10C to 14SG Migration Findings...

[jharkin]

(Edit June-5-13)

Renaming the title of this one. I'm going to document my findings as I move model setups from the 10C to the 14SG as a helpful reference to anyone else doing the same.

Because of the common programming this **should** apply to moving either 10C or 9C programs to either the 8FG or 14SG. Might also work for the 12FG, 12Z, 14MZ, 18MZ sets, but I don't have any to test (donations welcome )

Summary:

• Channel assignments: same as long as you configure your function assignments in the 14 to match channels in the 10
• servo reversing: same in both radios
• Servo neutral: Servo protractor tests and servo programmer signal measurement prove its exactly the same @ 1520uS.
• Sub trim: The 14SG has double resolution (however you cannot just double the values since the actual position signal varies slightly, see post #10)
  • 14SG (+/- 240 steps for +/- 12 degrees, 933-2108uS)
  • 10C (+/- 120 steps for +/- 12 deg, 943-2098uS)
• End Point adjust: Same in both radios (But the positions dont output the exact same values, see post #11) Note if you change your subtim while migrating it will require a compensating adjustment to EPA to get equal travels.
  • 100 EPA = 45deg
  • 110 EPA = 50deg
  • 120 EPA = 55 deg
  • 130 EPA = 60 deg
  • 140 EPA = 65deg
• Dual Rates - are multiplicative to EPA, and can go up to 140 in both TXs, but still obey the limit points. I do not reccomend using over 100 DR, because any combined value over the limit point will mean the servo stops moving before the stick does (see post #12). What do I mean? Say the limit point is 140...
  • EPA 140 + DR 100 = travel of 140
  • EPA 100 + DR 140 = travel of 140
  • EPA 140 + DR 140 = travel of 140, servo stops at 3/4 stick deflection
    
• Swash Mix: The 14 does seem to end up with different mix % for ail/ele/pitch. Not sure why yet
• Other P-mixes: should be same in both radios

(/Edit)

------------------------------------------------------

I searched a bit but didn't find much. Has anyone been able to transfer setups from a 9C/10C series radio to a 14SG or 8FG sucessfully?

Right now Im in the process of doing the 10C->14SG move for all my setups. The last time I upgraded radios it was a snap, just moved all my 9C setups to 10C via campac and only needed some minor tweaks.

Now for the 14SG there is no way to migrate programs from the older radios obviously so I tried on one model (A 450 flybar heli setup on the 10C in 7 channel mode on an R617) to just copy and manually enter in all the settings - sub trims, end points, etc.

Channel assignments, reversing, rates and mix configurations translate ok. But everything else is a mess.

The servo neutrals are all way off... on the heli everything is off and in the servo monitor I can see that its not transmitting the same center position with the same sub trim values. So I'm guessing the 14 uses a different resolution for sub trim steps.

End points and swash mix percentages are also off. In the 10C it only needed 46% pitch mix to achieve +/-11 degrees collective. On the 14 in needs about 65% to hit 11 degrees. Cyclic appeared to be off the other direction.

and so on.


Has anyone had luck moving settings or did you just start over from scratch? I wouldn't mind so much starting over but Ive also got 4 fixed wing to move including a giant scale 3D ship with dual ailerons, dual elevators, smoke, etc and a lot of tweaked compensation p-mixes.

[jharkin]

In the 14 hover pitch was on by default and throwing off the centers. Turned it off and the centers and sub trims look right visually. Need to check everything with the gauges but things are looking better.

[jharkin]

I spoke too soon! By eye things looked closer when I turned off hover pitch, however using the pitch gauge and leveling tool it was still quite a bit off.

I spent last night seting everything up again and had to change a bumnch of the sub trims quite a bit (I think one channel was +15 before and is +80 now! and no, moving a tooth on the servo spline didnt help).

I also ended up changing the swash mixes, pitch mix had to increase from 46 to about 60 to keep +/- 11 and the ail and ele mixes decreased a bit.

I suspect that the sub trim steps are much finer on the 14, for every click o nthe 10C you can actually hear the servo move, on the 14 it seems like you have to go about 5 points to see/feel any change. Tonight I will do some quick measurements with my Servo City protractor to see exactly what each radio is doing. I'll post findings in case anyone finds this info useful.

[jharkin]

Ok, to try and figure out whats going on I setup a test rig. I setup both TXs with a new blank test model. I used one RX (in this case an R6008 running in FASST 10 channel mode) and connected a single servo (an old Hitec analog) to the aileron channel. I setup the servo in the protractor so I could measure position precisely. I then bound the RX to each transmitter in turn and recorded the position of neutral, max subtrim, and the endpoints at various settings.

What I found:

• As best I can measure, both TXs are transmitting the same neutral.
• There IS a difference in subtrims. The 14 has double the resolution
  • On the 10C, there are 120 steps of subtrim available resulting in +/-12deg variation of neutral (10steps/degree)
  • On the 14SG, there are 240 steps of subtrim available resulting in +/-12deg (20steps/degree)
• Endpoints appear to be identical on both radios
  • 100 EPA = 45deg
  • ...
  • 140 EPA = 65 deg

So other than the difference in subtrim resolution, the radios appear to be sending the same position data (as I would expect) and it would seem that all one needs to do is enter the end point values point for point and double any subtrim values to get identical setups.


