Rich McGee

How about a wizard who has a serious speech impediment, so he can't cast spells with incantations? Easy to model in HERO, not so much in D&D.

How so?  Hero's approach is pretty obvious, but D&D is a point-free system where GM fiat covers a multitude of needs.  The GM can offer one of two obvious choices: 
 
1) Just don't let the PC cast anything with verbal components, and maybe they can't use items with command words either.  This is obviously quite challenging and likely to render them really ineffective, so don't run mini-maxed high-danger-level games with the poor wiz.  Think Rincewind, magician in name only, or maybe someone who's really into angsty tragedy characters can make it work as a backstory for a wizard who was tortured and mutilated to deny them they powers.  Tone your game down to a low-combat, high-social/politicking format and do something different.  You might want to to give the PC a familiar capable of speech, a simple magic item that lets them speak through it, or some other way to communicate without relying on NPC literacy and/or hand signs/pantomime.  Nothing that can cast spells that way, of course, that would defeat the original concept.
 
2) Make the impediment less severe, pick a failure percentage, and whenever to try to cast a spell roll and see if that attempt works or just fails without burning the slot.  Depending on how high you set the percentage that might still be a huge drawback or might just make you unreliable and erratic if it's down around 5-15%.  Maybe it only effects actual casting (letting them use non-scroll items with command words) or maybe it effects all speech, maybe checking for listeners failing to comprehend or just turning that percentage of words incomprehensible each time they speak.  You might need to write the whole campaign with the PC in mind (and perhaps a bunch of other PCs with similar issues that make their class more difficult - an "Iron Man" variant) as above, or if the percentage is low it becomes a quirk. 
 
Probably award extra XP for struggling through it, either irregularly as they accomplish specific things or just as a fixed percentage of XP earned.  I like that idea a lot better than the way older TSR editions rewarded extra XP for having terrific stats that fit your class best.  That was totally unbalancing, for what little the game ever cared about balance. 
 
Or just say life's unfair and do nothing for them.  The player presumably wanted this and will play it without rewards rewards.  I've seen plenty of people just choose to start PCs with a missing eye, hand, etc. to make them more unique, something TSR D&D was not great at.    

Kind of, but I wouldn't call it universally true.  The GM can use whatever tracker length they like, but longer ones are usually used for big, epic scenes - the climax to a multi-session story arc is an example the designers use - and shorter ones are often minor scenes with an obvious "win condition" that the heroes don't need as much time to complete.  In practical terms that usually translates to difficult scene (following their guidelines for scene construction) = longer tracker while easy scene = short tracker, but again there will be exceptions.  The encounter budget system seems optimized for moderate scenes with the 8-round 2/4/2 GYRO tracker, but it work for easy and hard scenes fairly well.  I think the design goal was to have the tracker almost equal the expected time for the heroes to win (or have a Total Party Knock-Out), with the ideal situation being that every hero has something meaningful to do on their turn (even if it's "just" Boosting a fellow hero) until the win/loss condition plus an extra round or two as allowance for PCs going Out (at which point their options are sorely limited).  Action economy is pretty important, as you might expect.  Makes challenges (which often require multiple Overcomes) more meaningful than they'd be without time pressure.
 
The spread of GYR rounds is also important for determining hero (and, rarely, villain) effectiveness as you surmised, and the more Red rounds there are the more impact those potent Red abilities will have.  Pretty rare for a scene to last more than two rounds once you're in Red, both because the team is now firing on all cylinders and because you may have a TPKO from heroes going Out back in Yellow.   By comparison, Green rounds really limit the PCs, and often get spent doing Overcomes on existing challenges, swatting some minions, or putting out mods with Boosts and Hinders rather than whaling on villains.  The book's suggested epic scene tracker has a 1/6/4 GYRO spread IIRC, and I don't think I'd ever bother with more than two Green rounds in an action scene unless something really weird was going on narratively.  Green is the calm before the storm in an action scene, Yellow is where the bulk of the action is, and Red is where the end (whatever that might be) is near.  Having the scene tracker run out is almost always beneficial to the villains, but what that benefit is can vary a lot.  Maybe the heroes get a fiat KO and wake up partially healed in a prison or deathtrap, maybe the villain's gadget/ritual/scheme seriously changes the world and everyone has to deal with that (Thanos snap, summoning Cthulhu, etc.), or maybe they just stole what they wanted and get away via a pre-planned escape scheme.
 
Worth noting that there are a few ways to use Red abilities before the tracker is in the Red, including just getting beaten up really badly early on.  A big part of the game's tactics involves around deciding when you need to (say) invoke a twist to use a key Red ability while still in Yellow.

