Why does nearly every tube circuit connect the screen bypass cap to ground not C

[ELS122]

...not the cathode. I've read in a few articles that say you should always connect the screen bypass cap to the cathode instead of ground. but every radio or audio amp circuit I've seen connects it to ground instead.
and radio books even show it connected to ground.

so are there circuits where you don't want to connect it to the cathode?

[CaptDon]

And what specific type of circuit did they not want the bypass connected to the cathode? At RF frequencies the bypass is always grounded to avoid the tube going into parasitic oscillations. In fact the screens are painstakingly capacitor bypassed to ground. What reason was given for capacitor coupling the screen to the cathode?? In some tubes the screen is internally attached to the cathode and generally in that type of tube the cathode has multiple pinouts that are grounded!! Who is the source of your articles??? You can find articles supporting flat earth also but I wouldn't try to navigate based on flat earth theories.

[ELS122]

And what specific type of circuit did they not want the bypass connected to the cathode? At RF frequencies the bypass is always grounded to avoid the tube going into parasitic oscillations. In fact the screens are painstakingly capacitor bypassed to ground. What reason was given for capacitor coupling the screen to the cathode?? In some tubes the screen is internally attached to the cathode and generally in that type of tube the cathode has multiple pinouts that are grounded!! Who is the source of your articles??? You can find articles supporting flat earth also but I wouldn't try to navigate based on flat earth theories.

no pentode has the screen internally connected to the cathode, it's the suppressor grid that often is, not the screen.
the article I read this one was here: http://www.valvewizard.co.uk/pentode.html

The screen bypass capacitor should always be connected to the cathode rather than to ground. Yes, I know a lot of circuits connect it to ground and get away with it, but it is bad practice.

I think the idea is that since the screen voltage is relative to the cathode, you would want to connect the cap to the cathode so weird distortion doesn't happen.
but I'm not sure, that's why I'm askin.

then there's articles like this: https://tubecad.com/2016/08/blog0352.htm
that show the bypass capacitor connected to the B+ rail so increase ripple rejection.
but this would be bad in an IF stage right? besides 100/120hz ripple wouldn't cause much problems in an IF stage anyway probably.

[vk6zgo]

And what specific type of circuit did they not want the bypass connected to the cathode? At RF frequencies the bypass is always grounded to avoid the tube going into parasitic oscillations. In fact the screens are painstakingly capacitor bypassed to ground. What reason was given for capacitor coupling the screen to the cathode?? In some tubes the screen is internally attached to the cathode and generally in that type of tube the cathode has multiple pinouts that are grounded!! Who is the source of your articles??? You can find articles supporting flat earth also but I wouldn't try to navigate based on flat earth theories.

no pentode has the screen internally connected to the cathode, it's the suppressor grid that often is, not the screen.
the article I read this one was here: http://www.valvewizard.co.uk/pentode.html

The screen bypass capacitor should always be connected to the cathode rather than to ground. Yes, I know a lot of circuits connect it to ground and get away with it, but it is bad practice.

I think the idea is that since the screen voltage is relative to the cathode, you would want to connect the cap to the cathode so weird distortion doesn't happen.
but I'm not sure, that's why I'm askin.

then there's articles like this: https://tubecad.com/2016/08/blog0352.htm
that show the bypass capacitor connected to the B+ rail so increase ripple rejection.
but this would be bad in an IF stage right? besides 100/120hz ripple wouldn't cause much problems in an IF stage anyway probably.

I think the fact he describes himself as "The Valve Wizard" & the sub-heading "how to design guitar amplifiers" is the vital clue--- guitar amplifier sites are full of "woo-woo" nonsense, which was never used by the "hard-nosed" designers of real HiFi amplifiers of the valve era.

Just looking at the example amplifier stage further down the page, what the screen is actually seeing is Cg2 & Ck in series to ground.
As Ck is so much larger in value than Cg2 the series combination is "near as dammit" to that of the latter----44.45nF as compared to 47 nF, so why bother?
As even CK is unusually small, the value of Rk may be of concern here----- as the whole idea of bypassing the screen grid is to have it as a "hard" DC voltage, unvarying with signal level, Ck would be much greater than 820nF, in a normal case.

As to returning bypass caps to HT-----in theory, DC anode/screen supplies (& , in fact, any DC supply to an active device) look like ground to ac, but in practice, nasty things like stray inductance & capacitance come into play at higher frequencies, so the most common connection of bypass capacitors is to ground.

