Paralleling PI output coupling caps with a switch

[seveneves]

Hi there,

I am slowly planning out a Vox AC30/6 top boost build and am thinking about putting in a few switches to change the voicing of the amp.

This may be crazy but I was thinking about putting in a switch (4 pole) to parallel the 0.047 PI output coupling caps C6 and C9 (as found in the "Treble" model) with 0.1 caps for a total of 0.15 (as in the "Normal" model).

Any reason why this would be a bad idea?

I also wanted to do a few more switches, these seem (?) less problematic:

1. Split cathode (seen in "Treble" models) switch for Brilliant channel (1k5/0.1 cap)
2. Bright switch for Normal channel (220pF)
3. Switch for "Normal" or "Treble" model voicing for Cut control (paralleling C10 as a 0.0022 cap with same)

I'm a newb and still learning a lot so if it all sounds ridiculous, you'll know why...

For reference, schematics for AC30 "Normal" and "Treble" versions

[Raja_Kentut]

you may import hum, and most likely it will * POP * loud when the caps behind the PI are switched when the Amp is on.
To prevent humming from being introduced, you could, for example, use a reed relay or other signal relay close to the capacitors for switching.
The switching Voltage for the relais could be taken from the heater supply. (rectify and use voltage regulator) or you take a 9V coil relais (just rectify and elco)

[cdemike]

I made a similar all-access type of Marshall build. Very fun learning experience, and I got a very versatile amp out of it. Problem I found is that the amp is much more versatile than I am as a player, so I usually set it up the same way, though it's nice to know that if I'm in an unusual recording situation I have a lot of options.

I suspect you may have some success reducing switch popping with series rather than parallel caps, though I think any conventional switch will probably give you some switch popping. I'm thinking something along these lines: https://el34world.com/Forum/index.php?a ... 3866;image
150nF and 68nF output couplers would give you the desired 47nF. A dual-gang 5K blend pot could also give you some pop-free choices, though that'd also amount to variable grid stoppers (probably wouldn't make a huge difference, IMO). The other way would be to use a 3 position 4PDT standby switch and just have it change when you switch the amp out of standby. That could be a 3 position 3PDT if you don't use the traditional standby and instead just have a switchable shunt/mute on across the output section, i.e., like a cross-line master volume set to zero. A standby switch is not necessary anyway unless you're considering a DC coupled cathode follower at some point (e.g. if you wanted to have a top-boost channel), and even then there are workarounds like described at the bottom of https://valvewizard.co.uk/dccf.html. The most noiseless solution might be an optocoupler, which changes resistance somewhat gradually (usually switches between near-zero resistance and very high resistance to give "open" and "closed" settings), so you'd definitely have less switching noise if you wanted to go down the remotely-operated switch road. Raja_Kentut's suggestions in terms of power supply would apply to an optocoupler as well.

Different values of top cut capacitors could also be done using a 3-position cross-line mute switch in lieu of a standby if you wanted to minimize switch nosie. If I understand the way the top cut control works correctly, it works as a shelving filter that allows phase cancellation across the phase inverter's outputs (https://ampgarage.com/forum/viewtopic.php?t=35508). Since the corner frequency of a high pass filter is equal to 1/(2πRC), using a 2.2nF cap with a 500K should give you an equivalent circuit that acts identically to the bassier circuit in the top half of the pot's resistance and like the brighter circuit in the lower half. Because Vox used a logarithmic taper pot, though, that won't correspond to the pot's travel. I suspect 4.7nF caps in series would be quieter than switchable parallel caps here too.

