Tipping point...and Casmo's checklist

[Raisuli]

3 minutes ago, MAXsenna said:

@Raisuli
In a perfect SIM world, you'd have a joystick with magnetic brakes, (could be Force Feedback), collective and non-centering pedals. NO CURVES!
Can't remember what you have, but a stick with an extension, NO CURVES, that's OOB non-centering or modded with rubber bands or RC dampers.
No matter the above, keep the trim button pressed when moving the stick.
If you have the Ka-50, have a go in it first, as it's easier, but sort of similar.

Cheers!

Sent from my MAR-LX1A using Tapatalk
 

White Russian Warbird base and CM50 stick (because you just can't have too many buttons), about 20cm of Croatian extension, Croatian rudder with the 3D printed 'tactical' pedals, width adjustment, and damper installed, French throttle (TM Warthog), Not sure who makes the Teensy boards I use for controllers, Japanese switches, a 38" Dell/Alienware monitor, and Sennheiser One headset.  Very international.  I keep looking at collectives, but there's just not enough room for one so the right throttle is my collective until and unless everything get a redesign.  Would be cool to get a new seat, which would also require that redesign.

I think one of my problems with the trim is I don't release the controls fast enough when I press it, so I double my inputs.  That's particularly true of the rudder/anti-torque. 

The other problem is the aircraft in which I have the most hours, and am actually semi-good at, has wings bolted to the frame.  I like helos, but I also like to go fast.  It turns out being about to do AAR and formation in fixed wing helps flying helos, and vice versa, so time spent catching a basket helps with hovering.  When I get frustrated with one I go to the other for a while.

My current operating theory is there are no unrelated skills when it comes to flying various aircraft, so there is no wasted practice time! Lately I've been doing long (for me) hops to build my ability to focus on fine motor skills; take off, tank, move to the operational area, tank, do a 250 mile. 30 minute formation flight, tank, back to base after dark.  About two, two and a half hours.  Total respect for the guys who do this for real and pull six or eight hour hops.  Of course they are a bit younger than me...

[Floyd1212]

3 minutes ago, Raisuli said:

I think one of my problems with the trim is I don't release the controls fast enough when I press it, so I double my inputs.  That's particularly true of the rudder/anti-torque.

Try the "Central Position Trimmer Mode" for the cyclic to eliminate the "doubling" of the inputs altogether.

You mention a damper kit on your pedals, but not if you have removed the spring or not.  If you have no spring on the pedals, then you should be using the "Pedals Without Springs and FFB" mode for those.

[Raisuli]

2 minutes ago, Floyd1212 said:

Try the "Central Position Trimmer Mode" for the cyclic to eliminate the "doubling" of the inputs altogether.

You mention a damper kit on your pedals, but not if you have removed the spring or not.  If you have no spring on the pedals, then you should be using the "Pedals Without Springs and FFB" mode for those.

The rudder also has light springs, because...fixed wing...  They return to center-ish, but aren't militant about it.  I keep a dead zone of 1-5 depending on the aircraft because the return is so light and it's not always easy to tell if I'm pushing one or the other with my feet on them.  Found out I have to keep my bloody feet on the floor during AAR because I tense up and start putting in rudder.  Two problems there; tensing up and putting in rudder.

Central position trimmer mode.  Going to have to find that!  All I've found searching for 'trim' is the 4 position trimmer switch.

I gotta say, for the record, you guys have been totally awesome helping me out with all this.  I'm leaning in a vacuum and the advice is really nice, if only to confirm what I already think is wrong, but also for all the hints and tricks to get better at it.  Thanks a lot!

[Floyd1212]

1 hour ago, Raisuli said:

Central position trimmer mode.  Going to have to find that!

This is in the Options > Special > AH-64D menu.  (You can only access this from the DCS Main Menu, not while in-flight.)

In Central Position Trimmer Mode, after you release the FTR button on your controller, DCS will ignore any changes to the cyclic axis until you have allowed your physical controls to return to their natural "sprung" center.  Then you can use the cyclic like normal to add roll and pitch to the currently' trimmed position.

1 hour ago, Raisuli said:

The rudder also has light springs, because...fixed wing...

I highly recommend removing the springs on your pedals and using the Pedals Without Springs mode.  It sounds like you are using the MFG pedals (from Croatia), so removing the spring is not a major chore.  Kind of a pain to swap back and forth if you hop between fixed-wing and helos in the same session, but your pedal work while flying the Apache will be a lot more intuitive with them unsprung.