[trobbins]

I'd suggest that the typical vintage amp has the screen bypass cap going to gnd/chassis due to practicality - that cap was typically by far the largest below-chassis part in that vicinity, and a chassis connection was the most convenient means for manufacture to secure and locate such a large part.  And some vintage first stage circuits connect cathode directly to ground, so by default the screen bypass is going to cathode. Nowadays such caps are physically much smaller so can be terminated wherever with ease, and grounded cathode circuits are rarely used.

I wouldn't dis people just because of their moniker - his first name is Merlin, hence the play on words.  He certainly does focus on the guitar community as that is bread and butter for him as he is not want to making a snake oil living from hi-fi or guitar hoop-la like many others - he also has a very detailed knowledge and practical experience of what does and doesn't matter with respect to tube circuitry.  As noted, some guitar focussed amps purposefully use low values of cathode bypass for specific tone shaping, so the connection point of the screen bypass can be of concern and well worth writing about.

[ELS122]

And what specific type of circuit did they not want the bypass connected to the cathode? At RF frequencies the bypass is always grounded to avoid the tube going into parasitic oscillations. In fact the screens are painstakingly capacitor bypassed to ground. What reason was given for capacitor coupling the screen to the cathode?? In some tubes the screen is internally attached to the cathode and generally in that type of tube the cathode has multiple pinouts that are grounded!! Who is the source of your articles??? You can find articles supporting flat earth also but I wouldn't try to navigate based on flat earth theories.

no pentode has the screen internally connected to the cathode, it's the suppressor grid that often is, not the screen.
the article I read this one was here: http://www.valvewizard.co.uk/pentode.html

The screen bypass capacitor should always be connected to the cathode rather than to ground. Yes, I know a lot of circuits connect it to ground and get away with it, but it is bad practice.

I think the idea is that since the screen voltage is relative to the cathode, you would want to connect the cap to the cathode so weird distortion doesn't happen.
but I'm not sure, that's why I'm askin.

then there's articles like this: https://tubecad.com/2016/08/blog0352.htm
that show the bypass capacitor connected to the B+ rail so increase ripple rejection.
but this would be bad in an IF stage right? besides 100/120hz ripple wouldn't cause much problems in an IF stage anyway probably.

I think the fact he describes himself as "The Valve Wizard" & the sub-heading "how to design guitar amplifiers" is the vital clue--- guitar amplifier sites are full of "woo-woo" nonsense, which was never used by the "hard-nosed" designers of real HiFi amplifiers of the valve era.

Just looking at the example amplifier stage further down the page, what the screen is actually seeing is Cg2 & Ck in series to ground.
As Ck is so much larger in value than Cg2 the series combination is "near as dammit" to that of the latter----44.45nF as compared to 47 nF, so why bother?
As even CK is unusually small, the value of Rk may be of concern here----- as the whole idea of bypassing the screen grid is to have it as a "hard" DC voltage, unvarying with signal level, Ck would be much greater than 820nF, in a normal case.

As to returning bypass caps to HT-----in theory, DC anode/screen supplies (& , in fact, any DC supply to an active device) look like ground to ac, but in practice, nasty things like stray inductance & capacitance come into play at higher frequencies, so the most common connection of bypass capacitors is to ground.

why would I ignore advice from an article because it contains a word like "guitar"?
besides, engineers make mistakes too, for example how RCA tube manuals recommended a heater wiring that connects to the power transformer at one end for one wire, and the other for the other wire, for more equal voltage across the tubes, but it creates an incredible hum loop.

As the tube would see, the screen voltage would vary less with it bypassed to the cathode, since you have to keep in mind that tubes don't have any idea what ground is, everything is relative to the cathode
the only problem I see with bypassing the screen to the cathode is reduced gain, and perhaps it could in the rare occasion cause problems when the screen behavior inverts and you get positive feedback, but I doubt this would ever happen in a normal circuit...

[vk6zgo]

why would I ignore advice from an article because it contains a word like "guitar"?
besides, engineers make mistakes too, for example how RCA tube manuals recommended a heater wiring that connects to the power transformer at one end for one wire, and the other for the other wire, for more equal voltage across the tubes, but it creates an incredible hum loop.

Why would you need the screen to be an unvarying set DC voltage, that is almost never needed, and also a sturdy screen supply would most likely melt the screen if you drive the tube too hard.
the only problem I see with bypassing the screen to the cathode is reduced gain, and perhaps it could in the rare occasion cause problems when the screen behavior inverts and you get positive feedback, but I doubt this would ever happen in a normal circuit...

Out of interest, I Googled for "Mullard Amplifiers", & found this:-
https://archive.org/details/19400164MullardCircuitsForAudioAmplifiers/mode/1up?view=theater

Looking through it, I found they do return the screen bypass to the cathode, but it seems to be only in EF86 input stages where the tube is operated with very high value anode & screen resistors in what they term the "starvation" mode, combined with a split cathode resistor with a bypassed part around 1K & an unbypassed section of around 100 , with negative feedback from the secondary of the amplifier's output transformer applied across the latter.

