[1.0.x] [V1.9f] Kerbal Foundries wheels, anti-grav repulsors and tracks

[Climberfx]

Man, that i can do. Can you share the original model of it with me? I use Maya, can be Obj, Fbx, or Mb.

I ask, but i see you are not the creator of it. Have you a close relation ship with him?

I can download de mod and do the trick with Blender Mu importer. But will the autor like this?

Hey Lo-Fi, you agree to have some especial repressor parts for my mod Omicron? I can release as an extra pack.

That way i don't need to download it and get it without colors. ()

[Gaalidas]

Yeah, so I'm just a vocal supporter and side worker for this mod. I don't have any power over the distribution of anything. It's just an idea. I can't necessarily get you the source myself, but lo-fi has been rather cooperative with other addon developers in the past. Even without the model source, however, you could probably pull it off using the current configs and models as a reference. I believe they're not all that much different from a wheel, except that there's simply some form of transform instead of a wheel fixture. It may be more complicated than that, however, since I also know there are ray-casts involved in the KF modules. I honestly don't understand it well enough to explain it further. As the very least, though this is rather hacky, you could likely create your own part, then weld it in the config with a current repulsor model, then use a combination of the standard repulsor module along with lo-fi's object removal module to basically leave the repulsor functionality available while removing the visual aspect of the original model, allowing your own model to be the primary visual element.

I'm not sure all of that even makes sense to me, but you can look in my signature for my standard disclaimer which applies to just about everything I say.

[occ]

I am having a problem with the repulsors. I created a simple hovercraft (less than 50 parts) and put it on the runway. I found that the craft's repulsors did, in fact, react to the ammount of weight on the craft. I was also able to adjust the height of the repulsors. However the "extend/retract repulsors" function did not work, and even when I had no part of my craft touching the ground, (I knew this because the repulsors were drawing energy, and I was able to verify that the craft was not weighed down so much that it was negating the repulsor's function) the craft would not move. It was on the runway, so it should have at least slid forward. It did not. I even tried placing a pair of lv-t engines on the back end. I do run somewhere around a metric ton of mods, but I do not think that was the problem because KSP's RAM usage did not go above ~2.5gigs (that is with aggressive texture management and half-res textures) Please ask if you would like any more information about this problem.

[Gaalidas]

Hmm... one thing to pay attention to is whether or not the repulsors can actually reach the ground. We've seen cases where the repulsors were placed too high up on the craft and couldn't repulse high enough to work. Also, the extend/retract functions are poorly named, since all they do is increase/decrease the height. Also, the repulsors will draw energy if any of them are actually turned on, regardless of repulsion activity (I'm pretty sure of this, not entirely though) and with no regard for what parts of the graft might be having collision issues with the ground.

You are right, however, that the repulsor field does mimic wheel suspension in that it responds not only to the height setting, but also to craft weight. Another thing to keep in mind is that even if a repulsor is as zero, there is still a collider under it that can still collide with the ground and, even when not actively repulsing, may scrape the ground even if the appearance is that the craft is hovering.

so, check the repulsor attachment positions to make sure they aren't too high (if not using the surface attached repulsors) and, when in doubt, add more power and repulsor units to the mix and try again. You might also see if your logs are pumping out any errors while attempting to hover, but I have a feeling this is not an error with the plugin's function here.

[Madrias]

Occ, the extend/retract function used to be like an on/off switch, but after much testing revealed the destructive power of suddenly-engaged repulsors, it was set up to raise or lower the repulsor power in increments of 5, I think. Tapping Retract should eventually get you to the ground and shut the repulsors down, and tapping extend will gradually raise your vehicle back into the air.

[Gaalidas]

Occ, the extend/retract function used to be like an on/off switch, but after much testing revealed the destructive power of suddenly-engaged repulsors, it was set up to raise or lower the repulsor power in increments of 5, I think. Tapping Retract should eventually get you to the ground and shut the repulsors down, and tapping extend will gradually raise your vehicle back into the air.

What he said... IF all of the things I mentioned are ironed out of course, which is all on the designer's end.

[occ]

Ok, I have been playing around with the repulsors a bit more, and I now see how weak they are. So, here is my question: if I wanted to design say, a fairly large ssto, how realistic would it be to give it repulsors instead of conventional aircraft wheels?

Fyi, the craft I have been working with is currently a little over 13 tons, has one repulsor on the front, and 16 on the back and the center of mass is 2/3 of the way from the front... I have gotten the one on the front to able to lift the front end up, but the back end isn't budging, even with all 16 at full power the back end barely budges, and I am still not able to lift the craft. (The weird effect of the entire craft off the ground is still there as well.

I have even messed around with the strength slider. (That's how I got the front end to levitate)

Edited April 7, 2015 by occ

[Gaalidas]

Well, you'd need to give it a lot of repulsors, depending on the size/weight. As for their weakness, did your tests include adjusting the strength sliders in the context menu?

[occ]

Well, you'd need to give it a lot of repulsors, depending on the size/weight. As for their weakness, did your tests include adjusting the strength sliders in the context menu?

Yes I did. By the way, what does the "dampening" slider change?

