The Nickel Blues™ - a Homebrew High-Gain, 5-Watt Tube Amplifier

Updated 12 Oct, 2008

So, you ask, "why yet another single-ended 5-watt tube amplifier?" (I considered naming the amp YASE-Five but settled on Nickel Blues, instead.) With a number of commercial and home-brew single-ended designs readily available that is certainly a valid question. The short answer is that I've yet to find a design that completely satisfies me.

Among the commercial offerings the Blackheart Little Giant comes closest to meeting my needs and were I looking for a low-gain amp I would modify and tweak a Little Giant rather than building from scratch. In fact, I actually own four of these amps, all modified, including a better output transformer in one case. You can find an article on modifying the Little Giant elsewhere on this site.

Of the popular DIY designs the High Octane over at AX84.com is a nice high-gain design but suffers (in my opinion, of course) from a lack of bottom end "grunt" and seems prone to oscillation if one is not very careful with lead dress and such. To be fair, I haven't built one and am basing my opinion of the amp on the dozens of clips I've heard. Also, I don't mean to imply in any way that it's a bad design – many people have successfully built High Octanes and love them. Some people like Fords, some prefer Chevys.

I set out to design an amplifier that would be similar to a Little Giant but with additional gain, a low-impedance effects loop, and with emphasis on the ability to get some good low and low-mid "chunk" when desired but without completely sacrificing "sparkle" on the high end. The design I came up with is actually superficially similar to an AX84 High Octane (there are only so many ways you can arrange four triodes and a power tube, after all). The overall gain of the Nickel Blues is probably a smidgen higher than that of a High Octane in spite of its lossier tone stack. The additional low-end is obtained by using larger filter capacitors, a massively over-spec'd output transformer, and by different voicing in the coupling circuits.

The Nickel Blues features a very warmly biased second stage that is designed to saturate early and smoothly. I've found that you can get about 90% or 95% of "cranked amp tone" even at apartment levels by biasing one or more preamp stages very warmly. If you first hear this amp with the volume low and the first gain control dimed you think, "hmmm, that sounds better than some off-the-shelf tube amps do cranked" – it is only when you then dime the volume control that you realize how much better it can be with the output stage being fully driven.

Theory of Operation

Throughout this section I make frequent comparisons to the AX84 High Octane amplifier. This isn't to criticize that design, nor to illustrate how the Nickel Blues is "better" than the High Octane, or anything of the sort. I make the comparisons simply because many in the DIY community are already familiar with the High Octane, so understanding how this design differs from that one will make it easier for these folks to evaluate whether they want to consider building a Nickel Blues or perhaps modify a Nickel Blues or High Octane to tailor an amplifier to something in between that meets their needs more completely than either of those designs.

This description begins at the input to the amp and proceeds stage-by-stage to the output section. This is followed by some brief notes about the power supply and filament supply. Experienced techs and engineers will find parts of this description hilariously simplified but that is intentional. Experienced folks can understand the amplifier from the schematic alone. Those who need a description of the theory of operation probably would not be able to follow a more thorough treatment.

Input and First Gain Stage:

• The input jack is wired like that on almost any amp. A switching jack is used to silence the amp when nothing is plugged in.
• R9 provides a ground reference for the grid of V1a. Placing R9 on the jack side of the grid stopper (R10) eliminates the slight (about 7%) signal loss that occurs on designs where the grid reference resistor is connected at the junction of the grid stopper and the grid.
• The somewhat unusual plate resistor (R11 - 150k) and cathode resistor (R12 1k5) combination provides slightly more gain (about 2db) than a 100k plate resistor but that's not the main reason why I chose it. I have found that I really like a 12AX7 biased this way for the first stage of an amp because it provides an ideal (to my ears) compromise between clean headroom and saturation. This only comes into play when you use a pedal or active pickups to push that first stage hard, but it makes the amp very "pedal friendly." I've found that a first stage biased more warmly doesn't leave enough clean head room while one biased colder is harsh when overdriven. At least, that's the story my ears tell me and they're sticking to it.
• This input stage does have one drawback, it is slightly more prone to leakage and thermal noise (hiss) and this is magnified by the fact that it is the first stage in a fairly high gain preamp. Using low-noise components is essential for the first stage, at least, of this amp. I used an 0.68u film cap for the cathode bypass (C15) in spite of my stated desire to emphasize bass response because using a noisy electrolytic can be problematic here. V1a and its associated components are not the place to cut corners. If you don't want a hiss-o-matic, use a good tube socket, top-quality metal film resistors and film and silver-mica caps, clean all of your contacts and leads really well and make sure you have excellent metal-to-metal contact before soldering the components. Also, avoid using tubes selected for extra high gain in the V1 position – they are almost invariably much noisier than lower-gain 12AX7s.
• As with any high gain amp if you do a lot of volume swells with the guitar's volume control you may want to add a series capacitor (.047u or larger film) between the input jack and R10 to eliminate crackling because of a changing bias on the grid. If you do add a series capacitor, make sure that you do not get it between R9 and the grid (i.e. don't break the grid DC ground reference provided by R9). Using a 68k resistor for R10 helps minimize crackling so you may not need a series capacitor even if you do volume swells – see the discussion of R33 in the next section.
• The 68k grid stopper resistor (R10) and the 39-47p capacitor (C32) in parallel with the small capacitance of the grid of V1a forms a low-pass filter that kills any tendency for the amp to oscillate. As you might surmise from the out-of-sequence component number, C32 was added when debugging and tweaking the prototype.