Ck in all instances is around 50uF---much larger than shown in the "valve wizard's" example.

In each instance where this is done, DC coupling of the EF86 anode to the grid of the following phase splitter is also used.
In other, more conventional EF86 stages, the screen bypass is directly returned to ground.
Unfortunately, Mullard don't go further into why they sometimes do this, & sometimes not.

As to a "sturdy screen voltage"--in high power Tetrode RF stages, the screen supply is regulated, so it has no adverse effects, seeing many such stages are driven at very high levels.
The only time the screen will be damaged is if the anode voltage is lost entirely, & the screen voltage is still present.
Variations of anode voltage with signal levels will not cause excessive screen current.

I was brought up in the valve era, I even built the 3 watt amplifier shown in the Mullard book.
Much of my work life was involved with "hollow state" devices---in Radio & TV Broadcasting, the "valve era" hung on a long time!

[ELS122]

why would I ignore advice from an article because it contains a word like "guitar"?
besides, engineers make mistakes too, for example how RCA tube manuals recommended a heater wiring that connects to the power transformer at one end for one wire, and the other for the other wire, for more equal voltage across the tubes, but it creates an incredible hum loop.

Why would you need the screen to be an unvarying set DC voltage, that is almost never needed, and also a sturdy screen supply would most likely melt the screen if you drive the tube too hard.
the only problem I see with bypassing the screen to the cathode is reduced gain, and perhaps it could in the rare occasion cause problems when the screen behavior inverts and you get positive feedback, but I doubt this would ever happen in a normal circuit...

Out of interest, I Googled for "Mullard Amplifiers", & found this:-
https://archive.org/details/19400164MullardCircuitsForAudioAmplifiers/mode/1up?view=theater

Looking through it, I found they do return the screen bypass to the cathode, but it seems to be only in EF86 input stages where the tube is operated with very high value anode & screen resistors in what they term the "starvation" mode, combined with a split cathode resistor with a bypassed part around 1K & an unbypassed section of around 100 , with negative feedback from the secondary of the amplifier's output transformer applied across the latter.

Ck in all instances is around 50uF---much larger than shown in the "valve wizard's" example.

In each instance where this is done, DC coupling of the EF86 anode to the grid of the following phase splitter is also used.
In other, more conventional EF86 stages, the screen bypass is directly returned to ground.
Unfortunately, Mullard don't go further into why they sometimes do this, & sometimes not.

As to a "sturdy screen voltage"--in high power Tetrode RF stages, the screen supply is regulated, so it has no adverse effects, seeing many such stages are driven at very high levels.
The only time the screen will be damaged is if the anode voltage is lost entirely, & the screen voltage is still present.
Variations of anode voltage with signal levels will not cause excessive screen current.

I was brought up in the valve era, I even built the 3 watt amplifier shown in the Mullard book.
Much of my work life was involved with "hollow state" devices---in Radio & TV Broadcasting, the "valve era" hung on a long time!

no the screen may melt if the tube is overdriven, esspecially in guitar amps but it can happen in any amp.
thats why screen stopper resistors are put in place (also they often can fix oscillating problems too)
in a push pull output stage it shouldnt really make a difference as to where you connect the screen bypass cap since the cathode will be more or less stable unless you dont connect both cathodes together and it is a cathode biased amp.
and I think it could cause instability in a push pull amp when the tubes go into class B operation
I think testing should be done to get clear answers on this

[gbaddeley]

In a typical common cathode voltage amplifier the screen grid should be bypasses to the signal ‘common’. For Vin and Vout, this is 0 V (ground). For Cathode bias using a resistor (and optional parallel cap - to increase the gain), connecting the screen bypass cap to the cathode creates feed back. AC Modulation of the screen current caused by amplifier operation will feed back to the cathode. This would have a small frequency dependent measurable effect on the Vout/Vin transfer function but probably not significant for most amplifiers.

[ELS122]

So from what I read it seems that connecting the bypass cap to the cathode maximizes stability and linearity in a single ended stage, but connecting it to ground maximizes gain.

[trobbins]

Perhaps some separation is needed for applications (and hence valves).  My previous comments related to input and low signal level stages. 

ELS122, it seems like your focus is on amplifier output stages, where the screen can conduct relatively substantial current (especially when the anode is driven towards saturation).  Datasheets for some output stage valves show representative distortion levels for a given circuit configuration, and they are normally based on a stiff screen voltage supply and hence decoupling to 0V.

