Board index FlightGear Development Spaceflight

Space Shuttle - Bugfixes

Discussion about development and usage of spacecraft

Re: Space Shuttle - Bugfixes

Postby Thorsten » Fri Mar 05, 2021 3:53 pm

Just the target display is showing wrong data’s if not reloaded after dump but the burn is nominal.


Okay - I don't quite see what we can do about that then...

I think I managed to fix some oddities in the manual MECO (at least my test today worked fine) and in the BFS->PASS transition with simplified avionics. So I'll continue my test mission in polar orbit and see what else I can dig up...
Thorsten
 
Posts: 12048
Joined: Mon Nov 02, 2009 8:33 am

Re: Space Shuttle - Bugfixes

Postby GinGin » Fri Mar 05, 2021 7:14 pm

I don't quite see what we can do about that then


Indeed, I had another run with the earlysts.xml and autoloaded OMS 1 works nicely despite that thing, so nothing to worry about.


I made some tests for the No Yaw Jet entry.
Outcome is quite the same for the Stable and Dev branch.

Before engaging the No Yaw Jet, I take Roll/Yaw in CSS and center the controls ( real procedure)
Activation above 21000 ft/s and qbar of 40, the beta is kept within an acceptable range and a no Yaw jet entry is manageable.
Below 17kfts up to 11 kftsish, it is the hard spot where it is really beta unstable and where I have almost all the time a loss of control.
Below 11 kfs, it becomes much more stable again yaw wise.

I will be out of the computer next week, but I will dig it in more details after if you want.
I need to do anyway several entries to test the Alpha modulation new pattern before pushing it.
GinGin
 
Posts: 1400
Joined: Wed Jul 05, 2017 10:41 am
Location: Paris
Callsign: Gingin

Re: Space Shuttle - Bugfixes

Postby Thorsten » Sat Mar 06, 2021 6:58 am

Below 17kfts up to 11 kftsish, it is the hard spot where it is really beta unstable and where I have almost all the time a loss of control.


Okay, so then our suspected scenario is that prior to engaging NO Y JET in that region beta is somewhat too large and while the DAP could hold small beta, it can't properly acquire small beta - so we need to help it a bit I guess. Either by tuning the coefficients more agressively, or by coding a special beta-acquire mode.

Thanks, I guess I'll have to give this one a look myself, I want to understand this in more detail.
Thorsten
 
Posts: 12048
Joined: Mon Nov 02, 2009 8:33 am

Re: Space Shuttle - Bugfixes

Postby Thorsten » Wed Mar 10, 2021 12:22 pm

Situation - I was coming from the south pole, flying nearly due north towards Hawaii as landing site. Initially DAz was 3 deg, and during the first phase it increased to 18 deg, then I had a roll reversal.

Now, the next time I checked, DAz was -345 deg and de-creasing - the AP steered me away from Hawaii, trying for a full circle to approach again. I initially assumed this was merely a formating issue on the PFD, but when I had the time to check on the trajectory map, I was actually going massively off course, so it turned out to be a genuine guidance issue.

Well, by the time I understood it all, it was of course too late to reach the site. :( So I aborted.

It would seem we have an unnormalized angle problem somewhere crossing the 0 deg line which we have to find.
Thorsten
 
Posts: 12048
Joined: Mon Nov 02, 2009 8:33 am

Re: Space Shuttle - Bugfixes

Postby GinGin » Wed Mar 10, 2021 2:19 pm

Interesting.
I never experienced that with Entry but last week when playing with TAEM , I had one time that delta azimuth issue.
Something to dig there Indeed
GinGin
 
Posts: 1400
Joined: Wed Jul 05, 2017 10:41 am
Location: Paris
Callsign: Gingin

Re: Space Shuttle - Bugfixes

Postby Thorsten » Wed Mar 10, 2021 2:42 pm

Okay, straightforward enough - the code just subtracts - which mostly works since we rarely fly polar orbits. So assuming I have a 2 deg course over ground, i.e. pretty north-bound, and the bearing to the site is slightly to the left, i.e. 355 deg, then by subtracting we indeed find a Delta Az of -353 deg - to which we'd like to add 360 deg to find out that it's actually only 7 deg off.

