[1.3.1] Ferram Aerospace Research: v0.15.9.1 "Liepmann" 4/2/18

[Wanderfound]

I was able to get a plane into orbit and land it after reducing the wing weight and control surface sliders to 0.11. The plane is now closer to its original weight of 86 tons. With the default weights, it was 96 tons. Here's the thing: Before the latest update, it could carry 38 tons of cargo into orbit, so the new wing values were definitely not overloading it. In any event, it's still nearly impossible to use SAS, due to the oscillation that I can't get rid of.

SAS before .25

http://www.animated-gifs.eu/war-bow/0044.gif

SAS after .25

http://netanimations.net/dolphin--2.gif

This is a problem with all aircraft. Not just the heavyweights. If I can figure out how to stop it, I'll be golden.

Same here. Reducing wing weight and getting the aero analyses perfect reduces the problem, but it never goes away. Deltas, big tails, canards, whatever: all of the planes are determined to fly in a sine wave. All of the designs I've shown as examples would have flown fine pre-25; now, they're flapping about like the crap I made when I first started doing this.

Active control of pitch is constantly required, whereas previously I could fly hands-free to orbit (literally: http://forum.kerbalspaceprogram.com/threads/94490-The-Hands-Free-Mission-Challenge?p=1433461&viewfull=1#post1433461).

Maybe some sort of unintended interaction with the body-lift stuff? Is the wing weight stuff getting somehow applied to fuselage parts as well?

I'm also finding that Smart A.S.S. has once again become suicidally useless, although I don't know if the problem is in Mechjeb or elsewhere.

[ferram4]

@UncleCrusty: Listed in the last changelog as a feature; wing masses are now based on the amount of area they need to support, and the numbers are based off of the approximate wing mass of an MD-83 (which happened to be about in the middle of the approximate wing-mass-per-area for a few jet transports). It certainly isn't multiply by 5.05, it's 0.065 tonnes per supported area.

@Wanderfound & Voculus: Like I said, I don't even know how you got SAS to be useful in the first place. Reduce the amount of control authority you have or make the planes more stable.

[Hodo]

JackCY's suggestion actually works for the SAS issue, a bit. It is still there just not as bad. i reduced the control surfaces range from the standard 15, to 6 and it has just about gone away, the craft still is maneuverable as I need it to be. But SMART A.S.S MJ is well bad again.... no clue on that one but that isn't a FAR problem.

[UncleCrusty]

It certainly isn't multiply by 5.05, it's 0.065 tonnes per supported area.

The numerous types of wings I tested (delta wing type things) all resulted in 5.05 (100kg -> 505kg, 20kg -> 101kg etc.). Now that I've gone back and tested further, I've seen it as high as 9.06. Most seem to lie in the range 5 - 10x. At any rate, while this is certainly more realistic, many aircraft which would have prior been sane designs are nigh-inoperable (could perhaps have something to do with these new smaller flaps as well?)

[Voculus]

With respect Ferram, I'm not flying brand new designs for the first time, and just ignoring good design philosophy. All the planes I built after falling in love with your mod are now chores to fly in the best case, and impossible in the worst case. I'm not trying to use SAS as an autopilot, but as a tool to reduce the workload, and in the past it's performed in that capacity with no issues. But this is not merely an issue of SAS. The flight characteristics in general are nothing short of abysmal at this point. Using SAS to bring it to heel only illustrates the fact that a screw came loose somewhere. I am not laying the blame for any of this at your feet. You say nothing substantial has changed on your end, and I take you at your word. Nevertheless, FAR is interacting with KSP .25 in a way that I'm absolutely convinced is incorrect.

I'm sorry if that sounds rude. It's not my intention, and I have nothing but respect and appreciation for all of your efforts.

Edited October 12, 2014 by Voculus
I spell gud.

[blowfish]

I launch the shortcut steam made, and it launches the 64bit automatically. So I guess i have to go into the file and make a shortcut for the 32bit version. I hope Squad fixes the 64bit version soon. ( ; _ ; )

Steam won't launch 64 bit automatically unless you set it to do so (you would have changed the command in the launch options).

[ferram4]

@Voculus: So they were built out of SP+ parts before they were re-balanced for inclusion in stock? Well, the wings were not given the correct parameters, prior to 0.25 and the masses of the fuselages changed, IIRC. The only flight mechanics that have changed at all are that intakes now have drag somewhat more than an equivalent-sized nosecone, so unless all your designs exploited incorrect wing parameters and no intake drag, nothing should have changed.

[GabeTeuton]

Install Active Texture Management and 32 bit will probably stop crashing. Also use -force-opengl. Both reduce RAM usage.

It is what it is and for good reason. Use OpenGL, that should help a lot. If you don't like it, please complain to Squad. They are the cause of this, the modders are only protecting the user (that includes you) from frustration.

