Electricaly what makes a digital servo?

[jamesatwork]

I am familar with conventional servos but what makes a digital servo?
is just internal to the servo or does it impliment a data bus to the recevier, and what are the compatablity issues with analog equipment

Thanks

James

[AndyOne]

Digital servos will take the usual 1-2ms/20ms signal but they will also take the same 1-2ms down to 3ms repetition rate. The rate at which the motor is pulsed is also much higher than 20ms even if the input rate is only 20ms.
I hope this makes sense.
There's nothing particularly "digital" about the input signal it's all in the internal processing.

Andy.

[jamesatwork]

Thanks
I suppose that leaves the question how does the receiver Know if the servo is digital or not to send the signals at 20ms or 3ms

James

[AndyOne]

James,

Receivers always send pulses at 20ms or close to. Gyros on the other hand can send signals at 3ms but these are the ones specified to work only with digital servos try hooking an analogue servo to one of these and you may find it gets hot.

Andy.

[vintage1]

Basically a servo gets an 'update' every 20ms or so - at least unless its coming out of a digital receiver...

If you like the digital servo puts a bit of intelligence in to 'guess' what the servos next update will be. So it can output a stronger signal to the motor as it gets near final position, without risk off overshooting.

'Analog' servos are a lot worse than they should be, but that's the price you pay (or DON'T pay really) for simplicity..

[Gary Warner]

In addition to what has been said, there is another difference in digital servos. Most digitals can have their internal micro controllers addressed to change the way the servo works (even the ones not advertised as such).

What I found:

JR has abandoned the D shaft on the potentiometer. This allows them to assemble the output shaft onto the potentiometer without great care about the pot's centering. Once the servo is assembled, the output shaft is centered and then the servo it "taught" that "that" is the center, as well as other functions.

I come to find out this because I was checking a servo with our driver, and at .98ms the servo ran out of control off to the mechanical limits. I had several of the same model servo and all the others ran ok down to .9ms. When the bad servo was opened, I could see that the pot was not centered when the motor would stop with a 1.5ms center pulse. By using a spline shaft on the pot instead of a D shaft, the servo was running at center with the pot off of center. Once the servo went too far to that end that was offset'ed, the wipers ran off the pot's carbon track.

It seems, in the case of this JR digital servo, that the manufacture can use cheaper pots by getting away from D shafts and simply teach the servo what center is.

[XJet]

In reality there's probably not a lot of price difference between the "guts" of a regular servo and that of a digital these days. Microcontrollers are now so cheap (in bulk) that there's probably less than $2 in it at a cost-level.

This is now starting to flow through to the market with a number of digital offerings out of China coming in at prices that are lower than other non-digital equivalents.

The Vigor VSD2 for example, is a digital standard-sized servo that sells for under $12 with pretty normal specs for such a unit. The non-digital version (VS2) costs $5, just $6 less.

Even the new TowerPro (actually spelt TowardPro on the label) MG996R servos, which are a much improved version of the hideous old MG995, have digital amps and come in at under $12 -- which isn't bad for a metal-geared 10KG/cm servo.

The next big shift in RC gear will be the "digital bus" which will send actual binary values rather than a pulse-width (which is in fact an analog unit) to the servos.

Such a system will allow servos to be either daisy-chained (ie: al run through a single cable) or simply pluged into any arbitrary connector on the receiver.

There are a number of existing microcontroller interface standards (SPI, I2C, etc) that could be used for this purpose with attendant benefits. I am aware of at least one manufacturer currently contemplating such a move.

[rmteo]

Take a look at this:
http://www.openservo.com/

[jamesatwork]

When a bus system is implimented it could have so much potental if it is made 'open' so a device instead of just telling the servo, motor controlers ect what it wants to be doing it can check what it is currently its doing reporting load it is experiancing postions ect back to a controler.

that would give lots of potential for 'avionics' on a model as real aircraft have which could prove useful for less stable craft like helis of some scale models.

