Footswitches

The circuit design on which I'm basing mine includes an overdrive (OD) feature: the input signal is amplified and shaped by a pair of preamp triodes and is either passed as a "clean" signal to the power section or diverted to a second pair of triodes, the OD section, before then getting routed to the power section. (It's important to note that the OD feature is most useful for lead/melodic playing, but not for playing chords to accompany a tune - at least not to my ear... the distortion created by the OD can make it hard to distinguish relationships between multiple notes, so it's less desirable for chords more complex than simple triads). If this amp is to be used for live playing, then I'd like to be able to easily switch between clean and OD during the performance of a piece, when reaching to a switch on the front panel of the amp would be impractical. A remote footswitch is a commonly implemented solution for this problem - all of the amps I've owned have had this feature... one I have taken for granted until having to design and build it into an amp myself.

Switches mounted on the front panel of an amp, such as those I included in my first build, typically carry and route the guitar signal this way or that through the amp. But, in the case of a remote foot switch design, we're moving the switch outside of the protection of the chassis, far from the delicate audio circuit, and adding a dozen or so feet to the length of the signal path. This would be likely to degrade the quality of the signal and introduce noise into the circuit - it's really just not done as far as I know, at least not in modern amps. Instead, we make the footswitch simply an actuator that controls an electronic switch that is inside the chassis, which keeps the signal path short and protected. To do this, we need an actuator for each switch and access to a power supply to drive them. It's not that hard to do, and it's been done a jillion times, but it IS a separate subsystem in the amp that requires it's own design and implementation effort, and wasn't something I had thought that much about before now.

In my design, even though I've got 3 switches in the circuit (Bright / Pre-amp Boost (PAB) / OD), I already have a two-button foot-switch that came with the salvaged Fender amp, so I'll just implement two of them, the PAB and OD. I also want to be able to actuate them in the usual way, with switches on the front panel, AND I'd like to have some visual indicators that tell me if each switch is on or off - LEDs will do. Here's the newly added section of the schematic that describes this additional subsystem.

Everything to the right of the lamp is newly added to support a pair of footswitches for the PAB and OD, as well as the associated panel-mounted switches.

Here are the bits and bobs needed to implement the switches: salvaged footswitch unit, relays and their mounting boards (one is assembled in the picture), power supply board with bridge rectifier and smoothing capacitor.  

Also, made the major mods to the chassis, cutting holes and installing mounts for the power transformer and tubes:

 

              

  

More on Screens

So, I went back to Merlin Blencowe's site and re-read (for about the 5th time) his example on a SE design (his, with an EL34 power tube). The part about the screen resistor now makes a lot more sense (though I still could not explain it to anyone). The thing I missed or didn't understand well enough the first 4 times, was that the location on the load line graph of grid voltage curves is dynamic with respect to screen voltage/current, which changes as the current flow shifts away from the plate to the screen, as in near-0 and positive grid voltage conditions. So, the trick is to calculate the screen resistor size ONLY based on a 0-volt condition, when the screen is flowing most of the tube's current... BUT, it's not doing that most of the time, so for clean running (the dominant condition) you still calculate bias and quiescent operating point based on zero signal (which means about -20 volts on the grid in my case). 

His three paragraphs on the EF86 pentode here are a good supplement to the SE example page with regard to screen current voltage, and a little more definitive, though pentode-general and not power amp specific (he also mentions by-passing the screen resistor with a capacitor, which I assume would work more or less like it does on a cathode to eliminate negative feedback and compression, but that's going to have to be another tutorial!).

I'll make a final determination on the exact value of the screen resistors once I get the real power supply up and running, but it'll be between 470R and 1k for these...   just like everyone else does it (go figure)  :-) 

Circuit Test Configuration 4

 Well... two steps back... but the path forward is now better. 

I wanted to be sure the power amp wasn't playing a part in what seemed to be tone and dynamics deficiencies of the "ODS" circuit, so I bypassed it and played around again with just the power section to be sure, and there are indeed problems with the behavior of the power amp, all through the range of volumes. In order to adhere to the generally accepted good practice of keeping my calculated plate load line crossing the 0V grid contour line above the "knee" of its curve (see plot in Feb 18 post), I had to use larger resistors than one usually sees on the tube screens. Apparently, this produces "... a significant degree of signal level dependent negative feedback across those massive screen grids...", and the poor sounding result is no surprise to amp forum member Pdf64 quoted here. I changed a few resistors around to mimic some standard power amp designs and that really improved its performance in all ways. But I haven't yet gotten an answer regarding the problems those changes have now created in my load line plot!  ("Sure, it may work fine in practice, but it will never work in theory!")

