4.232 & PMDG feedback

[cshark172]

Pete,

I looks like Cloud turbulence is causing the heading issue. I did some quick testing, flying out of San Francisco with the latest in the PMDG 747x, using ASX weather from today at 1900Z. I took off and dialed in various headings to circle the field up to about 12000' then climbed on course to FL380 to CYVR. Wind smoothing is at 3 secs per deg/knott, temperature at 50/pressure at 20. Here is a summary of my analysis (reloaded flight and weather each time):

Left all boxes unchecked (including the new variance) - Heading wandered aimlessly above about 3000'.

Checked supress variance only from wind tab - no change

Checked variance and cloud turbulence - Heading held

Unchecked variance - Heading still held

I then loaded up Weatherset2 and noticed the following for clouds:

layer 0 - 2100' - 2598' - Turbulence 2/Icing 1

layer 1 - 3301' - 7300' - Turbulence 2/Icing 1

layer 2 - 27100-27300 - No turb

What is interesting here is I had cloud turbulence at 10% and icing off in ASX. I wonder if there is something in the way ASX generates the low level weather that is causing this. My heading hold issues have always been at the lower flight levels.

The winds on climb were very smooth. I had some IAS issues between 10k and 12k feet, but wind turbulence was at 3, so that was expected. As a reference, here are the wind layers I passed through (some significant speed changes in there).

2989 185/9

5991 238/25

8990 260/23

11991 276/21 - Turb was 3 here

17898 304/49

23990 304/65

24006 308/68

29990 300/69

33990 320/96

38990 322/98

I am about 45 minutes into the cruise now, and have not seen any rapid TAT changes or violent wind shifts.

Things are looking very good with this version. Really appreciate all the hard work you are putting into this. Microsoft should give you a cut of the profits :roll:

Kyle

[Pete Dowson]

I looks like Cloud turbulence is causing the heading issue.

FSUIPC emulates turbulence identically no matter whether it is from clouds or wind inputs. There's no difference, it's the exact same piece of code.

Left all boxes unchecked (including the new variance) - Heading wandered aimlessly above about 3000'.

Checked supress variance only from wind tab - no change

Checked variance and cloud turbulence - Heading held

Unchecked variance - Heading still held

With turbulence suppressed the variance suppression is doing nothing unless the winds are variable (i.e. non-zero variance in WeatherSet2), set in METARs by nnnVnnn or just "VRBL" for light variable winds.

So, basically, what you've found is that the PMDG autopilot somehow reacts badly to even small flickering of the windspeed (even without direction changes). Weird. PMDG said they'd come back to me about this. I hope it can be explained.

I then loaded up Weatherset2 and noticed the following for clouds:

layer 0 - 2100' - 2598' - Turbulence 2/Icing 1

layer 1 - 3301' - 7300' - Turbulence 2/Icing 1

layer 2 - 27100-27300 - No turb

With turbulence level 2 (moderate) you'd get wind speed variations randomly about the current setting (or smoothing changing setting) of +/-2 knots, with the new default TurbulenceRate of 1.0. It is very hard to understand why such changes can affect the A/P so badly.

What is interesting here is I had cloud turbulence at 10% and icing off in ASX.

You haven't got the random features enabled in FSUIPC I take it?

The winds on climb were very smooth. I had some IAS issues between 10k and 12k feet, but wind turbulence was at 3, so that was expected.

And didn't the A/P go wrong then too? For a level 3 turbulence the wind speed would be changing +/- 3 knots, that's up to 6knot changes in all, That should have been worse than the ones you got in cloud.

Regards

Pete

[cshark172]

I just reflew this 2 more times and the only box that is having any effect on the Heading bug is Cloud turbulence. Each time, I have left all the boxes in wind page unchecked. It seems wind turbulence has no effect, since I flew through the layer with Wind TURB 3 and there was no problem with heading hold.

Also, I do not have random icing set, so ASX is still setting it with the slider set to 0.

[Pete Dowson]

I just reflew this 2 more times and the only box that is having any effect on the Heading bug is Cloud turbulence. Each time, I have left all the boxes in wind page unchecked. It seems wind turbulence has no effect, since I flew through the layer with Wind TURB 3 and there was no problem with heading hold.

