Re: (Split topic)Interfacing my motorized 737 TQ

[Sudden81]

This is nice movement of TQ handles: West Star Aviation's Auto Throttles for the Hawker 800/850/900XP

This is the best movement I have seen: Auto throttle and auto trim wheel demo of the MPS Boeing 737-800 flight simulator.

[fordgt40]

The real plane is better

You are comparing commercial very expensive simulators with ones that are home built on a limited budget. Do you not think that is a little unfair and discourteous to fellow forum members.

When will we have the opportunity to review your work?

Nick

Great work, thanks for sharing - David

[Maurice]

Do you not think that is a little unfair and discourteous to fellow forum members.

When will we have the opportunity to review your work?

Could not agree more.

Nick, this is amazing work. You should seriously consider marketing your program when you are done with it since it could easily be used in homemade TQs as well as real TQs.

Way to go!

Maurice

[Sudden81]

The real plane is better

You are comparing commercial very expensive simulators with ones that are home built on a limited budget. Do you not think that is a little unfair and discourteous to fellow forum members.

When will we have the opportunity to review your work?

Nick

Great work, thanks for sharing - David

You defiantly will.

When I can come up whit a way that get me to what I want.

I have a real 737 TQ converted to NG interfaced by northernflightsim but it is an old conversion with servos and GoFlight cards.

This works bad and does not really work. I talk to IvarHestnes how also told me that he tried to interface with servos that it did not work well.

I am waiting pationtly for ProSims to have a better support for home built TQs.  And I do not want to by new expensive interface cards that I then have to sell.

My plan is right now as this:

Use Pokey cards for buttons and axis. This gives me super fast Update frequency and 12 Bit Pot Resolution. Use heigh quality Pots. Maybe with dual pots in one pot. Withs gives me the possibility to absolutely sync input to FS and the input to the servo card. 
Use Pololu Servo Card in combination with there DCmotorcard(with a special cabel between them (http://www.pololu.com/product/1393)). This give me the possibility use Prosims support for servos but with DC motors

Off course Slip clutch with ONLY wire no chain and a wire wheel instead to fit it iside the TQ not outside. 

Maurice? Point of that?

Nice Program Nick

I will manage to accomplish what I want in the end

[fordgt40]

Jacob

"This gives me super fast Update frequency and 12 Bit Pot Resolution. Use heigh quality Pots. Maybe with dual pots in one pot. Withs gives me the possibility to absolutely sync input to FS"

Some thoughts

12 bit resolution in itself is not the answer. The two key factors are the stability of the 5volt reference voltage to an A/D converter and more importantly, the physical distance that the lever travels for each "software" unit. That is why I suggested in an earlier post to get a pot with say three turns at least for full scale plus physical gearing to the lever.

Fast update frequency should not be an issue with modern I/O cards - they are more than capable of keeping up! Ease of programming software control is more important

If you go the dual pot route to maintain the "god" of synchronised lever movement, how are you going to deal with asynchronous lever movement to cope with control problems, differential thrust when taxiing etc etc. Or do I misunderstand what you are suggesting by a "dual pot" to serve both levers

[Sudden81]

http://se.mouser.com/ProductDetail/Honeywell/D53C150K/?qs=sGAEpiMZZMtC25l1F4XBU1ssG7966jzizo9c0vdpOoE%3d

Something like this.

What is 8 bit 10 bit 12 bit?

http://labjack.com/support/faq/what-does-12-or-16-bit-resolution-mean

This is how I understand it:

If we have 8 bit that means 255 different steps
If we have 10 bit that means 1024 different steps
If we have 12 bit that means 4096 different steps

How many steps do fs have? About 30000= -16000 +16000

But how far can the handle move about 34 cm and radius abut 31 cm on my TQ. I just measured fast.

This gives me a travel of about 47-48 cm. Lets say we divid 480mm/255= that give accuracy of 1,88 mm that is really bad.

(10 Bit)Ok lets say 480mm/ 1024= 0,469 mm That is better.
But with 12 bit you will have absolutely smooth input with a good pot. (12 Bit) 480mm/4096= 0,117 mm per step witch gives you a theoretical possibility to 30000/4096= 7,324 steps in fs.

With OC and 8 bit it will be 30000/255= abut 118 per step in FS.

This is only theory.

But You have to have a stable ground with good hardware to stand on. Filter in Pokeys of course.

I have herd that the most important thing is that that the handle is good balanced.

My friend Hans Westman have don a perfect job with his one built slip clutch, perfectly balanced, continues rotating servo, Pololu card, 2 Pots, One directly connected to the servo(removed the pot inside the servo)

This is one step closer to perfect motion:
Test TQ servo Prosim rejected to
Test TQ servo Prosim rejected to

This is prof of what you can do with not that costly material if you know what to do.

Small Servo Modified for Continuous Rotation with Motion Tracking

y-5-1-continuous-rotation-servo-power-gearbox.html

[fordgt40]

Perhaps this should be moved to a new thread to avoid detracting from Nick`s efforts - Jack, could you help here, if you agree?

Your maths are incorrect, I suggest you recheck your assumptions and measurements. On my throttle with OC 8 bit resolution A/D, I get .66mm lever movement for each software unit. Perfectly adequate for smooth operation.

[XOrionFE]

Perhaps this should be moved to a new thread to avoid detracting from Nick`s efforts - Jack, could you help here, if you agree?

Your maths are incorrect, I suggest you recheck your assumptions and measurements. On my throttle with OC 8 bit resolution A/D, I get .66mm lever movement for each software unit. Perfectly adequate for smooth operation.