Other folks chimed in with comments related to my schematic: the V3 shared cathode causing problems with the stereo field, the placement and values of the R/L level pots, and one crucial rule of thumb from a builder that sorts out an issue that has been vexing me since the start: I have been making an assumption that for a clean sounding amplification stage - including in the power section - the design goal is to limit the max incoming signal to remain at or below the clipping point of the tube. It's logical, and might be true for a hi-fi music amp, but that's not how it works with guitar amps. This builder says he usually aims for allowing a max signal to the grid of 1.5x the tubes theoretical clipping point. So, if the power tube's bias point is -20v, organize your signal driver to deliver a maximum peak signal of about 30v (coincidentally, about 20v  rms) to the power tube's grid. He says this gets you just into a distorted sound at max levels, but clean for most of the input range. This makes me feel a bit more relaxed about exceeding calculated grid limits, and gives me a new reference point for prototyping and modeling. 

(updated power section schematic)

Circuit Test Configuration 3

I got the power amp drivers set up and tested using two configurations: (1) same guitar signal going to both output channels, and (2) guitar signal going into a stereo effects processor, split into left and right effected signals, with each going to a separate output channel. Sounds great at lower volumes. There's good L/R separation (though not perfect) and the power tubes seem to start to breakup right near the top of the volume range. Here's a schematic of the portion of the circuit I'm talking about: 

(homing in on the final power section circuit)

It's been really instructive to be able to take measurements of changes in voltage when the amp is running, and to easily swap components in and out to see what changes. For example, it took me a while to realize that I really needed to separate the cathodes on the two power tubes (which in theory can share a resistor and capacitor with no problems, as in V3 in the picture above) in order to even begin to understand how they work in relation to the plates and screens... and to be able to correlate my readings with spreadsheet calculations, which relies on being able to isolate the power tubes and the current running through them.... and having them separate will be essential to optimizing and balancing their respective biases, or operating parameters, which will vary slightly from tube to tube. Also, after some testing and getting some non-stereo output when I thought I should get stereo, and then using the TINA modeler, I realized I had to place the signal "return" pots (essentially, the volume controls for each channel) after the 220k ohm grid stopper resistors just prior to the V3 triodes, in order to separate the channels. I still may need to increase these resistor values to improve the separation of R/L signals... but that may produce other issues to deal with.    

Next, I put together an "ODS" style overdrive preamp (based mainly on the work of a couple of amp builders on the EL34 World forum - "DeGeezer" and "Tubenit"), and ran it through the power amp. 

(test board at work)

The final design will have a "clean" preamp section - not designed to produce heavy distortion - that will send its signal either directly to the effects loop, or to another circuit that will push the signal through two additional triodes configured to easily produce overdrive distortion ("ODS"), prior to going on to the effects loop. The mock-up of this idea sort of works as intended, but it's hard to tell for sure without also setting up the "clean" circuit. Because I can't produce a test signal that I'm sure is similar to that of the "clean" circuit in both quality and magnitude, I don't know for sure whether the "ODS" circuit is behaving as it ultimately should. Cranked to full output, the clean preamp as designed will be able to send about 8 volts (peak amplitude) to the "ODS" circuit, which I'm pretty sure is way more than I'll want to send there in practice - it will probably want no more than 3 or 4 volts to run right.  Also, while I am able to produce a clean guitar signal using a digital preamp, I know that digitally produced signals and tubes don't play that well together, so the tone quality of test output may be way off... or not, I don't know.  In testing, the "ODS" circuit sounds just OK but not great. I've tried using 12AX7 tubes as well as 12AY7 (which provide only half the gain of the 12AX7 and have different tonal characteristics), and from what I can tell I prefer the 12AY7s. There may not be much more major testing that is worth doing unless I build the "clean" preamp circuit as well, but that will require pulling the whole test board apart to make space, and I'm not sure that's worth doing, especially considering that the overall tone quality of the test system will still be lot different than the thing that gets built.   

