EA Vibrato “SMOOTHING MODS” // jcm(c)2020
The “EA Preamp and Vibrato” by Anthony Leo, published by ELECTRONICS Australia (Nov/1968) is a very simple Tremolo (rd. amplitude modulating) circuit. It is, yet, another effect circuit that relies on a jFET operating as a Voltage Controlled Resistor (VCR).
An online search of demos (YT) shows a choppy sounding Tremolo circuit no matter what mods people bring to the circuit. IMO, this is yet another jFET circuit begging to be “fixed” …
SOME BACKGROUND ON JFET VCR’s
When attempting to use jFET devices as controlled resistors – ie., as seen in jFET based phasors, limiters, tremolos, VCA’s, etc. – it is a good idea to review basic “control” principles …
Vishay “jFETs as Voltage-Controlled Resistors”
in a nutshell: jFETs operating as voltage-controlled-resistors (VCRs) are meant to be “driven” by a Vgs voltage landing inside the Vgs(off)-to-0volt range … noting that Vgs(off) is negative for an n-channel jFET
THE STOCK DESIGN
now, let’s take a closer look at what that jFET is trying to do, and should not do, in this circuit /// … that is, seen from the point of view of this Vgs – the real control voltage, properly speaking …
here’s the logic:
Seeing that the jFET is Source grounded in this circuit means the Gate wants to see a strictly negative voltage relative to ground (since Vgs = Vg – Vs = Vg, in this case) … this clearly imposes a condition on Vg that needs to be met if we are to strive for smooth’ish VCR operation (ie., Vgs>0 is a no-go zone in this game) …
But since the output (pot) of the LFO, being ground referenced itself, puts out a bipolar voltage (ie., it goes above and below ground in its sweep) then clearly, for half the time the LFO won’t be doing squat to the response … herein lies the problem
In other words, the controlled-resistance response simply cannot avoid being choppy no matter what we do outside of this constraint – it is inherent in the original design … the shape of the final response therefor has nothing to do with the type of n-channel device used (or rather, specifically, the Vgs(off) value associated with the VCR device) as the LFO output can get padded down as desired – a point that was clearly mentioned in Anthony Leo’s article!
SMOOTHING MOD #1 // BACK BIAS
the path to a smoother EA Tremolo is thus rather simple, all we have to do is lift the jFET’s Source potential to the right level … (ie. somewhere around |Vgs(off)|/2, see below … therefor, it’s a good idea to measure Vgs(off) for your VCR device
This is the same technique used in the “Orange Squeezer” and “1176” jFET limiters … same technique as used in the MXR “phase 45/90” jFET phasors … etc. etc… it’s so obvious, everybody in the DIYstumped community should now this by now …
My cheap/low-noise solution involves stringing several Si diodes together (dependent on the VCR device’s Vgs(off) value), a biasing resistor, a low’ish value trimpot (for making the back-voltage variable), and a big cap – everything else remaining the same … all with little more than an added mA of draw on the battery …
note, an LED or two can be used in lieu of the diodes with the assumption that Vgs(off) isn’t beyond 2volts in size … if Vgs(off) is larger still just add more diodes or LEDs till you get sufficient back-bias range …
the idea is to produce a back-voltage of around |Vgs(off)|/2 volts from the trimpot
SMOOTHING MOD #2 // SINE-WAVE LFO
The SMOOTHING MOD presented above will bring the behavior of the circuit more into a pseudo-linear operating mode of the jFET. To smooth things further we can make the LFO put out a better behaved (more ideal) sine-wave.
To do this we can take a page out of the Shin-Ei Univibe LFO and stick back-to-back diodes across the middle phase-shift capacitor. That’s it!
Using a single pair of diodes will reduce the output swing of the LFO significantly. Employing, say, 4 or 6 diodes instead (2 or 3 on each leg) will increase the output if one desires.
BALANCING OUT THE RE-DESIGN
To tie it all together, and to help cut to the chase, you would ideally want to know the Vgs(off) value of your chosen jFET. The maximum control range of the jFET is basically |Vgs(off)| volts wide, which then gives us an idea of how much LFO signal is needed when transitioning into full-chop territory.
Obviously, you would want to choose a jFET device with a Vgs(off) of no more than -2volts or so to get a good match between the LFO output and the range of modulation depth. Of course, this depends on how the LFO is modded, or not. This is where the padding resistor (in series with DEPTH pot) comes into play.
Good Luck!