I agree,

Nick,your work is incredible.   Bravo!
Lets get this thread back on track.

Scott

[Sudden81]

Perhaps this should be moved to a new thread to avoid detracting from Nick`s efforts - Jack, could you help here, if you agree?

Your maths are incorrect, I suggest you recheck your assumptions and measurements. On my throttle with OC 8 bit resolution A/D, I get .66mm lever movement for each software unit. Perfectly adequate for smooth operation.

But how is the math then?

[fordgt40]

"But how far can the handle move about 34 cm"  Remeasure it is circa 170mm
"How many steps do fs have? About 30000= -16000 +16000"  Check your FSUIPC Docs it is -4096 to + 16384. The negative values are used for reversers.

[FredK]

All this is interesting discussion.

This is what I get regarding measurements for my TQ:

Arc (radial sweep of levers) = ~ 82 degrees
Radius (center fulcrum to outside end of white handle) = 330 mm

Therefore...

Arc traveled = 3.14(330x2)x(82/360)
Arc traveled = 472mm (that figure checks with Jacob's)

Now......

My pot travel across this range is only 108 units (225-117) since (a) I am not utilizing the full sweep of pot movement in my engineered setup and (b) part of my pot travel is used for the reversers.

So......

I in fact get about 4.4mm handle travel per pot unit. So admittedly that would be noticeable (albeit very slightly) and a higher bit resolution setup should improve on that.

However I contend that is still only a minor aspect of the problem.  If you look at the video I posted for a full sweep of the levers on a descent retard I typically will get about 1-4 noticeable "hesitations" along the way......no where near 108 step hesitations as one may nominally expect to see by the argument logic.  In between those hesitations the motion is in fact perfectly smooth.  This is because the lever position at any moment lags the target position values. This is why I have suggested that using slower motion is favorable regarding enhancing smoothness.

So the way I see it is when the motion does come to a hesitation it is because the lever has caught up to or, worse yet, overrun the target value that is commanding it.  This instance could be due to several factors: the target value itself from FS, the avionics software in use, communications lag within the electronics, communications lag within the OS, program efficiency (I have attempted to simplify my SIOC script as much as possible), and a compounded combination of all of the above. A lot of that is really very difficult to pinpoint at least to the extent of my abilities, but some of the electronics communication could conceivably be improved by better bit resolution I would guess.

So I have to admit that there seems to be some logic in what Jacob is suggesting particularly if ProSim is looking at the software side of the equation integral to it all.

Let there be peace!  Would love to see it!

In the meantime I am enjoying my new TQ setup.  The lever performance does work better than my old Phidgets servo-based motion using DVATX (note Phidgets is 10 bit resolution).  That much I can say for sure.

Fred K

[fordgt40]

Fred/Jacob

I will hold my hand up and admit that I was measuring at the gate and not the top of the lever, however, I still do not get the same measurements. My Symulatory throttle gives a top of the lever travel of 300mm whereas my real 737 throttle gives only 400mm.

Currently, I have only motorised the Symulatory throttle, however, the pots are three turn ones and are geared so that full scale movement of the throttle levers utilising the full sweep of the pot. That gives me 1.17mm per software A/D unit or 64 FS units - significantly different to what was being suggested. I still maintain that approx 1mm variance is not material. Clearly the real throttle will be slightly worse resolution and I may indeed need to go to higher A/D resolution

I fully agree with Fred`s software comments.

Let peace reign 

David

[FredK]

One other observation.......

I went back and took a careful look at the ProSim video clip posted by Jacob.  The speed of lever movement there is very slow across the complete range of motion, certainly slower than you can see in mine as one comparison.  Again, my postulate is that the slower the speed, the better the smoothness for reasons I stated above.  So that simple fact alone will make a difference.  I am running at the lowest possible speed setting for the motors that I have installed....retrospectively perhaps a lower gear ratio motor would have been a better choice. 

FredK

[fordgt40]

Fred

You are right about the lever speed, I found a similar effect. I designed my motor/gearing around a travel time of 5 seconds from idle to firewall. Not certain if this is correct. Jacob do you know? What is the timing on yours Fred?

David

[Sudden81]

Fred

You are right about the lever speed, I found a similar effect. I designed my motor/gearing around a travel time of 5 seconds from idle to firewall. Not certain if this is correct. Jacob do you know? What is the timing on yours Fred?

David

Will try to find out.

It is more to TQs than I antically thought! Ned some Fcom reading

[FredK]

Fred

You are right about the lever speed, I found a similar effect. I designed my motor/gearing around a travel time of 5 seconds from idle to firewall. Not certain if this is correct. Jacob do you know? What is the timing on yours Fred?

David

Full sweep on my levers is about 8-9 seconds.

Fred K

[Sudden81]

Fred

You are right about the lever speed, I found a similar effect. I designed my motor/gearing around a travel time of 5 seconds from idle to firewall. Not certain if this is correct. Jacob do you know? What is the timing on yours Fred?

David

Full sweep on my levers is about 8-9 seconds.

Fred K

In this vid: Antalya to Katowice

I timed it to 2,81 sec, 40-60% to N1 so I think About 3-4 sekonds is a good number.
Any way I think that we ned a ramp upp sped and a ramp down speed? Acceleration And possibility to control that. A possibility to control MAX speed.

[fordgt40]

Jacob

Thanks, see the new thread I have started. The facility to ramp speeds up/down is incorporated in our SIOC programs. Perhaps you may need to learn SIOC 

David