For my Concept SRX and OS32sx, I use 100% cyclic to throttle mixing. Yes, you read that right. One day I figured I'd add more and more mix until it started to overspeed during rolls, and then back off a little (sorta like dialing in gyro gain - you don't know you have enough until you've seen too much). I never saw the overspeeding happen during aerobatics, so I just left it at 100%.

I DO see a little bit of overspeeding when I wiggle the cyclic stick in a hover, but it's no big deal. I'm also running a tuned pipe with that engine though - a muffler might allow enough overspeeding to be annoying.

For my Futura SE and YS61st2, I let the governor take care of the throttle for me, and I don't use any mixing for the throttle.

For my Concept SRX and OS32sx, I use the tuned pipe and cyclic-to-throttle mixing to get a consistent head speed.

The narrow powerband of a tuned pipe makes them work pretty well without a governor (or even without mixing, if you ask some folks). As the engine overspeeds, power drops off to keep the overspeeding in check. If the 'normal' speed is on the high-rpm side of the curve, bogging the head speed down a bit will open up more power to keep the speed from dropping further.

Governors aren't perfect... Even with a governor, my Futura overspeeds in descents. The YS midrange is plenty rich, it's the idle that I can't get rich enough. Or rather, when I do set the low-end valve rich enough to stop overspeeding, the engine loads up at idle. I have mixed feelings about this engine. It's strong, but the mixture curve kinda ***** . At least mine does. Maybe I should try a new throttle barrel.

Anyhow, if you fly 3D without a governor or tuned pipe, you will benefit from cyclic->throttle mixing. If you have a tuned pipe, you probably don't want a governor, but you will probably still benefit from mixing. Try 50% mixing. If it helps, keep it. If it doesn't, set it back. You've nothing to lose by trying it for yourself. Then draw your own conclusions.

There's a widespread (mis)conception that the throttle servo should be really fast. I think this is bogus.

The way I figure things, the GV-1 servo doesn't have to be any faster than the collective or cyclic servos, since they are the ones that end up changing the rotor torque that the GV-1 has to respond to.

And besides, the throttle servo is going to be centered at around 65%, moving between 40% and 100%. That gives it about half as much distance to cover as the collective servo. So effectively the throttle servo is twice as fast as the rest of them to begin with.

Tail rotor servos need to be extremely fast, but a) the tail rotor mostly just responds to collective and cyclic loads, and b) the tail rotor has a relatively small effect on engine load, and it's biggest demands are transient anyhow (starting/stopping pirouettes, basically).

And, if you're not using a governor... no matter how fast your throttle servo is, it will never move faster than you ask it to. That is, it will never move faster than your thumb moves the throttle stick. There are people who suggest using a really fast throttle servo because then the throttle will 'lead' the collective, and this is why I think they're wrong. Throttle will only lead the collective if you're slamming the stick faster than the collective servo can move. How often does that happen?

For my Futura, I'm using a non-digital throttle servo with .18 transit time and 40 oz/in torque, and it works just fine.