re-configurable amp - Princeton, Top Boost, JCM, yada

[BackwardsBoB]

Kevin O'Connor is your soulmate

Yeah. There are lots of very successful amplifiers that sound great and use switching. It's not uncommon to switch in another tube to increase drive. Channel switching, boost switches and bright switches are all popular and useable.

It's not too hard to grok how preamps work, but there's a second, longer step, where you learn how things sound, and the amp you're working on can really help with that.

I point you to Kevin O'Connor's TUT series of books. I think it's TUT1 that spends maybe a hundred pages on switching techniques and footswitches. Any switch configuration you might need is covered using JFETs, BJTs, MOSFETs and transmission gates. He generally winds up with JFETs. They're much lower power than relays, more reliable, etc. You need to consider what happens when you flip the switch. There's frequently a place to stick a cap that will help. The books aren't cheap, but they transfer lots of experience for less than the cost of a tube screamer. Kevin uses lots of switches. He also comes up with power stages that are true chameleons. You can call him at LondonPower.com and he'll point you to the right one, though switching is a theme throughout the series.

I'd also like to point out my babies, the Carvin X-series. Their power transformer kicks out +/- 15V, pretty much just for switching. The X100Bs use JFETs for channel switching, reverb bypass, triode input muting on the unused channel, and other purposes, using CMOS inverters running between ground and +15V to drive the JFETs based on the setting of various footswitches and pull-pots. There's a neat circuit in there that's a capacitively coupled gate that generates a pulse when noisy switching occurs that mutes the output for the duration of the pop. You can find schematics in the tech section at CarvinMuseum.com.

I think you should go for it, but figure out what the amp's supposed to be. Does it do smooth jazz? Fender clean? Blues? Rock'n'Roll? Rock? High-gain Rock? Metal? Is it for a club, a church, or a bedroom? The traditional sound for all these applications force different speakers, rectifiers, output tubes, power output, etc. Switchable sag, power scaling and output tube types/configuration can get you close to most of these sounds, and if you do a head, you can choose the right speaker for the application, but if you can figure out what you want the amp to do, knowing that you can build another for other things, you can optimize more. Take a look at your guitars and what you want to play, and go for an amp that meets your needs. Most of the guitar gods had their own tone, and don't deviate much - Richards, Gibbons, Santana, Iommi, Satriani, King, Hedrix... you get my point. And remember that you probably play guitar with two hands, and find it challenging, at least sometimes, so you really don't have the resources or time to be a sound man while you're actually playing.

[Steve Conner]

BackwardsBob, are you saying that you're the guy who designed the Carvin X-Series? Cool

there's a second, longer step, where you learn how things sound...
...you really don't have the resources or time to be a sound man while you're actually playing.

I totally agree with that. I'm an EE and programmer in my day job, and spend all day building complicated gizmos for industry. I also used to run a home studio producing drum'n'bass. What I love about electric guitar is its simplicity. It goes strings-wood-tubes-speaker and that's it. You really don't need that many options.

I got burnt out on the electronic music stuff just because of the sheer complexity of it. Sometimes I could spend an hour just figuring out why nothing was making any sound today. But I kept on playing guitar purely because I like it.

I've been homebuilding amps for years, and I finally settled on two setups. A happy amp and an angry amp. I think amp designs polarize quite neatly along that line. My happy amp is currently a 40 watt unit, more or less to a classic Fender/Marshall/Vox type circuit with spring reverb, in a 5E3 cabinet with some Eminence speaker with a seam cone. The angry one has a high-gain Boogie-style preamp, an active EQ circuit, 6550s, and an EVM12L reissue in a closed-back cab. They both sound like what you would expect, although Mr. Angry Eyes cleans up remarkably well.

[BackwardsBoB]

No. I'm the guy who bought them all.

I've got maybe 16 x-series amps. I've got a Fender DRRI I'm going to sell, a Marshall acoustic amp, and a Peavey bass amp, but everything else is x-series.

