FRMT Sustainer

A Greatly Improved LM386-Based Sustainer Circuit

Overview

After reverse engineering the Sustainiac Stealth, I was contacted by several people who lived in foreign countries that, for various reasons, were unable to find the PCB's or parts necessary to make the circuit themselves and the cost of PCB assembly services were prohibitively high. Given these circumstances, I decided to challenge myself to come up with a sustainer circuit that would work really well but would be buildable with the most common parts available. I decided that what I would attempt would be to use an incarnation of the Sustainiac Stealth opamp stages, but simplified as much as absolutely possible, with an LM386 power amp. Additionally, I wanted to use through hole parts and layout so that people could use perf board, vero board, or fabbed PCB's depending on their circumstances and preferences.

The LM386 amplifier is the most common amplifier used in DIY sustainers, but frequently performance is poor because most circuits use a single JFET preamplifier and then straight into the LM386. Harmonic mode is typically done by just reversing the phase of the driver. Though this can result in a sustainer that "works", there are three really big issues in my opinion. These are:

No automatic gain control (AGC) to keep performance consistent
No independent gain controls for fundamental and harmonic mode
No phase lag circuit for harmonic mode enhancement

Without an AGC, the system gain might be set up for sustaining with a loud signal, but then it's not able to maintain sustaining with a weaker signal. If the gain gets boosted to sustain with lower signal levels, then stronger signal levels can overdrive the coil and cause interference with the bridge pickup.

Independent gain controls for harmonic and fundamental mode allow for dialing the system in to the instrument's electronics and the player's preferences.

The phase lag circuit provides both a boost to harmonic mode signal as well as introducing some phase shift that helps with the production of strings vibrating at a harmonic instead of the fundamental.

Now, the circuit is only one half of the equation. An improved driver design for this type of sustainer is described here. The cool thing about the two coils is that one could do two FRMT circuits and wire one to each driver coil. This would allow for independent gain controls for high and low strings as well as the ability to run low strings in fundamental mode and high strings in harmonic mode, or vice versa.

How It Works

The building blocks of the FRMT are will be familiar to those acquainted with the Sustainiac because I started with the basic building blocks and then stripped away anything that wasn't completely necessary. As a refresher, the Sustainiac stages consist of 5 parts: an input buffer, AGC, modified attenuverter, phase lag circuit, and reference voltage buffer. The power amp stage is a pretty bog-standard LM386 amplifier implementation.

The input buffer is a simple JFET buffer. I decided to do a discrete buffer here so that I could execute the rest of the circuit with a single quad opamp IC. This is a common source amplifier, which is a non-inverting, unity-gain buffer. Note that 2N5457 is specified, though J201 would also work. I strongly recommend against J112/J113, as they are chopper transistors and, when testing them in this circuit, the interactions with the AGC caused all kinds of bad problems. Stick with JFET's that are N-channel general purpose amplifiers.

FRMT Input Buffer

The output of the input buffer is then fed into the AGC. The AGC is actually doing two things simultaneously. First, it is applying a static gain by virtue of the components in the negative feedback loop. The second function is the automatic gain adjustment performed by the positive feedback loop. The static gain for this non-inverting opamp stage is calculated by the standard gain equation, which in our case is A=1+Gain/R5.

The positive feedback loop creating the AGC is where the real magic happens. As the signal comes out of the opamp stage, it goes through R7 and then through Q2, which is set up as a kind of switch where loud signals will make it through while quieter signals don't satisfy the Vce requirements for conduction. R6 in conjunction with C5 and R5 creates a low pass filter with a roughly 5 Hz cutoff frequency. Eliminating the audio frequencies in the feedback loop ensures that the AGC isn't trying to compensate for audio frequency changes in level. Rather, the filter acts like a simple envelope follower and allows the AGC to track slower changes in signal level. That filtered signal is then fed to the gate of a JFET. The larger the signal on the JFET gate, the lower the resistance from drain to source. R3 and Q1 form a voltage divider, where the resistance of Q1 depends on how loud the signal is. With a louder signal, R3/Q1 will attenuate the signal while weaker signals will not and it will feed the fully amplified signal downstream. The JFET selection here should be governed by the same considerations as the input buffer. I tried J112 with some success, but its characteristics resulting in slightly poorer performance then a 2N5457. J112 can be used in a pinch, but it isn't ideal.

FRMT Automatic Gain Control

The modified attenuverter is similar to the Sustainiac, but note that the harmonic gain pot is placed in the feedback loop for the inverted signal case. This allowed me to eliminate some components, such as some transistors and their gate resistors. Those components are good practice, but they don't fit the bill of absolutely necessary. The attenuverter also pulls double duty by allowing for the phase inversion of the signal while providing an independent gain for harmonic mode. With the component values specified, harmonic mode can have a gain (relative to fundamental mode) between 1/3 (-10 dB) and 3.333 (+10 dB).

FRMT Modified Attenuverter

The next stage is the phase lag circuit. This utilizes the same values as the Sustainiac circuit, but I got rid of the mode switching transistor and gate resistor for simplicity, along with the input high pass filter. I didn't find the filter strictly necessary and the switching transistor isn't absolutely necessary, so out it went.

FRMT Phase Lag Circuit

The final piece of the audio signal path is the power amplifier. This is a pretty basic implementation of an LM386 amplifier, though there are a few things worth mentioning. The large cap from pins 1 to 8 sets gain to maximum while filtering out the very lowest frequencies to avoid eating up headroom. This cap could be changed to a smaller value to further restrict bandwidth, but I don't think there will be much tangible benefit there.

Cap C11 from pin 7 to ground is critical in this application. Without it, the amplifier will not work properly. This is a power supply stabilization cap and, even though we run from a battery, its absence can cause issues as the cap acts as a reservoir of charge.

For the external connections, note that DR- is grounded due to the LM386 being a class AB amplifier. For class D amplifiers, that is a big no-no, but here it is actually proper practice. Also, note that the ground point is labeled Jack. That is because the ground to this board should be wired to the ring of a stereo jack for the guitar's output. This means that ground is only connected when a cable is plugged into the guitar. This is a failsafe against the sustainer being left with its power switch enabled. Of course, if you leave the switch on and the guitar plugged in, it will still suck your battery dry.

FRMT Power Amplifier

The last block of the FRMT is the reference voltage buffer. I have tried the circuit without it and it doesn't work very well. I suspect this is due to the fact that Vref is used not only for opamp biasing but as the AC signal ground throughout, so any variation in it can cause the circuit to not work as intended. For that reason it is left in, though with the bare minimum of components.

FRMT Reference Voltage Buffer

So now that we know how it works, you should build one! If you like fabbed PCB's, you can find the Gerber files and BOM here. If that doesn't work for you, feel free to build up a perf or vero layout. If you get one that is verified, I will happily post it here. Enjoy!