I have been happily using whatever value is specified in tube amp schematics I find on the web, but I'm curious as to how the values specified are arrived at by designers, especially considering that sometimes schematics using the same tubes and topology use different values for coupling capacitors, seemingly at random.
So I'm used to seeing values from .01 to .47 uF and everywhere in-between, but I'm not clear on how they are calculated.
I did some searching online and I'm still a tad confused. I see that one must take into account the plate resistance of the voltage amplifier tube and the 'corner frequency' of the desired output.
So in the case of a 1633 tube (25 volt heater 6SN7, essentially), the datasheet shows plate resistance of 6900 ohms. I'm not sure what this tells me.
I also don't know what 'corner frequency' means when trying to calculate the coupling capacitor value.
If someone could provide a dumbed-down layman's explanation for a newbie like me, it would be sincerely appreciated!
So I'm used to seeing values from .01 to .47 uF and everywhere in-between, but I'm not clear on how they are calculated.
I did some searching online and I'm still a tad confused. I see that one must take into account the plate resistance of the voltage amplifier tube and the 'corner frequency' of the desired output.
So in the case of a 1633 tube (25 volt heater 6SN7, essentially), the datasheet shows plate resistance of 6900 ohms. I'm not sure what this tells me.
I also don't know what 'corner frequency' means when trying to calculate the coupling capacitor value.
If someone could provide a dumbed-down layman's explanation for a newbie like me, it would be sincerely appreciated!
Download the Radiotron Designers Handbook 4th edition from here:
http://www.tubebooks.org/Books/RDH4.pdf
Chapter 12. "Audio frequency voltage amplifiers" begins at page 481 and the answers to your questions can be found from page 483 to 485.
There is also a lot of other useful information...
http://www.tubebooks.org/Books/RDH4.pdf
Chapter 12. "Audio frequency voltage amplifiers" begins at page 481 and the answers to your questions can be found from page 483 to 485.
There is also a lot of other useful information...
This explaination probably qualifies as being dumb. You have to take the parallel value of the plate load resistance (not the internal plate resistance) and the following stage grid resistor and put that in the formula for finding the coupling cap value when you already know the corner frequency you want. The corner frequency is the frequency 3dB down from the start of rolloff point.
Last edited:
It does look fascinating, and thank you for the link to that document. I will study it.
In the meantime, however, at first blush, the math is way beyond me. I guess by 'newbie' I mean 'stupid'. Talk to me like I'm an idiot and we'll probably be closer to my understand. For example, the document you linked too talks about calculating 'tangents'. Yeah, no. I gave up on math at basic algebra. I'm sure there must be a way to figure such things without knowing physics and cosigns and phases of the moon?
I guess not dumb enough. I understand what you said about the plate resistance, that's in the data sheet. I do not know what the parallel plate resistance is (is that double?). I do know what a grid resistor is, but what if you're not using a grid stopper? And I have no idea what a corner frequency is. Also, what is '3dB down'? I understand dB means decibel, but otherwise, I'm clueless here.
I realize I'm showing my idiocy. Forgive me. Jargon just passes directly over my head for now. I'm sure I will understand it someday.
The grid stopper is generally not considered because it is past the grid resistor. So then the corner frequency is the low frequencies you want to eliminate, like everything below 20Hz. The cap and resistance create that filtering. You have to know how to figure a parallel resistane value from two or more resistors.
Hmm. Not sure what a grid resistor is if it is not a grid stopper. The only two resistors I know of in that part of the circuit are the grid leak resistor and the grid stopper. Do you mean the grid leak resistor?
So if the corner frequency is 20Hz (and by the way, for an audio amp, would it ever be anything else?), then we could just say '20Hz' and not 'corner resistance' to avoid confusion.
As to knowing how to calculate parallel resistance from two or more resistors, I can do that. The question I guess I would have would be why. If there is one input tube, and therefore one plate (and hence one plate resistance) then where does parallel resistance come from?
I appreciate your patience with me here.
FYI, I have been playing with this calculator I found online:
https://www.ampbooks.com/mobile/amplifier-calculators/coupling-capacitor/calculator/
I suspect it is intended for guitars, but in any case, I'm not quite positive how to fill it out. It asks for a value for the volume pot, but if I have no volume pot in that position, what do I enter? Sorry, clueless...
https://www.ampbooks.com/mobile/amplifier-calculators/coupling-capacitor/calculator/
I suspect it is intended for guitars, but in any case, I'm not quite positive how to fill it out. It asks for a value for the volume pot, but if I have no volume pot in that position, what do I enter? Sorry, clueless...
