For this post, I’d like to take you through the process of designing a pedal, from initial idea to completed PCB and pedal. Everyone has their own way of doing this, so I thought it might be interesting for people to see that way I work on something like this. Pedal (or Synth) design is a combination of engineering decisions and aesthetic choices, and engineering in the service of those aesthetic choices. That’s what makes it so fascinating to do!
Where did the idea for the Hard Bargain come from?
The pedal started as my thoughts about tone controls for dirt pedals. There are various different tone controls that appear on pedals, and probably most things have been tried at least once. Overdrive/distortion/fuzz pedals are a specific application because the tone controls have so much to work with they can shape the entire sound of the pedal – they’re a key part of the voicing of a particular circuit. So what options are there?
- Baxandall Bass/Treble or Bass/Mid/Treble
- Sweepable controls, with Frequency and Cut/Boost
- Full Parametric EQ, with Frequency, Q, and Cut/Boost.
- Amp tone stacks (Fender, Marshall, etc etc)
- Big Muff Pi and derivatives
- Simple RC lowpass (Proco RAT and similar)
While the Big Muff control is a masterpiece of doing a lot with very little, in general I prefer active controls – no volume drop, bigger cut/boost range, more control, no interactions with other circuits around them. For some of the passive circuits, you finish up needing either a buffer to prevent interactions or an op-amp to provide make-up gain afterwards, and if you’re going to do that, you might as well have an active circuit in my view. The problem with some of the active circuits is that they’re complex to build and require several controls, so they’re complex to use too.
What exactly are we trying to do?
I decided I needed to come at the problem from a different direction, so I started thinking about what we’re generally trying to achieve by twiddling a tone control. I tried to boil it down and simplify it as much as possible, so I finished up with four basic options:
- Smoother or heavier sound (more bass)
- Brighter sound (more treble)
- Cut through more (mid range peak)
- Classic scooped tone (mid range notch)
It occurred to me that this could be done with two controls – one that provides a Bass/Treble balance, and one that offers a mid range Cut/Boost. The Hard Bargain design came about as a test platform to see if this idea was practical and offered a good range of control for just two knobs.
Designing the tone controls
Ok, so I’d established the general concept. Now I needed circuits to implement it. I vaguely remembered something about a rarely-used tone control known as a “Tilt” control which offered the sort of Bass/Treble balance I was looking for. This is pretty much an active version of what the Big Muff tone control does. The only reference I could find was on the Elliott Sound Products (ESP) audio pages. The circuit looks like this:
Rod Elliott is rather rude about the circuit, but he’s thinking about it in a hi-fi context. For our sound-sculpting purposes, it might be just the thing.
The other part of my two-control scheme was a Mid Cut/Boost. There are lots of ways this could be done. For some reason, initially I decided that a constant-Q design would be a good idea, and again I started with a design from the ESP pages, this time from a graphic equaliser project.
Putting these elements together in an LTspice simulation allowed me to see the resulting frequency responses. The Tilt control as I found it was centred on 800Hz, which is a pretty good starting point, so I left it alone and designed the mid-cut/boost to match. This was a question of finding a good set of component values for the Multiple-feedback bandpass filter that the circuit is based around. I use the extremely useful OKAWA web-based filter design tools for this kind of task. This gave me a Rev.1 schematic:
This gave me a working tone controls, but there’s still a lot wrong with it. It’s far too complicated for a start. Five op-amps just for two knobs? That’s pretty wasteful! However, the response graphs I was getting looked good, so I thought I was on the right lines:
At this point I decided I needed some feedback from my contemporaries, so I posted a thread about the design on DIYStompboxes.com. While I got some positive feedback for the design, it also confirmed my suspicion that it was over-complicated, and Scruffie suggested an alternative Baxandall-type Mid control instead. So I went back to the simulation and swapped the constant-Q stage for the much simpler Baxandall style mid control. This is not constant-Q, but perhaps that wasn’t really important. I wasn’t clear in my own mind why I’d made that choice initially, so it wasn’t too hard to let it go. That gave me the following design:
Comparing the frequency response with the previous design shows how similar they are in overall character, even if there are a few differences of detail. Mostly the mids on the Baxandall design aren’t so “peaky” – a lower Q.
