Chasing the autopilot

[Paraffin]

Is the Sperry autopilot supposed to drift this much on the roll axis?

After I get the plane trimmed straight and level as closely as I can in cruise mode, I enable the AP and it engages just fine. But it never stablizes in roll axis. It will drift very slowly to the left or right, so every minute or two I have to manually click the roll knob to regain my set heading (and then an equal number of clicks back to the "center" on the knob). The drift is something like 5 degrees every minute and a half. This is with zero winds and turbulence.

I can understand an imperfect, early system, but that's much more pilot input than I would think is necessary. Especially if it's supposed to be coupled to a gyrocompass. The way it's working now, it's acting like just a fine-tuned aileron trim control, with no feedback loop from a gyro or other heading reference.

Maybe this is the way the real one worked, but if so, then I might not be able to use it like was planning to in FSEconomy for long-distance flights. It's just too much drift and too much constant attention needed. By the way, the pitch function works okay. It can be a little finicky, but eventually I can get it to hold a level altitude.

Another odd thing is the way the AP locks out all control input from the yoke. I don't know how the real one responded to control input, but I don't think it would have been possible to mechanically disable the yoke this way. This isn't as bad as the lack of the AP to hold a stable heading, it's just something that feels weird in a vintage airplane like this.

P.S. in every other respect, this model is a lot of fun to fly.

[Goran_M]

The way it's supposed to be flown under autopilot control, is to trim the aircraft, as you have done, and to ensure it is flying straight and level for at least a minute or 2. The way you confirm this is to make sure the adjustable Heading Selector Gyro on the Autopilot Panel (The heading you wish to fly) is still aligned with the Directional Gyro after that minute. If the aircraft is on a slight roll, you will see the DG slowly moving out of alignment with the adjustable heading selector.

If you are drifting, it's likely you haven't trimmed it enough.

If you're satisfied with the trim, and the 2 gyro's remain aligned after about a minute or 2, then engage the autopilot. It takes a bit of practice.

The early Sperry's, as far as I can see in the info I have, do not have a wing leveling function when activated (Although I could add one, but that would take away some realism). They hold the pitch, roll and yaw that the aircraft is currently flying.

Regarding the yokes not being able to move while under autopilot control. When the Sperry Autopilot is on, the yokes have no control whatsoever over the aircraft. They also do not override the autopilot.

Now, it's possible that in the old vintage versions of this aircraft, the yokes did move with the autopilot, but I have to say, in all the research I did, I was not able to find out for sure. What I could find regarding this is that "the autopilot, while operational and in control of the aircraft, cannot be overridden with the control column."

Edited April 23, 2012 by Goran_M

[Paraffin]

If you are drifting, it's likely you haven't trimmed it enough.

This could spark a whole discussion about how difficult it is to trim any X-Plane model so it "settles" in straight and level flight, but I won't go there.

As close as I can get it, and stabilized in cruise mode, the autopilot still won't hold heading for more than a minute or two without drifting. I'd like to hear what anyone else here is experiencing, especially on long flights. This was tested on FSEconomy assignments that were one or two hours long, where constantly fiddling with the AP really isn't an option. If you use time compression (which I sometimes do, on long flights), the problem is magnified. Time compression works fine on every other model in X-Plane, because the AP holds heading the way you expect it to.

The early Sperry's, as far as I can see in the info I have, do not have a wing leveling function when activated (Although I could add one, but that would take away some realism). They hold the pitch, roll and yaw that the aircraft is currently flying.

Are you sure that's how they worked? I'm no expert on this stuff, but a little Googling turned up this old 1933 reference to the Sperry autopilot being driven by a gyrocompass:

http://books.google....tosphere&f=true

That would act in a feedback loop to maintain a steady heading: In other words, not just "fixing on where your control input was" when you engaged it.

Aside from just my personal wish to fly this plane hands-off for long periods of time in FSEconomy, I think that anyone coming from the MSFS side of the fence to X-Plane will be expecting at least a steady heading hold when AP is engaged. This is such a great DC-3 model... it needs to be practical for long, real-world flights, not just zooming around the pattern at a simulated air show.

About the yoke input being disabled, that's a lesser issue. It just looks weird because moving the yoke or joystick doesn't move the 3D control column in the cockpit when the AP is engaged. It's not that big a deal.

[Goran_M]

I don't want to go and point the finger at the x plane trim function just yet.