No. You take the parallel combination of internal anode resistance and anode load resistor, then add this in series to the following stage grid resistor. In most cases the grid resistor will be sufficiently higher in value than the other resistances that you can get a good approximation by just using the grid resistor value on its own.20t020 said:
When someone says 'grid resistor' you can usually assume he means 'grid leak resistor' or 'grid bias resistor'.Wigwam Jones said:
There are reasons why people might want a different frequency.
The anode internal resistance and the anode resistor are in parallel, for the purposes of calculating stage output resistance.
This was your big mistake. Here's the secret: it ain't that hard! School makes it harder than it needs to be for whatever reason. Once you understand this, there's no reason to be afraid of the maths. It's not like you're trying to prove Fermat's Last Theorem or Riemann's Hypothesis (though you can get yourself a cool $Million if you pull it off).
Yes there is, but I'm really way to lazy to do that: Wolfram's. Complex number arithmetic isn't that hard, and once learned, is a helluvalot easier than memorizing all the various formulas you need to get around it.
As for the original question as to why so many different values for interstage coupling capacitors, there are a variety of reasons:
*) Selecting a specific cutoff frequency. Depending on the specifics, you may not want the low frequency response going too low due to infrasonic interference.
*) Staggering the time constants so's these don't interfere. This might be for the purpose of letting one RC coupling set the overall frequency response. It could also be necessary if gNFB is applied for phase shift considerations. If time constants cluster, that can compromise the low frequency stability when NFB is applied. Spreading out the time constants is why you sometimes see progressively larger coupling capacitors
Just need to point out that the plate resistance (or internal tube resistance, usual synbol: rp) is a function of the tube plate current. In most designs the value of plate current is quite lower than the value given in a typical parameter table - note the anode current!
Some datasheets give graphs for the important parameters vs. plate current. It will be noted that rp and S (transconductance) can double and halve, respectively, at lower plate currents. In practice, if one has a component stache, it might be easier to test frequency response using several coupling capacitor values close to the calculated value.
Some datasheets give graphs for the important parameters vs. plate current. It will be noted that rp and S (transconductance) can double and halve, respectively, at lower plate currents. In practice, if one has a component stache, it might be easier to test frequency response using several coupling capacitor values close to the calculated value.
I don't disagree with Miles' advice to learn some math.
But it is rarely essential.
And first you want some Basic Concept.
The plate resistance "does not matter".
Any series grid-stopper "does not matter".
What you typically want is the Grid To Ground resistor!
And a reactance chart. (There's other ways, some with Pi, some with JavaScript.)
If you figure a 0.01u cap into a 1Meg grid-ground resistor and get 15.9Hz, and then try it in-circuit, with precision parts, you may get 15.3Hz. That is because 20to20 is correct, the thing driving the R-C network also counts. Typically a plate and its DC resistor. 12AX7 near 1mA and 100K is ~~60K in the tube and 100K resistor, in parallel (as far as the load knows), so about 39K. So the true R-C resistor is 1,039K. But this is same-as 1Meg for practical purpose. It "almost always" is, because we try to select load resistances 3 to 10 times bigger than source resistances.
But it is rarely essential.
And first you want some Basic Concept.
The plate resistance "does not matter".
Any series grid-stopper "does not matter".
What you typically want is the Grid To Ground resistor!
And a reactance chart. (There's other ways, some with Pi, some with JavaScript.)
If you figure a 0.01u cap into a 1Meg grid-ground resistor and get 15.9Hz, and then try it in-circuit, with precision parts, you may get 15.3Hz. That is because 20to20 is correct, the thing driving the R-C network also counts. Typically a plate and its DC resistor. 12AX7 near 1mA and 100K is ~~60K in the tube and 100K resistor, in parallel (as far as the load knows), so about 39K. So the true R-C resistor is 1,039K. But this is same-as 1Meg for practical purpose. It "almost always" is, because we try to select load resistances 3 to 10 times bigger than source resistances.
Last edited:
- Status
- Not open for further replies.