Best way to make this true-bypass?

[DuoSonicboy]

I tried converting this some time ago with a DPDT, the problem is that there was then a huge volume drop when engaged.  Stock, this is compensated for by reducing the bypass volume (with consequent loss of all clarity, midrange, etc.).

I'm looking for a way to do true bypass and get matched volume by:
1. Boosting the effected signal (suggestions given that there isn't a lot of room inside the box for a booster)
2. Cutting the bypass signal while maintaining most of the high-end
3. Modifying the circuit itself to put out more signal (no idea where to start on this one).

Any suggestions?

[Mark Hammer]

This circuit is crying out for an input and output buffer.  Most other phasers have one and this baby, for whatever reasons, doesn't.  Is there something special about it that makes you feel compelled to build it, as opposed to something with buffers that will lend thsmelves towards TB without level drops?

[DuoSonicboy]

It's a vintage piece I'm trying to make more useful.  If there's no easy way to add a buffer stage, I might just sell it and get a uni-vibe.

[slacker]

Like Mark said the easiest thing to do would be to add a high impedance buffer to the front of the effect between the input jack and R1. That would get rid of the tone sucking so you could keep the original switching, no need for true bypass.
You can build a simple Jfet buffer pretty small so you should be able to fit it in the pedal somewhere.

[earthtonesaudio]

Yeah there's gotta be room for two jfets, two caps, and 4 resistors in there, even if you have to point-to-point wire them from jack to board.

[Mark Hammer]

The problem with this pedal is that it has some attenuation already built in via R20/R22.  So, even if you cancel the phase-shifted signal, you're cutting back the clean.  Bypass the whole effect and you go from a straight wire to a considerable attenuation.

Looking at it more closely, though, you may have some salvation in R3.  That component is 470k, yielding a gain for the unit of about 1.2x.  If R3 is dropped to around 100k, you increase the gain at the front end to 2x.  Dropping it to 47k gets you a gain of a little over 3x.  It is a riskyt proposition, though, since the FETs may not be able to take that much signal without clipping.

The other problem is that the clean signal is taken from before the op-amp, so you'd have to fiddle with the values of R20 and R21 to get an appropriate balance.

[DuoSonicboy]

You guys are awesome - thanks for all the replies!

It looks like I can squeeze a JFET buffer in between where the two boxed jacks sit - I've got to make a small bear order soon anyway so I'll get the parts and make it happen.

[earthtonesaudio]

You might want to do an amplifier on the output instead of a buffer, considering Mark's observations about the gain of the circuit. 

[Mark Hammer]

Alex is right.  In an ideal universe, one sticks the brunt of the gain at the input so that any noise accumulated throughout the circuit is not amplified at the output.  This assumes, however, a tolerance for high input signal levels, and in the case of JFETs, they can be incapable of that.  So the optimum is to provide juuuuuust a bit more gain going into the sequence of phase-shift stages to squeeze all the S/N ratio you can out of it (and that is why I was not suggesting that R3 be dropped to 22k or less), and then tack on a wee bit of "make-up" gain on the output, via a single-transistor circuit.

Finally, when it comes to noise in phase-shifters, a strategy that shows up often in commercial designs is sticking a small-value cap in the feedback path of one or more phase-shift stages to roll off the upper treble where the hiss lives.  So, if the phaser used a 10k input and 10k feedback resistor for each stage, you'd stick something like a .0012-.0018uf cap in parallel with the 10k feedback resistor to produce a 6db/oct rolloff aboves ome frequency.  For instance, 10k and .0015uf gets you a rolloff above 10.6khz.  The general idea is to aim for a rolloff that trims some hiss but still lets you hear the notches created in the upper ranges.  Normally, one would only reach for a solution like this once every 4 stages.  So, if it was an 8-stage phaser, you might see an extra cap on stage 4 and 8.

The alternative is something like what was used in the old orange Ross phaser ( http://www.home-wrecker.com/ross_phaser_orange2.png ).  There, an additional 5th phase shift stage (fixed rather than swept) was inserted in the recycle/feedback path to further enhance the regen sound.  Because regeneration runs the risk of further exaggerating any hiss that has accumulated over the phase-shift stages, Ross elected to stick a .001uf cap in the feedback path to keep hiss and and HF oscillation in check.