On real model configurations this is definitely NOT working. Ive moved 2 models so far, the TRex 450 and a 30oz electric 3D plane. In both cases I had to make wildly varying changes to sub trim (some values almost x4), and a a few small adjustments to endpoints.

It **might** well be that my old mechanical setup wasn't as precise as I thought and I'm just now fixing errors. We will see when I move my 85" Edge. That plane has all Hitec digital's that I had set to exactly 90deg horns with the hitec servo programmer so its running zero subtrims across the board in the TX - if that setup does translate exactly it may well mean I just didn't setup the other aircraft very well.

[jharkin]

... for your time.

Other than the smaller screen and the sensor touch interface Im really liking the feel of the 14.

[jharkin]

The fact that I am sing differences when I move actual aircraft setups is really bugging me.With the exception of the narrow band high speed tail servos, ALL our hobby servos are still designed to work on the old legacy servo pulse signal of 900uS - 1500uS - 2100 uS. That 900->2100uS signal equals the -60 -> +60 degree servo travel range that all published servo specs are based on. So each degree = 10uS

I know there is a very slight variation of the neutral signal between brands (Hitec I think uses exactly 1500uS, Futaba 1520uS. Don't know the others). However within brand I think the neutrals should be exactly the same for any radio so its bothering me that I am seeing difference when I move model setups.

Using the servo protractor last nigh it appeared that visually I was getting the same signal from fresh programs on both radios. The fact that real model setups didn't work so well could be that I'm correcting minor mistakes in the geometry or correcting for trim that I was flying with before to get everything zeroed. Once I move a few models I'll do a comparison of the final values to see if that's the case.

I'malso going to try a more precise instrumented test than the servo protractor. I dug out my old Hitec HFP-20 servo programmer. The nice thing about the programmer is it will read the actual servo pulse being output by the RX and display the raw uS value on the display. I will do some tests with the two transmitters and report. (might be tricky as this older programmer wont read the low latency "high speed" output from FASST/10CH and FASSTest. So I have to do the test in FASST/7CH mode on an R617)

[jharkin]

[Math warning]

I was thinking about this even more. Knowing how the EPAs equate to degrees of travel and how subtrims equate to degrees of travel, we should be able to calculate what our EPAs need to be for a given throw with a given subtrim, given a perfect mechanical setup.

What do I mean?

Well consider an fixed wing example. Say we have a perfect setup like so:

• Servo horn length (center of gear to pushrod link) exactly equals the control surface horn length (hinge line to pushrod link)
• Control rod length set so when servo horn is exactly at 90deg, the control surface horn is exactly at 90 degrees and the surface is exactly centered

If we want exactly +/- 45 degree throws and we have to use some sub trim we can just do some math to figure the end points. Lets assume we are using the old 10C.

• If I have right 30 points of subtrim to get 90deg, that equals a neutral offset of right 3 degrees (30/10 steps per degree= 3)
• So to keep +/- 45 I have to decrease the left endpoint 3 degrees and increase the right endpoint 3 degrees.
• Since we know 100EPA = 45 deg, we do the math 45/100 = 0.45 degrees/point. 3 deg /0.45 = 6.7 points of EPA adjustment needed.
• So I would round up and then assume that I would need these end points:
  • Left 100-7 = 93 EPA
  • Right 100+ 7 = 107 EPA

Another subtle take away you might notice here. You know how Futaba recommends that you keep your EPAs between 80 and 120 at max throw? Well 120 EPA leaves you with 20 points or about 10 degrees of EPA headroom to adjust which is pretty close to the 12 degrees of sub trim range available.





Why am I doing all this? Well the above analysis got me to thinking about subtrims and the endpoint adjustments we do to mix out "CCPM interactions" I suspect the end point adjustments are not really due to any mixing interaction from CCPM but rather just an artifact of differential travel when we need subtrim to achieve true perpendicular servo horns at neutral. I think you can actually calculate how much is needed and be very close (knowing that some mechanical differential is introduced if your mechanical setup is not exact). I'm going to sit down and work out the math on my setup and see if the EPA adjustments that give me a level swash at the ends actually calculate out proportional to the subtrims I needed this way.


Another related thought - This analysis also tells me that one should NOT get hung up obsessing about having close to zero sub trim values. Its great if you can do it, but given that a typical servo spline is 25 teeth, there is 14.4 degrees of spacing between teeth, or 144 steps! (288 steps on a 14SG) of subtrim between teeth!! . I think the odds of finding a servo horn that gives a perfect 90deg every time without subtrim is pretty low.




Final thought - all of this is why programmable servos are so nice. You can just leave your subtrims at zero and use equal endpoints and just dial in the neutral on the servo and have everything exactly perfect Its been a while since I used the Hitec programmer and I'm interested to see if the new Futaba programmable BLS servos let you do the same.