[Syndrome]

11 hours ago, Hiob said:

I doubt that curves on the input axis' create any noticable input lag. If so, it is far below my perception treshold. (Edit: Come to think about it. Smoothing the raw input data by applying some kind of moving average would introduce a (very small) amount of input lag - but that is done on a system level by the windows drivers probably; The input curves just apply an offset or an factor to the input. That is, what, half a dozend instruction cycles? Measured in nano seconds. So no - no input lag is added by the curves).

Anyhow - there are different methods of tinkering with the input curves. Linear and non-linear, and they only mess up your muscle memory, if you change them all the time. How can they mess up your muscle memory if you apply them once and then leave them alone?

They are meant to allow for all different kinds of different input hardware. Long throw, short throw, strong springs, soft springs, no springs....and so on. It is/was a brilliant move of whoever came up with the idea to implement such an interface in this kind of software.

I think to encourage people to not to make use of those possibilities is bad advise.

 

When you flatten the curves you reduce dx/dy but you increase dx/dt.  Which is to say that the physical motion to get from 0-10% is larger, and hence it takes longer to get from 0% deflection  to 10% deflection.  This increased length of time (to get to 10% for example) is a form of input lag, and can be on the order of 100ms.  This may seem small, but most people can notice input lag of 16ms (about 1 frametime of lag on a 60hz monitor).  

If you flatten the curves by lowering the saturation, then the response is as consistent as default.  However if you have a non-linear curve then the degree of spring tension won't be related to the virtual deflection when using centering trim.  This will confuse muscle memory training, and is likely a major reason why so many people using curves go cattywampus when using or holding trim, since the bird isn't responding how they would expect at a given deflection.

You are correct that the large range of input hardware makes it difficult to give advice on which setup is "best".  Personally I don't believe there is a best setting.  There are simply positive and negatives for any kind of setup.  I was pointing out the downsides of curves for more the common spring loaded hotas stick of any length, but there are also upsides to curves for sure.  I use them on certain warbirds that are difficult to aim precisely without them, and I suck up the input lag and make do with the trim issues.  

Edited July 4, 2023 by Syndrome

[Raisuli]

1 hour ago, Floyd1212 said:

This is in the Options > Special > AH-64D menu.  (You can only access this from the DCS Main Menu, not while in-flight.)

In Central Position Trimmer Mode, after you release the FTR button on your controller, DCS will ignore any changes to the cyclic axis until you have allowed your physical controls to return to their natural "sprung" center.  Then you can use the cyclic like normal to add roll and pitch to the currently' trimmed position.

I highly recommend removing the springs on your pedals and using the Pedals Without Springs mode.  It sounds like you are using the MFG pedals (from Croatia), so removing the spring is not a major chore.  Kind of a pain to swap back and forth if you hop between fixed-wing and helos in the same session, but your pedal work while flying the Apache will be a lot more intuitive with them unsprung.

FOUND IT!  My current setting was "FFB Friendly", and being an unrepentant FFBist (only because I don't have one) it got changed to Central.  Seems much better, but didn't stop me from doing BadThings(tm) to engine 1.  I'm pretty sure they can repair the airframe, maybe, at a depot level.

Removing the spring, where my rudders are, is more than just kind of a pain.

1 minute ago, Syndrome said:

When you flatten the curves you reduce dx/dy but you increase dx/dt.  Which is to say that the physical motion to get from 0-10% is larger, and hence it takes longer to get from 0% deflection  to 10% deflection.  This increased length of time (to get to 10% for example) is a form of input lag, and can be on the order of 100ms.  This may seem small, but most people can notice input lag of 16ms.  

If you flatten the curves by lowering the saturation, then the response is as consistent as default.  However if you have a non-linear curve then the degree of spring tension won't be related to the virtual deflection when using centering trim.  This will confuse muscle memory training, and is likely a major reason why so many people using curves go cattywampus when using or holding trim, since the bird isn't responding how they would expect at a given deflection.

You are correct that the large range of input hardware makes it difficult to give advice on which setup is "best".  Personally I don't believe there is a best setting.  There are simply positive and negatives for any kind of setup.  I was pointing out the downsides of curves for more the common spring loaded hotas stick of any length, but there are also upsides to curves for sure.  I use them on certain warbirds that are difficult to aim precisely without them, and I suck up the input lag and make do with the trim issues.  

