In this schematic what is going on here with the grounded cathode stage? I get that C1 is the stage decoupling capacitor correct? Its purpose is to prevent motorboating right? It wraps the whole B+ just like a power supply filter except closer to the tube right? But what is happening with C2, R6 and R7 here? Is not C2 also a stage decoupler of sorts? I don't remember ever seeing a plate load divided in half like this. I thought when you put an electrolytic all the way across the tube and its two resistors, but near it, then that was your stage decoupler. Is this some other way To decouple a stage? R6 and R7 are not dividing any DC here, the plate resistor is still their sum 44k. Confused.
Here is the article where I found this:
Welcome to the most complete do it yourself guide on the D. T. N. Williamson tube amplifier., chapter 3 , The Circuit
Here is the article where I found this:
Welcome to the most complete do it yourself guide on the D. T. N. Williamson tube amplifier., chapter 3 , The Circuit
Attachments
Oops I think I figured it out, that stage is the splitter, duh. So my question becomes was this done to get more accurate phase splitting? Why divide the top resistor in two and why the capacitor to ground?
Last edited:
That's a decoupling circuit also, R6 and C2. There's no real need for R6 and R7 to be the same value.
With more 22k resistors to choose from, R5 and R7 could be more easily matched for the phase splitter.
With more 22k resistors to choose from, R5 and R7 could be more easily matched for the phase splitter.
Last edited:
R6 drops the voltage to the value the designer chose and R7 is the load resistor. C2 is a decoupling capacitor. So, R6 works as a voltage divider along with R27.
Wow thanks. Didn't know that the the outer extents of a cathodyne can be set with a simple decoupler like that. I was thinking the whole plate herevhad a 44k load which didn't match the cathode resistor. Thanks.
Mmm… (re: drops to design voltage), I really don't think that is the primary purpose of R6. It serves same purpose as R2, which is along with shunting capacitors C2 and C1 (respectively), to act as power supply ripple filters, yielding much quieter DC B+ to each valve stage.
The choice of '22 kΩ' or '33 kΩ' is a bit arbitrary, but the point is to have a working B+ as high as necessary to achieve good stage linearity, but also with a large enough RC constant to seriously filter out hum, buzz, and especially interstage cross talk.
At least that's my drift.
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅
The choice of '22 kΩ' or '33 kΩ' is a bit arbitrary, but the point is to have a working B+ as high as necessary to achieve good stage linearity, but also with a large enough RC constant to seriously filter out hum, buzz, and especially interstage cross talk.
At least that's my drift.
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅
With a CAP in the middle! 8uFd is "huge" around 22k resistors. About 1k at 20Hz and much less throughout the audio band.
And a cathodyne has about NO power supply rejection in the top side. While a simple triode stage may have PSRR "about half", the huge cathode resistor makes the working plate resistance far higher than 22k; here, maybe 400k. So the PSRR is like 20:19 or 0.95, and "all" the power supply crap comes into the top output of the cathodyne.
We may also distinguish between "big" and "small" cathodynes. When driving power tubes directly we may have to swing 30V to 60V peak and we need lots of supply voltage (OTOH at high level a little PS buzz matters less). But in the Williamson there is another amp stage before the power jugs. Signal level is a couple volts. We don't need a high supply but it should be very clean.
Last edited:
Decoupling R6 is indeed its primary purpose. Without it the stage doesn't work correctly. The extra RC filtering looking from the PSU end is nice to have but not essential. Decoupling R6 is essential.
I finally understand what decoupling capacitors truly can do, i think. I used to think they were just little reservoirs around stages. But here its actually making the AC "forget" that the series resistors into the plate don't match the cathode resistor. Since a cathodyne says that the cathode and plate resistors must be equal, the blocking cap here is making that happen, but that only has to happen for the AC. the additional plate resistors then just adjust the DC point. So by decoupling you can take the phase off of the matching 22k resistors because the AC no longer "sees" the other plate resistors? Most phase splitters I've seen go straight to matching resistors so I never noticed decoupling around the whole cathodyne before this one.
It depends on how the schematics are drawn. Follow the B+ from the output transformer and you will usually see a resistor and a decoupling capacitor before the phase inverter. It's there but just drawn differently.
Windcrest, it has nothing to do with a cathodine or any other stage topology. A capacitor, in particular a large one, is a short at AC. So for signal purposes, the point where B+ is decoupled by a capacitor is an AC ground.
Z = 1 / (2πFC)
is the primary 'way of thinking about it'. This is algebraically rearranged toF = 1/(2πZC)
which then predicts a 'corner frequency' when a resistor-in-series-with-a-capacitor is envisioned as a low-pass filter, or, as a resistively fed 'power supply'. Let's use that. 8 µF and 22 kΩ.