I set the to full strength, and now they spin when I run them at full power. It makes the back end of my craft sort of bounce up and down like a low rider.

Edited April 7, 2015 by occ

[Gaalidas]

I can't really tell you what everything means specifically, but my guess is that strength is exactly that, how strong the repulsion field is (higher means more... umm... "ooomph" it has) and dampening deals with how it responds to impacts. This allows you to have the repulsors respond to a landing after catching air over a hill by giving a little instead of suddenly impacting with a hard bump. This also allows the craft to go over a tiny little bump without actually bumping itself into the air the way a wheel would. I think anyway... lo-fi is the guy who knows things here, I'm just very good at typing a lot and saying very little in the process. It's a gift.

I'm unsure what you mean by "spin" because these parts are not meant to spin... but the bouncing I can definitely see happening if strength and damping are set to certain levels. As for what level of damping is considered smoothest of a ride, I don't really know. I usually leave it alone. You just have to fiddle with the values and see what happens to really get a feel for it.

The main problem with repulsors as landing gear is the lack of directional control, or braking. Repulsors have absolutely no friction with the ground, so the only thing slowing you down is the air and gravity, the latter of which the repulsors are actively trying to counteract in the vertical direction, but will gladly let you slide uncontrollably down any slight incline your craft sets itself upon.

On the up-side, if you want to land on the water, or on rough terrain, and your descent is shallow enough, repulsors will allow you to land safely most of the time.

[Madrias]

From having used repulsors as anti-lithobraking devices on lighter crafts, I could give a bit of information.

Maximum Altitude is not what you want, contrary to popular belief. I've found it can make craft a little bit unstable. Around 75 to 80% of maximum is preferred.

Strength, set somewhere around 80% or so. Powerful, but not pushing those risky boundaries (again, I've noticed some respond fine to 100% power, but others get touchy and do stupid things, like spinning or auto-destruction.)

Dampening, around 20% to 70%, depending on the craft. Lower values are better for heavier craft from my testing. Higher values are good for light craft traveling rapidly across rougher terrain, where the ability to absorb that bump, terrain seam, or momentary loss of ground is more critical than your ground clearance.

That many repulsors, I'd be worried about power draw concerns. Try removing about half of those rearmost repulsors, make the needed adjustments, and see if the ship is more stable.

Also, keep in mind that repulsors will not steer you, nor will they slow you down. You'll need an action group to slowly turn them down and eventually off, and you'll need landing gear to touch down on, preferably some with brakes.

Note: I'm not a developer for the mod, so my mentions of what things do is more based on what I've gotten from playing about with the settings. I've done my testing, but I'm not always 100% right.

[Gaalidas]

It's a better description than I could provide, and I've worked on the code a bit too. Thanks for clearing that up for us.

[Thorbane]

Does the max suspension strength increase with the cube of the wheel scale? I've found that a lot of heavy craft need an absurd number of wheels to avoid bottoming out the suspension.

Edited April 11, 2015 by Thorbane

[Gaalidas]

That's an excellent question... and I don't know the answer. I'll do some digging around in the code...

I know for certain that the tweakscale exponent settings don't take that into account at all, but we have a special variable that we set along with the scale which modifies a lot of other variables within the code, so it'll take some digging to figure out if the suspension is modified by it.

However, absurdity with wheel numbers is a rather realistic situation with extremely heavy craft. Considering the vessels in KSP are likely a lot heavier than what we would build in the real world due to the fact that we would build parts that are made specifically for the purpose of that vessel rather than using rocket and space station parts to fill in the gaps... well, we'd likely be using an absurd number of wheels too.

EDIT: So, I did some poking around and it turns out the "springRate" variable, also a config parameter, is indeed multiplied by the "tweakScaleCorrector" variable which is set to the equivalent of the current scale. "springRate" is a badly named variable for "Spring Strength" in the code. However, part mass (and other settings) is covered by a completely different exponent which is outside of this mod's specific settings, which means the strength may not be updating to compensate for any other changes such as the part's own mass. If the wheels are having to compensate too much for their own mass, then the vessel mass may not be taken into account correctly. I will aim to try and override that setting for the KF parts and see if that makes any difference.

EDIT2: So, as a test, I'm adding a second modification on the springRate and damperRate in the tweak scale settings. If you haven't modified your local file, then you could try modifying your own local file and see if it makes any difference. In the code, those settings are already multiplied directly with the tweakScaleCorrector variable which is made to equal the scale level. However, if we also scale the base value that the scale corrector multiplies with then, in theory, we can increase the spring/damper to better match the new part scale without touching the original code. Granted, this is a work-around that we would want to update in the code itself later, but for now this is a good way to test the results. Since I don't have any really heavy rovers built at this time, it would be quicker if you were able to test the results. I'm going to try and explain what to edit now.

So, in the file named "KF_Scaletype.cfg" you will be looking for this:

TWEAKSCALEEXPONENTS
{
    name = KFModuleWheel
    !tweakScaleCorrector = 0.8, 1.0, 1.2, 1.4, 1.6, 1.8
}

Except there will be different values matching what scale levels that size-class can take. You'll want to make two copies of the line starting with "!tweakScaleCorrector" and place them under that line, then rename the two parameters to "!springRate" and "!damperRate" respectively.