 

Actually wouldn't that be dx/dd and dy/dd where d=displacement, because x is not a function of y in this case.  Those are independent axis.  d could be a considered a function of time, but d'd/dt' is not constant and depends on the input rate from the user, which means if you have quick hands dd/dt is shorter.  Of course then we have to take overshoot into consideration and return rate, so there's going to be a dampening function k in there and spring equations are not out of consideration.

What I know, apart from calculus, is flatting the curves cost the Army three helicopters and me a few days in the hospital with neck and back injuries from the hard landings today.  George is a blow up doll with a helmet, so I'm not worried about George.

There is probably some point in my future where precise 1mm deflections of the stick are a thing, but I'm not there yet.  I watched Casmo's CrotchCam(tm) (available without an OnlyFans account, so that was a bonus) and it almost seemed like he was deflecting the stick more than I was, but the end result was somewhat different (and the Amy got that one back in one piece).

So back to my 20 curves and practice.  I simply can't not think of that passage in Chickenhawk, which I haven't read in decades, where he had to hold a hover pointed at a tree.  And totally sympathize with him.

 

[Hiob]

1 hour ago, Syndrome said:

  This increased length of time (to get to 10% for example) is a form of input lag,

 

No, sorry, that is by no means the definition of lag.

You get less reaction for your input, which is exactly the purpose of it.

Lag is a pause between action and reaction, but here the reaction is immediate and just reduced by a factor.

[Syndrome]

 

 

On 7/4/2023 at 1:40 PM, Hiob said:

No, sorry, that is by no means the definition of lag.

You get less reaction for your input, which is exactly the purpose of it.

Lag is a pause between action and reaction, but here the reaction is immediate and just reduced by a factor.

I was actually choosing my words very carefully, hence why I said a "form of lag".   You are correct that there is no extra delay in the initiation of a deflection, but there is a time delay to get to the end of a desired deflection short of 100%.  This is true no matter how small the deflection, because you are defining a curve where below 100% deflection: dy < dx, and hence dy/dt < dx/dt, and hence dty > dt_x for any Δx=Δy.  This time differential delay is measurable as t=(Δt_y-Δt_x).  Again likely measured in mere milliseconds.  Whether you notice a delay or not will come down to how fast your nerve conduction is.  It reduces with age (by about 1m/s per decade) so it's possible that many people that play DCS won't notice.  However, I'm no spring chicken and I still notice a pretty big delay at a curvature of 10. 

Now you might say that this matters less for small deflections, but actually the opposite is true.   For instance at curve = 10 an input of 1% will result is a 0% output.  To reach 1% deflection output, you'll need 2% physical deflection and 3% physical input to reach 2% output. So the smaller the deflection, the longer the relative time delay.  100% to 50% time delay within a deflection of 3% of the center of the stick.   3% is about is about the max you'd want to deflect in a hover correction, so a time delay of even 50% is pretty significant.   Eg if it takes you 100ms to deflect the stick to a physical 2% and virtual output of 1%, then you could have deflected it the same amount in approx 50ms by removing curves.  That's equivalent to a ~3fps input delay (or "lag" if you will).  Keep in mind this example is using a very minor curve of 10.  For higher curvatures like say 20, the resulting differential time delay for a 3% deflection will be dramatically longer. 

I would encourage you to test the different curvatures about the center of the axis in small deflections, and then compare them to the response feel of the default 1:1 or zero curvature.  See if you notice the difference in time to get the output to read 3% from zero curves to curvature 10 and curvature 20.  The degree of curvature that you don't notice a time delay is probably your sweet spot.  Mine is between 0 and 5.

Edited July 6, 2023 by Syndrome

[Syndrome]

On 7/4/2023 at 12:38 PM, Raisuli said:

Actually wouldn't that be dx/dd and dy/dd where d=displacement, because x is not a function of y in this case.  Those are independent axis. 

 

The x and y refer to the x-y curve plot in DCS, but yeah you can label the variables however you want.  In this case, x is the physical input displacement and y is the game's registered output displacement on that same virtual axis, as defined by the curve: x is the independent variable, and y is the dependent variable.   This is not the same x-y axis labels on the cyclic axes (pitch and roll).  

