Switchable bypass cap/resistor combinations

[jack2911guitarblues]

Hi to all,

Could anyone help with a schematic of how to properly connect 2 or 3 combinations of bypass caps and resistors on a single switch?  I have seen such drawings in the past, but now can't find anything like that.
I have a cascade gain mod in my self build 1987 plexi, which is switchable and am still able to blend the normal and bright channels in the Plexi setting. Bypass caps/resistor values are:  V1a 820/ 2.2uf,  on V1b 2.7 k/ .68uf . The 820/2.2uf (from 820/330uf) has been great for getting a bass tone that's not too boomy, but in the cascade setting that combo could still be lower. ie. 820/1uf or .68uf in an effort to control boomy bass.
That's why I would like to be able to switch over the V1a to a lower cap in the cascade setting.
Any suggestions would be helpful!
Much appreciated, Jack

[VMS]

Here is two options, second one is easier but you have to compromise with the values.

100k resistors are there to prevent popping noise.

[jack2911guitarblues]

To  VMS

I will try that 1st option as soon as I get the chance

Thanks for the suggestion
Cheers

[HotBluePlates]

And if the 1st option is a SPDT On-Off-On switch (center off, no connection to either outer lugs), then you get reduced gain/flat, full gain with some low roll-off, and full gain/flat.

[jack2911guitarblues]

Hi HotBluePlates
 Appreciate your response!  any advice I can get, all helps with the learning curve.   Only i'm not really familiar how caps in parallel and series work, which is what will happen if I use a SPDT switch in place of an ON-ON switch I guess?  sorry about being a bit thick, but this is my first build., When you say " reduced gain/flat, does that mean circuit is running through just the 1uF cap and 820 ohm resistor? would it be possible just to show a little diagram how the On-Off-On would wire into the circuit?
just trying to get my head around it all!  must say it's wonderful being able to tap into experienced guys knowledge. Hopefully I can help less experienced guys down the track.

Thanks, Jack

[VMS]

It is wired the same way with both switches.

With on - off - on switch in middle position you only have the 820 resistor connected so the gain is less and you don't have that 3db roll-off point.

You can play with this calculator and see how different size capacitors effect the frequency response:

https://ampbooks.com/mobile/amplifier-calculators/cathode-capacitor/

[sluckey]

I think switching between 1µF and 2.2µF is going to be very subtle. May I suggest a 4.7µF and 10µF on the switch? And if you use a center off switch, hard wire a .68µF across the cathode resistor. That will give you three choices, 5.38, .68, and 10.68. Get a handful of different value caps and experiment using your ear to find the combination you like.

[2deaf]

Would you consider a pot. instead of a switch?

[Fresh_Start]

I think switching between 1µF and 2.2µF is going to be very subtle. May I suggest a 4.7µF and 10µF on the switch? And if you use a center off switch, hard wire a .68µF across the cathode resistor. That will give you three choices, 5.38, .68, and 10.68. Get a handful of different value caps and experiment using your ear to find the combination you like.

+1 on sluckey's suggestions. 

The only time I installed cathode bypass cap switching, I used a 2-pole switch to change the values on the first TWO preamp triodes.  First stage was like 25uf or 4.7uf.  The second stage was more complicated, but essentially like 25uf or 6.8uf.  This was after a LOT of experimenting, and the change still isn't huge.  The amp's owner really likes his "bright" switch.

The pot is an interesting idea at first; however, DC voltage on a pot generally isn't a great idea - even low cathode voltage.  Also, as you reduce the resistance "below" the 2.2uf cap as shown, the frequencies affected change because you have a high-pass filter and you're changing the R in the formula.

Hope that helps,

Chip

[Willabe]

The pot is an interesting idea at first; however, DC voltage on a pot generally isn't a great idea - even low cathode voltage. 

The cap can only shunt ACV to ground through it and the pot is on the ground side so there should be no DCV on that pot?

[Fresh_Start]

The pot is an interesting idea at first; however, DC voltage on a pot generally isn't a great idea - even low cathode voltage. 

The cap can only shunt ACV to ground through it and the pot is on the ground side so there should be no DCV on that pot?

You are correct.

Chip

[2deaf]

The pot is an interesting idea at first; however, DC voltage on a pot generally isn't a great idea - even low cathode voltage.

No DC on that pot.

Also, as you reduce the resistance "below" the 2.2uf cap as shown, the frequencies affected change because you have a high-pass filter and you're changing the R in the formula.