That is really weird, because there's only one piece of code in FSUIPC doing turbulence, and both the cloud and wind settings use it.

Maybe FSX itself is doing something different with cloud turbulence which interacts badly with what FSUIPC is doing, and it doesn't do the same for wind turbulence. I am mightily puzzled!

Pete

[Pete Dowson]

Maybe FSX itself is doing something different with cloud turbulence which interacts badly with what FSUIPC is doing, and it doesn't do the same for wind turbulence. I am mightily puzzled!

As an experiment, I've made some changes to the way the turbulence is handled. Please see if version 4.233, now available in the FSX Downloads announcement, is better.

Regards

Pete

[cshark172]

Hi Pete,

Sorry, but I think I may have steered you in the wrong direction wth Cloud vs Wind Turbulence and the PMDG 747. I tried the new 4.233 this morning, and still had heading hold issues with all suppressions off. I then looked at my screen shots of WeatherSet2 from the other day, and noticed there was no turbulence in the wind layers. I then had the bright idea to globally set weather in ASX for different situations of turbulence. Level at 15000' I cleared all clouds and set wind turbulence to Severe in the 12000-18000 wind layer. The heading tried to maintain, but struggled mightily, banking left/right and missing the mark by up to 30 degrees. After a while of this oscillation, the plane would overbank and become uncontrollable under Autopilot. Then I cleared wind turbulence and set a cloud layer with turbulence Moderate--the plane started to wander, and struggled as with wind turbulence. With all turbulence suppressed, all is fine.

The bottom line is the wind/pressure/temperature smoothing appear to be working great! If the PMDG directional issues cannot be figured out, I would not be too concerned, as just using the suppression will fix it. It looks like the other PMDG users are reporting similar results as well.

Kyle

[Pete Dowson]

I cleared all clouds and set wind turbulence to Severe in the 12000-18000 wind layer. The heading tried to maintain, but struggled mightily, banking left/right and missing the mark by up to 30 degrees. After a while of this oscillation, the plane would overbank and become uncontrollable under Autopilot.

I suspect that's pretty realistic. In severely turbulent conditions I'm pretty sure the pilot has to take control -- or if it looks like continuing long, seeking avoidance by climbing or descending (with permission, of course).

The bottom line is the wind/pressure/temperature smoothing appear to be working great! If the PMDG directional issues cannot be figured out, I would not be too concerned, as just using the suppression will fix it. It looks like the other PMDG users are reporting similar results as well.

Yes, okay, but I may just have fixed it in 4.233, so please give that a whirl for a while first -- with all suppressions off please, and preferably avoiding "Severe" weather for now! ;-)

Regards

Pete

[cshark172]

OK! Just completed climb out of Seattle with ASX live weather (2000Z). All suppressions turned off and had no problem with directional control. WeatherSet2 was still reporting Turb 3 for cloud layer between 2700-6699. The METAR was reporting surface winds 170/3 with bkn clouds 2600 and 3100 and OVC 8000.

The autopilot held heading with some variations left/right about 5-10 degrees. The rest of the climb up to FL390 was very smooth. Wind layers all had 0 turb.

Cruising along to DFW now. Something odd just happened. IAS is jumping up/down about 30 KTS! I looked at weather set and the wind layers are 112/33 at 38996, and 322/34 at 39006. The wind vector is swinging back and forth. It seems I am caught between those layers--cruising at 390. The reported winds are 309/23 at 340 and 304/35 at 390. Not sure where that 112 degrees is coming from?

Probably won't finish the flight as I am going to a Superbowl party. You know us Yanks and American Football :lol:

Thanks for all your hard work.

Kyle

[Pete Dowson]

OK! Just completed climb out of Seattle with ASX live weather (2000Z). All suppressions turned off and had no problem with directional control. WeatherSet2 was still reporting Turb 3 for cloud layer between 2700-6699. The METAR was reporting surface winds 170/3 with bkn clouds 2600 and 3100 and OVC 8000.

That's excellent news! Thanks!