It appears that the X-series designs were generated by a small and changing team over a decade. They did a really good job. The amps are solid, and I definitely approve of the semiconductor tone stack, JFET switching, etc. There's a +/-15V supply in there that will be really handy for what I want to add.

The amps tend to have a mediocre drive channel, but I think they're just a few tweaks from something better.

I figure I need all of the X30s, because nobody loves, them, three of the X60s because they're so little and cute, you need 3 XT or XV 112s in case two get stolen, the XV212s make great boat anchors, and the heads bring down the wrath of God. Of course, you need them in both Tolex and grey fuzzy, and 6L6 and EL34.

I'm drawn to them because all my guitars sound really good dry, and I play music to fall asleep by. Carvin clean justs adds a bit of warmth and reverb. I can do one PC board with stuff options, and get a whole family of even better amps.

That and they're about $350 on ebay. There's no way you can buy a cabinet, chassis, reverb, PT, OT, knobs, fuseholder, speaker, corners, rubber feet, etc. for that much.

[pvsage]

It's not too hard to grok how preamps work, but there's a second, longer step, where you learn how things sound, and the amp you're working on can really help with that.

I point you to Kevin O'Connor's TUT series of books...He generally winds up with JFETs. They're much lower power than relays, more reliable, etc.
...
I'd also like to point out my babies, the Carvin X-series...There's a neat circuit in there that's a capacitively coupled gate that generates a pulse when noisy switching occurs that mutes the output for the duration of the pop. You can find schematics in the tech section at CarvinMuseum.com.

I think you should go for it, but figure out what the amp's supposed to be. Does it do smooth jazz? Fender clean? Blues? Rock'n'Roll? Rock? High-gain Rock? Metal? Is it for a club, a church, or a bedroom? The traditional sound for all these applications force different speakers, rectifiers, output tubes, power output, etc. Switchable sag, power scaling and output tube types/configuration can get you close to most of these sounds, and if you do a head, you can choose the right speaker for the application, but if you can figure out what you want the amp to do, knowing that you can build another for other things, you can optimize more. Take a look at your guitars and what you want to play, and go for an amp that meets your needs. Most of the guitar gods had their own tone, and don't deviate much - Richards, Gibbons, Santana, Iommi, Satriani, King, Hedrix... you get my point. And remember that you probably play guitar with two hands, and find it challenging, at least sometimes, so you really don't have the resources or time to be a sound man while you're actually playing.

Wow - so many sub-threads to respond to here!

First, one thing I'm having a bit of trouble grokking is the parallel RC tone shaping circuit, like the 470p cap/470K resistor combo found in Marshall amps and similar pairings in other amps, or the "treble bleed" cap across a preamp volume pot. I understand *what* they're doing (basically frequency shelving), but I'd like to be able to predict, mathematically, how changing these affects performance. Is there a formula for this, like the one for cathode bypass caps?

JFETs are a great alternative to relays - thanks for the reminder! I briefly looked for info on this when I was toying with distortion pedals, but for some reason, the "boutique" pedal community is married to "true mechanical bypass" switching...while they try in vain to capture the sound of tube distortion with solid state circuits. Another major upside of JFET switching is the things are dirt cheap!

The Carvin Museum? I *definitely* need to check this out! Thanks for the info.

As much as I respect the "guitar gods" and envy their technical prowess, I don't necessarily love their tone (except for Carlos - that man's lead lines are frakkin' tasty!) My favorite band right now, as far as guitar sounds are concerned, is Bowling For Soup - everything they play, from clean to crunchy to searing, just sounds so sweet to me, and I'd be very happy with and amp that gives me that range of sounds. Other bands that I love for distorted guitar are The Muffs, King's X, Kenickie, and Collective Soul, and I like the clean sound Phish has on some recordings. I'm focusing on low-power SE amps right now, but I'm not ruling out a PP amp in the future.

So...based on my musical tastes, what kind of preamp should I focus on? I think BFS would probably have the most influence on my choice of power tube and speaker, so...anyone know what amps they use?

