crossup

My Feedback: (1)

Since Im using a Futaba 8UA system I posted this in the Futaba forum but apparently the moderator and Futaba people are dead as no one is receiving answers.

Can some one explain under what conditions we do and dont obtain maximum resolution? Specifically servo travel in degrees and the impact of expo, use of digital servos(Hitec in particular). Please feel free to ramble on as Im interested in all aspects of the subject.

HarryC

My Feedback: (1)

It means the TX will split the full possible range of its travel signal into 1024 spots. The Rx then has to convert that back to a PPM signal for the servo. The full travel range of a servo is 0.9 to 2.1milliseconds, a range of 1.2ms or 1200 microseconds. So for every 1 point of the 1024 resolution, the Rx output to the servo can change 1.17microseconds. Most digitals and some good ordinary servos will respond to a minimum of 1 microsecond so they will respond to the 1 point step of the 1024 PCM. Most ordinary servos have a deadband of 2 to 4 microseconds so will take a few more steps of the 1024 system before they respond to a change.

When in the default travel of 100% a Tx is not sending the maximum travels, it is well below it. For example, JR allows up to 150% at which point it reaches the full 1024 positions, its default 100% travel uses 683 of the 1024 positions. Futaba uses a slightly different mapping, 100% travel uses 731 of the positions which is why Futaba’s max travel is 140%.

The resolution is always there, regardless of using expo, rates etc. These just affect the curve mapping the stick position to the output value, but the resolution of the output value remains the same. Each brand or type of servo uses a different max rotation in degrees, so you can’t say precisely that one step is so many degrees. But a typical servo would rotate approx 100 degrees at max, so each of the 1024 steps would equate to around 0.1 degrees.

crossup

My Feedback: (1)

Thanks Harry, the second and third paragraphs were what I was looking for.. as I thought its best to setup to use 140% travel especally if doing 3D

Geistware

My Feedback: (16)

This from Hobby Services:

The PCM1024 means that the transmitter basically takes one extreme of stick travel plus trim to the other extreme of stick travel plus trim and breaks it into 1024 descrete bits. This then is used to generate the PCM code for each servo's position. You won't get this level of precision at the servo, though, since there's a bit of slop in the gear train. Even so, most good servos will be about +/- 0.1 degrees or so of positional accuracy.

The highest level of precision at the servo will take place if you set the ATV of each channel to its maximum, and then use the pushrod position on the servo and control horn to determine the control surface's travel.

HarryC

My Feedback: (1)

As so often in aircraft, you have to find the best compromise. As Geistware pointed out, resolution is lost in gear train slop, linkages and so on. To get maximum travel you could use long servo arms, but that exaggerates any slop in the gear train and allows the surface to flutter. It also means you get the lowest force from the servo, since the constant torque means the further out the pushrod is, the less force it receives. You can use the innermost hole in the control surface horn but that allied to a long servo arm that can really induce slop and a serious loss of resolution. So it looks like turning the travel up to the max is the perfect solution, why don't we do that as standard? Because all servos are now rotary, sadly, the big names no longer make linear servos. With travel at 100% it is designed to make the servo rotate to around 40 - 45 degrees. Once you go above 100% then the servo's effect on the pushrod is mostly side to side instead of pushing/pulling so it is an inefficient way of getting more travel. The higher the travel value is, the more and more you have to increase it to get smaller and smaller increases in pushrod travel. The advantage is that by increasing the travel, you keep the pushrod at the same distance from the centre and retain the force that the torque can generate. Increasing the travel value also slows down the response time, since the servo now has to rotate further to get to the max travel, and that takes it a bit longer, though with the speed of many modern servos this is difficult to notice.

So there is no one simple answer, you have to compromise between slop, force, speed, and loss of travel from the servos rotary motion, and that will vary from plane to plane depending on the servos and the linkage. My preferred compromise is to turn travel up so the servo rotates about 45 degrees each way and adjust the linkage to get the required maximum from that, unless too much slop appears in which case back off the linkage a bit and accept having to turn the travel beyond 45 degrees.

mick15

H why do you think it is we modellers have not maintained pressure on the manufacturers to produce linier servos when they are so obviously superior in just about every respect??

M

Geistware

My Feedback: (16)

Good question. I would imagine that it would be difficult to get the displacement you get in rotory servos in a linear package using the existing servo size. A standard servo is smaller than the available displacement or throw.

HarryC

My Feedback: (1)

It's a shame they don't still make linear servos so that at least we could choose. The linear gives you no choice about the force/distance that you get with a rotary arm, as the force applied by the linear is fixed, and to get less travel you may have to turn down the Tx travel rather than move the pushrod in a hole thereby getting more force that you can do on a rotary. I wouldn't say linear are superior to rotary, but there are applications in which a linear is better than a rotary. Throttle, and elevator Y pushrods come to mind as great places for linear servos. I still have an old Futaba linear servo which I bought for a model with a Y elevator pushrod so I did not get differential travel. When we did not have travel adjust in our Tx it was a bit harder to use a linear servo, now that we have travel adjust we can make best use of linear servos and they stop making them!

Another thing against linear servos is that the linkages generally mean that a linear push gives an exponential effect at the control surface, which is often not wanted. The irony is that a rotary servo gives a more linear response at the control surface. The reason is that as the control surface and its horn rotate, it takes less travel of the pushrod/horn to make the surface rotate a given amount. At the same time, the rotary servo is giving less travel and the two tend to balance each other, but a linear servo will still be giving the same amount of travel to a control horn that needs less travel so you get exponential. In these days of the more advanced computer sets allowing both positive and negative expo, the Tx can take care of it if it becomes a problem for the handling.