Gain Control and Coupling to 2nd Stage:

• C14 is the coupling capacitor that isolates the high DC potential at the plate of V1a from the components that follow. Good-quality capacitors that are not prone to leakage or breaking down are essential for the coupling capacitors of all stages. I use "Orange Drops" but any good-quality film capacitors (rated 400V or better) are fine.
• VR1 is a 1M audio pot as used in just about every tube amp design you'll find. The value is by no means critical but such a high resistance avoids loading the signal and leaves lots of "possibilities" for shaping the frequency response with other components.
• C16, R33, and R13 form a network that reduces the signal by about 33% but serves two important functions. First R33 significantly reduces any tendency for the amp to "pump" or crackle when the gain control is changed. Without R33 the grid reference resistance for V1b varies from 0 ohms (short to ground) to around 500k as the gain is turned up. With R33 the grid reference resistance changes from around 250k to around 595k.
• Second, R33 and C16 even out the frequency response through the second stage as the gain control is moved from near zero to maximum. Without R33 the second stage attenuates almost none of the highs when the control is dimed and drastically cuts the highs when the gain control is reduced to very low levels (because almost the full 1 megohm of the pot is effectively in series with the grid capacitance). R33 alters this by reducing the ratio of minimum to maximum resistance in series with the grid capacitance from 1:1000 or higher to something more like 1:3. C16 prevents the overall high frequency loss from being too severe.
• Some amps use a "bright" cap across the high and wiper terminals of the gain pot. I'm not a real fan of this approach because you can end up actually making the response brighter as the gain control is turned down very low – and typically when I'm playing clean I want all the bass I can get, I only want to cut bass to reduce “splattery” intermodulation distortion when overdriving. The circuit used in the Nickel Blues (and the High Octane) still allows for some cutting of highs (relative boosting of bass) as the amp is cleaned up.
• Note that C16, R33, and R13 are similar in value to the equivalent components in the High Octane. This circuit cuts a little less signal (33% vs. 50%) and emphasizes the "bright" component a little less than does that in the High Octane.
• SW3, if used, defeats the "bright" circuit when the switch is open. R33 remains in circuit to prevent pumping and crackling. The amp will be pretty dark with the switch open, so I'm not sure how useful it would be.
• A more useful approach might be to wire the switch as shown for SW4 – "boost" would allow one to boost the overall gain and further emphasize the highs if the switch is closed when the gain pot is dimed. If the switch is closed with the gain control low the amp will be quite dark.
• Note that I haven't built an amp with either SW3 or SW4, and probably won't. My gut feeling is that the amp will be very dark with SW3 open and I simply don't need the gain boost that would happen with SW4 closed. When I boost gain I prefer to use a pedal to push the first stage harder.

Second Gain Stage:

• V1b is biased very warmly and saturates early, providing very smooth preamp overdrive as discussed in the first part of this document. The small cathode resistor (R15, 560 ohms) causes a fairly small bias on the grid and that in turn increases current at idle, dropping the plate voltage to about half the supply voltage. When a signal is applied to the grid the tube goes into saturation (maximum conduction) much earlier than cutoff (minimum conduction). Triodes enter saturation gradually (especially when fed from a high-impedance source) and thus the signal compresses smoothly at the bottom of the output waveform. By contrast a tube goes into cutoff quite suddenly, leaving very sharp square-edged distortion similar to that of a typical transistor circuit.
• As mentioned in the introduction, you can get about 90% to 95% of "cranked amp tone" from warmly biased preamp stages like this one. In situations where volume has to be severely limited I have found that I prefer the tone from a warm preamp stage to the tone from a power amp heavily "strangled" by any of the attenuators I've tried.
• Ordinarily a warm stage like this has one major drawback, it seriously limits clean headroom. In this design we have stages with good headroom before and after this saturation stage and individual gain controls so when headroom is desired we can easily reduce the level of the signal going through this stage while maintaining adequate overall gain to fully drive the power tube.