Kind of, but I wouldn't call it universally true.  The GM can use whatever tracker length they like, but longer ones are usually used for big, epic scenes - the climax to a multi-session story arc is an example the designers use - and shorter ones are often minor scenes with an obvious "win condition" that the heroes don't need as much time to complete.  In practical terms that usually translates to difficult scene (following their guidelines for scene construction) = longer tracker while easy scene = short tracker, but again there will be exceptions.  The encounter budget system seems optimized for moderate scenes with the 8-round 2/4/2 GYRO tracker, but it work for easy and hard scenes fairly well.  I think the design goal was to have the tracker almost equal the expected time for the heroes to win (or have a Total Party Knock-Out), with the ideal situation being that every hero has something meaningful to do on their turn (even if it's "just" Boosting a fellow hero) until the win/loss condition plus an extra round or two as allowance for PCs going Out (at which point their options are sorely limited).  Action economy is pretty important, as you might expect.  Makes challenges (which often require multiple Overcomes) more meaningful than they'd be without time pressure.
 
The spread of GYR rounds is also important for determining hero (and, rarely, villain) effectiveness as you surmised, and the more Red rounds there are the more impact those potent Red abilities will have.  Pretty rare for a scene to last more than two rounds once you're in Red, both because the team is now firing on all cylinders and because you may have a TPKO from heroes going Out back in Yellow.   By comparison, Green rounds really limit the PCs, and often get spent doing Overcomes on existing challenges, swatting some minions, or putting out mods with Boosts and Hinders rather than whaling on villains.  The book's suggested epic scene tracker has a 1/6/4 GYRO spread IIRC, and I don't think I'd ever bother with more than two Green rounds in an action scene unless something really weird was going on narratively.  Green is the calm before the storm in an action scene, Yellow is where the bulk of the action is, and Red is where the end (whatever that might be) is near.  Having the scene tracker run out is almost always beneficial to the villains, but what that benefit is can vary a lot.  Maybe the heroes get a fiat KO and wake up partially healed in a prison or deathtrap, maybe the villain's gadget/ritual/scheme seriously changes the world and everyone has to deal with that (Thanos snap, summoning Cthulhu, etc.), or maybe they just stole what they wanted and get away via a pre-planned escape scheme.
 
Worth noting that there are a few ways to use Red abilities before the tracker is in the Red, including just getting beaten up really badly early on.  A big part of the game's tactics involves around deciding when you need to (say) invoke a twist to use a key Red ability while still in Yellow.

So it is, like many narrative systems, very important to set up the contest: what victory means, what failure means and the consequences of each, with a handful of fuzzy edge elements you might throw in to qualify both victory and failure.
 
You say normal 8 round tracker. I presume longer teackers are easier for PCs, shorter ones harder.  I also presume that turning red sooner makes it easier because PCs get access to those abilities for longer in the scene?  Though that feels a bit counter-intuitive...

Hmmm.  How many players do you expect to have?  That value (H in the rulebook) determines how big your scene budget is, which is kind of a starting point.  It sounds like you'll want a moderate or even easy action scene for the rescue and escape (since they're secretly supposed to get away).  The normal 8-round tracker is probably fine, giving you two rounds in Green (where the tension is low), four in Yellow (as the action heats up) and two in Red (hopefully the climax where the heroes have to get away with the prisoners before overwhelming reinforcements trap them).  Since they're intended to get away, having tracker run out should maybe just give them some disadvantage going forward - or perhaps they get a reward in the form of those two "doomed by plot" prisoners surviving if they finish early, while a tracker exhaustion leads to them being gunned down horribly bravely delaying pursuit for the heroes?
 
From your description you'll want to spend one of your H "elements" on an environment ("Sinister Nazi Laboratory" or somesuch) whose twists will gradually put NPC guards and researchers into play, apply effects to one or both sides (eg an alarm might Boost the baddies with the Min die, while breaking a bunch of chemical flasks in a lab might Hinder or Attack everyone there), introduce a new challenge (eg a security gate slides shut, requiring an Overcome to bypass or force open).  Elaborate to suit, but remember that the environment will do something every round so you want to think about a variety of potential twists so you don't have to repeat them too often.  Environment twists usually start out with very small game impacts in Green, are roughly equivalent to a modest hero ability in Yellow, and can really produce huge changes in Red, with Major twists being stronger than Minors at each tier.  So don't be surprised if them seem trivial initially - they mean more and more as time goes by.   
 