H

Geistware

My Feedback: (16)

Does anyone know the mechanics of a linear servo?

ZAGNUT

My Feedback: (1)

it's the basic jack screw with one of those "slide pot" thingies under the output arm. look for pics of the tiny wes-technik servos and you'll see how simple it can be.


dave

HarryC

My Feedback: (1)

It's like rack and pinion steering. The final output gear is in the middle of the servo rather than at one end. Instead of having a splined top sticking out of the case, a straight arm with gear teeth in it lies on either side of the gear so that when the output rotates one arm gets pushed one way and the other is pushed the opposite way. The linear servo has force and speed rather than torque and speed, its a simple matter of varying the gearing to get a range of forces and speeds just as with rotary servos.

In this photo you can just see the teeth of the output rack. The tiny hole conatins a screw connected to the feedback pot, so with a jewellers screwdriver you can adjust the centering of the arms.

ZAGNUT

My Feedback: (1)

guess futaba did it differently the little wes-technik looks a whole lot simpler

here's a link with pics and video [link]http://www.rcmicroflight.com/nov99/technik_servo.asp[/link]

dave

mr_matt

My Feedback: (10)

Actually longer arms (servo and control surface) minimize the effect of slop in the connection links.

I agree though that very large rotational angles (with large ATV settings) increase the non linearity of the output arm. Expo can of course help this to some degree.

HarryC

My Feedback: (1)

A servo arm will rotate a fixed number of degrees due to any slop in the gear train. The longer the arm is, the more travel this converts to in the pushrod. A linkage with slop elsewhere must be free to move something, the more it can move the point at which it is connected to the servo arm the more it can "slop", so again the longer the servo arm is the more slop you get.

mr_matt

My Feedback: (10)

But in my experience most slop comes from the linkages (clevises, etc) not the gear train. Linkage slop gets better with longer arms.

(See below)

mr_matt

My Feedback: (10)

Not sure what this means. Let me try again.

Take clevis slop as an example. If you have a 1/2 inch servo arm and a 1/2 control horn on an elevator, and some fixed amount of slop in the clevis(s). Jiggle the elevator up and down and it will subtend some angle.

Now increase both arms (servo arm and control horn) out to 2 inches. The same clevis (linkage) slop exists, except now the angle the elevator will subtend from the same clevis slop is much less.

Servo drive train slop has the same effect in either case. Say .2 degrees of servo slop will cause .2 degrees of movement in the elevator with the 1/2 inch arms. Make the arms (both servo and control horn) twice as long it will still be .2 degrees.

RTK

My Feedback: (1)

I remeber in the old days when there were no ATV's etc. on my Kraft radio. Unfortunately you did not have to worry about resolution, just adjusting the linkage to work correctly.

ShamelessMale

This problem isn't limited to PCM systems, but also applies to any radios that utilize computer mixing.

The Multiplex EVO 9 and EVO 12 systems are really big offenders in this regard. The mixing appears to be done with 440 addressable steps for total possible servo travel.

In other words, you can set maximum servo travel to be +/- 110 %. This covers an angle of about +/- 65 % with a brand new Hobbico CS-59 analog servo on which I attached a protractor. The smallest increment the servo can move is about 0.3 degrees! If you reduce the servo travel limits to +/- 55 % it now only moves 1/2 as far at full stick deflection, yet the minimum increment of movement is still 0.3 degrees, so 1/2 the resolution at the control surface.

I set the Expo to –100 % so that I could command the servo to move in very tiny increments. I found that I could move the stick at least 1-2 mm before the servo would jump 0.3° when the output display showed an output of 1/2 %. Move the stick something like another millimeter, the output display shows 1 % and the servo moves another 0.3 degrees. So in effect I created a huge deadband in the center of the stick travel by selecting maximum Expo!

My conclusion from the above is that the software mixers in the EVO have a minimum addressable resolution of 1/2 % and a range of +/- 110 % corresponding to the full range of servo travel. My guess is that they have limited themselves to this 440 step resolution due to computer processor limitations. I suppose it could also be done to mask hardware limitations with their sticks, as if you select the opposite setting 100 % expo the sticks can only then return to center with about +/- 1-2% servo output accuracy!

These radios may have world-class prices ($512 for the EVO 9), but I don't know how they compare to other maker's high-end products...

HarryC

My Feedback: (1)

You may have measured more about the response, resolution and deadband of the servo rather than of the radio.

ShamelessMale

Still getting used to the EVO. I may be able to adjust the "servo calibration" so I get the whole 440 steps with less servo travel, which would help. The 440 steps is definately the max. Your point about the servo performance is well taken. However, the steps are a digital thing, nothing to do with the servo.

HarryC

My Feedback: (1)

The steps are due entirely to the servo! It is waiting for a certain size of change in the timing pulse before it will react. You could be changing the timing pulse by 1 microsecond at a time, but if the servo has a deadband of 3 microseconds then it is going to jump in 400 steps, which is roughly what you are observing. You are telling us about the steps your servo works to, not the radio.

To make claims about the radio, you need to measure the actual timing pulse with a meter that reads to at least 0.000001 seconds. Using your servo as a meter is not valid because it is probably reading to a minimum of 0.000003 seconds.

H

OhD

Consider this:

We want to control lift linearly with elevator stick movement which is an angle. Lift is a function of angle of attack which is an angle. Angle of attack is a function of control surface angle. With a rotary servo, control surface angle is a function of the angular output of the servo.
Converting a servo output to "linear" will make the system non-linear.

I put linear in quotes because the way it is used to describe a servo implies a straight line output. In the real definition of linear, a rotary servo is linear if the output changes the same amount for a given change in input throughout its operating range.