This should now be fixed...

Probably there's similar issues lurking in the TAEM logic, only less apparent because while the corridor of problem during entry is 2* 18 degrees wide, during TAEM we rarely fly such large deviations from the desired course.
Thorsten
 
Posts: 12048
Joined: Mon Nov 02, 2009 8:33 am

Re: Space Shuttle - Bugfixes

Postby GinGin » Wed Mar 10, 2021 2:56 pm

Ok good.
I had that in Taem while testing guidance resilience with very late change from ovhd to stin / mep to nep after sturn and large delta azimuth ( coherent with what you said )
GinGin
 
Posts: 1400
Joined: Wed Jul 05, 2017 10:41 am
Location: Paris
Callsign: Gingin

Re: Space Shuttle - Bugfixes

Postby Thorsten » Thu Mar 11, 2021 1:26 pm

Some fixes:

* the angular issue in both entry guidance and TAEM guidance (that actually was fixed in two places, just one was omitted) should be addressed - awaits testing, but it seems straightforward

* the freon controller now is of PD type and less oscillatory - it still seems to trigger the alarm once due to startup transients, maybe one can set the initial temperatures differently, but afterwards it seems to manage temperatures nice and stable (basically I forced the valve to cool the AVBay too much and then observed how it managed to get back and adjusted the D-coefficient to make that smooth). Also awaits a bit more long term testing.

* the save/load feature now includes new switches - that kinda breaks old saved states, so if you want to resume things you have saved, don't update for now...
Thorsten
 
Posts: 12048
Joined: Mon Nov 02, 2009 8:33 am

Re: Space Shuttle - Bugfixes

Postby GinGin » Thu Mar 11, 2021 3:14 pm

Nice bug fixes .
I will run the Taem I did last week to see for the delta az .
GinGin
 
Posts: 1400
Joined: Wed Jul 05, 2017 10:41 am
Location: Paris
Callsign: Gingin

Re: Space Shuttle - Bugfixes

Postby Thorsten » Fri Mar 12, 2021 1:39 pm

I've tried to record some data with regard to the NO Y JET issue.

So it seems that till around Mach 24 we find beta < 0.01, then down to Mach 20 beta < 0.02, at Mach 17 I recorded beta as large as 0.05, that persisted down to Mach 13 where I saw beta 0.06, then dropping rapidly to < 0.02 again.

Which would explain why we have a problematic region in which engaging NO Y JET is a problem. On the other hand, the NO Y JET mode seems to be managing beta as large as 0.3 fine to generate the roll moment.

I still didn't succeed though, engaging it late led to an oscillation with ever-growing beta amplitude and loss of control. engaging early at around Mach 25.5 led to decent control. but to loss of vehicle due to TPS failure. Given that I've tested it a handful of times on old milestones and never saw a failure, I don't think this has been a pre-existing issue, all points to the FDM changes having influenced the failure...

But basically that'd mean that the actual failure mode is not so much a problem in containing beta but a transient that leads to an oscillation, and it's the failure to dampen the oscillation that leads to disaster (the TPS failure on the other hand is a higher-level issue - the entry guidance commanded a maneuver that the NO Y JET roll simply can't deliver....)
Thorsten
 
Posts: 12048
Joined: Mon Nov 02, 2009 8:33 am

Re: Space Shuttle - Bugfixes

Postby GinGin » Fri Mar 12, 2021 6:41 pm

For the TPS failure due to AP roll authority , normally no yaw jet entry is flown in Roll/ Yaw in CSS
I have time next week to test if you want to have some feedback on upcoming changes .
GinGin
 
Posts: 1400
Joined: Wed Jul 05, 2017 10:41 am
Location: Paris
Callsign: Gingin

Re: Space Shuttle - Bugfixes

Postby Thorsten » Sat Mar 13, 2021 5:47 pm

Meh - making no progress with this. It doesn't seem 'simple' in any way - probably the whole scheme has to be re-tuned :evil:
Thorsten
 