Switchable split and shared cathodes are not uncommon and work well, though this is one area where there's almost no way to avoid popping with conventional switches since you're changing the operating point of the tube. Raja_Kentut's suggestion of a reed relay I suspect would encounter the same issues with popping here since you're still suddenly changing the operating point of the tube. One way around that would be to wire a parallel relay to mute the amp at the same time as operating the relay on the V1 cathode, but things get complicated quickly if you want completely silent operation since it'd be switching the operating point simultaneously with the mute switch, so you may still get switch noise. Best (simplest?) solution to avoiding switch noise that I'm aware of again would be to use an optocoupler. I think the simplest solution would be to just live with the switch noise or to set up half of V1 with equivalent values as the shared cathode on one side (3k with 12.5uF bypass) with the other being the usual 1.5k/25uF or 1.5/100nF values.

I used the simple Fender-style approach built around a SPST switch for my bright cap switch (see https://schematicheaven.net/fenderamps/ ... _schem.pdf). If you want extra insurance against popping, you can set it up so your switch shorts a resistor in series with the bright cap (1M is fine) rather than just opening and closing a circuit with the bright cap.

[Raja_Kentut]

the relays are only mentioned for avoiding hum picked up by wires in high impedance paths. You are right, they will pop like a switch.

[cdemike]

the relays are only mentioned for avoiding hum picked up by wires in high impedance paths. You are right, they will pop like a switch.

Ah, yeah that makes sense. Definitely agree on the noise.

I also just re-read the first post and saw you planned on a top-boost channel, i.e., will have the cathode follower. You can still get around the traditional standby switch using the resistor and diode combination from the link to Merlin's site, Valvewizard if you wanted to go down the standby switch road. I'm actually not sure how common arcing is in 12AX7s in DC coupled cathode followers on startup. PRS's Marshall clone doesn't have a standby switch, and I assume they considered potential warranty claims stemming from toasted V2 tubes, so on that basis I gather it's a non-zero but unlikely possibility. Still, I'd rather avoid it, esp. since a switch or a diode/resistor combo is cheaper than a new tube and a lot less frustrating than an amp going down before a practice, rehearsal, show, recording session, etc.

[R.G.]

A reliable way to stop pops from switching in capacitors is to install the switchable caps in series with a resistor. Do the switching by shunting the resistor. The resistor keeps the cap at the same DC voltage all the time, so there is no sudden DC voltage change to pop.

Relays are handy, but you can also use MOSFET-output optocouplers if you'd rather switch the LED instead of the relay's coil.
No, the MOSFETs don't introduce distortion.

[seveneves]

I made a similar all-access type of Marshall build. Very fun learning experience, and I got a very versatile amp out of it. Problem I found is that the amp is much more versatile than I am as a player, so I usually set it up the same way, though it's nice to know that if I'm in an unusual recording situation I have a lot of options.

Thanks for your reply. Lots of great info/food for thought. My questions and comments below.

I suspect you may have some success reducing switch popping with series rather than parallel caps, though I think any conventional switch will probably give you some switch popping. I'm thinking something along these lines: https://el34world.com/Forum/index.php?a ... 3866;image
150nF and 68nF output couplers would give you the desired 47nF. A dual-gang 5K blend pot could also give you some pop-free choices, though that'd also amount to variable grid stoppers (probably wouldn't make a huge difference, IMO). The other way would be to use a 3 position 4PDT standby switch and just have it change when you switch the amp out of standby. That could be a 3 position 3PDT if you don't use the traditional standby and instead just have a switchable shunt/mute on across the output section, i.e., like a cross-line master volume set to zero. A standby switch is not necessary anyway unless you're considering a DC coupled cathode follower at some point (e.g. if you wanted to have a top-boost channel), and even then there are workarounds like described at the bottom of https://valvewizard.co.uk/dccf.html. The most noiseless solution might be an optocoupler, which changes resistance somewhat gradually (usually switches between near-zero resistance and very high resistance to give "open" and "closed" settings), so you'd definitely have less switching noise if you wanted to go down the remotely-operated switch road. Raja_Kentut's suggestions in terms of power supply would apply to an optocoupler as well.

Er... I'm probably going to pass on this. I understand the difference in the PI output coupling caps between the "Normal" and "Treble" models will probably be hard to discern. That, and the complexity involved with doing this. And the hum/pop.