Something odd just happened. IAS is jumping up/down about 30 KTS! I looked at weather set and the wind layers are 112/33 at 38996, and 322/34 at 39006. The wind vector is swinging back and forth. It seems I am caught between those layers--cruising at 390. The reported winds are 309/23 at 340 and 304/35 at 390. Not sure where that 112 degrees is coming from?

Such close layering is the result of the weather interpolation bug which is causing all these problems in the first place. I have got code built in which, when it reads station weather with spurious layers, tries to remove them and writes the corrected stuff back. But to avoid a performance hit it doesn't do this too often, and it has to cycle around all of the surrounding WX stations before it will "correct" the weather at your position -- by which time you might have moved into another set.

When I added the wind smoothing I reduced the frequency with which i do these corrections. Maybe I should restore it, but I am always concerned with performance. Additionally, it may be that the gap between those layers (10 feet) escaped my net. I don't think so, but I'll check...

... Ah, yes. It does slip through. I remove layers up to 100 METRES thick if their windspeed and direction are reasonably close, but currently I only remove layers less than 2 metres thick for such different winds as yours. I think I made this decision based on observation of lots of examples which had such thin spurious layers. Evidently I didn't review enough, or maybe it's a result of the way ASX is setting things.

I'll increase the any-wind removal check to catch layers of 20 metres or less. I think that should do it. I'll do the same with the temperature layers which suffer similarly.

What puzzles me, though, is why the winds produced by this silly thin layer weren't smoothed by my wind smoothing? Before I remove the very thin layers I shall need to do some experiments I think.

I'm afraid I'm out most of tomorrow (Monday), so the next improved version will probably be Tuesday.

Regards

Pete

[Pete Dowson]

I'm afraid I'm out most of tomorrow (Monday), so the next improved version will probably be Tuesday.

I got back earlier than I expected, so FSUIPC 4.234 is now up and waiting for you. The main change is that it tries much harder to eliminate the silly spurious thin wind and temperature layers.

Regards

Pete

[cshark172]

I tried 4.234 with my KSEA-KDFW flight using the weather from yesterday where the spurious wind layer got me. On climb out, I had the heading hold issue again. WeatherSet2 showed Turb2 in the lower cloud layers. Wind was a full 90 degree crosswind at 18 knotts varying +/10. The plane overshot the heading and never recovered. Yesterday must have been sheer luck. I turned off cloud turbulence and got back on track. At 39000 cruise, I again had the +/- 30 kts IAS issue. Here are the wind layers from WeatherSet2 when that happened:

38553 - 314 degrees @ 27 kts

38999 - 68 degrees @ 34 kts

43553 - 309 degrees @ 34 kts

Reported winds were 304/34 at 34000. I could not hold speed, so I descended to 37000 and all smoothed out for a short while. A little later, I again had the IAS problems. Now, the wind layer up to 38553 was holding speed but varying from 354 to 111 to 11 knots. Reported was 283/27. This was happening around the BKE VOR SW of Seattle using ASX weather from 3 Feb at 21Z.

I did capture the following screenshot from WeaterSet that is showing some wierd informaion in the temperature layers. These -65513 values would keep going for a while and then clear up and come back.

[cshark172]

I tried again with live ASX weather for this evening (02Z) for the same route KSEA-KDFW. I still had cloud turbulence suppressed, so can't comment on heading control. Climb to FL390 was very smooth. Cruise started out OK, then at about the same point--about 45 minutes in, IAS went crazy again (actually increased to Mach 1.01 in the 747!) I see the same strange direction change in the winds for the layer I am flying in. Before the IAS problems, WeatherSet2 showed consistent directions and speeds. I have attached the WeatherSet2 screenshot (notice the direction at the 43,553 layer). Reported wind at 39000 was 354/72. I also had the strange numbers in the Temperature layers. -131 thousand--can't remember the second half of the number. It seems that the winds are smoothed very well for climb and descent--I just have these problems in the cruise.

[Pete Dowson]

I tried 4.234 with my KSEA-KDFW flight using the weather from yesterday where the spurious wind layer got me.

I hope you've still got that. Looks like I'm going to need some logging done to get to the bottom of this.

