PMDG 747-400 trouble with turbulence with 4.28

[jase439]

It could be related to a recent update. I am running with their latest (2.10 build 40 or something). BTW, I should qualify my earlier - and unfair and overly broad - characterisation of the 47's "poor control logic". In all other respects, the aircraft is behaving admirably and is a fine product. I merely "suggest" it may be a flaw in the 47's control logic that is elicited or exacerbated by the new turbulence effect as other aircraft are seemingly unperturbed by it. But to qualify it as "poor" was unfair (and equally inaccurate).

Also, to your earlier question. While I have been testing this problem in the context of cloud turbulence, I don't think it's restricted to cloud-triggered turbulence. This problem first surfaced for me in seemingly clear skies (unless I was passing through some unseen ultra-sparse cloud layer). I suspect people are simply encountering random cloud-generated turbulence "more often" than its wind-triggered counterpart, hence the perception that it's only cloud turbulence.

Regards,

J

[Pete Dowson]

It could be related to a recent update. I am running with their latest (2.10 build 40 or something).

Hmmm. where do you find that info? I suspect mine is older. I don't think I have access to their downloads system at all, so I'd need to ask them.

Maybe they have changed something and never thought to re-check it in turbulence.

Pete

[jase439]

Here's the latest update info (that I know of):

http://forums.avsim.net/dcboard.php?az=9454&page=

For you, I imagine they would probably just point you to a brand new installer if you ask.

[Pete Dowson]

However, wind direction was steady here, and only wind speed varied (+/- 3 knots or so) and the aircraft went bananas. Indeed, suppressing turbulence via FSUIPC immediately corrected the problem. Unsuppressing it, the aircraft began its slalom maneuvers.

Adjusting TurbulenceRate only affected the magnitude of the wind shift. Not surprisingly this is what the documentation says it does :) Dialing it down, didn't appear to have much impact - it just took a little longer for the plane to begin rocking uncontrollably. Dialing this value up just made the aircraft roll over on its head that much sooner.

Adjusting TurbulenceDivisor had no perceptable impact on this problem that I could tell. I tried adjusting the first two divisors from 20,20 to 30, 40, 60, and 100. I couldn't tell you 100 was any better or worse than 20 with respect to this issue.

If you are interested in following this up further (please?), I have added a lot more logging to FSUIPC in an interim version 4.292. This provides logs like this:

   298687    Wind values set: 	32.5	8.2	+0.1	[32.6	8.1	+2.0]
   298734    Wind values set: 	32.8	8.0	+0.3	[32.8	8.2	+2.0]
   298797    Wind values set: 	32.5	8.2	+0.4	[31.0	8.3	+2.0]
   298812    Wind values set: 	32.2	8.3	+0.5	[31.0	8.3	+2.0]
   298875    Wind values set: 	31.9	8.2	+0.6	[31.0	8.1	+2.0]
   298937    Wind values set: 	31.6	8.3	+0.5	[31.0	8.3	+1.2]
   298953    Wind values set: 	31.3	8.4	+0.4	[31.0	8.5	+1.2]
   299000    Wind values set: 	31.6	8.3	+0.2	[31.8	8.4	+1.2]
   299062    Wind values set: 	31.9	8.4	+0.1	[32.6	8.9	+1.2]
   299062    Wind values set: 	32.2	8.6	-0.0	[32.6	8.9	+1.2]
   299094    Wind values set: 	32.5	8.7	-0.1	[32.6	8.9	+1.2]
   299156    Wind values set: 	32.2	8.8	-0.0	[32.5	8.9	+1.6]
   299172    Wind values set: 	32.5	9.0	+0.1	[33.8	9.0	+1.6]
   299219    Wind values set: 	32.8	8.8	+0.3	[33.8	8.4	+1.8]
   299281    Wind values set: 	33.1	8.7	+0.4	[33.8	8.4	+1.8]
   299297    Wind values set: 	33.4	8.6	+0.5	[33.8	8.4	+2.2]
   299344    Wind values set: 	33.7	8.4	+0.7	[33.8	8.4	+2.2]
   299406    Wind values set: 	33.4	8.6	+0.8	[32.1	8.5	+2.2]
   299406    Wind values set: 	33.1	8.4	+0.7	[32.1	8.5	+1.1]
   299453    Wind values set: 	32.8	8.6	+0.5	[32.1	8.8	+1.1]
   299500    Wind values set: 	32.5	8.7	+0.4	[32.1	8.8	+1.1]
   299516    Wind values set: 	32.2	8.8	+0.2	[32.1	8.8	+1.1]
   299562    Wind values set: 	32.5	9.0	+0.1	[33.9	9.4	+1.1]
   299625    Wind values set: 	32.8	9.1	-0.0	[33.9	9.4	+1.1]
   299625    Wind values set: 	33.1	9.3	-0.2	[33.9	9.4	+1.1]
   299687    Wind values set: 	33.4	9.4	-0.3	[33.9	9.4	+1.1]

