• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Why does no-one use the Anode Follower?

Thanks for that. There's something I'm not understanding, though.

In the examples, the feedback resistor Rag is connected from the plate of V2 straight to the plate of V1. That means Rag is also the plate load resistor for V1, which also means ra of V1 forms the 'series resistance' that makes the voltage divider defining the level of negative feedback applied to V2. (Rg is in parallel with all that as well, so I assume those resistance values need to be made as large as possible.)

What if you have a physical series resistor to form the negative feedback voltage divider? My understanding is that this resistor (Rfb1 in the attached schematic) defines the input impedance in parallel with the tube's own internal input impedance. If you use a large enough value for Rfb1 (let's say 150k ohms) won't that keep the load on the preceding stage light enough to keep the preceding stage from loading down too much?

I modeled this in LTspice and got really good results, which is why I'm persisting with this line of inquiry.
--

Rongon,

I am relatively inexperienced, but I dont think there is anything wrong in the circuit you posted- in fact, most, almost all my triode stages are designed this way. I've built 3 little prototyped amplifiers, and all have used local feedback in this way, and in ALL stages.

I have no doubt it sims well, it also seems to work well, in my limited experience. I'm sure there are reasons not to do this.

From a practical point of view, I have used values of Rfb2 down to 150k, with good results. Better setting 180 to 220k as a minimum though. Gain of somewhere between 2 and 3, with the medium mu triodes I've been using, a practical minimum.

But I dont try an build a follower this way, A=1.

Truth be told, I haven't built a decent low thd follower yet.
 
Last edited:
I suppose what I call an 'anode follower' isn't purely the A=1 version.

What do you call the circuit if you adjust the NFB resistors to get a gain of 3? Is that a 'common cathode stage with plate-grid feedback'? Or is it an 'anode follower adjusted to have 10dB gain'?

Does it really matter?
--

Either way, it's a nice tool to have in the kit. I'm surprised I don't see it used more often.
--
 
Last edited:
I suppose what I call an 'anode follower' isn't purely the A=1 version.

What do you call the circuit if you adjust the NFB resistors to get a gain of 3? Is that a 'common cathode stage with plate-grid feedback'? Or is it an 'anode follower adjusted to have 10dB gain'?

Does it really matter?
--

Either way, it's a nice tool to have in the kit. I'm surprised I don't see it used more often.
--
I'd just call it a common cathode stage with parallel applied voltage feedback, not that 'anode follower' is wrong. It's just a name that someone came up with that isn't that descriptive and isn't universally known.

You might find my recent experiments interesting here.
 
I'm trying to use up a stash of 6DJ8s (I know, lucky me). I was thinking of using them for a MM phono stage, but you can't get more than about 34dB of gain from two 6DJ8 sections.

--

My E88CC MM Phono Cascode has 46dB of gain. 😉

I use two E88CC's per channel, feeding into a mosfet follower that can drive VERY long cables (if required). If you have a stash (like I do) why not use them?

Cheers,
Ian
 
Do you use two cascode pairs per channel?
I could see a cascode as the input stage, for sure.
Not sure what would be the best thing to do for the second stage. That's the stage with the highest signal levels, so where the most signal distortion could happen.

What about a cascode pair 6DJ8 into a mu-follower 6DJ8 and thence to the MOSFET source follower? Or...

Aikido? That uses LOTS of tubes!
--

Speaking of cascodes for the input stage...
Like pentodes, the top tube's plate resistor (Ra) sets the output impedance of the stage, correct?
Since (like a pentode) the ra (plate resistance) of the cascode stage is very high (hundreds of thousands of ohms), and Ra is going to be much lower in value (like a couple of tens of thousands of ohms), couldn't one use that output resistance as the Rseries for the RIAA? That would be slick, right?

So, for the first stage of an RIAA preamp, let's say I use a 15k Ra on a 6DJ8 cascode pair.
The output resistance of that stage will be just a little less than 15k ohms, as long as the load resistance is high... Right?
It also won't change much, since the ra of the cascode is much higher than that 15k ohms.
Therefore, it should be safe to use that 15k output resistance as R1 in the standard passive RIAA filter, instead of an actual, physical R1.

No?
--
 
Last edited:
Therefore, it should be safe to use that 15k output resistance as R1 in the standard passive RIAA filter, instead of an actual, physical R1.
--

In the typical RIAA filter, Rseries is R1, as in the schematic below:

PassiveRIAA.JPG
 
The last example shows two passive RIAA filter networks, with no NFB.
Are you referring to the 'anode follower' itself?

There are several circuits shown in this thread with NFB employed. Which one(s) are you referring to?
--
 
Yes, understood. That was mentioned as one of the limitations of the circuit.
If the source impedance is steady at about 5k ohms, and the series NFB resistor is 150k ohms, how much would the source impedance influence the gain and level of NFB?
--

It looks like MerlinB has already answered a lot of these questions, and online as well:
The Valve Wizard
--
 
Last edited:
The anode follower is misnamed. There's no drain follower, or collector followers, since a follower has (close to) unity gain, and the anode/drain/collector has inverted polarity.

Furthermode as the anode-grid feedback is shunt feedback, the input impedance is lowered, which is very much not what you expect from a follower (ie its not a high input impedance buffer).

And the circuit topology doesn't imply (negative) unity gain at all, the gain (or attenuation) is programmed by two resistors (plus the source impedance).

A shunt feedback stage is the generic label for this kind of circuit. And I think you'll find its widely used all over the place.
 
I like 'shunt feedback stage.' That's a better, more accurate name than 'anode follower.'

One still doesn't see it used much in line level stages. I suppose that's because of the low input impedance. However, if you take a 6DJ8 (for instance) and reduce its gain to 3x (10dB) using shunt feedback (plate to grid), you could use a 220k ohm resistor for the series grid resistor and 1M for the feedback (shunt) resistor, and still get about a 250k ohm input impedance. That's not onerous.
--
 
But would you really want to use 220k in series with the grid?

I get that you're trying to push Rseries as high as possible to keep Rin high, and Rfb high also; but you could as easily use Rfb of 520k, and Rseries of 120k. Maybe even halve those values again...

Like I often profess, I am no expert, most of my learning in valves comes from doing. But in that limited experience, I havent found an issue with 2nd stage loading the first due to low input impedance, when the Rfb was as low as 220k.
For clarity, this is the input stage and driver stage of a power amp that I refer to, with the input stage triode operating at a quiescent current of perhaps 1-1.5mA only.
 
Last edited:
Ok. I don't mean to tell you something you already know 😀

Of course the valves I have experience using have rp somewhere in the range 4k to maybe 10k at most, are generally medium mu types, so even with the first stage run open loop, output Z must still be adequately low.

I wouldn't like to say for higher mu types, 》25, since I haven't much experience there (besides the little 6N17B-V, mu of about 70).

In general I found better results using 2 stages of medium mu triodes, rather than a single high mu stage achieving the same gain
 
Another way to look at the Anode Follower, with its FB network it is a simple Operational Amplifier from which differentiators & integrators can be built. Op Amps were & still are found in analogue systems. Some not with electronics at all. For example, early cruise control used vacuum & mechanical parts as a simple Op Amp application. One of the guys in the lab I worked in circa 1960 knew a EE who sold non-electronic Op Amp components.😀
Many control systems run on other formats, for example compressed air.
 
Fascinating to see exactly the same futile arguments about the name 'anode follower' going on again. It seems to come round every 10 years or so, rather like the fallacious arguments about split-load phase splitter output impedance. Humans are a strange race. I used to think it was a bad name too, but I'm more relaxed about it now. It's just a name, and it has stuck.