With the full 10K ohms in series, the "S" shaped frequency response curve is about the same as a 0.68uf cathode capacitor by itself.  With the pot. at the other extreme (0.0 ohms), the "S" curve moves to the lower frequencies and is the same as a 2.76uf cathode capacitor by itself.  In between the two extremes is a series of intermediate "S" curves.

The effect that the pot. has on frequency response is logarithmic, so most of the effect is from 5K to 0.0 ohms.  That means that you won't notice much difference from 1 to 5 on the control and then most of the difference from 5 to 10.  A reverse-taper pot would be good in this application.

The bigger a cathode resistor gets, the larger the range of db change for cathode capacitors that affect frequency response..  So changing the cap. values will have a more noticeable effect with a 2.7K cathode resistor than with an 820 ohm.  Also, the smaller a plate resistor gets, the larger the db range.  Just a couple of things to keep in mind when playing with cathode capacitors.

[jack2911guitarblues]

Thanks so much everyone, for all the useful information!
 I will certainly try some of those combination as suggested with the switch.  I did try a 1uF Bypass Cap just to compare it with the 2.2uF i originally put in, and indeed I had difficulty hearing the difference. So I might try some larger ones like sluckey suggested as well as the 0.68uF

I was just looking at some of these charts available for calculating frequency attentuation with certain cathode bypass cap/resistor combos.
 So  with an 820 resistor/2.2uF on V1a on a 12AX7,  anything below 133hz gets attenuated.  But then the same chart shows that using another combination of a 10k resistor/0.68uF, has attenuation below 169hz.  so both combinations are nearly identical in shaping the tone. I guess the only difference is the gain factor then?  820 ohms will give more gain than a 10k resistor. Is that right?

[2deaf]

So  with an 820 resistor/2.2uF on V1a on a 12AX7,  anything below 133hz gets attenuated.  But then the same chart shows that using another combination of a 10k resistor/0.68uF, has attenuation below 169hz. 

Can you show me the math?  Or at least what constitutes a point of attenuation? 

so both combinations are nearly identical in shaping the tone.

Those two combinations will have a radically different effect if you put them in an amplifier.

[Willabe]

so both combinations are nearly identical in shaping the tone.

Those two combinations will have a radically different effect if you put them in an amplifier.

He's referring to the bass roll off and not overall gain when he said "shaping the tone".

[2deaf]

You need to plot both frequency response curves on the same piece of graph paper.  Then you can see how radically different they really are. 

[Willabe]

Music is in the ears not the eyes. Yes of coarse you can plot things and do the math but how it sounds in the end is going to be the final answer, as long as you don't burn something up.

With something as easy as swapping a couple of bypass K caps and K R's why bother with the math? Let your ears decide.     

[jack2911guitarblues]

Yeah sorry, just trying to come to grips with the types of possibilities in cathode cap/resistors combos. I worded that wrong I think.
Willabe worded it just right.  " I was referring to the bass roll off, and not gain". Looking at this Cathode Resistor Bypass Chart you can see how some of these combinations  get really similar with Bass roll off, (check out the attachment)  which is what I was struggling a bit with the Plexi Cascade Mod  in the first place.
And I think just trial and error, and those 2 flaps we call ears, is often the best way to decide!

[2deaf]

So the values in that table are for "Lower -3db frequency points".  I'm having a little trouble with the literal sense of this and even more trouble with the resulting numbers.

I use the point where the phase angle of the parallel RC circuit on the cathode is 45 degrees as the cutoff frequency for this triode arrangement.  It's 45 degrees when the ratio between the current through the capacitor to the current through the resistor is equal to one.  This happens when the impedance of the capacitor equals the resistance of resistor.  A quick manipulation of the equations shows that the frequency for this event is equal to one divided by (2*pi*R*C).  Since the gain of the triode is inversely proportional to the impedance of the cathode circuit, this point appears on the frequency response curve where the gain starts to rise at the lower frequency end.  Is that how you guys see it?

Can anybody explain that table to me?

 

[HotBluePlates]

Also, as you reduce the resistance "below" the 2.2uf cap as shown, the frequencies affected change because you have a high-pass filter and you're changing the R in the formula.

With the full 10K ohms in series, the "S" shaped frequency response curve is about the same as a 0.68uf cathode capacitor by itself.

Without knowing it, you're each talking about different effects.