Using another element or two on free-standing (ie not generated by the environment turn) challenges to represent the perimeter defenses or maybe breaking the prisoners out of their cells or whatever would give people some Overcomes to deal with, which is usually a good idea.  It helps generate twists to complicate things, and it lets heroes use the Principles and overcome abilities rather than just punching things all the time.  Experienced players frequently come up with crazy ideas all on their own that are best treated as Overcome actions, but for newbies challenges act as signposts for "think of a cool way to use your powers and abilities" points.
 
The rest of your budget can be spent on the baddies.  Sounds like you want mostly regular soldiers and maybe some tougher leaders, so a mix of minions and a few lieutenants as leaders is best.  Doesn't seem like you'd want an actual full-blown villain here.  They don't have to be actual infantrymen - the lieutenant mechanics are a good way to represent armored cars, halftracks and tanks, for ex.  Durable, but not so tough a hero can't deal with them in a few Attacks.
 
One thing the rules gloss over is mapping.  The game's very theater of the mind when it comes to ranges and positioning, but I still find it's a good idea to split a scene into at least a few different broad locations.  In this case, maybe a couple of different outer perimeter sections (maybe one backs on a forest and the other drops off as a cliff so the players can choose to use one or both to fit their powers?), a vehicle park in front of the lab building, and a few different interior sections (actual lab space, holding area for the prisoners, maybe a barracks or personal quarters/offices if they might matter).  Maybe a guard tower, Nazis love guard towers.  Usually best to have somewhere between H-2 and H+2 discrete (albeit often abstract) location so the heroes can split up or concentrate a bit.  Movement rules are pretty abstract as well, but I wouldn't let it get in the way of drama - don't count how many locations it is to the perimeter on the way out if a hero is just running like mad to beat the scene tracker, especially if they have mobility powers.
 
A single moderate difficulty action scene usually takes my groups under two hours to play, maybe three if there are a lot of distractions.  If you want a longer session you could throw in social scenes as endcaps or even just hit pause on the tracker and shove a short one into the middle of the action when the prisoners are first found (especially if the heroes have been fast and sneaky and the scene's still Green or low Yellow).  Remember to give them their Hero Point for social scenes, even if it doesn't matter in a one-shot.  You could also endcap with montage scenes before or after, although tend to go pretty quick.
 
If you really need to stretch the time, you could split things into an initial action scene where they fight past a patrol getting near the lab, montage to recover and sneak into the lab itself (using Overcomes) without alerting the guards, then start a second action scene as they find the prisoners just in time for a second patrol that found the remains of the first one to pull up and sound teh alarm.
 
No matter what you do, the usual GM proviso about no adventure plan surviving contact with the players applies, of course. 
 
That's a whole other kettle of fish, but you can do a lot more than just super-brawls with the GYRO system and different time scales on what a "round" is.  I ran a three-month election campaign for city mayor as a single very extended action scene (with other more traditional-format scenes nested inside it) and it worked out dandy.  All you need to make an action scene is some kind of deadline to provide urgency and justify a scene tracker, and election day is as good for that as the countdown timer on a bad guy's doomsday device.
 
Hope that helps some.  Best of luck if you do try running it.  Very different system from Hero, but it has its merits. 

Hmmm.  How many players do you expect to have?  That value (H in the rulebook) determines how big your scene budget is, which is kind of a starting point.  It sounds like you'll want a moderate or even easy action scene for the rescue and escape (since they're secretly supposed to get away).  The normal 8-round tracker is probably fine, giving you two rounds in Green (where the tension is low), four in Yellow (as the action heats up) and two in Red (hopefully the climax where the heroes have to get away with the prisoners before overwhelming reinforcements trap them).  Since they're intended to get away, having tracker run out should maybe just give them some disadvantage going forward - or perhaps they get a reward in the form of those two "doomed by plot" prisoners surviving if they finish early, while a tracker exhaustion leads to them being gunned down horribly bravely delaying pursuit for the heroes?
 
From your description you'll want to spend one of your H "elements" on an environment ("Sinister Nazi Laboratory" or somesuch) whose twists will gradually put NPC guards and researchers into play, apply effects to one or both sides (eg an alarm might Boost the baddies with the Min die, while breaking a bunch of chemical flasks in a lab might Hinder or Attack everyone there), introduce a new challenge (eg a security gate slides shut, requiring an Overcome to bypass or force open).  Elaborate to suit, but remember that the environment will do something every round so you want to think about a variety of potential twists so you don't have to repeat them too often.  Environment twists usually start out with very small game impacts in Green, are roughly equivalent to a modest hero ability in Yellow, and can really produce huge changes in Red, with Major twists being stronger than Minors at each tier.  So don't be surprised if them seem trivial initially - they mean more and more as time goes by.   
 