[pvsage]

OK...here's my updated schematic

I've removed *a lot* of switching, removed the CF, moved the TMB before the gain control, and a few other yadas.

First, the preamp: I removed the impedance switch. This is intended as a dedicated guitar amp, so I figured HiZ should be acceptable across the board for this use. I changed the cathode bias resistor and bypass cap switching from "A or B" to "A || B" to increase safety (the relays (and I suspect SPDT JFETs) are "break before make") and reduce pop.

This is followed by a standard BF tonestack. I know the Princeton has a fixed resistor instead of a mid control, but since the extra pot doesn't increase insertion loss, I figured, "Why the heck not?" The 200p treble bleed cap on the volume control is actually inspired by the Marshall/Vox amps, but I realize some larger Fenders also have a switchable "bright" cap here.

Next is the tonestack recovery stage. Fairly self-explanatory - the cathode bypass switch is there because some Princetons have it and others don't; it also gives me just a little bit more control over the gain.

The next stage is bypassable tone-shaping, lifted from the Marshall JCM 800 2204. As you can see, I've managed to deal with the "path to ground" issue without resorting to an alternate bypass wiring that would have to break the grid path before shunting to ground, yet there's no change in the voltage divider.

Since the TMB has so much insertion loss, I don't expect to get much tube saturation until the lass stage, which is basically the same as the one that feeds the JCM's cathode follower. Again, I have a cathode bypass cap in case I want a little more "kick".


I have the first three stages at a lower voltage than the last stage because that's what's in the Marshall and Vox schematics. Could I safely run these stages at the same voltage as the final stage? What effect would this have on the sound?

[BackwardsBoB]

Parallel RCs are easy and hard. We know that the impedance is half the value of the resistor when 1/(2*pi*f*C)=470K, so for your 470pF||470k combo, this happens when f=1/(470k*470pF*2*pi)=720Hz. So we know the impedance of the circuit starts out at 470K at low frequencies, starts dropping near 720Hz, is half the value at 720K, and winds up dropping 3dB/octave at higher frequencies.

Now the hard part is what does that do?

If you look at your second stage cathode resistor, and you want to figure the effect of the 25uF parallel capacitor, you have to use a value of 776 Ohms for the cathode resistor. You can see that if you drive a little current into the cathode resistor at the cathode, the voltage will go up, the bias will go down, and the current from the tube will drop, so the current you send in will be partially cancelled by the drop in tube current, and the resistor will look smaller (smaller than expected voltage change from the current you injected). Following the hairy formula, with a mu and Rp and stuff in it, you get 776 Ohms. So the cap has to be larger than you'd expect. The design needs 20uF to be effective down to 8 Hz. On the other hand, that 22nF cap does a good job of passing most of the signal in its higher impedance circuit.

That 470pF||470K we were looking at is driven by the signal at the cathode with a source impdance equal to the plate resistor in parallel with Rp, through a 22nF cap, and terminated with 470K. The math is straightforward, but there's an alternative, and it comes in really handy when you want to change the cross-over points in your tone stack.

Wonks use analog simulators - Spice and its variants. They're generally not cheap, but Linear Technology gives away SwitcherCAD, and it can be used for tube circuit simulation. You can find a starting point at duncanamps.com/technical/ltspice.html.

If you don't simulate or calculate, you'll be forced to clone circuits from amps you like, and if you don't clone the whole amp, you won't really know the system gain, and how hard you are driving the power tubes. That's not a big problem. You can probably tweak the gain here and there and get where you want to go.

But when you go switching out a cathode bypass cap in the first stage, later stages won't be driven as hard, and the amp will get softer. Is that your goal, or are you doing it for tone? Will it do what you want?

Simulation could be big fun and worth the bother. Just don't ask me for advice. I haven't tried it yet. I don't know how well the tubes' models model saturation, cutoff, etc.

[pvsage]

That 470pF||470K we were looking at is driven by the signal at the cathode with a source impdance equal to the plate resistor in parallel with Rp, through a 22nF cap, and terminated with 470K. The math is straightforward, but there's an alternative, and it comes in really handy when you want to change the cross-over points in your tone stack.