Here the number on the left is the elapsed time in milliseconds, and the numeric columns after (which are tabbed) are: Direction (degrees T), Speed (knots), Vertical speed (knots) then in [] the three corresponding Gaussian-distributed (hopefully) targets. You'll see that the targets are recomputed whenever they are met OR when the smoothed wind changes enough to invalidate the previous target. this is done individually for the three components.

You will find that although the speed or direction can sometimes change maybe by 2 or 3 units in a second, they do so via a number of smaller increments -- maybe 20 or 30. I think that's dependent upon frame rate. Maybe you are getting such a good frame rate that I am applying the increments too quickly, but i doubt that -- however, you would be able to see from this. Most certainly the increments used ARE affected by those turbulence divisor values, so they determine how fast things change. I don't really want to have to put time constraints in this as the computations are done "in line" between a call from the Weather module in FSX to the Sim engine (SIM1).

All the results I get here are, so far, in accord with the way it was designed and follw guidelines on turbulence emulation I have stuied. they are also exactly what the expert in PMDG says would be right.

Anyway, let me know and I'll supply a link to this version and instructions on enabling the logging. Maybe I should do some statistical analysis on the data and log that too, though I'm not sure what would be useful at present.

You also will find it useful to use the FSUIPC monitor to display the actual ambient windspeed and direction on screen in real time (more accurately than the Shift+Z or ND wind indicators, but from the same values). Do this by Monitoring 2DE0 (direction) and 2DE8 (speed) both as FLT64's, and checking "FS window" (all this on the right-hand side of the FSUIPC Logging tab).

Regards

Pete

[jase439]

Sure thing. I'll give this a go and report back with my findings.

EDIT: I have the new build (I guessed the URL :^), but I don't know how to enable the logging you're showing in your previous post. Let me know and I'll give it a look.

[KeyserSoze]

Hi,

I have the same problem with PMDG 747, FSUIPC and ASX. When I get into a turbulence, my aircraft just swings to one side and the autopilot does not react on it (when I try to level it with a yoke, the autopilot disconnects and I get an error message "Extensive column *something*, I do not remember exactly). In that case I have to fly the aircraft manually as long as the turbulence occurs and when it is gone I can connect the autopilot again (and it puts the airplane back to LNAV path).

I am using:

FSX w. Acceleration Pack

PMDG 747 w. SP 2.1

FSUIPC 4.28 (wind smoothing enabled)

ASX w. SP3 (wind smoothing and stabilization disabled)

[Bill Casey]

In that case I have to fly the aircraft manually as long as the turbulence occurs and when it is gone I can connect the autopilot again (and it puts the airplane back to LNAV path).

I have now found that on the rare occasions that this uncontrollable deviation from path occurs that by dropping your next waypoint to the scratchpad then re-inserting as the next waypoint (then Execute) your a/c will recover and LNAV can be resumed. You have to be quick before the deviation has gone too far but it does work (for me anyway!).

[Pete Dowson]

Sure thing. I'll give this a go and report back with my findings.

Slight delay whilst I figure out how to change the logging so it doesn't affect the timings.

EDIT: I have the new build (I guessed the URL :^), but I don't know how to enable the logging you're showing in your previous post. Let me know and I'll give it a look.

