Let's say I have my XF in Voice Mode. I go to my new Sax Growl Ctrl voice, which has the Foot Controller controlling E-LFO Depth, which is how I control vibrato/tremolo of the non-growling elements (Waveform 1097).

When I start to play, the vibrato/tremolo is all the way up (i.e. the E-LFO Depth is maxed out). I can zero it out with the foot pedal, and then it works like I want it to. But I can't figure out how to configure the voice so that the E-LFO Depth is zero when I first get to the voice.

I'm sure there must be a way to do this, but I haven't found it yet. Where is it?

In order to configure the Voice to start with E-LFO Depth at zero, you must configure it to follow a controller that defaults to "minimum" as its starting point.

Minimum-to-Maximum
Certain physical controllers (Source) are minimum-to-maximum controllers that send 0-127 when active. Some reset to minimum (0), others reset to maximum (127).

For example, on your XF, while in VOICE mode, the assignable min-to-max controllers include the MW(01), BC(02), FC2(04), FC(11), AT; the minimum or maximum controllers are AF1(86) and AF2(87) - as switches they are either On or Off. FS(88) is the quintessential assignable momentary Foot Switch, selecting this will apply 127 when engaged and return to 0 when released

By MIDI convention Controllers will RESET upon Program Change and/or Mode Change. The convention serves the ultimate goal that when reset a program should be returned to a condition that is ready to sound. This may or may not serve your goal. The value that they reset to is as follows:

MW - minimum 0
BC - maximum 127
FC - maximum 127
AT - minimum 0
AF - minimum Off
FS - minimum Off
SUS - minimum Off (includes Sostenuto)

The reason your Sax Growl defaults to growling initially is because you have selected a physical controller (FC) that defaults to maximum value when you recall a Program. Since you want to start with the growl not being applied...
You can either:
_ Move the pedal to heel down position before you begin playing
_ Select another physical controller one that defaults in the direct you require
_ Bias the controller to work in reverse (toe down equals no-growl, heel down equals growl). This involves reversing your thinking.

Zero in the Center
Other physical controllers have the same range of 128 steps but have their zero point in the middle, so you can go above/below, or left/right of a center point

AS1(16), AS2(17), PB, RB(22)
These each have a story... The value for the General Purpose 1 and 2 (cc16 and cc17) reset to zero, but Yamaha allows you to store an offset value for these programmable devices. The PB is spring loaded and therefore resets itself physically to center (zero). The RB (Ribbon) is initially recalled at center but is programmable on a per VOICE basis to Reset to its nominal zero (center value) when released, or to HOLD the last value when released.

Choosing Controllers
Choosing controllers requires that it not only be available but that it works as you require. This includes as intuitive a gesture as is possible and that it works in the direction you require.

You must recognize that the MIDI convention that governs controllers and their use, has a set of rules that determine the initial setting. See the DATA LIST booklet, page 125 "3-2-2 Reset All Controllers". This table gives you the conditions mandated when a Reset All Controllers message is sent (like when you initially recall the Voice)

Alternatives
Alternatives to applying growl with an FC7 (obviously not optimum for your goal), could include a FS plugged into the Assignable Foot Switch jack. Will work like a sustain pedal (momentary) growl applied when you step on the pedal, stops when you release the pedal. The Assignable Function 1 or 2 buttons that would bring in that Element when pressed (individually programmable as to momentary or latch), RB - growl applied when you touch one or the other edge of the Ribbon. AfterTouch, growl applied when addition pressure is applied to a key.
And of course, MW.

You will need to discover and judge for yourself which is the most intuitive gesture for you, personally. We could discuss which ones work or don't work, but in the end controller assignment and Depth settings are very personally. Having "played" the VL1 where "programming" an instrument from scratch was for professors, adjusting controllers was the "programming" for the user; how you set your controllers is what allowed each player to find their individual sound. Same as it is in the real world... Or closer to the real thing, any way.

We are not certain, exactly, how you've set things up to bring in the growl, but try several controller assignments (given the knowledge of the default conditions and directions that controllers naturally work)