Wonks use analog simulators - Spice and its variants. They're generally not cheap, but Linear Technology gives away SwitcherCAD, and it can be used for tube circuit simulation. You can find a starting point at duncanamps.com/technical/ltspice.html.
...
But when you go switching out a cathode bypass cap in the first stage, later stages won't be driven as hard, and the amp will get softer. Is that your goal, or are you doing it for tone? Will it do what you want?

Simulation could be big fun and worth the bother. Just don't ask me for advice. I haven't tried it yet. I don't know how well the tubes' models model saturation, cutoff, etc.

Thanks for the formulas, and for the link to LTSpice! I can't wait to try it out. I already have the Duncan Tonestack Calculator, which is what I had used to try tweaking the slope resistor and tone caps in the Fender TMB to make it a little closer to a Vox TB without changing the pots. I *think* I have a game plan...maybe compare virtual models of my amp with models of the big guys' amps. It's cool that Duncan provides these tools for us amateurs!

As for switching bypass caps...yes, I mainly intend to use the caps on the second and fourth stages as a gain boost - I understand some manufacturers call this a "fat boost". In the first stage, on the other hand, my main goal is Fender (1K5||22u, shelf at 9.3Hz) vs. Marshall (2K7||680n, shelf at 232Hz) - I got the component values from published schematics and the frequencies from a chart that I picked up at Aiken Amps.

By the way, those are supposed to be 22u; I'm not sure where I got 25...maybe the GA-5 Reissue? You're not going to find that value in any IEC 60063 decade...

[Bruce / Mission Amps]

As for switching bypass caps...yes, I mainly intend to use the caps on the second and fourth stages as a gain boost - I understand some manufacturers call this a "fat boost". In the first stage, on the other hand, my main goal is Fender (1K5||22u, shelf at 9.3Hz) vs. Marshall (2K7||680n, shelf at 232Hz) - I got the component values from published schematics and the frequencies from a chart that I picked up at Aiken Amps.

Remember... although fairly accurate, I drew that chart for Randall Aiken with respect to a grid driven, common cathode 12AX7 and only a 12AX7.

[Bruce / Mission Amps]

I've removed *a lot* of switching, removed the CF, moved the TMB before the gain control, and a few other yadas.

I have the first three stages at a lower voltage than the last stage because that's what's in the Marshall and Vox schematics. Could I safely run these stages at the same voltage as the final stage? What effect would this have on the sound?

I think the biggest effect you'll have with this circuit is the very high cascaded gain and the difficulty in controlling it.... and all that before the additional gain of the phase inverter.
If you think about this with respect to how much gain each stage will have, it would not be unreasonable to expect a gain of around 30 for each stage except the one with the 10k cathode resistor.... call that one 10.
Do the math... using a simple 100mv input signal.
Yes, the tone stack will kill of a lot of the signal but you will could end up with a nasty sounding monster on your hands.
Now this could be the type of sound you want so I won't tell you not to do it.
But be prepared for having to mess for a long time so that isn't just a one trick pony... or should I say a 100 bumble bees in a paper bag trick.
Truthfully, I am astonished by the number of kids that like that juvenile sound so maybe I'm just told old to comment about the thing.

[BackwardsBoB]

It's interesting that the designers of the amps you are taking the best of didn't have overdrive in mind. It still works very well. I'm sure Bruce's amps sound really sweet, both clean and overdriven, and the originals are wonderful as well.

But your favorite of the "gods" I listed was Santana, and his Boogie pre is something else all together.