Thinking that perhaps I could get a higher Q sound by using an alternative circuit, I also experimented with a gyrator-based circuit, which is another common option for an EQ section. It’s possible to build gyrators with very high Q if required. The circuit is also pretty simple:
However, looking at the frequency response for this mid control (and this time, just the mid control on its own) showed up a defect.
This graph shows the response every 1/10th of the way around the knob’s travel. As you can see, most of the action is at the extremes of the range, with almost half the response between 0-10% for the cut and 90-100% for the boost. The response of the Baxandall circuit was more even and so was the preferred design.
At the end of these initial experiments, I’d tried three different designs of mid control, decided on one that looked to provide the best balance of performance versus complexity, and settled on initial component values based on a crossover frequency of 800Hz. There’s nothing that’s magical about this number, and it could be higher or lower. That is not something that can be decided in a simulation, so it was time to get a circuit together to test it.
What kind of distortion stage?
The basic question with dirt pedals is how “hard” you want to make it. Are you aiming for a soft overdrive that just gives the guitar sound a bit of bluesy grittiness, or are you aiming for something more crunchy? Maybe you want an all-out wall of death noise?! All of those options are possible.
For this design, I wanted a fairly hard distortion, something that would provide plenty of distortion harmonics for the tone control to work with. I hoped we’d be able to get meaty scooped-mid tones and screaming-mid leads, so I wanted something that really rocked, but at the same time I wasn’t aiming for extreme metal, which is a sound of its own.
Like with tone controls, there are various options when it comes to creating distortion, and you need to know which to choose to do the job you want to do. Here’s a few:
- Op-amp with diodes in the feedback loop, Tubescreamer style.
- Op-amp for gain with silicon diodes to ground after the amp, Proco RAT/Boss DS-1/ MXR Distortion+ style.
- Transistors with diodes in the feedback loop, Big Muff PI style.
Of course, there’s a lot of variations here. “Diodes” is a pretty big group, and you have a choice of silicon versus germanium, or LEDs, or power diodes, or transistors of various types wired as diodes, etc etc. Many of these aren’t particularly interesting, but some produce a distinctive sound.
I was aiming for a fairly hard distortion sound, so I chose the second option, an op-amp providing gain with diodes afterwards to hard-clip the signal.
Input buffer with gain
I like to start a pedal circuit with a high impedance input buffer, so that the effect doesn’t load the guitar (the notorious “tone suck”) and so that we have reliable pedal that can cope with whatever you plug into it. If you use op-amps, the typical buffer design is a unity gain voltage follower. This voltage follower design is easily tweaked to add gain, so that’s what I did.
Since this was not the only gain I was going to put in the circuit, there was no need to go crazy, so I picked a couple of conventional values (10K and 33K) and had a look what I got. For the non-inverting op-amp arrangement, that gave a gain of (33/10)+1 = x4.3, +12.7dB. That’s a reasonable boost for starters, and acts as the foundation of our sound. It also gives us a bit more level to work with so that we don’t have to provide massive gain in our distortion stage, which tends to get a bit noisy. If you wanted a bit more boost, swapping the 33K for 47K would be a simple change to make.
The next step was to add a proper variable-gain drive stage to this. With no particular reason not to, I chose another non-inverting op-amp stage. (Note that the +ve input on pin 5 of the opamp is the input from the buffer above)
The values here were selected to give a good gain from a easily-available 100K pot. A log pot is chosen because the gain should be linear in decibels. If we aim for a gain of x100, we need 1K resistors. We don’t want the gain to go right down to unity, so we can include a resistor in series with the pot to provide a lower limit to the gain. Given that we already have a gain of x4.3, we should consider that any gain here is multiplied by that factor. On this basis I chose a low value of 1K, which gives a minimum gain of x2 for this stage. With the 100K pot, the maximum gain is (100K+1K / 1K) + 1 = x102. With the gain from the input stage, that gives a total gain of x438.6. That’s +52.8dB, which is plenty of gain.
What’s more tricky than the gain is the frequency response of this stage. C6 acts as a highpass filter and rolls off the bass below its cutoff frequency. In this circuit I chose the TubeScreamer value of 720Hz to avoid muddy bass notes. Similarly C5 rolls off the treble, and interacts with the gain control with more treble being lost at higher gain. Since higher gains tend to produce more distortion and harmonics, this “compensation” effect is quite useful. With the given value, at minimum gain the capacitor has no effect except on ultrasonic frequencies of several hundred kilohertz. As the gain is increasing the cut-off frequency moves down until it reaches 3.4KHz at maximum gain. This is low enough to smooth the op-amp clipping a little at the extreme and to prevent too much treble reaching the next stage.