It's true that if you have quick hands you can reduce dx/dt, however dy/dt will still lag behind it by design (unless you remove the curves).  So it will always take longer to deflect dy the same amount as it would dx if the curves are flattened.  Also, if you have faster nerves, then you'll also notice more time passing per second so while the lag may be shorter, you'll notice it more. And because of the nature of the curves, the smaller the deflection the bigger the % gap between input and output, so hence the longer the relative delay.  See the post above this one for more details.

 

On 7/4/2023 at 12:38 PM, Raisuli said:

There is probably some point in my future where precise 1mm deflections of the stick are a thing, but I'm not there yet. 

 

You'll get there man.  AA refueling for instance

[Floyd1212]

2 hours ago, Syndrome said:

See if you notice the difference in time to get the output to read 3% from zero curves to curvature 10 and curvature 20.

It sounds like you are describing the benefits of adding a curve to your axis in the first place.  The same benefit is gained when adding an extension: you can be more precise with your inputs around that zero point

If the goal were to thrash the stick back and forth to a 3% deflection as fast as you can, yes having a linear response without a curve would get you there faster.  But with the curve/extension you can be more precise and smooth with micro adjustments, to more easily input the 0.50 percent deflection you need that moment, or then another smaller correction to only 0.47 percent deflection.

I run a curve of 15 with a 20cm extension on a Virpil base with very light springs and the least-aggressive cam available, and I am still amazed at how little I need to move my controls to get her to do what I want.

[Syndrome]

10 hours ago, Floyd1212 said:

It sounds like you are describing the benefits of adding a curve to your axis in the first place.  The same benefit is gained when adding an extension: you can be more precise with your inputs around that zero point

If the goal were to thrash the stick back and forth to a 3% deflection as fast as you can, yes having a linear response without a curve would get you there faster

Yup, this is another way of stating the restating a nuance about the same thing.  There are definite benefits to slowing down the input.  There also potential negatives as you gain more precision in your muscle movements with experience.  It's a balance of two positives and two negatives.  Precision vs responsiveness, and delayed degree of deflection vs twitchiness.  

Different pilots will have different sweet spots.  There is no one-size fits all solution here.  It like saying "cake is good" so more cake must be better right?  If you want an extreme example of going outside this proverbial sweet spot, go ahead and set your curves to something way too high (like 50-70), overload on "cake" until you can't help but notice that response feels too slow about the center and then you will end up thrashing, drifting, and PIOing.  Clearly there can be too much of good thing.  The limit for "too much" will vary from person to person, and over time as well.  When I started flying the Apache, my sweet spot was 20-25 curvature.  After several months of hovering practice, my curvature tolerance is now down to 0-5 with the same 10cm extension.

It's funny that you bring up thrashing, because after practicing hovering for a couple weeks, I noticed that the more precise I could get my hover, it became counterintuitively more prone to slight drifting and micro PIOs.  That is actually how I realized that the source of the instability was the differential time delay from having too high curves for my newly acquired muscle precision.  Around the same time, Casmo had a video that briefly mentioned reducing the curves as you improve, so as an experiment, I lowered the curves, and the thrashing, PIOs, and overcorrection, and drifting all but vanished.  Iirc, Casmo didn't really talk about why you would want to reduce curves.  I guess it should be intuitively obvious as to why, but had to figure that out on my own and it's why I'm sharing it here because I was kind of surprised how low the curves could be and still notice the negative side effects, and I only have maybe 100 hours on a simulated Apache, and 1500hrs in DCS.  Not the best pilot by any stretch, so if I could notice this, then others surely will too. 

Edited July 6, 2023 by Syndrome

[Leg2ion]

If you have head tracking - bin it when you are learning until you have basic flight manoeuvres nailed.

I found the IHADSS was messing with me, in that the eye to horizon to HUD was skewing things, so once I fixed the view things became a lot easier.

[Raisuli]

So, I've found force trim is not my friend in a ground effect hover, or any kind of hover, really.  It's really not my friend when it comes to anti-torque/rudders, and will take me for a serious spin sometimes and I end up full-boot to get pointed the direction I want to go. 

I can hold a dead hover at a fixed altitude if I use the auto modes, otherwise not so much.  Working on using auto to get me stable, then turn it off and try to maintain that, and if I don't use the force trim I drift around but not wildly.

All of which tells me force trim isn't what I think it is.

Casmo needs to add a new item to his checklist after getting massively frustrated in the AH-64:

Go get a book (currently SPQR by Mary Beard) and read for a while.