On climb out, I had the heading hold issue again. WeatherSet2 showed Turb2 in the lower cloud layers. Wind was a full 90 degree crosswind at 18 knotts varying +/10. The plane overshot the heading and never recovered.

That's weird. That's a trivial thing for the A/P to deal with.

PMDG have offered to look at this from their end for me, if I can provide a reproducible setup. Do you think that is possible? If you save the flight just when it starts to go wrong, will it do the same when you reload that flight (without ASX or whatever running)? If so maybe you could Zip up the FLT + WX + FSSave files and send them to me (petedowson@btconnect.com)?

At 39000 cruise, I again had the +/- 30 kts IAS issue. Here are the wind layers from WeatherSet2 when that happened:

38553 - 314 degrees @ 27 kts

38999 - 68 degrees @ 34 kts

Groan! I increase the "thin layer" removal from 2 metres to 20 metres, and now we find that we get 500' layers also causing problems! At that rate I shall remove all layers! (Actually, that's the way I did wind smoothing back in FS2000 days -- removed all layers and controlled the one layer left. Maybe that's the only way ...:-( ).

I need to understand why the smoothing can't cope with this. If you think this is reproducible at 39000, could you do another test with the same weather, please? First edit the FSUIPC4.INI file, adding

Debug=Please

to the [General] section.

In FSX, before you get to the 39000 cruise, say 1000 feet below, sace a Flight, then go to FSUIPC's Logging tab and enable Weather logging AND enter the number '65' in the box for "Extras", below. This sets special debugging logging data on. Carry on the flight with the worsening directional fluctuation until it settles or you leave the level, then go back to the Logging and turn both off again (Extras to 0).

Zip up the Log plus the saved FLT, WX, FSSave files and send to petedowson@btconnect.com, please.

I did capture the following screenshot from WeaterSet that is showing some wierd information in the temperature layers. These -65513 values would keep going for a while and then clear up and come back.

Ouch! That is nasty. That seems to indicate really bad corruption somewhere in the weather stuff. If you think you can repro this could you apply that Logging above to this phase too, please? Maybe you could save a Flight at the time? Zip up the LOG, FLT + WX + FSSave file and send them to me?

Thanks,

Regards

Pete

[cshark172]

When I get home from work this evening (I'm in US Central Time), I'll refly the scenario and get you some logs and saves. The Cruise issue seems to be pretty consistent. I have had it happen on 3 different flights now from 3 different airports (KDEN, KSEA, and KSFO). As far as the low level issues with heading hold, ASX must be randomizing the lower level turbulence, because the cloud layers are not always showing the same for the same archived weather. I will try a couple of takeoffs and see if I can get one to come up. One thing I did notice, it seems to be more prevalent when there are overlapping layers with turbulence.

One thing I was thinking today, I fly with frames unlimited, and maybe the weather engine is having trouble keeping up? I am getting 50-60 fps at altitude. I will also try some flights frame rate locked to see what happens.

A couple other things of my setup. I run ASX over the network and have been using wideclient to grab the WeatherSet2 informtion. FS box is running Vista 64 bit and client box is XP.

Kyle

[Pete Dowson]

When I get home from work this evening (I'm in US Central Time), I'll refly the scenario and get you some logs and saves.[/quotes]

Thanks. I'm going to try and set some similar tests up here in the meantime.

One thing I was thinking today, I fly with frames unlimited, and maybe the weather engine is having trouble keeping up? I am getting 50-60 fps at altitude. I will also try some flights frame rate locked to see what happens.

I wouldn't have thought that to be a problem. If FSX can generate the frames that fast it can certainly process the weather at the same rate. With 50-60 fps though those FSUIPC logs are going to be really big as some of the smoothing log entries are being done several times per frame. Maybe the act of logging will noticeably slow FSX somewhat too.

A couple other things of my setup. I run ASX over the network and have been using wideclient to grab the WeatherSet2 informtion. FS box is running Vista 64 bit and client box is XP.

Sounds good. I am getting a new FSX box built and am going for Vista 64 also. I'll only have FSX + outside scenery views on that -- everything else, panels included, are on Networked XP boxes.