On a different front, I picked up a cheap used solid state amp (Fender Frontman - FM212) that is physically the right vessel for my design - it's got two 12" speakers and plenty of room in the chassis (which I had already removed prior to taking the photo below). I'm playing the 5F6A/JTM though the two 12's - they're OK speakers for now, but based on comparisons with other speakers I have, I'll probably end up swapping them out. I'll see how they sound in the final build and decide then.    

(Fryette power station, 5F6A/JTM head, sitting atop a Fender FM212 combo, that will become the home of the new stereo single ended amp)

Here are draft drawings of how I'll organize the amplifier components inside the existing FM212 cabinet. The chassis on the this model amp is a funny shape (see side view below), not a standard box. and will require some cutting, bracing, and reshaping to accommodate tubes and a big transformer. The knobs and jack spaces on FM212 front panel pretty much match what I'll need for my amp, so that's a plus. 

(top view)

(front view)

(side view)

 

 

 

Circuit Test Configuration 2

Have successfully set up a basic one-stage pre-amp driving a power section configured as two fully separate single-ended outputs, driving two transformers and two speakers. 

Next, will be setting up separate drivers for the two output channels, each fed by separate (right/left) FX loop incoming signals. 

 

Circuit Test Configuration 1

My testing environment is coming along... 

(generic circuit mostly complete)

I've got a Fender Champ preamp (basic two-stage) set up there, and the back-end is coming along. 

I think I've figured out the power section values I'm shooting for, for a first test. With a class A setup there's a fairly narrow set of choices about how to do things because, based on what I've read, (a) I want the tube(s) to run at full power (at the plate) all the time, to a target output in watts (e.g. 23W for a 5881 tube), and (b) for reasons I don't yet fully understand, I want to leave a margin of positive voltage swing on the grid (meaning, big signal, overdrive situation) to protect the tube and circuit. Achieving both, given the constraints and flexibility of my PT and OT, means there are just a few options.  I found an online calculator that really helps visualize the relationships, and home in on the optimal values. 

Here's a graph of the configuration I'll try first, showing plate dissipation (straight red line) remains below the max rated dissipation for this tube (red dotted curve), and that at a 2500 ohm output transformer load (affecting the slope of the plate dissipation line) and screen voltage 30 volts lower then the plate (affecting the blue grid-voltage gradient lines) should allow operation up to around 13 peak signal volts (which can be easily produced by a two-stage pre-amp) to the tube's grid:

Test Bed

I'm setting up a test/development environment to make it easier to experiment as I figure out what I want my circuit to be. It's a generic and configurable space where I can easily modify circuit pathways, components and configuration, without soldering and unsoldering connections, which are instead made across terminal strips, screwing and unscrewing wires and parts. 

(figuring out what should go where on the board...)

(Setting up power supply, tube sockets, cap cans)

(heaters on the tube sockets are wired up and working!)

Turns out, the output transformer I already have won't work well for a single ended (SE) design - the one I have is meant for a push-pull (class AB) amp, which has different requirements than SE, so I've got a pair of transformers on order: 

 

Next Amp Considerations

First, the 5F6A / JTM that I recently completed building, includes a two-tube power/output section that is a class AB "push-pull" design, known for stability and power efficiency. A class A power section is a simpler design that can work with just one tube (and therefore doesn't require a phase inverter stage driving it), and although it is less efficient, this circuit is known to produce different audio/tonal characteristics than AB designs. So I'm interested in building something that uses a class A power section design, which would be lower power/volume than the 5F6A / JTM, but perfectly fine for my purposes. 

Next, I've always liked the depth of a stereo guitar sound, I used a Roland JC-120 (head version) for years, then later, a Fender Princeton Chorus, both stereo amps that allowed for time-based and textural effects - reverb, delay, tremolo, vibrato, chorus, etc. - that sound so much better in stereo. I've been intrigued by vintage Magnatone amps (especially the 96-10041 from the early 60's), and the recently resurrected company does have a new-ish stereo amp that utilizes their special versions of tremolo and vibrato. I considered building an imitation of that, but I think the analog tremolo, vibrato and reverb circuits would not be worth the trouble, considering the variety of great sounding digital versions of those and other effects (as external floor boxes or rack mounted gear) that one could just plug into an effects loop (like on my pervious amps). 