Drat! I'll have to update the version number again then. :-)

I looked at the results I was getting and decided they were being affected too much by the sheer act of logging (in-line between WEATHER.DLL and SIM1.DLL calls), so I am working on a way for the logging to be to memory during those times, catching up after. I'll upload 4.293 when I have it working. It'll be this weekend, but maybe not today.

Please don't use 4.292 in any case. it'll drag your frame rates down a bit -- I tried changing the way I was getting the Time for the logging, to make it more accurate, and it is way too inefficient.

Regards

Pete

[Pete Dowson]

Sure thing. I'll give this a go and report back with my findings.

Okay. You need 4.293:

http://fsuipc.simflight.com/beta/FSUIPC4293.zip

Edit the FSUIPC4.INI file. Remove all Logxxxx=Yes lines (same as unchecking any logging in the Logging tab). We don't want a load of unnecessary logging spoiling the tabulation of the wind data.

Add Debug=Please

LogExtras=256

This enables those "Wind values set" lines I showed.

Some changes I made since I showed that:

1) The logging is actually done OUTSIDE the hooks into the WEATHER/SIM1 link I mentioned, so that it doesn't affect the timing.

2) When I did (1) I noticed that sometimes (about 8 times a second) the log entry was being instigated twice at virtually the same time. Upon checking I found this was because I also do the same intercept from VISUALFX.DLL to SIM1.DLL. I actually forgot I did that too -- if I miss those calls incorrect wind values creep in and mess up the smoothing (causing jitters).

So, I added a little more code to eliminate any increment when no apparent time had passed. In fact I do more than that -- if the time since the last increment is less than 40 mSecs, I reduce the increment in proportion to the time difference/40. This should prevent it changing the winds too fast during periods of high (internal) frame rates -- I'm effectively limiting the changes to 25/sec. This means with the default divisor of 40 that it takes 1.6 secs to reach each target.

I don't increase the increment for slow frame rates -- all that happens then is you get less wobbly turbulence.

3) I fixed the logging of the Vertical speed targets. The log extract I showed above had the wrong values in that column.

I've downloaded the latest update to the PMDG 747 and will try it later, maybe tomorrow. But I suspect the reason I've not seen any problems is that my frame rates have always been around 20-25 on my test machine. If the problem is due to faster changes because of higher frame rates I think my limiter, now added, should fix it.

Regards

Pete

[jase439]

Is there supposed to be a space between Add and Debug?

EDIT: Well, my world has gone from bad to worse. I am unable to get any kind of turbulence generation at all (is there any way to set the odds to 100%?) Even with situations that were saved with turbulence, I get nothing. My frame rates are in the basement. I'm not getting any of the logging you described (but maybe that's because there is no turbulence). And my panel is flickering violently. Gone back to 4.285 for now.

Incidentally, I'm thinking about switching back to my old tried-n-true FSMeteo. It still uses FSUIPC's NWI for programming weather parameters, right? I feel like FSUIPC and ASX are constantly fighting for control of the weather. Moreover, what I see in the ASX UI for the active METAR and ALOFT data doesn't correlate to what's happening in the sim (often, there appears to be no relationship between the two whatsoever). ASX has a much sexier UI, but I don't feel like I'm in control of my weather environment - not to the same degree as I did with FSMeteo/FSUIPC. I just purchased the upgrade to FSMeteo v8.0 and hoping it still uses the same tried-n-true FSUIPC wx interfaces as v6 and v7.

[Pete Dowson]

Is there supposed to be a space between Add and Debug?

Sorry, it was supposed toi say

Add:

Debug=Please

LogExtras=256

EDIT: Well, my world has gone from bad to worse. I am unable to get any kind of turbulence generation at all (is there any way to set the odds to 100%?)

I just set up the weather in the FSX weather menu. You can set a nice thick cloud layer and set the turbulence level in it, or same with winds. FS is quite flexible in that regard. Surely you are not relying on turbilence added at random by FSUIPC?

My frame rates are in the basement. I'm not getting any of the logging you described (but maybe that's because there is no turbulence). And my panel is flickering violently.

That sounds like 4.292 (the frame rates) but I've no idea what you could have their causing panels to flicker. Are you sure you downloaded 4.293? Maybe you made a mistake in the INI? Show me a Log please.