Tone Stack:

• This is a typical tone stack as used in many preamps patterned after the "Marshall style" (including the High Octane) with the major exception that it is not driven by a cathode follower. What difference does that make? A cathode follower lowers the source impedance driving the stack from around 40k or 50k to around 2k or less. Lack of a cathode follower means that this stack is considerably "lossier" than a stack driven by a cathode follower. Overall, across the guitar spectrum this stack will average about 6db more signal loss than the almost identical stack in the High Octane.
• So, why did I choose not to use a cathode follower to drive the stack? For two reasons. First, recall that one of the design goals was to have a good low impedance FX loop. The only practical place to put an FX loop in this amp is after the tone stack and that is not by any means a low-impedance point in the circuit, even if the stack was driven by a cathode follower. So, I used the cathode follower to drive the FX loop directly – ensuring an output impedance around 2k.
• Second, remember that another design goal was to have good low-end "grunt" when desired. The tone stack actually cuts less bass (in relation to the overall spectrum) when driven by a high impedance than when driven by a low impedance. Essentially, the bass control becomes more effective. Another way of looking at this is that the tone stack cuts about the same amount of bass when driven by a low or high impedance, but that the stack cuts as much as 6db more from the top two thirds of the guitar spectrum when driven by a high impedance.

Cathode Follower and Effects Loop:

• The effects loop is unique to this design - so far as I know, anyway. It's awfully difficult to claim originality when people have been doing interesting things with tubes for seventy years! Though at least one commercial amp manufacturer seems determined to patent every circuit he digs out of 50 year old magazines and tube manuals!
• J2 is a switching TRS jack that provides both the send and return for the FX loop. When nothing is plugged into J2 the cathode follower is not in the audio path, the full signal from the gain control (VR5) goes directly through the switched ring terminal and to the final preamp gain stage (V2b). When a stereo plug is inserted, the FX send (buffered by the cathode follower V2a) is on the tip and the return is on the ring. If a mono plug is inserted the buffered send is on the tip and the return is grounded – in this configuration it is safe to operate the amplifier as a preamp with no speaker attached due to the grounded input and the safety resistor (R29) across the output transformer.
• LEDs D5 and D6 protect external equipment by clamping the sent signal to about +/- 1.5 volts (approximately, depending on the forward voltage drop across the LEDs which varies with color and type). LEDs also clamp fairly smoothly (they are used as clippers in some distortion pedals). So, whenever the FX loop is in use you have a built in overdrive pedal, sort of.
• R20 and R19 form a voltage divider to pull the cathode follower well above ground to ensure that the signal is not clipped hard by the cathode follower. C20 and C21 prevent this bias voltage from affecting the rest of the preamp. See the schematic for additional notes regarding the FX circuit.

FX Recovery / Final Preamp Stage / Master Volume:

• V2b is biased just like the first stage – see that section for what that means. This is done because when the FX loop is used (or if the FX return is used as a power-amp input from some other device being used as a preamp, such as a POD or what have you) I wanted it to overdrive the same way the first stage does, for the same reasons. The only difference is that here noise is not as critical because this is not the first of three gain stages, but the last. Therefore, I used a 22uf electrolytic for the cathode bypass capacitor (C25) for good bass response.
• R30 cuts down some unneeded gain that can be troublesome when the FX loop is not being used. Without it the volume control is very sensitive and it is easy to push the power amp into blocking distortion. C23 acts with R30 to form a low pass filter that trims high frequencies and "rounds off" sharp edges from the preamp distortion, making the overall amp smoother (this was mentioned in the section on the gain control and second-stage coupling). This is an area where you can easily tailor the amp to suit your tastes – from no capacitor at all up to about 220p, maybe even 470p for those who like really, really smooth output and don't mind sacrificing bite to get it. I found that 220p took away too much bite (and I like fairly dark amps). I ended up using two 220p caps in series for 110p.
• R25 is the grid resistor for the output power tube. A more typical value here is 5.6k but when I was scoping the prototype I noticed that the power section became unstable and did weird things when the volume control (VR6) was turned all the way down (shorting the wiper to ground). This happened regardless of any other control settings. I checked the volume pot with an ohmmeter (after allowing the circuit to drain, of course) and found that the instability occurred when the total resistance between the EL84 grid and ground fell below about 8k. Changing to a 10k resistor solved the problem. Why? I dunno – 'cuz it's tubes?