Posts: 12048
Joined: Mon Nov 02, 2009 8:33 am

Re: Space Shuttle - Bugfixes

Postby GinGin » Sat Mar 13, 2021 11:01 pm

Alright , tough aerodynamic :/

Good news :the workaround you made for Delta Az works well, I tested again exotic TAEM scenarios were I encountered weird Delta Az and it is ok now.
GinGin
 
Posts: 1400
Joined: Wed Jul 05, 2017 10:41 am
Location: Paris
Callsign: Gingin

Re: Space Shuttle - Bugfixes

Postby Thorsten » Sun Mar 14, 2021 3:21 pm

So, I'm finally making some progress. I think the problem is that the original design of the no Y jet mode was flawed.

The core issue is that 'roll' has two meanings: First, we want to have a target roll rate which we aim to reach by the beta-induced roll moment. So we compare current roll with desired roll to decide what to do. But second, we command roll to set a beta-dot to reach a certain beta value. While we do that, the roll rate is obviously very different from what we target in the longer run.

So the second roll may not feed substantially into the decisionmaking process for the first kind of roll or this ends up as a fight re-setting the beta targets all the time and oscillating around.

The original design has 'solved' this by introducing enough slack for the processes to roughly decouple through mere inertia. Apparently that's not good enough to continue to work.

So now I'm trying to separate the two roll meanings more explicitly - the beta targeting code no longer refers to roll rate at all but directly couples beta-dot with the roll command. At the moment I seem to be able control the Shuttle fine by manually hacking beta-targets into the property tree to generate suitable roll acceleration - as long as the roll rate is kept small enough, this works - if it rolls too fast it becomes unstable though.

So the part that's missing is to tell the 'stick input to roll rate' controller when it is meaningful to compare with the current roll rate and set a beta target and when this makes no sense because we're in the process of managing beta and temporarily roll against the intended direction. Probably some kind of sample-hold might work here, I'm not sure yet.

Might still be some time till this is thoroughly tested, it's one of the most tricky systems of the Shuttle.

***

But the good news is that I think this is the main roadblock to a stable state.

To recap - my current 'To Fix' list is:

* add fuel cell efficiency to saved state
* fix the NO Y JET DAP
* fix the low energy entry scenario
* let the selection of a payload write the attachment coordinates to the notepad

and on the 'To be fixed' list is

* font issue that floods the console with errors
* some dust for the lakebed landing

Of course if people keep checking in new features, these also have to be tested... On my 'to test' schedule is

* single engine TAL with droop logic test
* contingency abort with bailout
* at least one test of the new entry/TAEM fixes
* COAS attitude reference with star observations
Thorsten
 
Posts: 12048
Joined: Mon Nov 02, 2009 8:33 am

Re: Space Shuttle - Bugfixes

Postby GinGin » Sun Mar 14, 2021 4:32 pm

I'm finally making some progress


Nice, it seems indeed to be quite complex by reading your findings.Really interesting part of the Shuttle fdm.

font issue that floods the console with errors


Ah yes, I will be on it next week.

single engine TAL with droop logic test


I made a couple of them, it was good fun :)


On my side:

1)I finished with all the display minor things I wanted to implement/tune.

2)I am finishing to test two fixes:Entry alpha modulation and JSB sim tables for a better pitch schedule linear interpolation for first Stage ascent instead of nasal conditionnals with aux flag in the autolaunch.nas
I will push that tonight.

3)Square bracket into SSU font

4) Distance to Wp1 during last part of Entry instead of distance to landing site to have a better accuracy at TAEM transition (almost done)
GinGin
 
Posts: 1400
Joined: Wed Jul 05, 2017 10:41 am
Location: Paris
Callsign: Gingin

PreviousNext

Return to Spaceflight

Who is online

Users browsing this forum: No registered users and 1 guest