Look at the "Cathode Bypass" picture below. It gives the graphical form of results for a tube stage set up as shown (the plate load and tube operating point matter, as well as simply Rk & Ck), roughly similar to the numerical chart. The graph is drawn assuming the cathode resistor is constant (because you picked it for a specific d.c. operating point) and the cap value is varied; however, if you hold C constant and vary R, a similar graph results. So on that point, Jack is right.

Gain stage cathode local feedback: If no cap is present (or has an infinite resistance/reactance in parallel with the cathode resistor), the gain stage in the diagram will have a gain of 29dB. With a given bypass cap, if the applied frequency is high enough to see the cap as essntially a short-circuit (0Ω in parallel with the cathode resistor), the resistor (and its local feedback effect) is bypassed, and stage gain rises to a bit over 35dB.

The -3dB point that is being presented in the numerical table is shown on this graph at ~32dB; this is the frequency when the cathode resistor is equal to the capacitor reactance. So given the 1.8kΩ cathode resistor in the diagram, the cap appears to the applied frequency as 1.8kΩ. This is in parallel to the actual 1.8kΩ cathode resistor, gives a net impedance to ground of half that (or 900Ω), and also has half the gain reduction due to feedback as the unbypassed 1.8kΩ resistor.

What 2deaf is describing, though, is a control that allows Partial Cathode Bypass, in a variable circuit which can be dialed from full-bypass down to zero bypass. Look at the "Partial Cathode Bypass" picture below. If you could smoothly vary the impedance to ground bypassing the cathode resistor from zero up to some value, you could also smoothly vary the gain up from a low of 29dB to a max of 35dB. The easy way to implement this is 2deaf's variable resistor between the bypass cap to ground.

At full resistance, gain will be 29dB across all frequencies because the impedance through the cap to ground is very much higher than the resistance through the cathode resistor (you'll need a pot around 10x the cathode resistor value to achieve full gain reduction). As you turn the pot resistance down, the maximum gain rises up from 29dB to the black line. The bypass cap value determines what frequency the rise from 29dB to the higher gain occurs, but the series resistance of the pot determines how much gain is increased. Reduce pot resistance some more and you'll move the gain boost up to the red line. Reduce the pot resistance until it is a bit less than 1.8kΩ and you arrive at the green line (I should have drawn the green line on 32 dB to make the example easier). Lower the resistance some more to get up to the blue line, and lower it all the way to 0Ω and we're back up to a bit over 35dB as the maximum gain of the stage.

So what 2deaf's circuit does is not change the turnover frequency of the bypass (that would be done with the switched circuit shown earlier), but changes the amount of boost realized. If you use a small bypass cap, which would make the amp sound bright, it then allows you to have "no brightening" or "a little brightening" or a good amount of brightening" or "a LOT of brightening."

[2deaf]

Hmmmm.  That's not the way I do my graphs.  I use voltage out divided by maximum voltage out to calculate db's and this graph uses voltage out divided by voltage in.  Should be the same db spread and curve, though.  With the bypass cap. in series with a pot., my curves get squished-up from the bottom as the resistance increases.  With your graph, the curves get squished-down from the top.  I'm gonna have to poke a calculator a little to convert my graphs.

[Fresh_Start]

I completely missed that the pot in 2Deaf's drawing is in parallel with the 820 cathode resistor.  Thanks HBP!

I was going to call this a "daydream" until I realized (a) how weird the math gets, and (b) I ought be asleep before now.  Anyway, this crazy circuit tidbit popped up as a possible way to shift the effective cathode cap without changing net gain.  It doesn't work that way though.

Chip

[2deaf]

So I converted a couple of my graphs and I decided I need a key that is Log times twenty.

My graphs with -db's squish up from the bottom and shift to the left with increased series resistance.  Your graphs with +db gains squish down from the top and shift to the left with increased series resistance so that the curves cross when plotted on the same paper.

The overall effect with this setup is decreasing relative bass as gain increases.  But this isn't the setup that I attached earlier.  That one has a variable series RC in parallel with a parallel RC circuit.

I have a shortcut for determining the total impedance of a parallel RC circuit when the capacitor and resistor have equal impedances:  I divide the resistor value by the square root of two. 

[2deaf]

Anyway, this crazy circuit tidbit popped up as a possible way to shift the effective cathode cap without changing net gain.  It doesn't work that way though.

I've used that one on the first stage of several hi-gain amps and it really gives them a lot of versatility with one knob. 