Using another element or two on free-standing (ie not generated by the environment turn) challenges to represent the perimeter defenses or maybe breaking the prisoners out of their cells or whatever would give people some Overcomes to deal with, which is usually a good idea.  It helps generate twists to complicate things, and it lets heroes use the Principles and overcome abilities rather than just punching things all the time.  Experienced players frequently come up with crazy ideas all on their own that are best treated as Overcome actions, but for newbies challenges act as signposts for "think of a cool way to use your powers and abilities" points.
 
The rest of your budget can be spent on the baddies.  Sounds like you want mostly regular soldiers and maybe some tougher leaders, so a mix of minions and a few lieutenants as leaders is best.  Doesn't seem like you'd want an actual full-blown villain here.  They don't have to be actual infantrymen - the lieutenant mechanics are a good way to represent armored cars, halftracks and tanks, for ex.  Durable, but not so tough a hero can't deal with them in a few Attacks.
 
One thing the rules gloss over is mapping.  The game's very theater of the mind when it comes to ranges and positioning, but I still find it's a good idea to split a scene into at least a few different broad locations.  In this case, maybe a couple of different outer perimeter sections (maybe one backs on a forest and the other drops off as a cliff so the players can choose to use one or both to fit their powers?), a vehicle park in front of the lab building, and a few different interior sections (actual lab space, holding area for the prisoners, maybe a barracks or personal quarters/offices if they might matter).  Maybe a guard tower, Nazis love guard towers.  Usually best to have somewhere between H-2 and H+2 discrete (albeit often abstract) location so the heroes can split up or concentrate a bit.  Movement rules are pretty abstract as well, but I wouldn't let it get in the way of drama - don't count how many locations it is to the perimeter on the way out if a hero is just running like mad to beat the scene tracker, especially if they have mobility powers.
 
A single moderate difficulty action scene usually takes my groups under two hours to play, maybe three if there are a lot of distractions.  If you want a longer session you could throw in social scenes as endcaps or even just hit pause on the tracker and shove a short one into the middle of the action when the prisoners are first found (especially if the heroes have been fast and sneaky and the scene's still Green or low Yellow).  Remember to give them their Hero Point for social scenes, even if it doesn't matter in a one-shot.  You could also endcap with montage scenes before or after, although tend to go pretty quick.
 
If you really need to stretch the time, you could split things into an initial action scene where they fight past a patrol getting near the lab, montage to recover and sneak into the lab itself (using Overcomes) without alerting the guards, then start a second action scene as they find the prisoners just in time for a second patrol that found the remains of the first one to pull up and sound teh alarm.
 
No matter what you do, the usual GM proviso about no adventure plan surviving contact with the players applies, of course. 
 
That's a whole other kettle of fish, but you can do a lot more than just super-brawls with the GYRO system and different time scales on what a "round" is.  I ran a three-month election campaign for city mayor as a single very extended action scene (with other more traditional-format scenes nested inside it) and it worked out dandy.  All you need to make an action scene is some kind of deadline to provide urgency and justify a scene tracker, and election day is as good for that as the countdown timer on a bad guy's doomsday device.
 
Hope that helps some.  Best of luck if you do try running it.  Very different system from Hero, but it has its merits. 

I think I did and got no uptake, the forums don't seem particularly active and I may not gave hung about long enough.
 
I was thinking of setting up a one-off, possibly using Sentinels for an arc of the Golden Age HERO game I am playing with my group.  I converted the characters but when I thought of the first scene I froze.  This first encounter us supposed to be the characters escaping from a lab where they had been getting experimented on.
 
Obviously the environment us important, lots of things that might interrupt or complicate the scene. There are the guards to defeat and a perimeter defence to breach.
 
So, the scene tracker.  How long? When to change G to Y to R? What happens if the tracker ends? Do I need another scene with higher stakes?  How long should that be?
 
I reckon this would become more obvious with experience of the game but I wanted the one off to go well. Any hints and tips would be useful, I want that first scene to get easy but, for plot reasons, not too easy (they are being allowed to escape but should not know that).
 
I was also wondering about whether and how the GYRO system might be used for the meta plot of the adventure.
 
Two heroes were captured,  the scene begins with the other heroes breaking in to free them, there will be three other prisoners, one of whom will escape with them, the other two will die/be recaptured. They need to get to London and return to their secret base.
 
Actually, this is all in their minds, the nazis are watching to find the location of the secret base and the surviving prisoner the captors way to communicate with the heroes. I want each scene to have the potential of giving away the secret, possibly using an additional GYRO counter.
 
Or am I being too ambitious? 🙂

You might better off asking over on the publisher forums, but as long as I'm here maybe I can help.  What are the question(s)?  If it helps any I did a two-part article on how I design an action scene over on my blog, the first part of which is here.