Regards

Pete

[Pete Dowson]

I tried to reproduce the wind direction swings without success. But in the process i spotted a couple of areas where I could improve things somewhat.

So, please move on to version 4.235 for further tests.

This is becoming a daily thing, isn't it? It's certainly a lot of work!

Regards

Pete

[cshark172]

Hello Pete,

I was able to duplicate all 3 of the issues I have been having this evening. I just sent you the saves and fsuipc.log files for each situation. Keep a lookout for email from Kyle Broadbent. Hopefully they will get to you OK. The .wx and log files can get pretty large. Hope these help you figure out what is going on.

Thanks for all your hard work,

Kyle

[Pete Dowson]

I was able to duplicate all 3 of the issues I have been having this evening. I just sent you the saves and fsuipc.log files for each situation.

Thanks! Got them.

Regarding your covering message:

“Wandering Heading” is a save when the heading lost lock and started to wander. This is repeatable with ASX turned off.

But even with ASX off, would the weather set in FSX still have been from ASX? I am thinking that the weather patterns in which this happens are peculiar to the way ASX is generating them.

I have also included the panel state files from PMDG in case someone needs them to fly the PMDG aircraft.

Thanks! Yes, that should be useful. If I can see the problem here with these files I'll shoot them over to PMDG for their opinion.

“Negative Temperature Layers” has a save and FSUIPC log when I was getting the large negative numbers in the temperature layer of WeatherSet2.

I can't see why in the log. I'll try to reproduce it. Else I'll need more logging -- more below on this.

I ran this for about 12 minutes waiting for things to settle down, and I got a fatal error shown below. I hope this has nothing to do with FSUIPC. I did get this fatal error twice tonight while testing and both times were when the wind was acting up. I see some of the others (jordanal and Bill Casey) testing these updates had CTD recently as well.

I think it must be down to one of the changes I made in 4.235. I really can't believe it is mere coincidence. I'll do a thorough check today and upload 4.236 this evening. Please look out for it.

And could you change the "Extras" logging number to 67 instead of 65 please? In fact place these lines into the [General] section of the INI file:

LogSimC=2DC8-2DE8,3470-3490,0580,2DF8

Debug=Please

LogExtras=67 (i.e. change 65 to 67)

LogWeather=Yes (this should be there already).

I think I'll put a note to this effect in the Downloads announcement. I missed asking all in the 5 or so threads to do more logging. The logs will get huge, but it'll give me the data I need in case I can't repro these events.

It would be interesting to see what the event log shows for them. I had a very large number of the negative wind layers showing up in this one as well. Hope this information helps.

Not sure how I get to an "event log". This information you provided:

Problem signature:
  Problem Event Name:                        APPCRASH
  Application Name:                             fsx.exe
  Application Version:                           10.0.61637.0
  Application Timestamp:                     46fadb14
  Fault Module Name:                          ntdll.dll
  Fault Module Version:                        6.0.6000.16386
  Fault Module Timestamp:                  4549bdf8
  Exception Code:                                  c0000005
  Exception Offset:                                0006089e
  OS Version:                                          6.0.6000.2.0.0.256.1
  Locale ID:                                             1033

simply says it was an access protection error, way down is a system module (ntdll.dll). I'll need to repro it here to get the stack so I can see who called what to get there.

Thanks. I'll be back later.

Regards

Pete

[cshark172]

But even with ASX off, would the weather set in FSX still have been from ASX? I am thinking that the weather patterns in which this happens are peculiar to the way ASX is generating them.

Yes I guess the weather that is in the saved flight is from the last time ASX wrote the information to FSX.

And could you change the "Extras" logging number to 67 instead of 65 please? In fact place these lines into the [General] section of the INI file:

LogSimC=2DC8-2DE8,3470-3490,0580,2DF8

Debug=Please

LogExtras=67 (i.e. change 65 to 67)

LogWeather=Yes (this should be there already).

I think I'll put a note to this effect in the Downloads announcement. I missed asking all in the 5 or so threads to do more logging. The logs will get huge, but it'll give me the data I need in case I can't repro these events.