[jharkin]

Originally Posted by jharkin

Why am I doing all this? Well the above analysis got me to thinking about subtrims and the endpoint adjustments we do to mix out "CCPM interactions" I suspect the end point adjustments are not really due to any mixing interaction from CCPM but rather just an artifact of differential travel when we need subtrim to achieve true perpendicular servo horns at neutral. I think you can actually calculate how much is needed and be very close (knowing that some mechanical differential is introduced if your mechanical setup is not exact). I'm going to sit down and work out the math on my setup and see if the EPA adjustments that give me a level swash at the ends actually calculate out proportional to the subtrims I needed this way.

Still thinking out loud here...

I thought through the above some more and I think I was only partially correct before. In the Finless video on eCCPM setup where he demonstrates "correcting for interactions" at high and low pitch using the bubble level on the paddles; it appears to me that is actually correcting for differential throws as I described above. I did the same but rather used the swash tool at either ends with a feeler gauge for accuracy. (Nice feature in the 14SG vs. 10C is to be able to lock collective at high/neutral/low in the swash menu and set everything from there).

In addition, thinking it through I can see now that once you have some differential travel set to achieve level collective though the range, that aileron and elevator cycling inputs could create some interaction. Now I can see the value in all the extra compensation mix fields in the 6 pages of swash menu that the 14SG has.

The more I think through this all the more I appreciate how much a FBL simplifies setup Still glad I tried a FB once so I at least have a good understanding of whats going on. On my next heli....

[HilliHeli]

Another related thought - This analysis also tells me that one should NOT get hung up obsessing about having close to zero sub trim values.

I agree, especially with a FB'd Heli (and even more-so with a FB'd Micro-Heli)

The more I think through this all the more I appreciate how much a FBL simplifies setup Still glad I tried a FB once so I at least have a good understanding of whats going on. On my next heli....

[jharkin]

... because we have our answer.


Tonight I setup my "test lab" with the trusty Hitec HFP-20 programmer. Using a blank program in each TX and the same RX, I measured the actual servo position signals in uS for various settings.

I found some interesting things!

First off, both radios are sending the exact same 1520uS neutral. But that is the only thing that is the same.

On the Futaba 10C, the sub trim values look like this:

• -120 = 1380uS
• -60 = 1450uS
• 0 = 1520uS
• 60 = 1590uS
• 120=1660uS

Now I mentioned that the 14SG has 240 steps and measured the same degree value on the protractor, so aits value at 240 should equal the 10C value at 120, right? Well close, but not exactly:

• -240 = 1370uS
• -120 = 1445uS
• 0 = 1520uS
• 120 = 1595uS
• 240 = 1670uS

Doing the math, the 10C sub trim range is perfectly symmetrical about neutral, but the 14SG values are offset 10uS to one side. I almost wonder if its a small bug in the firmware...

So it seems that my difficulties where a combination of slight differences in the subtrim values and correcting prior mistakes in the mechanical setup.


Continued...

[jharkin]

I found similar slight differences in the endpoint values.

10C EPAs

• -140 = 943uS
• -120 = 1025uS
• -100= 1108uS
• 100 - 1933uS
• 120=2015uS
• 140 = 2098uS

14SG EPAs

• -140 = 933uS
• -120 = 1016uS
• -100= 1100uS
• 100 - 1940uS
• 120=1595uS
• 140 = 2108uS

Keep in mind this test was done in the same modulation (FASST/10CH), with the same RX, on the same channel.

Also of interest is that when I tested this setup with a Hitec servo in the protractor I got -65deg -> +65 deg travel range, however Hitec's published servo specs list all there servos as +/160deg at 900-2100uS. So given the uS values Im seeing I shouldnt be seeing 65 degrees. But it is an old standard analog...


But whats really most annoying is that the values are not all shifted by the same amount between radio, you see that most of the values are shifted by 7-10uS except the 120 positive EPA thats shifted by 20uS for some reason (???).


All of this and the subtrim changes means that to move a model you either can:


#1 - Buy a Hitec programmer and read every single value in uS and fine tune the outputs of the new one.

OR more likely...

#2 - Copy everything over and then fine tune every neutral and end point using your gauges.

[jharkin]

... what happens when we use Dual rates over 100?

So both radios will allow both EPA and DR to go to 100. I measured what happens when you max them.
On the 10C:

• 140 EPA / 100 DR = 2108uS
• 100 EPA / 140 DR = 2108uS
• 140 EPA / 140 DR = 2159uS

So we can see that the values multiply, but still obey the hard limit, which appears to equate to a 144 value. Also of note is that if you have some combination of EPA and DR that multiple to a value above 143, as you move the stick whenever the value reaches that max the output stalls. So if you have 140 EPA & 140 DR set the servo will stop moving at about 3/4 stick deflection (i.e. it does not stretch the value to fit the joystick travel).

The 14SG had a similar behavior but slightly different values.