[HotBluePlates]

But this isn't the setup that I attached earlier.  That one has a variable series RC in parallel with a parallel RC circuit.

I now see that.


Forget calculating; build and listen/measure. It would be enormously faster to breadboard it and plot a graph from measured output of a swept signal than to calculate

[2deaf]

Forget calculating

I can calculate one of those curves up to the point that it breaks into a curve at the top, but about the only value in that would be to test the limit of my attention span.  However, knowing one very simple calculation and the significance of it can give you a real leg up over blind trial and error.

[shooter]

knowing one very simple calculation and the significance of it can give you a real leg up over blind trial and error.

But the graph HBP posted takes the *blindness* outta the equation, sorta like the 'ol adage; "a picture is worth a thousand calculator strokes" 

[2deaf]

Anyway, this crazy circuit tidbit popped up as a possible way to shift the effective cathode cap without changing net gain.  It doesn't work that way though.

I've used that one on the first stage of several hi-gain amps and it really gives them a lot of versatility with one knob.

I'm pretty sure I attached this one before, but this time it is for an example of a variable cathode.  You should take into account the rest of the pre-amp circuit and the volume I was playing this thing at.   

[Fresh_Start]

Thanks 2deaf.  Hadn't seen that before.

Am I correct in thinking that gain is highest with the wiper in the middle of the pot? (1/5 + 1/5 + 1/1.5 = 1/R)

Really appreciate knowing "my" late night idea wasn't crazy after all. 

Chip

[John]

2deaf, do you mind explaining the input a little bit, and also your first 250kl gain pot?


Input: Is the .022 in series with the grid to protect against possible leakage (if it happens) from the 47pf coming off the plate? Or does it also serve a tonal function?


Gain: Is that a lower loss way , instead of the normal pot /volume? Does it increase "gain" instead of just "volume"?


I always figure there's a good reason for more parts.    Thanks in advance!

[2deaf]

Am I correct in thinking that gain is highest with the wiper in the middle of the pot? (1/5 + 1/5 + 1/1.5 = 1/R)

The gain is the lowest when the wiper is centered, but the gain is 1 to 1.5 db more than completely unbypassed and the frequency response is flat.  At the 22uf extreme, the gain is 5db or so more than the centered setting and the frequency response is flat.  At the 0.47uf extreme, the gain is almost the same as unbypassed at 40Hz and does the "S" curve up to a little less than fully bypassed gain.  The curve is more similar to a 0.56uf by itself than to a 0.47uf by itself.

I think this control is better suited for high gain amps where you are still overdriving the pants off of the tubes at the lowest gain setting.   

[HotBluePlates]

Forget calculating

I can calculate one of those curves up to the point that it breaks into a curve at the top, but about the only value in that would be to test the limit of my attention span.  However, knowing one very simple calculation and the significance of it can give you a real leg up over blind trial and error.

We can't seem to get on the same page; not sure where I'm screwing up...

Yes, the simple calculation is set X_c = R_c, and calculate the resulting frequency. That's the -3dB point, assuming the response drops forever to zero output.

But it doesn't; there is a lower-limit defined by the feedback due to the cathode resistor. And if you have a series resistance between the bypass cap and ground, there is an upper limit to the gain boost due to partial bypassing of the cathode resistor. But how much boost? Where? How does it curve during the transition from with-feedback gain to slope due to Rk/Ck, and from that slope to the partial-bypass gain? Do you really want many calculations if you're needing to plot this by hand (your mention of converting/plotting graphs is what I was responding to when I said "forget calculating"). It's very easy to breadboard the circuit, run a swept test signal into the tube grid, measure output and plot the results to see what happens and where.

[2deaf]

Input: Is the .022 in series with the grid to protect against possible leakage (if it happens) from the 47pf coming off the plate? Or does it also serve a tonal function?

The .022uf cap. serves no tonal purpose.  The speakers were jumping out when I touched the strings and this cap. stopped that. 

Gain: Is that a lower loss way , instead of the normal pot /volume? Does it increase "gain" instead of just "volume"?

The gain control progressively destroys the negative feedback at the second stage, but it doesn't do it equally at all frequencies.  The lower frequencies take a bigger and bigger hit as the gain is turned up. 

[2deaf]

Yes, the simple calculation is set X_c = R_c, and calculate the resulting frequency. That's the -3dB point, assuming the response drops forever to zero output.