[Ralph Walton]

the hideous old MG995, :rolleyes:

I've got a couple of those. Clunky,noisy, but quite a bit of grunt when I test ran them 2 years ago

Went to use them the other day, and both had developed a mysterious illness called "he no workee"

Wonder if I can give them an "open servo" transplant?

[deh6]

A key feature implemented in the so called "digital" servo is the addition of "integral" feedback. In the so called "analog" servos the feedback is "proportional," which means that when there is a load on the servo there is an error between what the position command coming from the transmitter via the receiver and the actual position. The bigger the load, the bigger the error. You can see this by twisting the servo arm with your fingers and watching the increased error and increased current to the servo as well. The amount of error is function of the gain of the feedback loop, so it is tends to be small, but nevertheless important in some situations. (For the r/c systems the "command" is sent as a pulse width, roughly 1-2ms, every 20ms or so.)

By putting a microprocessor in the servo it is feasible to add "integral" along with the "proportional" feedback. When there is a load on the servo any error is accumulated (i.e. accumulation is in effect integration) so the amount of signal applied to the motor keeps increasing. The result is that the error is nearly zero for varying loads. This explanation is greatly simplified, as there are important dynamic aspects (such as servo overshoot) as well as non-linear effects (for example, saturation (limits on the motor), gear backlash), however the overall result is that the performance of a servo with a given motor/gear-train can be improved by going to a more sophisticated feedback scheme.

There is "ton" of stuff on the web about proportional, integral (and derivative) in control loops. For example--
http://en.wikipedia.org/wiki/PID_controller

From this one can see why the digital servo is said to draw more current. The motor drive is increased until the error is zero, whereas in the old servo the current is proportional the error. Certainly if the servo throw is such that the control arm drives up against a hard stop, the digital servo is going to run the current up to max trying to get that error to zero, whereas the analog servo draw is whatever the amount of error dictates (one hears the "buzz").

Whether it is useful is another matter. If a plane is not pulling up fast enough because under the heavy control surface load the error is large, our response is to pull back on the stick, which does exactly the same thing that the "integral" does in the digital servo, namely with accumulated error we increase the motor drive. The digital servo does this faster than we can by eye, so in some situations it is a definited improvement. In others it hardly matters.

BTW, all of the features of the "digital servo could be implemented with analog circuits, but it would be get unwieldly, heavy, expensive. With the advent of rather powerful microprocessors, with costs in the $0.50 range, the additional cost is almost nil...and with all the nifty features that can be advertised they can get away with charging a lot more!

The marketing lingo for r/c has ended using the term "digital" for this. The big advantage comes not from the digital aspect, but the improvement in the basic feedback loop, which a cheap digital processor made practical for r/c.

Beyond the basic servo loop principle, the faster pulse rate for the motor drive has limited benefit. The transmitter controls how fast the servo command is updated, so even if the servo response was instantaneous the overall control response would be limited to the 20 ms or so frame rate. For things such as gyro stabilization in helicopters there is the potential for making use of the increased rate since that can take place without an update from the transmitter.

Oh yes, the voltage for these digital servos is lower than the analog ones since it has to accomodate the microprocessor and some have max ratings around 5.5v (which means in practice they will probably withstand a bit more if everything else, such as temperature, is nominal). IIRC, the analog servos have a capacitor that is rated at 10v.

[XJet]

Oh yes, the voltage for these digital servos is lower than the analog ones since it has to accommodate the microprocessor and some have max ratings around 5.5v (which means in practice they will probably withstand a bit more if everything else, such as temperature, is nominal). IIRC, the analog servos have a capacitor that is rated at 10v.

Not often true -- most of the digital servos run a microcontroller from a regulated 2.8V-3.3V supply which is derived from the 4.8-6.0V (nominal) provided by the flight pack.

Lots of folks now run all their servos on 5-cell NiMH or 2-cell A123 packs, both of which are about 7V off the charger and 6.5v when loaded.