[Xerxes]

Servo motor construction can be really simple. I attached few photos of my tiny 20W AC servo. It doesn't even have a slotted stator but windings are moulded in resin probably in some kind of lamelled iron tube that isn't visible. The motor has distortion free sinusoidal commutation waveform.

[vroemm]

I want servos! But i sure don't pay the horrible prices for them here in Sweden! :P I think it would be a very fun and interesting project to build a servo!

Very interesting idea.
I would love to see it made possible.

Mean while, scrapyards are a hobby cnc router builders good friends.
I found lots of cheap stuff on scrapyards.

Like recently a 42 volt 7.7 ampere AXEM servo motor.
For almost no money.
I bought that day 10 industial strenght push buttons, one emergency stop push button, big aluminium bar of 6 kilogram, and the AXEM servo motor.
Total price : 22.50 euro.
Like i said, almost no money.

I hope this does not torpedo the idea to build your own servo motor :-)

[JBV]

Hehe nope! I'm a perfectionist and a symmetry freak! I NEED 4 identical servos or none at all Also i want them "repetable", i don't want to have to chase a replacement

I'm thinking of building a hexapod later, then i would need 6 more servos :P

[vroemm]

Hehe nope! I'm a perfectionist and a symmetry freak! I NEED 4 identical servos or none at all

If thats the way you want it :-)
We are very different, i have 4 different servo motors, and no problem with it. :-)

Also i want them "repetable", i don't want to have to chase a replacement

I would adapt the cnc router to fit any suitable servo motor i could get.

Vroemm.

[JBV]

Made a housing & brush design...

100*100*120 mm
80 mm Rotor, 0,5 mm clearance (don't know whats appropriate)
Ten 40*10*6 mm Neodynium magnets...
12 mm radial ball bearings

Comments?

[jans123]

Hehe...

I made a simple (I would use more poles for servo use!) model of a brushed motor to see if the parts will be easy to make.. Not finished yet. Need brushes, bearings and endplates.

More poles? Hmm... Weel you can actually get an indefinite number of "poles" by skipping the rotating iron and let the magnetic field rotate and "drag" a drum of copper around. It only takes very thin walls and if you think copper is to heay go for aluminium, but then the losses will be higher. Of course you will have to fiddle more on the electronics, but you wanted something new or?
:-)
/jan
:cheers:

[JBV]

I was thinking of using the UHU Servo controller So that limits me to brushed servos :P

But perhaps it's possible to build in a (active) commutation circuit in the motor that switches the coils...? Could be interesting

[vacpress]

neato. interesting. i am sooo glad i dont have to try and build that thing... lurk on ebay!

[JBV]

I just mailed for price on the neodynium magnets...

How about not stopping at building the motor, but also make encoders? Would feel cool to have built everything Have anyone done this?

[JBV]

I got the prices for the magnets now! about $25 for 10 NdFeB magnets!

I belive it will be enough for the application, Ferrite magnets need 80 times the volume for the same magnetic flow!

$25 magnets
$10 ball bearings
~$50 materials (housing/rotor/wire)

About $85 for a diy servo... I'll try to make som calculations/simulations to see what i should expect from it

[JBV]

I used the calculations from this
page, i also used the conservative value for magnetic flux of 1!

"The power F from the magnetisized pole working on the rotating magnet is proportional to the magnetic flux B in the air gap and the area of iron A and equals: F = A x 4 B ^ 2"

Magnet area: 4x1 = 4 cm2 * 10 magnets = 40 cm2

F = 40 x 4(1^2) = 160 N

The rotational moment of the motor (80 mm rotor):

M = 160 N x 0,08 m = 12,8 Nm

As i understand the motor give that torque (theoretically) at maximum load and any rpm... Sounds a bit better than my current setup

Power for 12,8 Nm (excluding losses):
500 rpm = 670 W
1000 rpm = 1340 W
1500 rpm = 2010 W

Seems resonable?

[Carel]

I have here an alternative encoder. It's a Sony magnetic encoder. This one sends out A-B signals, and because all the Sony measuring devices are sinoid, and need thus the Sony-interface to read them. So Sony doesn't make them anymore. These are bomb-proof. Basically it is comparable whit a cassette player (the black disk) with an decoder.

Carel

[WilliamD]

I am very interested in this thread and want to contribute! I'm am experimenting first with using brushed DC motors, with an encoder and gecko's. But I would love to build my own motor also. This should be very interesting.

[Carel]

WilliamD: Roughly looking at the replies the Americans wanted to buy the motors (at a low price), and the Europeans want to make their own. You're spoiling our potential market! Just throwing a little oil on the fire .

Carel

[Carel]

OK, it's history night. Here is a stepper motor with an encoder on it's back. The encoder signals are going into a specialised chip. Out of this chip come Phase signals and PWM. The nice thing is, that the chip gives settable, rpm depended Phase Advance. This means that the coil is switched on a little earlier than logically needed, but you win the magnetising time. This set levelled out at 5000 rpm. Due to this all it is not a stepper anymore, but a brushless servo. And you can't miss steps.

Carel

[vroemm]

How about not stopping at building the motor, but also make encoders? Would feel cool to have built everything Have anyone done this?

I think it is very well possible to make your own encoder.
I have looked into it, but not build it yet.

Making a encoder wheel:
Draw the encoder wheel of 50 mm on the computer.
Go to copyshop, let them print it on transparant folie.

This is 40 mm, 500 lines.


Cut a 40 mm disk from a CD.
Glue the transparant folie on the disc.
The 50 mm transparant encoder wheel will stick out 5 mm.
There you can put the led's and light sensors.

Here is what i wrote last year.
I might have changed my mind on some parts. :-)
http://home.wanadoo.nl/vroemm/cnc/encoder/index.html

Here is a picture of the electronics in use in a very early test setup..
It does not have the home made encoder wheel.
It seems to work, not tested extensively.
Perfectionists : Please brace yourself for impact :-)

[vroemm]

Power for 12,8 Nm (excluding losses):
500 rpm = 670 W
1000 rpm = 1340 W
1500 rpm = 2010 W

Seems resonable?

Isn't 12.8 Nm much to strong for a hobby cnc router ?
I though 1 or 2 Nm was allready very good.

[JBV]

It's just how much torque the magnets could handle Nice to have a little overhead!

I think i will try to design the coils for 5 Nm and 1000 rpm (about 650 W @ 80% efficency). With direct drive on my mill it would give 2000 mm / minute on X & Y and 4000 on Z...

[vroemm]

It's just how much torque the magnets could handle Nice to have a little overhead!

I think i will try to design the coils for 5 Nm and 1000 rpm (about 650 W @ 80% efficency).

Nice values :-)

Does 5 Nm mean you can use smaller cheaper magnets ?

How much MAINS power can users pull from their home wall socket ?
4 servo motors + the motor which drives the cutter should stay within the power a wall socket can give.
This is the reallity for most hobbyist.

In the netherlands one mains group can give 16 ampere 220 volt.
Which is 3520 Watt.
But i do not know how much one wall socket can give.

With 650 watt per servo motor the user might need to set a current limiter, to make sure it always stays within the ratings of the mains wall socket.
Or use a lower servo motor voltage, and lose some power.
Then it would be no problem.

[JBV]

We have 380V 20A and 230V 10A here I don't think it would be a problem, i can't belive that i would run all axes and the spindle att full rpm and load at the same time hehe.. Especially not the rotary axis :P

I guess you could use smaller magnets (or ferrite) but i like to have a good margin The magnets is specified to 150 degrees celcius... I wonder if i would need some cooling!? :P