Finally, I've been intrigued by the Alexander Dumble sound, and have read a lot about the architecture of his original amps and amp mods (probably won't be buying one anytime soon), and the many designs that folks have created (Sebago Sound, Custom) to imitate the tone, so am interested in implementing some of that in the preamp section of my next amp. 

So that's the general idea: 

• Class A power section.
• Stereo for richer effects, which means:
• two-channel effects loop (buffered) between the preamp and power sections
• two single-tube output sections in parallel, to drive two speakers independently.

• Dumble-inspired preamp design.

Considering it will be a stereo amp, I also plan to build a 2-speaker cabinet to match, that can be run as stereo, with the new amp, or mono with the first amp. 

Calling the modded 5F6A done

I've spent a lot of time joyously playing this amp in the past 2 months, exploring the wide array of cleans, crunchy and overdriven sounds. The amp works well for the variety of styles from which I draw to get my own sound, from jazz, to roots-country, to pop and funk. I'm still trying to figure out whether there is anything for me in the the "stock" channel dirty sounds of the Tweed Bassman - it's just a bit ragged sounding for my tastes when cranked up. For output tubes I've tried 5881's, 6L6's and KT66's and so far prefer the 6L6's, but I have some other sets of 5881's and 6L6's still to try. 

Link to final schematic and layout.

For the past few weeks I have focused exclusively on better understanding the cascaded input mode and Cold Clipper bias setting of the V1A triode. With the MC5, I located the sweet spot where the amp maintains a full dynamic range, single notes ring and shimmer, double-stops crunch a bit but aren't fizzy, and when I switch the pickup from single coil to humbucker, the guitar output pushes the whole pre amp section into a nice, heavier but still articulate, fairly smooth overdrive. 

In the cascaded input mode, the Telecaster (both guitars picture below) with its slightly lower output single coil pickups also has its own sweet spot, where it maintains its Tele voice, but has a thickness and shimmeriness I like. To get a thicker overdriven sound here I use the ChaseTone Secret Pre pedal to boost the signal from guitar to amp input. 

To control overall volume, the amp's speaker output is routed to a Fryette Power Station, which drives a single Celestion G12H speaker, mic'd with a Shure SM57. The Dirty Tele and MC5 clips were recorded dry and a little reverb applied in the rough mix down, and the High Gain clip utilized the Fryette effects loop into a Line 6 effects unit for reverb and delay during recording. 

This short clip of the Tele uses cascaded mode, and progresses through light comping and a little lead playing on both pickups, then I switch to a clean boost and bridge pickup for the third half of the clip.  

Dirty Tele Clip on Soundcloud

This short clip of the MC5 uses the same amp settings, though I dialed the gain back slightly just to clean it up a bit; it also progresses through comping and lead playing just on the neck SC pickup, then switched over to humbucker, then through the boost pedal, and finally a few seconds playing just through the bridge humbucker.  

Dirty MC5 Clip on Soundcloud

I can get higher gain out of it, but the character of the guitar and amp both tend to diminish as gain increases, so I haven't spent a lot of time yet in that realm... but here's a sample of a high gain sound, using the Tele with boost peal and/or DOD 250 (I don't remember). 

High Gain Tele on SoundCloud

So, tone-wise I think I have a grasp on it and am usually able to dial up a specific sound for a specific application. A lot of folks-in-the-know say good amps are usually just really good at doing one thing, and don't expect it sound great at a clean setting as well as overdriven...but I'm pretty delighted with the versatility of this amp.

The tolex finally arrived and I spent the better part of a day learning and doing that bit.  

Left - Fender Telecaster (American Elite, 2014)
Right - Ibanez Musician (MC500, 1980)

 

  

Nearly complete

So, electronically, I think it's complete and where I'd like it to be in terms of tone. 

There has been a delay in receiving the tolex vinyl for covering the cabinet, so it's still bare wood, but the faceplate is done and in place. I read on a forum somewhere that someone had reverse printed their labels and such on a piece of clear plastic, so I used that approach. Generally, it went like this:

Use the layout from hole drilling template to locate the places for text, etc.