The wind value logging I put in are operative all the time if the parameters I specified are added -- as long as Wind Smoothing is enabled of course (without that none of this is relevant). The logging is not specific to turbulence, it logs all of the wind settings all of the time.

[LATER]

It occurred to me that, just possibly, your flicker could have been caused by the Monitoring of those two wind values on-screen in real time. It uses one of FSX's messaging facilities, but I have sen flicker in full screen mode with some video drivers. If this is what is doing it just don't bother with that -- I just thought it would be informative for you. Alternatively switch to a maximised Windowed mode, where I've never seen any such flicker.

So far, with all the turbulences types and modes I've tested all the figures look good, exactly as designed. I'll not get to testing the 747 update till tomorrow now.

Regards

Pete

[Pete Dowson]

I've released 4.294 in the FSX downloads announcement above. Please read the notes.

I tested this with the PMDG 747X version 2.00.031 and found that, with moderate turbulence, flying with 20-40 knot winds at 13000 feet or so, the autopilot, though wavering about, never got more than 10-20 degrees off the intended course. I think this is excessive, and you can actually see it turning the wrong way (i.e. turning left slightly even though the heading bug is to the right). I have no idea why, but no other autopilot i've seen behaves like this so I'm afraid it is a question for PMDG.

I updated to version 2.10.0040 of the PMDG 747X and this seems better, never getting more than 10 degrees off, and usually only 1-2 at most. At least, this is how it behaves once it is on course -- it seems to have a lot more difficulty actually getting TO the right course, often apparently turning the wrong way for a little while.

I cannot explain the odd A/P behaviour, but I have examined the detailed logs of the turbulent wind details, and they are always exactly as indented, with small increments and the speed and direction changing smoothly up and down over the intended periods. To tone them down any I think would make them unrealistic, though you can experiment with the Turbulence parameters, as documented.

In the end I'm afraid that, for the PMDG aircraft, if you still have problems, you have two choices -- either contact PMDG and supply them with the logged wind data and a description of what went wrong with the A/P, or simply set the FSUIPC4 options to prevent the turbulence emulation. The latter would merely make wind smoothing in FSUIPC act like it did in FS2004, so it isn't such a loss, relatively speaking.

Regards

Pete

[JSkorna]

Hi Pete,

Is this with ASX weather? I just need to know because there are a few reports of this at our place.

[Pete Dowson]

Is this with ASX weather? I just need to know because there are a few reports of this at our place.

My tests today were with weather set up with moderate turbulence via the Weather dialogues in FSX. I have previously run the PMDG 747X with ASX, but not with your latest update (SP3). It was okay (well, the same as with user-specified weather).

Scanning through the reports earlier in this thread I really find only one thing in SP3 which one user found affected things, to quote:

"are you using the new Wind stabilization (not smoothign) feature in ASX SP3? My first flight with ASX SP3, I had that enabled, and the PMDG 747x went all over the place when crossing a wind layer. The wind vector was twitching like crazy. I disable the wind satbilization feature, and have now completed 3 long flights with no problems."

However, this whole thread is strewn with inconsistencies, so I would necessarily take any of it on faith. For instance one chap is absolutely sure it is only "cloud turbulence" in FSUIPC, not "wind turbulence" which cases a problem -- yet there is only one turbulence emulation provided, so that makes no sense. The difference between cloud and wind is which weather record the "turbulence" instruction comes from, that's all.

Another inconsistency is folks absolutely sure that it was all okay in 4.26 and not in 4.28, yet I've compared the code and there is absolutely no difference -- the turbulence is identical in both. And this is borne out by examining the actual wind fluctuations. From the sounds of those particular reports it seemed as if the FSUIPC turbulence emulation was actually turned off (by having smoothing off too0 when they were using 4.26, as the type of turbulence they reported was exactly FSX's own, which cannot happen when smoothing is in operation (the whole reason for the emulation in the first place).

I have spent a lot of time examining exactly what my own turbulence emulation code does, and it is now doing exactly what was intended and what was agreed with PMDG as realistic. I've had to make one change, and that is to remove a dependency on the FSX internal frame rates to regulate the wind fluctuations. I suspect that the instability on some systems may have been due to them "running away" a bit -- and that might also vary due to other things, even different versions of ASX or FSUIPC.