Will do if I get a chance this evening. Might be a bit short on time tonight.

Not sure how I get to an "event log".

Just the standard windows application log contains the information. Control Panel -> Administrative Tools -> Event Viewer on XP. Vista also but there is another layer below event viewer to get to the Application Log. I am not at a Vista box now, so can't remember wat it is. Even if you get a drop to desktop with no error message, the application log usually contains some info, though not very informative (usually points to unkonwn module ver 0.0.0.0). I used to get this on FS9 on XP frequently. Vista seems to have better error handling. The few crashes I have had all produced an error box with the information I provided. I have had about 4 crashes this week while testing, all pointing to the ntdll.dll module.

[Pete Dowson]

Will do if I get a chance this evening. Might be a bit short on time tonight.

Me too, there's an England friendly on (first one with our new italian manager).

Anyway, here's the progress:

1. I've definitely found and fixed the problem which was causing the weird temperature layers listed with just a big negative number. It was a silly typo (a missing '(') in a mod for temperature and wind layers I did for 4.234.

2. I think (pretty sure) that this same problem would have been the cause of the spate of FSX crashes. There's a sort of 50:50 chance whether you get the wrongly numbered layers first or the crash.

3. I can repro problems with the PMDG heading hold during turbulence using the files you sent. After the short start of this scenario, the aircraft climbs out of the turbulent patch and all is well. So to test more I descended again, to 11,000 and also went into the Weather menu, Advanced, to make sure we had cloud there with 'moderate' turbulence. The winds were light, around 7-8 knots. Doing this you can see the A/P first making small adjustments to try to stay on track, but its adjustments actually aren't necessary, and instead of getting smaller, the drift gradually gets bigger -- in this case with the light winds it gets as far as 30+ degrees out. It gets into some sort of unstable state. The wind arrow on the ND is fluttering only very slightly. Maybe the A/P is reacting too strongly to small wind changes, and maybe too fast -- possibly it should average over a few frames.

Anyway, the logging in FSUIPC shows that there's nothing I'm doing which is wrong here, but I have enough now to ask PMDG to take a look. Maybe it's something they can fix in the A/P or advise me of a way to avoid it. If neither then we shall probably have to just tell folks to suppress turbulence when flying PMDG.

I'm sending your files with my notes to PMDG today.

=========================

That just leaves the "FL390" problem where you were seeing violent changes of wind direction. I still don't understand that one, and I can't get anything to go wrong here with what you supplied. Maybe it only happens with ASX still connected and manipulating weather?

Anyway, I'll will upload FSUIPC 4.236 soon, so for any further tests please use that. Maybe the "FL390" problem will show up in the fuller logging, when you do get a chance.

Best Regards

Pete

[cshark172]

Hi Pete,

I did have a chance to test the flight again this evening. I did get the wind shift at FL390. Massive log file and autosave just before the shift is on the way. Good things to report are no more negative wind layers, and I flew the route once with no crash (I was not logging then). Unfortunately, I did get a crash the second time I flew with the logging on, but as I mention in my email, there was a great deal of activity at that instant (autosave, weather update, and logging). Hopefully there is some information in the log file to help out.

Kyle

[Pete Dowson]

I did have a chance to test the flight again this evening. I did get the wind shift at FL390. Massive log file and autosave just before the shift is on the way.

OkayI see it in the Log, but I cannot reproduce it. I can only think it is related not to static weather setups in saved WX files, but dynamic weather -- were you getting weather from ASX or FSX downloads?

Anyway, I've found I need more files in any case for the saved FLT + WX + FSSave files, when you fly PMDG. There are other saved files in the PMDG\747400\PanelState folder. They'll have the same name (eg "AutoSave ...") but types .flt.sav, .flt.0.rte, .flt.1.rte (etc). So that I can fly exactly as you did, could you include those too next time, please?

Unfortunately, I did get a crash the second time I flew with the logging on, but as I mention in my email, there was a great deal of activity at that instant (autosave, weather update, and logging). Hopefully there is some information in the log file to help out.