Power Amplifier:

• The PA can use either an EL84 or a 6V6 tube. I used separate cathode bias resistors because typically a 6V6 requires quite a bit more bias voltage than an EL84. Keep in mind that this design is really optimized for best performance as a "lead" EL84 amplifier. An EL84 is biased just about perfectly and is also at its "sweetest" plate voltage. I've tested two builds with a variety of EL84 tubes and the plate dissipation has varied between 11.8 and 12.1 watts with mains voltage at 119VAC. Screen dissipation is about 1.25 watts.
• The 6V6 works fine for clean and lightly crunched material but it takes a lot more signal on the grid to overdrive a 6V6 than is required by an EL84. With a 6V6 in place the additional current drawn pulls down the supply voltage through the amp so there is even less voltage available to drive the grid. Also, the 6V6 is biased just a little cold (about 13W plate dissipation, 14W is the rated dissipation for a 6V6). You can overdrive a 6V6 hard but it just doesn't sound as good as an EL84 in this circuit.
• One could probably reduce the cathode resistor on the 6V6 to about 300 or 310 ohms for a little warmer bias, but this would draw even more current and lower the voltages through the amp even further. It would also increase the power dissipated by R4 to somewhere near its rated 5 watts (it's dissipating about 3.6W with a 6V6 as is).
• A better solution if one intends to use the amp mostly as a 6V6 amp might be to lower R4 to around 800 ohms and then increase the cathode resistor for the EL84 as needed to keep an EL84 at it's 12W bias point. This may, however, make the EL84 sound a bit strident or stiff (I'm not crazy about the sound of EL84s run at unusually high voltages and low currents). This will also reduce the effectiveness of the power supply filtering and may result in slightly more hum.
• The bottom line is that it is very difficult, though not impossible, to design an amplifier that will be a really good EL84 amp and a really good 6V6 amp. Doing that requires more complexity – such as switching R4 when the tubes are swapped so that each tube is running at its optimum voltage and current. It's very easy to compromise and build an amp that is a so-so EL84 amp and a so-so 6V6 amp, but I was not willing to go there so this design is a really good EL84 amp and only a fair 6V6 amp. If I really wanted to use the amp for "lead" with both the EL84 and 6V6 I would probably add a switch to change the value of R4 from 1.2k to about 800 or 900 ohms when switching to the 6V6.
• Finally, note the use of rather large cathode bypass capacitors in the power amp. Many amps use as little as 22uf or 33uf here and if you just run the math that would seem to be adequate. However, such small capacitors can be "pumped up" by unheard subsonic frequencies, changing the bias point and driving the amp "colder." One of the changes I recommend to almost any cathode-biased amp is to increase this bypass cap to at least 330uf.

High Voltage Power Supply:

• Power supply filtering is critical in a single-ended amp and also critical for good bass response. When built as shown and if the grounding notes on the schematic are followed hum is basically non-existent and bass response is superb. On the prototype, built in a16X8X2 aluminum P1X/High Octane chassis from Doberman, there was just a hint of hum detectable under the white noise of the high gain preamp with everything dimed. The next two amps were built on a larger 19-1/2X8X3 aluminum chassis and I can't hear any hum under the white noise at all with the controls dimed. However, I attribute the difference in hum to the better filament supply in the subsequent amps (see below), rather than the chassis size. Obviously, your mileage may vary depending on how careful you are with lead dress, shielded wiring, etc.

Filament Power Supply:

• I highly recommend DC power to the filaments on this amplifier but inadequately filtered DC can create more noise than AC filaments due to the sharp switching transients on the rectifier diodes. About 10,000uf of capacitance is required per amp of current to bring ripple voltage down to about a volt. Such a large electrolytic capacitor does a very poor job of filtering very fast transients such as those from the rectifier diodes so it is very important to bypass the electrolytic cap with a poly cap of around 0.1uf (this value is not critical).
• NOTE: If you do not adequately filter the output of the bridge rectifier you will have a very annoying hum or buzz – annoying because it is at 120hz instead of 60hz and that puts it well within the guitar spectrum. The resulting noise will "beat" against some of the notes you play and the amp will sound terrible!
• The amount of capacitance is a balancing act between adequate filtration and avoiding overloading the power transformer due to capacitive loading on the filament winding. 10,000uf is about as much capacitive loading as I'm willing to apply to a winding being used at about 50% of it's rated capacity (as is the case with 1.4 amps on the 3A winding of a Hammond 270DX). I built the prototype with a single 10,000uf capacitor and it worked okay, with about 1.2p-p volts ripple on the filament supply to all tubes. Then, I had a better idea...
• On subsequent amps I used a simple pi filter with two 4700u caps and a 1R1 2 watt (actually, 5 watt was what I had handy) resistor. The power tube filament is supplied from the first stage of the filter and the preamp tubes are supplied by the last stage. This resulted in slightly higher ripple on the power tube filament but reduced ripple on the preamp filaments to about 0.7p-p. This also reduced the capacitive loading on the transformer. The transformer doesn't run uncomfortably warm on either amplifier but the difference in transformer temperature between the two amps after idling in standby for a long time is detectable.
• This also reduces the filament voltage to the preamp tubes to right at the low edge of the nominal range (5.8VDC measured with mains voltage at 119VAC). This shouldn't be a problem unless you frequently play through brownouts – in which case you might get EVH's "brown sound!"

Schematic