Since the corner frequency of a high pass filter is equal to 1/(2πRC), using a 2.2nF cap with a 500K should give you an equivalent circuit that acts identically to the bassier circuit in the top half of the pot's resistance and like the brighter circuit in the lower half. Because Vox used a logarithmic taper pot, though, that won't correspond to the pot's travel. I suspect 4.7nF caps in series would be quieter than switchable parallel caps here too.

Not sure I get this here... are you saying I could use a 500k (log) pot with the 2n2 OR two 4n7 in series if I use a switch? I kind of like the latter (if I'm understanding correctly) as it seems that I'd get the full range of the cut pot with either the 0.0022 ("Treble") or 0.0047 ("Normal") cap.

I think the simplest solution would be to just live with the switch noise or to set up half of V1 with equivalent values as the shared cathode on one side (3k with 12.5uF bypass) with the other being the usual 1.5k/25uF or 1.5/100nF values.

Also not getting this one... I read this in M. Zollner's Physics of the Electric Guitar amplifier chapter:



Is that 3k you're recommending what JMI should have done instead? Does putting in the proper 3k value change the tonality compared to the 1k5/0.1 mistake?

I used the simple Fender-style approach built around a SPST switch for my bright cap switch (see https://schematicheaven.net/fenderamps/ ... _schem.pdf). If you want extra insurance against popping, you can set it up so your switch shorts a resistor in series with the bright cap (1M is fine) rather than just opening and closing a circuit with the bright cap.

I forgot about using a resistor. Thank you for that reminder.

[seveneves]

BTW, what do you all think of this solution?

[Roe]

BTW, what do you all think of this solution?

yes series might be better. note that the ac30/4 used 10nf PI coupling caps, which reduces blocking distortion and bass mud. Personally, I am using 15nf (had plenty of these caps) and it sounds great with more than enough bass for guitar.

[cdemike]

BTW, what do you all think of this solution?

yes series might be better. note that the ac30/4 used 10nf PI coupling caps, which reduces blocking distortion and bass mud. Personally, I am using 15nf (had plenty of these caps) and it sounds great with more than enough bass for guitar.

Agree with Roe here, and the more I think about R.G.'s post, I think he's right that you'd probably get minimal switching noise if you used an arrangement similar to the one shown in figure 2 in that link. I've attached a drawing of how I'd probably do it based on R.G.'s post. I didn't pick 10k for a specific reason -- I think as long as it's reasonably (but likely still arbitrarily) large enough to block signal from the switched coupling cap.

I made a similar all-access type of Marshall build. Very fun learning experience, and I got a very versatile amp out of it. Problem I found is that the amp is much more versatile than I am as a player, so I usually set it up the same way, though it's nice to know that if I'm in an unusual recording situation I have a lot of options.

Thanks for your reply. Lots of great info/food for thought. My questions and comments below.

I suspect you may have some success reducing switch popping with series rather than parallel caps, though I think any conventional switch will probably give you some switch popping. I'm thinking something along these lines: https://el34world.com/Forum/index.php?a ... 3866;image
150nF and 68nF output couplers would give you the desired 47nF. A dual-gang 5K blend pot could also give you some pop-free choices, though that'd also amount to variable grid stoppers (probably wouldn't make a huge difference, IMO). The other way would be to use a 3 position 4PDT standby switch and just have it change when you switch the amp out of standby. That could be a 3 position 3PDT if you don't use the traditional standby and instead just have a switchable shunt/mute on across the output section, i.e., like a cross-line master volume set to zero. A standby switch is not necessary anyway unless you're considering a DC coupled cathode follower at some point (e.g. if you wanted to have a top-boost channel), and even then there are workarounds like described at the bottom of https://valvewizard.co.uk/dccf.html. The most noiseless solution might be an optocoupler, which changes resistance somewhat gradually (usually switches between near-zero resistance and very high resistance to give "open" and "closed" settings), so you'd definitely have less switching noise if you wanted to go down the remotely-operated switch road. Raja_Kentut's suggestions in terms of power supply would apply to an optocoupler as well.