Thanks for the tips, i ended up reducing the mods count and got it to work perfectly well without crashes AT ALL... tried the ATM but it kind of breaks the toolbar icons and (it just breaks the icons not the functionality hehe) but i didn't like it, besides i got it to work... it runs so much better in 32 bits... I didn't mean to complain with the devs of the mod anyway, just wanted to share my frustration, but i ended up solving it... Anyway i'm gonna definately test the opengl setting, and if the graphics stay as they are i'm going to keep it...

Just as a side not, this are the mods i was using before and the ones i'm using now:

Before (x64)=

1) Astronomer's Performance Improvement

2) 0PinLine Fairings

3) Astronomer's Visual Pack

4) Better Science Labs

5) Chatterer

6) Docking Port Alignment

7) FAR (not working until i noticed it)

8) Hot Rockets

9) Infernal Robotics

10) KAS

11) Kerbal Alarm Clock

12) Kerbal Engineer

13) KSP Interstellar

14) Part Wizard

15) Precise Node

16) RCS Build Aid

17) Science Library

18) Station Science

19) TAC Fuel Balancer

20) TAC Life Support

21) Time Control

22) Toolbar

After (x32)=

1) Toolbar

2) Better Science Labs

3) Chatterer (this one is awesome, useless when it comes to gameplay, but awesome)

4) FAR

5) KAS

6) Kerbal Engineer

7) Infernal Robotics

8) Docking Port Alignment

9) KSP Interstellar

10) Procedural Fairings (awesome update)

11) RCS Build Aid (always useful)

12) Station Science

13) TAC Fuel Balancer (idem RCS Build Aid)

14) TAC Life Support

[Wanderfound]

I'm afraid that I have to agree with Voculus. Things built as stockified replicas of SP+ stuff that previously flew perfectly now fly badly. Things built as new from-scratch .25 designs with all-green analysis screens fly badly. I can make them work, but they're all a dangerous chore to fly and nothing I'd be willing to offer up as a design for anyone else to use.

This is with wing weight wound down to 0.12. That weight still seems to give wings much stronger (and presumably heavier) than ye olde Ferram. Given that 0.12 seems to give extra-strength wings, I really wonder why the scale goes up to five; that's well past the point of "ludicrously indestructible". There couldn't be a decimal point out of place somewhere?

Is it possible that just getting the weights back to what they were may solve the problem? Going from untweaked to tweaked-to-0.12 certainly reduced it.

[regex]

Regex is looking into how SAS works to see if anything can be done about it, but I don't know how long it'll be before I can consider messing with that.

I just released my PID tuning plugin, anyone who is interested can click the link in my sig. It'll take me some time to document the back-end of the SAS system, but it's pretty simple.

[ferram4]

@Wanderfound: I don't know where you're getting the setting going all the way up to 5; it only goes up to 2. Based on the problems you're having, I suspect that both of you simply have improper installs. I haven't been seeing any of the issues you are complaining about. If reducing the mass that much has no effect, I suspect that you haven't downloaded the updated FARAeroData.cfg, and so strength doesn't vary with mass at all. Remember to delete CustomFARAeroData if you that is the problem.

Edited October 12, 2014 by ferram4

[Wanderfound]

I've been deleting my previous Ferram folder and installing the Gamedata folder from the zip download at https://github.com/ferram4/Ferram-Aerospace-Research

The "up to five" thing was me misremembering; mea culpa.

It still seems like the scale is hugely right-biased, though. I can't imagine ever needing anything higher than 0.5, and mostly what I want is 0.05-0.1, with an easy way to set it. Picking the sweet spot between 0.01 and 0.13 at the moment is a nuisance.

regex: any chance of a bit of explanation as to what the assorted input values in your tuner mean?

[regex]

regex: any chance of a bit of explanation as to what the assorted input values in your tuner mean?

Look up how a PID controller works on Wikipedia. Clamp has something to do with how well it "sticks" to a heading, lower values seem stiffer.

[Wanderfound]

Cutting pitch kp to 1/3 instantly stopped the rocking-horse motion.

You've just invented the latest must-have spaceplane mod. Thanks.

[JackCY]

JackCY's suggestion actually works for the SAS issue, a bit. It is still there just not as bad. i reduced the control surfaces range from the standard 15, to 6 and it has just about gone away, the craft still is maneuverable as I need it to be. But SMART A.S.S MJ is well bad again.... no clue on that one but that isn't a FAR problem.

I had to reduce to around 7 on my test cargo plane. I can do more when the plane is heavier. But none so far seem to be above 10.

Just tried 0.24, fixed FAR there as latest MM didn't work or something with 0.24 but gave me no warning, replaced it with older MM.

A plane built in 0.24, looks fine, balance is easier, no need to reduce control surfaces, they were the default 15 or 20, no stalling, single engine pulls like mad since the weight is only 6t with fuel.

Now building the same plane in 0.25 I had to do minor changes to wing type and placement to get the CoL more rear as the "10x higher" masses of wings were messing with CoM and the plane turned into a lazy 10-12t beast due to the wing weights

Hacking it down to fly without fuel using infinite fuel I got closer to original weight and plane stats, but still somewhere around 8t. Plane was too sensitive to control, had to reduce all control surfaces to 10 and it still liked to go into minor stalling.