F = 1/(2 × 3.1415 × 8×10⁻⁶ × 22×10³ )
F = 0.90 Hz
This would be the –3 dB point. The fall off is classic Butterworth, so –6 dB per octave, or –20 dB per decade. At 50 Hz (lowest power supply ripple) thus:F = 0.90 Hz
50 Hz ÷ 0.90 Hz = 55×
log₂(55) = 5.8 octaves
5.8 octaves × –6 dB/octave = –35 dB
. See? log₂(55) = 5.8 octaves
5.8 octaves × –6 dB/octave = –35 dB
A very decent power supply ripple-and-buzz quenching low pass filter. Always remember that “low pass” also means “high cut”. So, everything above 50 Hz is more than –35 dB attenuated. sweet!
The 'cost' of this is whatever voltage drop across the 22 kΩ resistor this (or your) circuit decides to operate at, as quiescent operating point. To the valve, in A/C, the plate load looks like a 22 kΩ resistor connected to a lower B+. From a down-circuit noise perspective, each stage endowed with such a cheap little circuit enhancement will be MUCH quieter, noise wise.
And that — I already wrote above — is very probably 90% of why it is included in the circuit diagram. Noise abatement.
For the purists and theoreticians out there, apart from its spectacular job in squashing power supply ripple, and any 'upstream' interstage intermodulate, lowering the effective B+ has a distinct effect on the valve's transfer curves, and linearity. In some ideal sense (as versed by analyzing constant-current load devices — which can be idealized as infinite-value resistors attached to infinite + power supplies!) the larger the combo of plate load resistor and plate power supply, with almost no exceptions, the more linear the gain of the stage.
But even with that, all it takes is a precocious pre stage with not-so-linear characteristics to make a hash out of overall linearity. A very common way to 'fix' or address that is by running a first stage as shown, and a second stage as a splitter… again as shown, but with a reduced B+ to slightly enhance its internal local-negative-feedback cathode degenerate behavior.
And of course, you get the power-supply-and-interstage-feedthru noise reduction 'for free'. Which has always seemed not to be the 'side' benefit, but ther primary one, to me.
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅
Last edited:
The primary purpose is to define the plate load at AC. Without it, in this circuit, the plate load is 44K and the cathode load is 22K. This does not work as a phase splitter, as it delivers double the voltage swing from the plate that it does from the cathode. QED
Ultimate source: Radio Dwsigner's Handbook.
Ultimate source: Radio Dwsigner's Handbook.
Boy, this thread contains a lot of word salad over what is ultimately just an ordinary power supply RC filter.
Agreed....
Guilty as charged, I guess!
Thing is, (esteemed) MerlinB, I find so much of what is given to be earnest discussion as patent tripe.
With the loss of level heads such as AndrewT, DF96, Popilin, SReten, Davada, kegger, Sasuma… (RIP), the voices of hard-won reason are in the decline. Magic thinking (e.g. word salad) is continuing to rise.
I had hoped here to decrease the volume of the salad chopper.
Ah, well…
Be well this Christmas!
GoatGuy
Guilty as charged, I guess!
Thing is, (esteemed) MerlinB, I find so much of what is given to be earnest discussion as patent tripe.
With the loss of level heads such as AndrewT, DF96, Popilin, SReten, Davada, kegger, Sasuma… (RIP), the voices of hard-won reason are in the decline. Magic thinking (e.g. word salad) is continuing to rise.
I had hoped here to decrease the volume of the salad chopper.
Ah, well…
Be well this Christmas!
GoatGuy
Now I think about it, I haven't seen SY (Stuart Yaniger) around here for ages, either
AndrewT, SReten & DF96 were great stabilizers, I agree (I think the others are before my time)
Others have told me I waffle, no matter, they havent told me in the way one of those guys would (wishful thinking), instead trolls abound.
I miss those guys
Word salad, agreed. RC filter, no different than word salad. If you have a battery supplying perfect DC for B+, there is no need for RC filtering but you still need decoupling caps.
The primary purpose of the decoupling capacitor is just that, as ejp explained very concisely above. For a non-EE novice, that is the best way to help understanding.
This is what I was trying to say I guess in non-technical knows-just-enough-to-be-dangerous-hobbyist terms. Without that capacitor I see an unbalanced cathodyne where the cathode and anode circuits are not matched. So the DC operating point is being set as needed with no regard for matching the Rk resistor, but the AC is decoupled by C2 so it can "swing" symmetrically across two matching 22k resistors then? Without C2 it looks imbalanced.
Last edited:
In your mind… envision it as an RC unit. It must not be envisioned as a 'resistor from which one can analyze it without the capacitor'. Both, or nothing at all.
Indeed, once you take both imaginarily out of circuit… connecting the 22 kΩ anode load to B+ directly, and the 22 kΩ cathode 'load' to ground, it becomes an obvious 'simple' phase inverter. Insertion of the RC pair substantially alters the power supply B+ quality, by filtering out whatever 'crud' is on the B+ rail, be that power supply ripple, crud from other stages, picked up RF (external or internal!) or what-have-you. It is solid engineering for a vexing problem, especially for circuits which are point-to-point wired with common 'economically feasible' wiring.
The only real reason I'm going on and on about this, in this way is because once you come to view the RC pair as a pair and a single tunable circuit element, viewing other circuits with the R and C perhaps quite distantly placed on a diagram becomes so much easier.
⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
⋅-=≡ GoatGuy ✓ ≡=-⋅