My partner is handy with a pen, and was generous enough to swirl out a bunch of variations of letters and labels and so forth. 

I scanned the calligraphy into images, blackened them and clean them up and converted to SVG, then placed those images onto the hole drilling template image and had Staples reverse print that final image onto a transparent sheet (could only get 8 x 11, so had to do across two pages and join the parts)

Cut holes for switches and spindles and and spray paint over the reverse image

Adhere to the front of amp with silicone

And voila!

(The star and tree graphic is taken from the old 1901 version of the Maine state flag.)

Tone Analysis III

I've done a variety of testing: So interesting, yet so tedious to produce a comprehensive enough collection of data to make definitive pronouncements. But here's what I think I can say. 

The TINA modeler does a great job showing what is likely to happen with my tube-driven circuit, especially when it's operating in linear territory, and it's even pretty good at showing saturation or cut-off conditions, when the inputs to tubes are driving them beyond their design limits. 

For example, here's a comparison of a modeled vs. actual 1kHz/200mv signal after is passes through channel I (full wide-open gain setting), cascaded through channel II set up as a "cold clipper" (very cool bias to force the clipping off of one lobe of the waveform, set to 25% gain), and measured at the grid of the second pre-amp tube:

Modeled using TINA:

Measured using a tone generator and capturing the signal in Visual Analyzer (VA):

Phase differences aside, they're pretty close in shape and proportion, and I think in actual measured signal voltage compared to modeled (I did some testing with volt meter and guitar generated signals, which conformed to expectations of signal voltage levels - no pics - but I haven't followed up using the tone generator). VA also shows the actual higher order harmonics accentuated by the circuit up to this point at these settings, in the lower panel. 

I also set up a suite of tone testing, but all the variations are too involved to capture and explain in detail, plus I changed some things as I went, and didn't want to go back and regenerate prior results to match up... Never the less, here are some examples of what I did and some key results. 

Using channel I and the stock TMB tone stack, I plugged in my MC500, set to neck humbucker PU only, and dialed in a sound that like - a full, round, straight-ahead jazz comping and lead tone. Then I replaced the guitar input with the white noise generator, and captured the plot in VA (the left side of figure below, green on black). Then I recreated the very same amplifier settings in TINA and plotted the modeled result (the right side of figure below), where the green curve is the signal prior to entering the tone control part of the circuit and the dark blue curve is the signal after being shaped by the tone controls. Here they are side by side:

Comparing the green curve on the left to the dark blue curve the right, you can see the similarities in overall shape, but the actual output appears to amplify the higher frequencies a bit more and extends them another 1k or so before beginning to fall off. There are also some spikes around 100 Hz which I'm not sure I actually hear.

This set shows use of channel II with the TMB tone stack, single coil on the neck PU, with a much different, very nice acoustic tone for comping or finger style playing - different visual orientation to widen the scale a bit. 

Here, the modeled vs. actual plots appear quite similar. with a pronounced notch at 600Hz. 

And, as below with the James tone stack set up with maximum mid frequency notch (close to 300Hz, nearly as predicted), we can see it's similar with regard to frequency response but doesn't represent the levels quite right.

Also (and this is important) I'm just getting my feet wet with this. I'm a complete novice and I just pick up what I can from reading and asking questions of people who actually know things. So, I'm sure I have based some of my findings and conclusions on inaccurate assumptions, misuse of complex software tools I have just learned for this application, and just plain incorrect interpretation of facts, all due to my limited knowledge and experience. 

Results 

Overall, TINA the modeler did a GREAT job guiding choices with regard to both tone and gain management from one stage to the next, and predicting actual outcomes. Bravo Texas Instruments! Now that I'm playing around with the amp in high gain territory, listening to the tonal qualities and overtones of just-breaking-up and highly distorted signals, I think all bets are off in terms of accurate modeling. But TINA got me 95% of the way there, and that's pretty impressive. 

The real result, the range of "clean" and "on the verge of breakup" tones that the amp creates with my guitars, is really quite impressive and everything I hoped for - and it's damn loud, which is just so satisfying in ways that are hard to convey in words (although neighbors have come up with a few!). It's sonically vast, and while I've got a handful of distinct and delicious tones I can use right now, it also provides so many avenues to explore and a lot more potential, as I swap out tubes (I have to try 5881 power tubes again and also some 6L6's, and a GZ34 rectifier I haven't tested out yet), and try new combinations of built-in settings.