0.24 is so much easier to build and fly.

0.25 the new wings and such can be a pain. The new parts, not only wings are unbalanced. Squad messed it up.

Edited October 12, 2014 by JackCY

[NathanKell]

Do blame me for the scale, not ferram, though; I wasn't aware how much stronger wings were in this version, and so wanted to make sure that the "nooo far y u break mah wings!" folks would have recourse (at quite the price in mass).

That said: 0.05 (the lowest you can go) is lighter than modern fiberglass gliders; sounds more like ferram needs to make wings more breakable again if you can actually fly supersonic with wings that mass only 10% what a jet airliner's does. Or you just have an incredible amount of wing area, of course.

[ferram4]

@Wanderfound: If you're setting the wings to be 10% their default mass, but somehow they're still stronger than the previous build, then your wing strength is not varying with mass. Setting the wing mass to 10% of the default will result in the wing only being able to sustain 20% the forces of the default. Since the default wing strength doubled between the old FAR build and the current one, this means that your wings should fail at forces that are 40% the forces felt in the previous build. Since you're reporting the opposite, I have to conclude that something is wrong on your end. Delete all the custom config files in your FAR directory and re-download.

@JackY: As noted, these wing masses are more realistic and the wing properties for SP+ parts prior to 0.25 were completely wrong. You are complaining that FAR is handling things properly now.

[Wanderfound]

Do blame me for the scale, not ferram, though; I wasn't aware how much stronger wings were in this version, and so wanted to make sure that the "nooo far y u break mah wings!" folks would have recourse (at quite the price in mass).

That said: 0.05 (the lowest you can go) is lighter than modern fiberglass gliders; sounds more like ferram needs to make wings more breakable again if you can actually fly supersonic with wings that mass only 10% what a jet airliner's does. Or you just have an incredible amount of wing area, of course.

The current version goes in a step from 0.01 to 0.12, then up by 0.01 increments from there. 0.01 is too weak, 0.12 is usable but over-strong for many purposes. 0.12 is strong enough for supersonic aerobatics with both large and small winged ships (0.25 makes it easier, though). At a guess, it feels like ye olde Ferram SP+ eggshells were somewhere between 0.05 and 0.1, probably closer to the lower end of that range.

A numeric input box would be vastly preferable to sliders.

[Wanderfound]

Ferram: wing strength is varying with mass, it's just that unless you use the overly-fragile 0.01 the scale seems to go from "a bit stronger than before" to "adamantium".

I've just redownloaded the FAR dev build again, but it's been a clean install each time. It really does feel like an accidental factor of ten slipped in somewhere.

[ferram4]

There, try that build.

[Voculus]

I think I had a slight Eureka moment, and I believe the CoL indicator isn't quite accurate. I'm kicking myself for not taking screenshots, but it goes something like this: On my heavy lifter SSTO, I placed the CoL in relation to the CoM, such that it barely protruded from the rear of the CoM. No problem, I thought, since it's worked so often in the past. On the runway, I raised the flaps, went to full throttle, and never touched the flightstick. As soon as it drove off the end of the runway, it pulled itself into a near vertical climb, corkscrewing on the way up.

I reverted, and rebalanced the fuel, so that the CoM was about three CoM ball diameters (if that makes sense) in front of the CoL. In the past, such a configuration would have required quite a bit of pitch up control to maintain level flight. Only now, it required just the slightest stick pressure aft to do the same thing. Naturally, the handling was vastly improved, and I could fly this 90-ton beast where I pleased, including into orbit, and back down to the runway again!

The SAS still caused the ship to bob for apples, but it wasn't quite as as severe, and was entirely fixed thereafter, with Regex's PID tuner.

I'm off to test the newest build!

Edited October 13, 2014 by Voculus
It is - a mystery...

[Wanderfound]

There, try that build.

My wings just got a lot more fragile.

[Voculus]

Confirmed on the more fragile wings. Not that I blew mine up or anything.

I repeated the same test I described earlier, and this time, the plane did not pull up when it shouldn't have. Somehow, the CoL indicator accuracy has been improved, intentional or not. It still feels a bit off, but not to the degree of the previous build.

For reference:

In the last build, this thing spiraled upward out of control, when it left the runway. In the most recent build, it flew level for a bit, and slowly descended toward the water. Both tests were with no control input from me, and no SAS active.

Edited October 13, 2014 by Voculus

[blowfish]

@Voculus: You might want to check how the COL changes at different angles of attack.

[Voculus]

@Voculus: You might want to check how the COL changes at different angles of attack.

I knew that was coming. The point, is that there's a night and day difference between the last two builds, with respect to how the plane flew. In the configuration shown above, it was nearly uncontrollable before, and now, is a pleasure to fly, just as it was several builds ago.

One other thing, just so folks know, all of my ranting and raving is not based solely on this ship. The problems I was having were manifest in several planes, all with different wing configurations, mass, TWR, CoL/CoM relationships, etc. It just happened to be this one where the anomaly was found.

Edited October 13, 2014 by Voculus