

You're right as far as the grid itself is concerned Alan. But it will be held to ground by a grid leak resistor and current will flow through that. The 6L6 data sheets typically recommend that this resistor is never larger than 500k, and if anything other than auto-biasing is used then it should be 100k. In a four-valve amp each half of the phase splitter has to drive two of these resistors in parallel. If that was 250k then things might not be too bad. But if it was 50k then quite a bit of drive current would be needed.

Cheers,

GJ

Re: Stereo Amplifier
 

As I said, for some reason beam tetrodes need smaller grid resistors than genuine pentodes. This seems to be generally known, but I have never seen an explanation. It must be because they have higher grid current or are more prone to thermal runaway but I don't know why this is so.

Valve grids have never been "voltage driven", although this is a myth that even some textbooks repeat. The input resistance of the grid of a healthy valve is high but not infinite. When the grid is more positive than about -1V it collects electrons (yes, that's right - a slightly negative grid can still collect electrons!). This electron current shoots up once the grid goes positive; everybody knows about this aspect of grid current. A negative grid collects positive ions; there are always some of these because no vacuum is perfect. This ionic current gets worse as the valve ages, or gets too hot. The net grid current is the difference between the negative electron current and the positive ionic current, and so the valve itself has a non-linear input resistance. Fortunately the current is small. The upper limit on grid resistor is set by the expected ionic current, which has the effect of making the grid more positive (or less negative) than the designer expected thus leading to more valve current and, eventually, thermal runaway. This is all understood. The mystery is why beam tetrodes have more ionic grid current.

So a 6L6 may have higher grid current than an EL34, and needs a smaller grid resistor. This may be partly why, on American audio websites, people get laughed at for using ECC83 as output drivers - their beam tetrodes need beefier drivers. We Europeans prefer our pentodes and know that the '83 is perfectly OK for driving them.

Re: Stereo Amplifier
 

Yes the chassis is huge it is 500x500mm and weights a tonne. Thank you all for your information I shall go over my calculation again and see if I have made a mistake.
When I finalise my drawing/schematic I will post a copy. I am also trying to come up with a remote control unit, using digital potentiometers and a microcontroller, but I need to get the amplifier working first.
I hope I haven’t over looked something by rushing to get it done, well not at this stage anyway.

Re: Stereo Amplifier
 

4 x 6L6? that's a BIG amp! My new amp is 2 x 807 (very similar to 6L6) in triode mode and it fetches over 25 watts per channel into my speakers ( 93Db/W ) it will never run more than about 5 watts, if that :)

Just a thought and best wishes for success.

Regards
Joe

Re: Stereo Amplifier
 

I hadn't realised that we were splitting hairs quite so finely :)
Perhaps I should have said 'grids were assumed to be voltage driven'.
Of course the grid input impedance is neither purely resistive nor infinite, but the discussion is/was about audio frequencies. I would suggest that the required grid resistor plus anode resistor plus coupling capacitor have a much greater effect on the design process than the miniscule non-linear grid curent taken during normal operation.
Alan

Re: Stereo Amplifier
 

Not sure if the photo you presented is the expected layout, but wouldn't you want to orient all the transformers and chokes 90 degrees from each other? The two transformers on the back are at 45 degrees. Will this make a difference?

Re: Stereo Amplifier
 

Nice job, what transformers are you using?
Rob.

Re: Stereo Amplifier
 

Let's wave some ball park figures around. If the output grid resistor has to be limited to a few 100K then that suggests that the grid current could be in the region of a uA or so, so the bias does not move by more than a few 100mV. Assume that the grid current is all distortion. The anode impedance of an ECC83 is about 60K, but this doubles for a LTP phase splitter. It has the anode resistor in parallel, so assume an output impedance for the PS of 80k. This means a distortion voltage of 80mV. Assume the total grid drive to the output is 20V, so 80mV is 0.4%. This is small, but it is not negligible and it will probably get worse as the valve ages or if it overheats. Two beam tetrodes would double this figure. One output pentode could halve it. As I said, these are only ball park figures but they show that the effect of grid current cannot be ignored if you are trying to do high quality design.