I'd say I'm nearly done at this point. I will probably modify one last resistor value in the V2GL switch, to help manage signal level going into V2 in Cascade mode. The cabinet for it is basically done and in use, but I need to do the related upholstery tasks. And the art and layout for the faceplate is being finalized. 

 

Tone Analysis II

I've mainly been playing through the amp with with two guitars: an old Ibanez MC-500 with Super 88 humbuckers, and a relatively recent vintage Fender Telecaster with Seymour Duncan BG1400 stacked humbuckers. Both produce a relatively strong signal, but I also have an echoplex style preamp boost pedal I've experimented with for high gain settings. 

The recently updated schematic of selectable tone sections now look like this:

It's worth noting that a James tone stack doesn’t usually have a mid control, but since I had the lower half of the mid duplex pot available, I did some testing to see what happens when one modifies the value of the (usually around ~100k) “balancing” resistor between the Bass pot and the signal out. It turns out that raising the resistance shifts the notch between bass and treble controls progressively down in frequency, and also changes the shape of the overall curve, so I’m just calling it a mid-shift control in this application. The side-by-side plots below show how the Duncan Amps tone stack calculator, set up with my James circuit, predicts the frequency response characteristics across the range of bass and treble knob positions, for each extreme end of the balancing resistor range: the left (yellow) plot shows 277k and the right (green) plot shows 27k, representing my mid pot in full-open and full-closed positions.

You can see how the curves in the yellow plot generally dip a little deeper than their green counterparts, and that the location of those dips are a little further left, in a lower frequency band, than those on the right. Aside from the audible tonal difference, this also means more of the signal is being removed from the circuit, resulting in less signal strength getting pass into the next stage (the phase inverter). At any setting, the James stack passes less signal on to the next stage, so selecting the James stack means lowering the overall gain of the signal going to the PI and power section of the amp. 

Setup for testing theoretical vs. actual tone stack characteristics

There are just so many different variables associated with the shape and timbre of the sound coming out of the speakers (Celestion G12H, and EVM 15B) or as direct signal. Of course there is the signal being generated from the guitar pickups, which will vary from guitar to guitar and among different pickup types/configurations on a single guitar. And, as we know from an earlier post and this doc, each triode’s cathode components (resistor R_k and capacitor C_k), coupling capacitor (C_G) and gain bypass (C_BP) configurations of each channel produce signals with significantly different frequency response curves long before the signal even reaches the tone controls. In this figure below, the teal curve is the frequency response of the circuit measured at the channel II triode output, the blue is for channel I, and the grey is a blend of both channels (jumpered mode).

 

So, it helps to keep this in mind when comparing outputs of the two tone-stacks for each channel.  I’ll look at a variety of combinations, but not all of them, just to get a sense of things.

Also, up to now, the tone controls have not exactly behaved as I expected – they work, but they didn’t alter the sound in the way I expected through their ranges. So before testing and documenting how it all came out, I went over them again with the original Fender schematic, the Hoffman modified schematic, the Duncan Amps James tone circuit design explanation, my own schematic, and my layout to make sure I had designed and implemented what I intended… and of course, it turns out I had not.

For the stock Fender TMB tone stack, I mis-wired the Mid pot somehow, both in its connection to the bass pot and in the “knock-down” resistor I used to cut the resistance to ground seen by the bass pot from 250k (which is the actual R value of the mid pot) down to 25k which is what should be measured at the bass pot’s terminal 3. And for the James stack, I wired it as designed, but after playing with the circuit in TINA, I realized I could increase the “slope” resistor by a lot and deepen the available mid scoop – that will cause some loss of gain, but I can live with that since I have a second tone stack that loses much less. I had initially installed a 27k slope resistor, and now with a 200k I get a deeper mid notch at a lower frequency (~ 250Hhz). One reason I wanted to do a second tone stack in the first place, was to get something more like a Fender Blackface Twin sound, which I think comes from a mid notch at roughly that frequency. So, I made the corrections and modification.

But why bother?