Er... I'm probably going to pass on this. I understand the difference in the PI output coupling caps between the "Normal" and "Treble" models will probably be hard to discern. That, and the complexity involved with doing this. And the hum/pop.

R.G. is much more knowledgeable and experienced than I am, and if I'm reading his post right, it looks like he's less concerned about switching noise if you use a shuntable resistor arrangement than I was. I may be misinterpreting his post, but I'll defer to the more experienced members here.

Since the corner frequency of a high pass filter is equal to 1/(2πRC), using a 2.2nF cap with a 500K should give you an equivalent circuit that acts identically to the bassier circuit in the top half of the pot's resistance and like the brighter circuit in the lower half. Because Vox used a logarithmic taper pot, though, that won't correspond to the pot's travel. I suspect 4.7nF caps in series would be quieter than switchable parallel caps here too.

Not sure I get this here... are you saying I could use a 500k (log) pot with the 2n2 OR two 4n7 in series if I use a switch? I kind of like the latter (if I'm understanding correctly) as it seems that I'd get the full range of the cut pot with either the 0.0022 ("Treble") or 0.0047 ("Normal") cap.

I meant you can approach it two ways: either use 2.2nF with a 500KA pot, which would be equal to the stock arrangement using a 4.7nF cap with a 250K pot with the pot set to its full 250k resistance. So going between 250k and 500k will start to give you the range that the 250KA + 2.2nF filter would give you. That's probably what I would do, though I'd likely end up using a linear taper pot so I'd know where I'm setting it and likely change it down the road to whatever version I find I like more. It'd be harder to use the control that way since the audio taper pot allows much finer adjustment at lower settings, but if your build goes anything like mine, I anticipate you'll end up using one side of the knob much more than the other. The other way to do it would be to have the cap also get switched on a 250KA pot, which I'd also recommend being a series capacitor switched via a shunted series resistor (basically the same arrangement being recommended for the coupling caps).

I think the simplest solution would be to just live with the switch noise or to set up half of V1 with equivalent values as the shared cathode on one side (3k with 12.5uF bypass) with the other being the usual 1.5k/25uF or 1.5/100nF values.

Also not getting this one... I read this in M. Zollner's Physics of the Electric Guitar amplifier chapter:



Is that 3k you're recommending what JMI should have done instead? Does putting in the proper 3k value change the tonality compared to the 1k5/0.1 mistake?

That's correct. Think of it as running two paths of current through the 1.5k resistor vs. one path through a 1.5k resistor. Those paths are equivalent, being two basically identical triodes, so the current requirements are the same. Doubling the current through the 1.5k resistor means it's electrically equivalent to having a separate 3k resistor on each triode. So rather than switching the operating point of the triode, I'd suggest running each side at the different operating points for simplicity's sake. I also don't think there's really any way around serious switching noise short of an optocoupler or a MOSFET switch when you're changing the cathode resistor (read: operating point) of V1, so running each side differently kills two birds with one stone.

[cdemike]

Forgot to post this: https://www.youtube.com/watch?v=M_Uq6yzqBGQ

Obviously very different amp than an AC30, and I agree with Roe that you get plenty of bass with smaller coupling cap values. I'm not sure how much of what you can hear in that video would carry over, especially given the differences between EL34s and EL84s.

[seveneves]

That's correct. Think of it as running two paths of current through the 1.5k resistor vs. one path through a 1.5k resistor. Those paths are equivalent, being two basically identical triodes, so the current requirements are the same. Doubling the current through the 1.5k resistor means it's electrically equivalent to having a separate 3k resistor on each triode. So rather than switching the operating point of the triode, I'd suggest running each side at the different operating points for simplicity's sake. I also don't think there's really any way around serious switching noise short of an optocoupler or a MOSFET switch when you're changing the cathode resistor (read: operating point) of V1, so running each side differently kills two birds with one stone.