Tubes have three kinds of "overdrive" effect (counting second-order distortion as the third), and which one you get is determined largely by the operating point on the load line. The operating point is determined by the bias current, which is determined by the DC cathode impedance. With low bias current, an adequate input signal will encounter cutoff first. Cutoff yields compression and rounded clipping of the bottoms of the input signal waveform, and second-order harmonics. Increase the bias current to a middle value, and saturation and cutoff are both far from the operating point, for maximum clean headroom. High bias current makes it easy to achieve saturation distortion, which provides a sharp clipping of the bottom of the input waveform (if the input is strong enough). It's easy enough to put cathode bias under front panel control with a pot, opamp, NPN BJT, a few caps and resistors, and a low current single supply. Variable interstage attenuation is a no brainer, so a cascade of two triodes can be put under control of four pots, yielding compressive cutoff or raunchy saturation distortion of both sides of the waveform, or only one side, with the magnitude and onset of the distortion controlled by the input attenuators. You can also make it through the first triode clean with gain, and get both distortions in the second. Or you can run the first triode in its linear region, then bias the second stage a bit cool with moderate drive, and pick up second-order distortion without much clipping. A pentode in front to get the input gain up, followed by maybe a cathode follower driving a tone stack with a bright switch, and you're there. Lots of knobs though. Once you find out where you like them, you can switch to fixed resistors, leaving one of the attenuators in as a drive control. The Fathers never did this, because they were trying to avoid the whole thing altogether. They more-or-less centered the operating point, and we wound up with EMGs, and, thankfully, great power-stage distortion.

This is the direction I'm headed in.

[Steve Conner]

With low bias current, an adequate input signal will encounter cutoff first. Cutoff yields compression and rounded clipping of the bottoms of the input signal waveform, and second-order harmonics.

Yes. For instance, I believe this is what happens when you put a 12AT7 in a socket intended for a 12AX7. Could be wrong, though.

High bias current makes it easy to achieve saturation distortion, which provides a sharp clipping of the bottom of the input waveform

If I understand right, this kind of clipping happens when the grid is pushed positive and draws current. It is relatively sharp. But remember that as soon as grid current starts to flow, the tube starts biasing itself, shifting its own operating point. This is where a lot of the life in tube amp sound comes from.

It's easy enough to put cathode bias under front panel control with a pot, opamp, NPN BJT, a few caps and resistors, and a low current single supply.

I've done this, but with a simpler method. I basically hooked the stage's grid leak resistor up to an adjustable negative supply, that I got from the amp's main bias supply. So it's a combination of fixed and cathode bias. I found that when using a 12AX7, I prefer it with just the cathode bias. But I left the control anyway for use with other tube types.

[pvsage]

I think the biggest effect you'll have with this circuit is the very high cascaded gain and the difficulty in controlling it.... and all that before the additional gain of the phase inverter.

Actually, I'm sticking with a SE power amp for now. Hence, no PI.

If you think about this with respect to how much gain each stage will have, it would not be unreasonable to expect a gain of around 30 for each stage except the one with the 10k cathode resistor.... call that one 10.
Do the math... using a simple 100mv input signal.
Yes, the tone stack will kill of a lot of the signal but you will could end up with a nasty sounding monster on your hands.

I understand this completely! My primary reason for building that second amp - the one with all the wide-open gain stages - was because I had heard first-hand how much signal a tone stack takes out of the Blackheart Little Giant, and I wanted to see how much gain was too much for my use. With two gain stages, the preamp didn't overdrive at all; with four, the final stage clipped with even the tiniest amount of input signal, which made it virtually unuseable; with three, I start to get saturation with the gain at about 10 o'clock, which works just fine for me - this is about where my Valve Junior (mostly stock, but with HiZ input and a big Hammond OT) starts to saturate the power tube.

Now this could be the type of sound you want so I won't tell you not to do it.
But be prepared for having to mess for a long time so that isn't just a one trick pony... or should I say a 100 bumble bees in a paper bag trick.
Truthfully, I am astonished by the number of kids that like that juvenile sound so maybe I'm just told old to comment about the thing.

Actually, it was more like ten thousand hornets bouncing around inside an oil drum!

I was under the impression that the overdriven guitar amp sound was somewhat mature, almost as old as Rock music itself. :-) Just kidding - I know what you meant. Many younger rockers are actually gravitating towards acoustic guitar now! (At 38, while I don't yet consider myself a grumpy old man, I already find myself saying "kids these days..."!)