I think partly, I just like to feel that I have some idea what is going to happen.... in life: If the situation is X and these are the options for influencing it, what affect is each likely to have? If I have no prior experience on which to base some prediction, then a model or metaphor that looks similar to the situation at hand is really useful. I just really like conceptual models and I love testing them against observable phenomena, both to learn and also to make them more useful predictors to use in the future. And that idea applies to lots I do in life, including I suppose, building my first amp. 

From testing, I'd like to know two things: (1) how close the result is to the modeled design, and (2) what does good sounding (to me) look like in a model and as some quantified expression?

To test overall response at all frequencies (not just guitar sounds), I'll pass a signal from a white noise generator on my iPad into each input channel and examine the audio output of the amp with a variety of tone control settings, by running the amp’s direct output signal (captured by my attenuator) to my computer, and examine it using a tool called Visual Analyzer. To keep it simple I'll just set up a few sonic scenarios for each case:

• Notch: the deepest mid notch that can be dialed in, keeping the Bass and treble pretty close to balanced (the upper and lower bounds of a managed response curve).
• Flat Output: dials adjusted to create the curve that appears most similar to the modeled response curve coming into the tone stack.

To test good-sounding-ness, I'll use each of my two guitars on two typical guitar settings, one on each channel and also one on each tone stack, for a total of eight. Then I'll just pass a white noise signal through after I've set the tone controls where I like them.

It’s worth noting that all of the tone pots have a “log” taper, which means the sweep on the dial from positions 0 – 5 transits only the first 10% of the resistance of the pot, and the remaining 90% of the pot’s resistance value is transited from positions 5 – 10 on the dial. The potentiometers in the TINA circuit simulator require setting a value as a percent of total pot resistance. I'll use the blue curve on the graph below (from eepower.com) to guide my TINA settings when generating the simulated results.

 stay tuned for results...

Tone Analysis I

As I mentioned I might do, I looked into lowering the resistance on the "cold clipper" function, did some modeling and testing, and decided to make the change down to something more like 10k, which is typical for the Marshall JCM-800 and many others.  Now the total Rk (cathode resistance) is 11.7k when the switch is engaged (the sum of the regular resistor and the 10k addition), the output is a bit hotter, and it produces a lovely smooth distorted tone when driven by the channel I input at any setting over about 50% (7 or so on the dial). The updated pre-amp circuit now looks like this:

Before anything else, let's look at the switches that impact tone ahead of the second tube and the tone stack.

The input impedance switch (the “Z” switch) was included to give some options for signal management in front of the channel II triode. The 5F6-A and JTM-45 both include options for reducing the instrument signal here, and it seemed like a good idea to add a voltage divider in that part of the circuit for managing the high gain signal that would be coming from the channel I when in cascade mode, not unlike second stage circuit designs of higher gain amps from the 70’s. For similar reasons, I added a grid leak resistor on a switch just prior to the grid of V2 (the “V2GL” switch), which would be the 3^rd gain stage when in cascade mode. I later added by-pass cap to the resistor on the V2GL switch in the hopes of also draining off some of the upper frequencies from the signal at this point, if the switch is engaged.

Below are the relative frequency response curves at the V2 input grid, for these two cases. The purple curve shows the Z switch set to Lo (200k resistor to ground), with and without the V2GL switch engaged, and the teal trace shows the Z switch set to Hi, with and without V2GL engaged. 

(Cascade mode, channel I @ 80%, channel II @ 50%)

Other than simply cutting signal strength, there are some frequency response differences particularly in the low end, although they seem minor. However, what this plot can't show is the effect of the "grid leak" resistors provided by both switches, in preventing blocking distortion at high gain settings (tube behavior is a complicated thing, but Rob Robinette's explanation of distortion and grid current dynamics is about the most understandable I've come across). The by-pass cap on the V2GL didn’t quite work as intended, and modeling various different values doesn’t change the response in a way that I think makes sense to bother with, at least until I learn more. I’m glad to have both switches at this point, as they do provide a convenient way to manage breakup and some high end (via the channel I gain bypass cap) at V1B and V2 somewhat independently. They probably also help prevent blocking distortion, but I've never heard that from this amp, so it's hard to say.

Construction of the amp cabinet is underway....