I also found that, because my hooks in FSX have to operate on two routes (VisualFX to SIM1 and Weather to SIM1) there were on average about 8 occasions per second where two increments were applied at almost the same time. This could also have made too much of a "jerk" in the wind. The regulation I've included now seems to sort all that out and, as I say, the results look good.

Let's see what sort of feedback I get from folks using 4.294 (or later). I'll also be installing ASX SP3 this week and will do some more tests with that.

Best Regards

Pete

[jase439]

Thanks for the follow up, Pete. I'm glad you were able to recreate the phenomenon (at least to a limited degree). The magnitude of the A/P deviations could be a function of many things, including wind speed, rate of change, fuel and balance, or frame rates (in the case of 4.285 at least - hopefully fixed in 4.294). Moreover, the observation that this behavior is unique to the PMDG 747 A/P and only PMDG is also consistent with my own. I have not seen anything that suggests FSUIPC is doing anything but what it should be doing. This does - at the surface - appear to be a bugaboo in PMDG's control logic that is elicited by FSUIPC's turbulence modelling. I hope PMDG is following this thread, but for the time being, I'm content to use FS2K4-style behavior.

J

PS. The violent flickering was indeed caused by the FSUIPC variable display. BTW, I'm still having trouble getting FSUIPC to forcibly "generate turbulence" using only FSX weather. I've been setting a moderate turbulence layer (in both wind and cloud) and nothing happens. The turbulence effect only seems to kick in if I use an external weather program like ASX or FSMeteo in tandem with "generate random turbulence". I don't know if it's because I'm loading a saved situation file (whose WX environment was originally created by ASX) or if I'm just setting the turbulence at the wrong station (should I be setting "per-station" weather or global weather?).

[Pete Dowson]

BTW, I'm still having trouble getting FSUIPC to forcibly "generate turbulence" using only FSX weather. I've been setting a moderate turbulence layer (in both wind and cloud) and nothing happens.

I just select custom weather, advanced, set up the layers, and, there they are -- turbulence to order. I do set it for all weather stations (i.e. globally), not just a selected one.

Maybe you have FSX's weather change slider set to "fast" (or at least not "slow") and by the time you've reached the relevant layer its changed? I speed things up by slewing up to the layer I've enabled turbulence in, but it certainly hangs around for a long time.

The turbulence effect only seems to kick in if I use an external weather program like ASX or FSMeteo in tandem with "generate random turbulence".

If you have FSUIPC's random turbulence option set (and the option checked for altering FS's own weather), then this is just as likely to stop the turbulence as start it. Also, in any case, the FSUIPC random turbulence won't have any effect when using ASX in the recent versions of FSUIPC because it no longer tries to change FS's own weather when it is set to "custom", which will happen when ASX runs.

... should I be setting "per-station" weather or global weather?).

Well if you set it "per station", unless you set it for al the stations in the vicinity you may not see it -- FSX like FS2004 interpolates weather from three stations, but determining which three is not possible. The surest way is to clear all weather first, then set global weather.

Pete

[jase439]

If you have FSUIPC's random turbulence option set (and the option checked for altering FS's own weather), then this is just as likely to stop the turbulence as start it.

Ah, this is quite likely what's happening. I'll try again later with this option disabled.

EDIT: Well, despite the fact that I am still unable to produce that debug log output (even with the revised settings), I can say that 4.294 does indeed appear to be "kinder" to PMDG. Where before, turbulence would result in oscillating rolls of increasingly greater magnitude (which ultimately results in a burn mark on the countryside), I am now observing the same +/- 20 degree deviations that you noted. Disabling turbulence immediately quells the problem.

BTW, below is a snippet from my log:

Module base=61000000

Wind smoothing fix is fully installed

DebugStatus=15

140 System time = 01:29:25

140 FLT UNC path = REDACTED

140 FS UNC path = REDACTED

797 LogOptions=00001001

797 LogExtras=256

See anything here that would explain why I'm not getting any wind data written to the log?