There are so many models and so many variants of the DC-3, it's literally impossible to make it a "universal" version.

I had a specific set of documents, and although there were some things I wanted to add/change, I had to stick to what was in front of me. SOME things were not explained in detail, so I had to improvise. (Case in point, the yokes not moving when the autopilot was on)

I'll talk to Cameron and Theo about adding a wing leveling function to the autopilot once it is engaged. But only if it's certain that it was in the old DC-3's of the 1940's and 50's.

It certainly isn't a difficult thing to add, so if it does get in, it will be a part of the update that hopefully includes the auto rich and auto lean functions.

If wing leveling is not an option, I can slow the trim down to make it more finely adjustable.

[Paraffin]

I don't want to go and point the finger at the x plane trim function just yet.

Me neither, I mentioned it because it's only relevant if the AP is working as a trim adjustment without an outside reference, like a gyro. If there was a gyro reference with manual control lock-out, then X-Plane trim should be irrelevant, unless you engage the AP with the plane badly out of trim.

I'll talk to Cameron and Theo about adding a wing leveling function to the autopilot once it is engaged. But only if it's certain that it was in the old DC-3's of the 1940's and 50's.

It certainly isn't a difficult thing to add, so if it does get in, it will be a part of the update that hopefully includes the auto rich and auto lean functions.

Thanks for looking into this, it would make a big difference for those longer flights that this ship was designed for. I think the key is to figure out if it did indeed reference a gyro in a feedback loop for holding a heading. That seems to be indicated in at least that one reference image above, although maybe this is a different model.

To make things even more confusing, I found some references to the early Sperry using the rudder for directional (heading) control, and that seems to be the arrangement in the image linked above. Apparently one of the earlier versions of Flight Simulator had a DC-3 that worked this way, but they reverted to AP aileron control for the DC-3's AP in FS2004 and FSX? Maybe because it was easier to hook into the flight model, or it was just more efficient. Even if the real thing worked that way, I think it would be an acceptable fudge to use roll for heading control in the LES model.

If wing leveling is not an option, I can slow the trim down to make it more finely adjustable.

That would help, although even if the increments are 1/4 or 1/2 degree for the roll, there will still be a need for continual adjustment of heading if the AP isn't slaved to a reference like a directional gyro. Still, that would be worth trying if you don't want to use that approach. And again, thanks for taking a look at this.

Now the next thing I have to work on, is a 2-wheel landing without bouncing.

[jmiked]

To make things even more confusing, I found some references to the early Sperry using the rudder for directional (heading) control, and that seems to be the arrangement in the image linked above. Apparently one of the earlier versions of Flight Simulator had a DC-3 that worked this way, but they reverted to AP aileron control for the DC-3's AP in FS2004 and FSX? Maybe because it was easier to hook into the flight model, or it was just more efficient. Even if the real thing worked that way, I think it would be an acceptable fudge to use roll for heading control in the LES model.

I think I read a reference in a developer's notes for a DC-3 in FS9/FSX that in FS9 they could set up the Sperry they modelled to only control heading by rudder movement, which was authentic. The rudder control was connected to the gyro, but airleron control was not. FSX was programmed so the autopilot was unchangeably linked to both rudder and aileron, so they had to live with it for that version. I don't recall which Sperry they were simulating (but I think it was the MKIII). It may have been the DC-3 sim from Mid-Atlantic Air Museum, which was designed to be just like a real-world DC-3 they restored and operate.

[jmiked]

Aside from just my personal wish to fly this plane hands-off for long periods of time in FSEconomy, I think that anyone coming from the MSFS side of the fence to X-Plane will be expecting at least a steady heading hold when AP is engaged. This is such a great DC-3 model... it needs to be practical for long, real-world flights, not just zooming around the pattern at a simulated air show.

I've found that the AP will hold a steady heading (and altitude) if Plan G is running at the same time with a flight plan made up. If the Sync feature in Plan G is toggled on and the radio panel is visible, you can set a heading either in the DC-3 or in Plan G, and the AP of the DC-3 will keep to that heading. If you change the heading on the Sperry, the plane will change direction to follow the heading that you have just set. It will also hold an altitude if the ALT button on the Plan G radio panel is clicked at the altiutude you are at.

You can also set the nav and com radios, ADF, and transponder frequencies from Plan G.