Although we’ve added gain, we haven’t deliberately added any clipping. However, if the pedal runs on 9V, a typical op-amp might only be able to provide an 8Vpp swing before it starts to clip close to the supply rails. At maximum gain, that means that a signal of only 8V / 438.6 = 18mV would be enough to clip the op-amp. Pretty much any guitar can put out 100mVpp, and humbuckers and other hot pickups can produce much more, so this level of gain is enough to produce op-amp clipping already. This means that choice of op-amp in this circuit is significant, since some can “lock up” when driven hard to their rails. We need one that copes reasonably gracefully with being overdriven hard.
After the op-amp, we have the clipping diodes to ground. There are two sets of diodes, one set pointing down, and the other set pointing up, to clip negative and positive peaks of the waveform. For this design I decided to leave open the option of asymmetric clipping which tends to produce more even harmonics than purely symmetrical clipping. If this is not required it is easy to put a jumper in place of D2 or D4. Similarly diode choice is left open. These could be the typical silicon 1N4148 or 1N914 signal diodes, or they could be germanium diodes like 1N34s, or they could be LEDs of various types. Or different types could be combined for an asymmetrical arrangement, with a silicon diode in one direction and an LED in the other, for example.
Having made these decisions, I now had most of the schematic worked out well enough to be able to lay out a prototype PCB. Here’s the Rev.2 schematic.
Aside from the parts already mentioned, this schematic includes a potential divider and a capacitor after the clipping (R12,R13,C8) and a buffer IC2.1 to drive the Tilt control circuit. The addition of the resistors and capacitor was to give me an opportunity to remove some treble and reduce the level if required. I didn’t mark any values on these components initially, since I didn’t know if this was necessary, but I wanted space on the PCB to have the option in case it was. The components were marked “TBD” for “To Be Determined”.
The first PCB
The only other addition was to stick a volume control on the end. With the circuit complete, it was time to lay out a PCB so I could hear how it sounded and experiment with some of the component values.
This was the first board I built up. It worked first time, which is always nice. I used 3 x 1N4148 for the clipping diodes, so I went with asymmetrical clipping.
The final version, Rev.3
I played a bit with the R12/R13/C8 values and decided that R13 could be left out (loud is good) and that R12/C8 sounded good with a bit of high-end roll-off, and I chose 47K/470p, giving a lowpass action above 7KHz. This just trims a little harshness without losing any high notes or removing too much treble. You can adjust to taste, of course.
The other thing that bothered me about the Rev.2 design was the use of five op-amps. Not only does this offer no significant space saving over using six op-amps (it’s three 8-pin chips either way), but it also necessitates the use of two different chips that look virtually identical. That’s pretty much asking someone to mix them up, and given single op-amps have a quite different pin-out to duals, the possibility of something getting fried is fairly high. So I was thinking that I had an op-amp in hand that I could use in the circuit somewhere. This lead to Rev.3 of the circuit:
This updated version simply uses the extra op-amp as an output buffer after the volume control. This prevents the volume control from interacting with subsequent circuits and provides a solid, low impedance drive to anything that comes after it. It means the circuit uses three identical chips so you can put them in in any order and it’s all fine. I used TL072, but any common dual audio op-amp is fine, as long as it doesn’t lock up. The only other change was to add a couple of pads to make wiring a “clipping” switch easy. This allows you to select between symmetrical and asymmetrical clipping. The volume will change if you flick the switch, but it’s such an easy thing to add I decided this didn’t matter and it was worth having.
Can I build one?
Sure, go for your life! I haven’t done a Vero layout, but if you do, do please let me know and I’ll post a link to it here.
We’ll provide manufactured PCBs in the shop for this project in due course. We’re currently waiting on some parts to arrive so we can build up the latest Rev.2 PCBs and make sure they’re ok before we order a pile for the shop. Once that’s done, I might get round to boxing mine up, and we can have a few photos of the finished unit!
Comments and feedback
As always, get in touch if you’ve got any comments or feedback about the pedal. We’d love to hear.