In the end, the sample from above should look similar to this:

TWEAKSCALEEXPONENTS
{
    name = KFModuleWheel
    !tweakScaleCorrector = 0.8, 1.0, 1.2, 1.4, 1.6, 1.8
    !springRate = 0.8, 1.0, 1.2, 1.4, 1.6, 1.8
    !damperRate = 0.8, 1.0, 1.2, 1.4, 1.6, 1.8
}

This will have TweakScale make those two base variables match the scale level for that part, which will be further multiplied by the corrector in the code, thus increasing them even further that usual. This change could, however, have unusual results, so use caution during testing. I also don't know if it will affect already launched vessels or not. Usual backup procedures should be used in this case to avoid any permanent breakage.

Let us know, if you decide to try this highly experimental change in your own game, what happens (if anything).

Edited April 11, 2015 by Gaalidas

[krolcwan]

where is button for download?

[lo-fi]

The blue Kerbalstuff link at the top of the OP.

Just thought I'd stop by to say I haven't disappeared completely... Have a lot of non KSP stuff on my plate at the moment, but keeping an eye on squad dropping the next version in case I need to update.

Hope you're all having fun!

[DarthWall]

I was thinking... Would be great if has some kind of magnetic or gravitational RCS, that just consume electricity and are omni-directional. It will be possible to create amazing sci-fi spaceships much more controllable like in the movies.

[lo-fi]

It would be possible - and been suggested a few times now - but I'm afraid I'm not really doing any active development at the moment.

[Little Green Stone]

Love this mod, but I seem to have a wee bit of a problem here.

MechJeb's autopilot can't handle these wheels well.

All the crafts using them start turning left and right instead of just going straight, usually with a +/- 2/3° error, which annoys me greatly for some reason.

Is it a local problem, or is it a known issue?

[lo-fi]

Never even tried it! Anyone else?

Latest KF and mechjeb? I'll have a go myself if I get some free time, but this may not be easy to narrow down.

[Little Green Stone]

Never even tried it! Anyone else?

Latest KF and mechjeb? I'll have a go myself if I get some free time, but this may not be easy to narrow down.

Yes, latest KF and MJ also, just downloaded them a few days ago. I've tried setting group numbers to 0, I've tried manually stopping this, I've tried engaging SAS, MJ's stability control, but seemingly it just...happens, I can't prevent it. Stock wheels, for me, behave normally, also the problem only occurs with Rover Autopilot (or what) engaged, otherwise the wheels work pretty much great!

...Could you tell me how do you rove around without an autopilot? Do you put some weight on your keyboard or do you use another, similarly refined method? ^^

Because I'm not hell-bent of using MJ to keep my rovers in a set direction and speed if there is another way.

[smjjames]

Never even tried it! Anyone else?

Latest KF and mechjeb? I'll have a go myself if I get some free time, but this may not be easy to narrow down.

I can confirm it as well, MechJeb has a harder time controlling the KF wheels and especially the tracks, seems like it's because they turn faster than stock, and so, MJ has a slower reaction time or is turning them too much. Seems more like a MJ thing than a KF thing.

[Little Green Stone]

I can confirm it as well, MechJeb has a harder time controlling the KF wheels and especially the tracks, seems like it's because they turn faster than stock, and so, MJ has a slower reaction time or is turning them too much. Seems more like a MJ thing than a KF thing.

That is strange. I've built but 3 rovers using tracks, one using a pair, the other two using two pairs, those things too start wobbling a little but for me, all 3 stabilizes after a few seconds/half a minute tops.

But yes, it seems like MJ is overcompensating the steering, maybe because these wheels have less traction, maybe for some other reason, I do not know.

[lo-fi]

Thanks guys. I suspect this is wrapped up in control theory. Welcome to the world of negative feedback, phase changes and control loop stability.

I reckon I know what's going on (notice the kf wheels do not just snap left and right instantly? It'll cause a delay in mechjebs control loop. Too large a delay causes negative feedback to turn into positive), but it'll be a mechjeb fix if my theory is right. Might be best off asking over there. The hunting you describe is a dead giveaway, though.

[Gaalidas]

Nice to have you lurking around again, lo-fi. This thread has been dead for a while now.

So, on the topic of the magnetic RCS, that could be done by basically recreating the ion-RCS, but stripping out the xenon requirement, and using something like RCSFX to eliminate any form of visual effect to go with the RCS firing event.

As for the control issues, I have never actually tried using mechjeb with a track setup, but it would make sense that it would have issues. It's built to handle rovers that have to be moving forward to turn and tend to turn without half of the rover actually slowing down slightly to finish the maneuver. My suspicion is that it freaks out a bit when the vessel's velocity appears to change suddenly, which itself causes a trajectory change that it tries to compensate for.

The wheels, however, I have had no issues with. Granted, the softer steering tends to turn the craft a little less precisely (in the time that mechjeb would normally expect the turn to be done) and so its reaction time is a little slow, but otherwise it hasn't given me any trouble.