An ECC81 would have an anode impedance of around 12K (from memory) so everything is easier. Maybe 20K output impedance from LTP, so distortion from grid current is only a quarter of before.

Re: Stereo Amplifier
 

A good workman like amplifier, are you using any overall negative feedback?

Re: Stereo Amplifier
 

That's an interesting analysis Dave. I hadn't thought about the problem in terms of noise - I was simply concerned that the ECC83 wouldn't be up to delivering the current needed into (worst case) 50k.

But one of the steps in your argument intrigues me. It's where you say "Assume that the the grid current is all distortion". Personally I would have guessed that it was mostly DC with a much smaller noise component on it (having said that, I know next to nothing about how big the noise component would be). Also wouldn't the noise component see the output impedance of the PS in parallel with the grid leak resistor ? If it was just one output valve then this might be only a small correction to the 80k. But if it was two, each running with a 100k grid leak, then the noise current would be flowing through an effective impedance of just 30k.

Cheers,

GJ

Re: Stereo Amplifier
 

Sorry its a bit Scrappy but I hope it helps. I hope I haven't made any mistakes.

Re: Stereo Amplifier
 

No, not noise but distortion. Grid noise only becomes an issue at the upper end of the HF region. Yes, it would have a DC component so I may be out by a factor of 2 - but I did say ball park. Grid current is quite non-linear, so it is fortunate that in most cases it is small.

You are right to say that the grid leak resistor would load this current and reduce the distortion voltage - I forgot that. Trouble is, this loads the PS even more so there is a tradeoff between triode anode distortion due to loading and grid current distortion. The former is probably smoother than the latter.

Re: Stereo Amplifier
 

As for the transformers, I had them made by variable voltage technology. I sent them my spec and they made them up for me.
The main tx is 310-50-0-310
choke is a 6h 100mA
output tx is 90w rms 20Hz to 20KHz primary Ra-a 3k3 ohms, an a secondary 4 ohm or 8 ohm tapping

Re: Stereo Amplifier
 

The recommended grid resistor for a 6L6 seems to be 0.1M if fixed bias and 0.5M if cathode biased which seems to be the same for a KT66 and similar for an EL34 (although ignored by Leak who used 1M). Cathode bias provides -ve feedback reducing the impact of a high grid voltage. I did read the radiotron bible on this subject and vaguely recall that the causes of grid current were many, grid impurities, secondary emissions, gas etc and so I guess this is an emperical figure. As these effects get worse with increased anode current and temperature, an overly high resistance will result in thermal runaway.

In my experience, noise is not a problem in the output stage. It is the early stage in the amplifier that contributes most.

The issue with the 6L6 is more likely to be the lower gm and higher ra which will necessitate a higher drive voltage. The higher drive voltage will demand lower driver anode resistance to keep the valve operating in a linear region.

The last guitar amp I played with had 4xEL34 fixed bias driven from an ECC83 but of course distortion was probably a good thing!

Of course I could be wrong....!

Re: Stereo Amplifier
 

OK, I hadn't come across grid current distortion before, so I went away and looked it up in Langford-Smith. He describes what happens at the g-k voltage where electrons start to flow from cathode to grid and the positive signal peaks are "damped" (this is not due to net grid current, just to the nonlinear shape of the grid current characteristic at this point). He says that it "... introduces an objectionable form of distortion which is particularly important when the preceding stage has high effective impedance ..." (Chap 12, Sect 2(iv) para (D)). Is this the effect you mean ? If so I'll have to think more carefully about your numbers. But in any case, as Langford-Smith also points out, the problem can be avoided by increasing the grid bias sufficiently to avoid this nonlinear region. This is not quite the same as Alan's proviso back in post #4 "that the output valves are not running into grid current". But it's in much the same spirit I think.

Cheers,

GJ

Edit: I see from the above post that I'm not the only one who reaches for Langford-Smith when the going gets tough :-) !

Re: Stereo Amplifier
 

what do the chokes feed ? just screens ? or whole amp ?
after all the reason I ask is that 4 6L6,s will suck close to 300 mA if you supply the whole amp. if just the screens somewhat less ( maybe 40 mA a quad set), yeah I know the books say less, depends on how loud you play it.

regards

Joe