I’ve been looking at the theoretical principles that govern the voltage-controlled amplifier (VCA) recently and came across some simple VCA designs that caught my attention. A common technique involves a JFET and an opamp to achieve VCA type operation by using the JFET as a voltage-controlled resistor somewhere in the input path. I’ve seen a good deal of these circuits out there, but I had trouble getting good simulation results from what I had seen. I cam across an article by Rod Elliot of Elliot Sound Products which covered some nice history and interesting discussion of the VCA in general. After going over the information there, I came up with this design based on some of the more basic designs presented.
One of the things I’ve been most interested in is trying to accomplish this with a single supply which is common in most stompbox setups. In the schematic, there’s an emitter-follower stage just to act as a buffer for the input signal followed by an inverting opamp stage. To make this work, both C1 and C2 are required to effectively AC couple both input signals (Vin and Vctl) to the opamp stage. There’s an RC network that is supposed to tame the distortion in the output by taking a portion of the output and connecting it to the input of the J201. The article explains this in decent detail. In simulation, the it seems to smooth out the non-linearity of the JFET as Vctl changes. Using the J201, LTSpice gives a decent linear-like response over a range of around 500 mV (0 to -500 mV at the input) and operates decently with a 500 mVp signal.
For use with an LFO, I found that the best results happen with a slight negative voltage offset and a signal who’s amplitude peaks at 0V (i.e. 250 mV sinusoidal signal with a -250 mV offset). Of course, this is all highly dependent on the threshold voltage of the J201 which can range from -0.3 V to -1.5 V according to the Fairchild datasheet. It will be interesting to see the results of this circuit on a breadboard.
 Gray, P. (2009). Analysis and design of analog integrated circuits. New York: Wiley.
This is a circuit designed for a classroom project whereby the instructions were to “improve” a differential amplifier circuit. The differential amplifier design is essentially an exercise in understanding the inner workings of an opamp, and it effectively works in the same way. The figure below is a schematic for the differential amplifier without the cathode follower output stage. In the LTspice simulation, the input signal is connected to the V+ terminal and the V- terminal is connected to ground. There is no feedback loop between the output and either input terminal making this a high-gain open-loop configuration. However, adding a resistor from V- to ground and one from the output terminal to V- would accomplish the same results as a non-inverting opamp.
I wouldn’t say that adding a cathode follower “improves” the output stage of the amp. It was more an experiment in comparing different solid-state output stages with a vacuum tube stage operating at very low voltages. However, I will say that this thing sounded amazing with the couple of guitars I tested through it. Putting a potentiometer in the feedback network allowed me to play with different gain settings. It’s a very bright sound overall giving lots of high-end sparkle, but the breakup was quite remarkable. I suspect this might be a very usable configuration for a tube mic preamp or a number of audio applications. Hopefully, I will get an opportunity to revisit this before too long.
For those of you interested in the ins and outs of this experiment, you can download our full report here.
after our move to Oregon two months ago, i’m finally getting back to a spot where i can do more building and experimentation. that being said, my arduino-based sampler/synth is getting a rebuild to make it more capable and road-worthy.
the LED segment display is currently showing voltage out what should be 5v DC. i think the power adapter might be limited on the amount of current it can provide which is creating a voltage drop. either that or the 3.3v and 5v pins are reading the load from the BJT stage which stabilizes and adds sonic color to the PWM output. also, i’ve dropped in an ATMega2560 board and removed the 328p for added storage and memory for more sample time. with some code revisions, i’m hoping to keep the processor for locking up when adding lots of modulations.
Need a switch to be fully-on fully-off and carry lots of current -MOSFET
Need a switch that needs to have lowish capacitance – BIPOLAR
Need a cheap, dirty 2 or 3 component current source – BIPOLAR
Need a low voltage/noise amplifier – BIPOLAR
Need an amplifier with VERY low input/bias current – MOSFET
Need a low noise AND low input current amplifier – JFET
Need a one component current source – JFET
Need switch or amp that must cost almost nothing – BIPOLAR
Need multiple transistor package that has matching – BIPOLOAR
Need switch that may be over-voltaged – MOSFET
need switch/amp that sits in nasty RF environment – MOSFET (BIPOLARS rectify & cause offsets)
MOSFETs can generally switch faster (they certainly require less complex and less power to drive their gates). But if I’m not mistaken, BJTs designed for the task can switch very very fast since they have no gate capacitance to charge and can also operate in quasi-saturation mode for even faster switching at the expense of conduction efficiency. MOSFETs have less losses when used as a switch at “lower” voltages (lower as in industry’s definition which is <~200V).
MOSFETs act like a resistor when on while BJTs act more like diodes. The resistance can be modified by changing the “dimensions” of the MOSFET while the BJT’s “diode voltage drop” can’t be changed so easily unless the materials are changed. THis tends to make MOSFETs have less losses at the lower voltages but also means MOSFETs can be paralleled since current imbalances will cancel out.
With parallel BJTs, the best BJT will hog the current from the other “not so good BJTs” and burn out and the cycle repeats with the remaining BJTs until they are all burned. This is similar to parallel diodes. You can correct for imbalances by manually tuning resistors in series with each BJT, but for power applications that’s needing massive resistors and wasting lots of power.
here’s a short clip of the delay circuit that i intend to convert to a stomp box. i’m still working out the kinks in the pre-amp stage to solidify the input signal from the guitar though. just a sample though to hear how it might sound with some guitar tones. more soon.