[abaragones]

Hi,

Just wanted to add that I am experiencing the same behaviour with CaptainSim 757(FSX) latest version (PROPack) and the latest FSUIPC version (June 22nd). If I run FSX without FSUIPC in the modules folders this issue dissapears. If there is anything I can do to help find the problem please let me know. Thanks

-Abraham

[Pete Dowson]

Just wanted to add that I am experiencing the same behaviour with CaptainSim 757(FSX) latest version (PROPack) and the latest FSUIPC version (June 22nd). If I run FSX without FSUIPC in the modules folders this issue dissapears.

What problem?

If you mean the autopilot cannot cope with the new facility for emulated turbulence then why not simply turn it off? Why run FSX without FSUIPC? On the other hand, if you are not using FSUIPC for anything, why is it installed in the first place?

All of the evidence so far proves that the turbulence emulation in FSUIPC 4.294 is working as intended. It is more realistic than FSX's own, but evidently some non-FSX autopilot programming is not designed to cope with such. Since the only reason I added it was to provide realistic turbulence to those using FSUIPC's Wind Smoothing option (which, as it has always done in previous versions of FS, stops all turbulence, gusts and so on), the two ways around the "problem" are

(a) disable wind and cloud turbulence in the FSUIPC options, and simply not have any such effects, as before, or

(b) disable wind smoothing so allowing the FS turbulence effects to operate.

There are parameters provided in the INI file through which you can tone down the turbulence to less than its current realistic levels, though some report that this has little effect on the A/P which presumably simply cannot handle any small but frequent wind changes.

Regards

Pete

[abaragones]

slow down, just wanting to help here...

What problem?

The one describe in the previous 5 pages, I didn't want to write a long post but...

If you mean the autopilot cannot cope with the new facility for emulated turbulence then why not simply turn it off? Why run FSX without FSUIPC? On the other hand, if you are not using FSUIPC for anything, why is it installed in the first place?

Yes, I mean that problem...

I did turn it off, again just didn't want to write a long post, I tried turning it off and the probelm went away too, same as taking FSUIPC from teh modules folder. Thsi point was made to repeat that the problem only appears when FSUIPC is present.

I have 7 reasons why I am using a PAID version of your FSUIPC, same reasons why I paid for the FS9 version... I don't think that list belongs in this post, but you are welcome.

All of the evidence so far proves that the turbulence emulation in FSUIPC 4.294 is working as intended. It is more realistic than FSX's own, but evidently some non-FSX autopilot programming is not designed to cope with such. Since the only reason I added it was to provide realistic turbulence to those using FSUIPC's Wind Smoothing option (which, as it has always done in previous versions of FS, stops all turbulence, gusts and so on), the two ways around the "problem" are

(a) disable wind and cloud turbulence in the FSUIPC options, and simply not have any such effects, as before, or

(b) disable wind smoothing so allowing the FS turbulence effects to operate.

There are parameters provided in the INI file through which you can tone down the turbulence to less than its current realistic levels, though some report that this has little effect on the A/P which presumably simply cannot handle any small but frequent wind changes.

Regards

Pete

Thank you, that is all I was looking for.

Hope you are having a better day.

[Pete Dowson]

slow down, just wanting to help here...

Sorry. Maybe I misinterpreted. But please do go read what I've posted in the release notes for FSUIPC 4.294.

It's just that I'm just not sure what help there can be. I spent most of last week putting together some special logging, in a way which doesn't interfere with the timing as normal logging would, and examining the results of FSUIPC's turbulence emulation on the actual wind values, and they all show just what was intended, very small (fractions of a knot or degree) changes varying up and down at a suitable frequency, based on the frame rate (but not exceeding 25/sec), to assorted targets slightly off the main wind values (for example +/-1.5 degrees or knots for moderate turbulence, which is one away from the realistically unmanageable "severe" setting) with a pretty reasonable approximation of a normal distribution of those targets.

You can enable the logging yourself, as soon as you spot an A/P problem (or slightly before, better) and see the values for yourself. If you do think they aren't as realistic or mild as I say, let me see them please. In order to enable/disable this logging just put "Debug=Please" in the [General] section of the FSUIPC4.INI file, then change the "Extras" number on the Logging tab when needed, as mentioned in the Release notes.

Regards

Pete