Yep its the new technology direct drive with brush-less motor that cost an arm and a leg.

I want to build my own.
Any one have info of where to source these direct drive stuff?
what sort of programming will these need.

If you are planning to shoot video use a fixed gimbal. Forget about direct drive. It won't work. You either use a fixed gimbal or a properly designed and built stabilized gimbal with adequate servos. There is nothing in between. A badly performing gimbal is worse than not having a gimbal.

Originally Posted by seismiccwave

if you are planning to shoot video use a fixed gimbal. Forget about direct drive. It won't work. You either use a fixed gimbal or a properly designed and built stabilized gimbal with adequate servos. There is nothing in between. A badly performing gimbal is worse than not having a gimbal.

+1.

Buy Ecilop, stabilized gimbal and frame, 2 for the price of one.

Hexa. If you want to build you own direct drive gimbal, take a look at piezo motors. They ofter very fine grain control and should be well suited for this.

what about these
http://www.sparkfun.com/products/10551



Description: These small steppers are a great way to get things moving, especially when positioning and repeatability is a concern.

Features:

Stride Angle (degrees) : 7.5
2-Phase
Rated Voltage : 12V
Rated Current : 400mA
3mm Diameter Drive Shaft
4-Wire Cable Attached
In-traction Torque : 100 g/cm

Originally Posted by Estiavilla

Buy Ecilop, stabilized gimbal and frame, 2 for the price of one.

lol thats the easy way out.

Maxon EC45

hmmm 100gm/cm is 1.4 oz-in. to get enough torque, you need pretty large stepper (nema 23 ) and they are heavy. steppers just are not that great for this type of operation t. there is a reason why everyone is using brushless servo motor.

you can also go maxxon brushless like the big guys. but they are not cheap either. 3-400 bucks/axis just for motor/reduction gear and driver. plus you still need some form of controller that can take 1-2ms pwm signal and turn it into direction/speed or direction/pulse . throw in fairly high voltage setup as well. then there is the tuning. (have to do a PID optimization with well balanced load or you won't get the performance)

can it be done? sure. but I just don't find it that attractive at the cost or effort. there is a reason why zenmuse is zenmuse and if you need that type of capability, probably cheaper to just buy zenmuse.

Originally Posted by askman

hmmm 100gm/cm is 1.4 oz-in. to get enough torque, you need pretty large stepper (nema 23 ) and they are heavy. steppers just are not that great for this type of operation t. there is a reason why everyone is using brushless servo motor.

you can also go maxxon brushless like the big guys. but they are not cheap either. 3-400 bucks/axis just for motor/reduction gear and driver. plus you still need some form of controller that can take 1-2ms pwm signal and turn it into direction/speed or direction/pulse . throw in fairly high voltage setup as well. then there is the tuning. (have to do a PID optimization with well balanced load or you won't get the performance)

can it be done? sure. but I just don't find it that attractive at the cost or effort. there is a reason why zenmuse is zenmuse and if you need that type of capability, probably cheaper to just buy zenmuse.

That is enough to turn the OP into a poor man.

Stepers aren't good for that - you need controller (and finding one suitable for flying - PWM input, lightweight) will be hard, they also need dampening (so not really direct drive).

I would say - pick micro ESC with simonk firmware (forward/reverse support), pick brushless motor and build geartrain. Will be cheap, should work even better than pulley system (if built without slop), OMG WTH torque, almost no noise (better than brushed motors - they got sparks and strange noises when running) and what's even better - perfect linearity (compared to servos).

Wanna even cheaper? Go this way (completely homemade, pics and gimbal aren't mine):
http://www.rcgroups.com/forums/showp...45&postcount=1

Originally Posted by ABLomas

... and what's even better - perfect linearity (compared to servos).

What is a typical response of the servos we're using on gimbals? Faster from a stop and then slows at the ramp to full speed?

Originally Posted by seattle_helo

What is a typical response of the servos we're using on gimbals? Faster from a stop and then slows at the ramp to full speed?

Don't know anything about servo response, but typical servo gimbal is implemented with linkages. Linkages convert arm rotational motion to linear. That way, most implementations works like on expo - very fast in middle, then slow response (or even backwards movement, if using > 180° movement angle servos) at full deflections.

subscribed to zenmoose killer.

to get the response you need, you need high speed with high enough torque. servos are nothing more than motors with feedback loop and tunable driver after all. (PID driven) with multicopter firmware, that is not too bad. the trick is making the reduction gearing that is slop free and compact. there is a reason why maxon's inline gear system cost as much or more than their servo motor. there are several layer of issues that has to be overcome. each by itself is capable of being surmounted, so may be that is the approach that should be taken. for my own line, we have decided to push the edge with hobby grade servo, but we also know it won't hit zenmoose level but will get us close enough for lot of people.

to get brushless direct drive, you are talking about 3 main issues. 1) control 2) mechanical gear reduction and 3) tuning/optimization of the unit. doing this while keeping cost reasonable is not easy especially number 2. while we need high speed, we don't need it that high and you need to balance it with torque. optimization of this is not that easy either.

now, openservo project had a pretty good hobby grade servo control board that could accept encoder feedback, but that project is near dead, but could be a base for controller. it had PID loop, so could be tuned as well. the base code is atmel based I believe but C and could serve as base model.