Capi's Gyro 101

[JoelKatz]

That doesn't seem to make sense. An analog servo would then be continuously implementing new corrections, with its PID loop implemented in hardware. So there would be no advantage to digital servos at all other than frame rate. (Which would be no advantage on cyclic.)

Yes, digital servos do overshoot. The point is, they apply a reverse voltage before they overshoot (predictively). It is my understanding that an analog servo will always command the motor clockwise if it is counter-clockwise to its intended position -- thus they cannot apply reverse until they overshoot.

I think part of our disagreement may simply have to do with variation in analog servos. I'm sure manufacturers have done all kinds of tricks to improve performance.

I am 99.9% sure that the major difference between analog and digital servos, and the primary advantage of digital servos, is that if you command a small change in position on a stopped servo, an analog servo cannot command a quick motor movement, as it is responding to a small change and hence sends a small motor signal.

Obviously, on heli cyclic, if I command a slight movement, I want a slight movement *now*. It may have to overcome significant force to make that slight movement, and I expect it to do so promptly. A digital servo can apply a very brief and very extreme pulse followed by a brief and extreme reverse pulse even before it overshoots.

[DominicD]

Without knowing how each manufacture actually implements their servo controller its pointless to discuss semantics. When i have more free time in the summer i will look into doing a servo 101, ill sacrifice a few servos to the test gods.

BTW Do remember a PID controller even in the analog domain doesnt have to be a continuous system, i wouldn't be surprised to see a one-shot type circuit as a difference amplifier.

[JoelKatz]

Yeah, I actually researched it. They seem to fall into many different categories depending on when they were made and how much you paid for them.

The first generation didn't even do PWM. They simply integrated the input signal, compared it to the servo position, then made one of three decisions "too high", "too low", or "in the deadband". If either of the first two, they latched the motor until the next input signal.

Refinements came later in the form of proportional output (PWM-modulated based on how far off you are) and latching of the input analog signal. This would allow the servo to reach a higher top speed without oscillating as it approached its destination.

So it appears that at least the two generations of analog servos were unable to store the signal from frame to frame. In fact, all they could store from frame-to-frame was 'high', 'low', or 'close enough'. This means analog servos had to have a dead band that was at least as far as the motor could travel in one frame.

So that's at least one point for you.

But it is also true that analog servos cannot command a large motor current due to a small error, nor can they apply reverse before they overshoot. They must either respond proportionately to how far off they are or have a good-sized dead-band.

I've seen no designs for analog servos with any kind of memory. They simply respond to where the servo is, regardless of how fast it's moving.

So I still think the primary advantage of a digital servo for cyclic use is that when you command a slight movement, it can make that slight movement with maximum motor force and opposite force prior to overshooting. An analog servo cannot do this or it would oscillate.

It would be very interesting to experiment on some servos. One simple test is to send a position signal just once and see what the servo does.

[DominicD]

You can fill a library with control signal theory ... and they have I'll have to remember to drag your butt into helping when i find the time to do the Servo 101 (or rather 201 at this point). I should be able to scope a few points on the servos which will give us a small idea of what is going on.

[m_a_r_k_y]

Very nice! Looking forward to more of your writeups. Thanks!

[DominicD]

I have another one on ESCs, but i am trying to find the time and energy to do one on servos later this summer.

[ashkanr]

Very informative...Thank you!

I still can't see the "fork" part in the first magnified image....can you explain that?

[ace101]

did you find out why your choppre is spinning?

[skyzoid]

Aha...

[billbill242000]

Originally Posted by HFG

I have another one on ESCs, but i am trying to find the time and energy to do one on servos later this summer.

Summer is almost up...where is the servo article???

[DominicD]

Well normally i have 4months of slow time in the summer that i get so bored i write these things, this year i have maybe a month. It doesnt look like i will be TA'ing this coming term so i might get around to doing it then.

ashkanr i will be at the London Funfly on the 22nd, come ask me in person

[colin hoover]

dang so it aint magic

[flyoz]

Yep I spotted a couple of mistakes you typed Marry Go Round rather than Merry Go Round apart from that I didnt understand a word ...Doh!

[jockstrap]

Great stuff....

[arfwrecker]

I like, I have, It works.

[Laurens]

Thanks for clarifying some stuff Capi, I'm working on a paper about gyro stuff.

[vk3cq]

Hi Capi, I love the tech write up and as an elec tech myself I managed to understand it!
Maybe you can help with what I don't understand...
I have a Belt CP V2 and on the radio there's a switch that controls the gyro, everyone says this should be up. I can't seem to understand what each position of the switch actually does. I have been doing all my learning with the switch up. (I am a beginner with only 12 hours airtime)
But there has always been a bit of drift in yaw that no amount of adjusting servo position will cure.
Recently I tried flying with that switch down. I'm not sure what the names of the positions are but am guessing down is head hold?
Anyhow there doesn't seem to be a lot of difference either way, except when I tried it switched down I had to re-trim the servo position to get the yaw to stay still in hover.
Now I'm only guessing, but with the switch up, the gyro compensates for rotor torque and with it down it doesn't.

Hope you can explain. Along with reasons if possible.

[DominicD]

Sorry not a familiar with the RTF type helis, honesty don't spend too much time or energy on cheap gyros. The best thing you can spend money on is a good gyro, it will help you regardless of your skill level.

[Trevor McPherson]

I was going to do a PM, but not allowed. HFG has done an excellent job, but I'd argue that his explanation of Coriolis is tecnically wrong. Coriolis is a real force, the lateral force a "person" would feel as they walked radially outward on a rotating disc. The actual force needed for the person to walk along a radial line marked on that disc would be twice that force. Gyro's are an example of consrvation of momentum and this is done by adding mometum vectors. Each vector lies in the direction of an angular velocity.

[MikelG]

I kind of want to piggy back off of Trevor in this, but on a different point. For the MEMS ring, please explain how a rotation causes the nodes and anti-nodes to shift position by 45 degrees (why 45 degrees? and is it ALWAYS 45 degrees?). I have a pretty solid understanding of dynamics, but that one is perplexing me... My understanding tells me that the disk would behave a little differently than what you're describing, but I could simply be missing something or not understanding, hence my request for a more detailed explanation.