OK, I think I finally understand this a little better now. FWIW, I had to think about this, research and ruminate on it over the course of several days until it started to make a little more sense, which led to me only replying on this now.

So ideally, the treble model should have had each triode set up with a 3k cathode resistor and 12.5uF bypass cap instead of the 1k5/0.1 + 1k5/25 deal shown on the schematic.

And, seemingly, JMI failed to do that on each triode of V1 (at least per the schematic). So if I'm tracking, the way the schematic is drawn is that the bias point of each triode (and, in effect, of the tube) is shifted.

With your recommendation of only changing the bias point of ONE of the triodes, I am now trying to determine which triode (which corresponds to either the normal or brilliant channel) to shift the bias point on. Or... do I want to shift the bias point at all?

I understand that if we use 3k/12.5uF for each triode's cathode that this is electronically equivalent to a shared cathode having 1k5/25uF, wrt the operating point of the tube. But tonally, is it equivalent? If not, why not?

[cdemike]

Yeah, so if the treble version had an identical operating point as the normal version, it'd use separate 3k/12.5uF resistor+cap combinations on each side of the triode. If the normal version used an identical operating point as the treble version, it'd use 750R cathode resistors and double the capacitance value you'd chose (pick either side; the rule remains the same). Generally, when you're talking about a shared cathode arrangement, the rule is to double the capacitor value and half the resistor value as compared to an equivalent split cathode arrangement with an identical operating point. That's because the two triodes are sharing the one resistor and each pushing current through it. Per Ohm's law: twice the current, twice the voltage drop at the same resistance.

JMI's mistake (or design choice -- I'm not really sure what motivated or precipitated the change in V1's design) was to use the same cathode values, even though retaining 1.5k cathode resistors in a split cathode arrangement places the value much closer to center-bias. The 1.5k shared cathode arrangement has an unusually cool bias, IMO.

AFAIK, the difference in cathode resistor changes the sound two ways:
1. Cooler bias: The cooler bias will generally lend a more compressed sound and will have less headroom than a center-biased stage
2. Difference in the shelving frequency produced vis-a-vis the relationship between the cathode resistor and cathode bypass cap. As the cathode resistor increases in value, the shelving frequency moves higher in the frequency spectrum. So the treble boost from, say, the 680nF bypass cap in a Marshall-style V1 bright stage (the default setup in the calculator at the bottom of this paragraph) is more pronounced with the stock 2.7K resistor than a 1.5k cathode resistor. This gets complicated but I'd recommend playing with this calculator to see the effect: https://www.ampbooks.com/mobile/amplifi ... alculator/

BTW, I don't know if you've looked through Merlin's site or have any of his books, but I rely on these a ton when approaching more technical questions like the impact of a gain stage's bias on its sound. He posted the first chapter of his preamps book online, and it goes into great depth explaining why different bias points sound they way they do and how to anticipate the impact of cathode bypass resistors: https://valvewizard.co.uk/Common_Gain_Stage.pdf

Basically, I'd recommend your normal channel use a 3k/12uF arrangement and your bright channel use 1.5k and 100nF. You could use 25uF if you prefer the normal channel from the "treble" version, but I anticipate it'd sound pretty similar to the other side aside from the obvious difference in coupling capacitors. But you can pretty easily find what you like best by tweaking your build as it develops -- there's very little current or voltage at the V1 cathode, so you could use a pot to dial in the sound you like best (be warned: it'll be extremely noisy when you turn the pot), then take the pot out of circuit to measure the value you settled on by measuring the resistance between the wiper and whichever terminal you connected to ground. I'd just use alligator clips and a 5k linear pot. The commonly available full-size Alpha pots are rated for 1/2 watt so should be more than fine to use for that purpose.