Calculating the total impedance of a high pass filter doesn't do us any good.  It just helps illuminate the significance of the cutoff frequency and why it is defined as such.  I think it is a good thing to understand what f_c is, but I probably take it further than necessary just because I enjoy the math.
   
That -3db point at f_c is a point on the curve for the high pass filter.  It appears on the "S" curves when the gain starts to rise at the low frequency end.  Knowing the value of f_c and knowing about the approximate location on the curve gives us a very good idea of what the frequency response is going to be.  Are we on the same page?

[John]

Input: Is the .022 in series with the grid to protect against possible leakage (if it happens) from the 47pf coming off the plate? Or does it also serve a tonal function?

The .022uf cap. serves no tonal purpose.  The speakers were jumping out when I touched the strings and this cap. stopped that. 

Gain: Is that a lower loss way , instead of the normal pot /volume? Does it increase "gain" instead of just "volume"?

The gain control progressively destroys the negative feedback at the second stage, but it doesn't do it equally at all frequencies.  The lower frequencies take a bigger and bigger hit as the gain is turned up.

Ahhh. Thanks!

[PRR]

> when the impedance of the capacitor equals the resistance of resistor.

In cathode bypass, there are two resistors.

The one you spent 12 cents for, and the one internal to the tube (cathode internal impedance).

They are in parallel for our purposes.

So the impedance you are bypassing is *less* than the resistor you see.

If you know Gm *at your operating point*, 1/Gm is the cathode impedance. For 12AX7, the book-value is not far-off from the way Leo Fender designed. For other designers (not copy-catting Leo) and for other tubes, the book-value of Gm may be very far off from the way the tube is used in the amp.

But the way self-bias works, you can *usually* assume the cathode impedance is very-similar to the cathode resistor.

Then your calculation goes by *half* of the resistor you see.

That's still approximate. The gain does not fall-off to infinity. It levels-out (as seen in graphs above). The level-out part is a gain about half of the fully bypassed gain. (Again, because resistor and cathode impedance are similar.) That's -6dB. So your estimation of the -3dB point is skewed because the curve doesn't go a "lot" below -3dB before it levels-out.

This can all be calculated.

I do not think it is worth the trouble, because you do not "know" what frequency will work best, for your guitar, rest-of-amp, speaker, room, and genre.

The curves/calcs can guide you to "the right octaves". 10uFd if you want full bass, 1uFd if you want the mids to stand above the bass, 0.1uFd to get bat-screech. But then you'll want to go up and down by 2X to find "the octave" for the action to happen. It is unlikely that you need to go up/down any finer than 1.4X, because the action is so gentle that 2.2uFd and 2.7uFd are the same for all practical purpose. (It is a very blunt knife.)

[2deaf]

So the impedance you are bypassing is *less* than the resistor you see.

Ya know ..., when you say something, and I already know it, your posts make perfect sense to me.  When you say something, and I don't know it or I'm not sure I agree, I have a hard time determining exactly what you are saying.  It sounds like the internal cathode impedance has to be taken into account as a parallel impedance with the cathode resistor when calculating gain or high-pass filter impedance.

My knee-jerk reaction is that since the internal cathode impedance never appears in the equation for Gain, the calculation depends on the resistor alone.  But the calculation is for a particular cathode resistance and the partially bypassed cathode uses a high-pass filter to obtain this impedance.  Maybe the resulting impedance is actually the result of the cap. in parallel with a resistor and an impedance.

So I calculated the total impedance of a 1.5K/0.68uf parallel RC circuit at fifteen different frequencies and then did the same for 0.7K/0.68uf.  It generated two "S" curves and of course the 0.7K/0.68uf is shifted to the right.

The gain of the triode in db's has a reasonably linear relationship with cathode impedance (at least in the range that we are using), so the gain curve should match the impedance curve.  I already have the gain curve for a 1.5K/0.68uf 12AX7 cathode and it matches the 1.5K/0.68uf impedance curve and doesm't match the 0.7K/0.67uf impedance curve.  The gain seems to be dependent only on the impedance of the cap. and the resistor in parallel.

 

[PRR]

> internal cathode impedance never appears in the equation for Gain

It must. But for historic reasons it is usually hidden in Rp and Mu. Rk is Rp/Mu.

[2deaf]

> internal cathode impedance never appears in the equation for Gain

It must. But for historic reasons it is usually hidden in Rp and Mu. Rk is Rp/Mu.

Thanks.

Congratulations on the 10,000th post milestone.