Nothing helpful regarding any sort of crash. This is a bit worrying. I know you think it may not be FSUIPC, but I'm definitely a bit worried that I've not got enough checks in place to make sure the weather variables I'm writing are always there. I've examined it very carefully and can't see where an error could now occur, but it is still worrying.

From your email:

I don’t know if this means anything but when flying without logging, I had frame rate unlocked and when I got the anomalous layer, my wind layers in WeatherSet2 went from 12 to 23 layers. The second run when I was logging, I locked at 30fps, and when the anomaly popped up, the number of layers remained at 12. All my previous testing was unlocked and I always had the 23 layers, when things went awry.

I suspect that all that is happening is that with a locked frame rate the weather routines are given more time to screw things up!

I am testing FSUIPC 4.237 right now and will upload it soon. I've made two further changes to try to nail this:

1. I've resurrected my "wind smoothing thread", which I didn't think would be needed now that I am hooking into what I thought was the right place in WEATHER.DLL. This thread attempts to re-write the desired winds up to 200 times per second. I want to see if this stops your 39000' swings.

2. I am now automatically doubling the WX station METAR reads and corrective writes during times when the number of temperature or wind layers at the aircraft exceeds 15. I am hoping this will avoid a build up of such layers, many of which are spurious and unwanted.

THe problem with both of these changes is that they may impact performance, and I really can't have that happen, at least not noticeably. The thread one may even be a little dangerous, although again I've taken steps to prevent it accessing stuff when it shouldn't. I can't make it crash at all here.

So, please, if you have time and patience still, try that 39000' test again with 4.237.

Thanks!

Pete

[cshark172]

Hi Pete,

I have already had a chance to check out 4.237. As far as smoothing goes, outstanding. I did not have the sudden shifts at 39000 at all this flight. Using same archived weather (3 Feb/20Z), synched to simulator time from ASX, with updates every 10 minutes. I had WeatherSet2 up during the entire time, and there were a couple of minor shifts, but the smoothing responded well, and there were no sudden +/30 kt IAS changes. Within a few seconds, the winds were back where they were before the shift. I did not notice and performance issues, even with logging enabled.

I did get the ntdll.dll crash again. I have had this on every flight since last Sunday. I have the log and an autosave from about 5 minutes before the crash, if you would like to look at it. I am thinking there might be a underlying problem in my FSX installation. Two weeks ago, I had installed scenery that worked with SP1 but not SP2, and I installed/uninstalled Accelleration a couple of times--in the end I decided the SP2 updates were more beneficial than the scenery enhancement. Sometimes all that adding/removing can corrupt files somewhere in the installation.

[Pete Dowson]

I have already had a chance to check out 4.237. As far as smoothing goes, outstanding. I did not have the sudden shifts at 39000 at all this flight. Using same archived weather (3 Feb/20Z), synched to simulator time from ASX, with updates every 10 minutes. I had WeatherSet2 up during the entire time, and there were a couple of minor shifts, but the smoothing responded well, and there were no sudden +/30 kt IAS changes. Within a few seconds, the winds were back where they were before the shift. I did not notice and performance issues, even with logging enabled.

That's good, thank you. Very encouraging!

I did get the ntdll.dll crash again. I have had this on every flight since last Sunday. I have the log and an autosave from about 5 minutes before the crash, if you would like to look at it.

That is a big worry for meyes, please send the files. Please include the PMDG files as I mentioned too.

I am thinking there might be a underlying problem in my FSX installation ...

Well, maybe, but the coincidence of it happening whilst doing these tests does not sit well with me. Perhaps, if it looks like we've conquered the main smoothing problems (I'm still waiting for reports back from PMDG as to what might be the problem with the heading drift in turbulence), you should return to having wind smoothing disabled for a while. If you get no FSX crashes on several successive flights then I think you might agree it must be something to do with the smoothing? Let me know, please.

Best Regards

Pete

[cshark172]

Pete,

Files are on the way. I will try some flights with smoothing off and see if the crash goes away. Should I also turn off pressure/temperature smoothing? As a change of pace, I tried the same flight with the Level D 767 and results were similar. A couple of shifts that were manageable, and eventually crashed with the ntdll.dll exception.

Kyle