Some output stage configurations that use cathode and anode coupled output transformers (eg. unity coupled style) by necessity use a decoupling cap from screen to cathode as the cathode is not fixed to 0V (as per fixed bias) or typically cathode biased but with a strongly decoupled cathode.

[TimFox]

Note that when a pentode is used as the input tube in a feedback amplifier, the cathode is usually not bypassed to ground and the feedback resistor connects to the cathode.
In that circuit, the screen bypass capacitor should be returned to the cathode, not to ground.
The classic Dynaco amplifiers used a 6AN8-type pentode-triode at the input, with the pentode grid as the input, an unbypassed cathode resistor, and the feedback resistor to the cathode.
The pentode plate connected directly to the grid of the triode, used as a phase inverter to drive the output tube grids from its plate and cathode.

[eugene]

Good grief. So much guessing. Most of us could learn more than a little from John Broskie (https://tubecad.com)

The bypass cap on the screen is intended to bypass AC currents. So look at the circuit and decide where the AC signal current loop is. In terms of electrons (as opposed to conventional current) it's up through the tube, out the screen, and back to the cathode. In the case of a pentode used for gain, grid bias is usually achieved with a cathode resistor. That cathode resistor also has an AC bypass cap. So if the screen is bypassed to ground, then the AC screen current goes through its own bypass cap to ground and then also through the cathode bypass cap. The fact is, it really does make more sense to bypass the screen directly to the cathode, thus skipping the detour through the cathode bypass cap.

My guess is that it's usually connected to ground because (1) few people take the time to think about the current loop, and (2) because most circuits are just copies of some other circuit and that's what everyone else does.

[b_force]

I think the fact he describes himself as "The Valve Wizard" & the sub-heading "how to design guitar amplifiers" is the vital clue--- guitar amplifier sites are full of "woo-woo" nonsense, which was never used by the "hard-nosed" designers of real HiFi amplifiers of the valve era.

I understand where you're coming from, because there is a lot of woo-woo, not only with guitar amplifiers, but also with tube/valve amplifiers.

But I also know for sure you never really have read anything from Merlin Blencowe or talked to him as well.
Because he is most definitely not a "woo-woo" person.
In fact, he's pretty down the earth and open minded about stuff.

fyi, if you read a little further he also has a book about hifi amplifiers as well.
I have talked to him quite a bit, and I know that he get his books, explanations and resources checked by quite some people in the field.
With that I don't mean "woo woo" people.

And to quote Merlin about exactly this question/problem; (on a different forum)

Screen bypass should always go to cathode, but it doesn't make a lot of difference when the impedance in the cathode is small / bypassed.

Which would be also my answer.
So yes, to cathode is the technical correct answer, but since the impedance is so small, it doesn't matter an awful lot.

That being said, why most people connect it to ground; because the very vast majority of people just copy-paste circuits.
Or, depending on the physical connections, is just a little easier to mount.

[ELS122]

But I also know for sure you never really have read anything from Merlin Blencowe or talked to him as well.
Because he is most definitely not a "woo-woo" person.
In fact, he's pretty down the earth and open minded about stuff.

imo the only way to learn it to be as open minded as possible.
can't be sure that you're right when you close out the possibility of being mistaken 

well I think I have a much better understanding of the impact of connecting the bypass cap to the cathode instead of ground.

[TimFox]

I just remembered an example of proper usage of screen bypass capacitor to ground instead of cathode.
I have an ancient E H Scott short-wave receiver designed for merchant-marine service in WW II.
Since many merchant vessels had DC power, it is an AC/DC set with series heater strings and off-the-line rectification.
Since "DC ground" in the circuit is not connected to the metal chassis directly, the RF bypass capacitors are "bathtub" hermetically-sealed multiple paper capacitors with common connection to the case.
While the DC connections to grid returns, screens, cathodes, and plates come from the rectified voltage, their bypasses are all done to the chassis through these bathtubs.
(This particular receiver was optimized for low local-oscillator leakage to the antenna, for security reasons, and used a 6K7 input tube with grid cap to reduce the leakage.)

[eugene]

While the DC connections to grid returns, screens, cathodes, and plates come from the rectified voltage, their bypasses are all done to the chassis through these bathtubs.

OK, but... why?

[TimFox]

While the DC connections to grid returns, screens, cathodes, and plates come from the rectified voltage, their bypasses are all done to the chassis through these bathtubs.

OK, but... why?

The "DC ground" is just a wire net connecting the negative side of the off-line rectifier.
To properly RF-ground the cathodes, screens, plate returns (tuned transformer primaries), and grid returns (tuned transformer secondaries) to a real shield, the RF bypass capacitors must be connected to the chassis.