If the "Marshall 2204" stage proves to be too problematic, I could always wire the first two triodes in parallel, or cascode one of the stages, or use the spare triode as a "normal" input with a separate gain control...

I'm going for preamp distortion with the master volume turned down because I want to be able to practice with some distortion, but even a single overdriven EL84 is LOUD for a practice amp, especially when pushing a Celestion. I've read somewhere that if a triode's plate-to-cathode voltage is higher than a certain point (180V for a 12AX7, iirc), the wave form gets compressed at saturation, causing softer clipping. Anyone else heard this?

[BackwardsBoB]

Oh! That's really good!

Steve Connor wrote:

I've done this, but with a simpler method. I basically hooked the stage's grid leak resistor up to an adjustable negative supply, that I got from the amp's main bias supply. So it's a combination of fixed and cathode bias. I found that when using a 12AX7, I prefer it with just the cathode bias. But I left the control anyway for use with other tube types.

You change the bias current by changing the grid voltage and leaving the resistance fixed. That might be easier sometimes.

[pvsage]

It's interesting that the designers of the amps you are taking the best of didn't have overdrive in mind. It still works very well. I'm sure Bruce's amps sound really sweet, both clean and overdriven, and the originals are wonderful as well.

But your favorite of the "gods" I listed was Santana, and his Boogie pre is something else all together.

Santana's sound is truly unique - overdriven, yet very toneful. I'm sure his choice of guitar and amp has something to do with this, but I never underestimate the effect of the player's technique itself. I'd never presume to try to copy his tone! Besides, I see myself as more of a crunchy rhythm person.

Tubes have three kinds of "overdrive" effect (counting second-order distortion as the third), and which one you get is determined largely by the operating point on the load line. The operating point is determined by the bias current, which is determined by the DC cathode impedance. With low bias current, an adequate input signal will encounter cutoff first. Cutoff yields compression and rounded clipping of the bottoms of the input signal waveform, and second-order harmonics. Increase the bias current to a middle value, and saturation and cutoff are both far from the operating point, for maximum clean headroom. High bias current makes it easy to achieve saturation distortion, which provides a sharp clipping of the bottom of the input waveform (if the input is strong enough). It's easy enough to put cathode bias under front panel control with a pot, opamp, NPN BJT, a few caps and resistors, and a low current single supply. Variable interstage attenuation is a no brainer, so a cascade of two triodes can be put under control of four pots, yielding compressive cutoff or raunchy saturation distortion of both sides of the waveform, or only one side, with the magnitude and onset of the distortion controlled by the input attenuators. You can also make it through the first triode clean with gain, and get both distortions in the second. Or you can run the first triode in its linear region, then bias the second stage a bit cool with moderate drive, and pick up second-order distortion without much clipping. A pentode in front to get the input gain up, followed by maybe a cathode follower driving a tone stack with a bright switch, and you're there. Lots of knobs though. Once you find out where you like them, you can switch to fixed resistors, leaving one of the attenuators in as a drive control. The Fathers never did this, because they were trying to avoid the whole thing altogether. They more-or-less centered the operating point, and we wound up with EMGs, and, thankfully, great power-stage distortion.

This is the direction I'm headed in.

WOW that's a bit to digest. As for the pentode input stage, I'm fairly wedded to the 12AX7 - it's good enough and flexible enough for The Fathers, so why should I need anything else? I'm also shelving cathode followers for now, as I've heard other amps get into Marshall territory without one. (I'm sure the EL34 power tube had a lot do do with that...) The rest of your suggestions...I need to give myself a refresher course in op amps, but it definitely looks like something I'll experiment with for my *next* project.

[pvsage]

I think the biggest effect you'll have with this circuit is the very high cascaded gain and the difficulty in controlling it.

Actually, what I meant is "What if I left out the last RC filter from the power supply and ran all the triodes at the same plate voltage?"