[higgrobot]

Larger servo motors are pretty rare where i live, you can either buy them new for a small fortune or you can spend almost as much trying to import surplus ones. I'm wondering if i could just make bigger ones using what i can get here now?


If i neatly mounted a 500 ppr encoder from a smaller 50w servo motor to the end of a Brushed PM 36V 500W dc motor such as the ones that come out of China on electric scooters, would i have myself a 500w servo motor?

I'd find it easy to pop a hole, fit a seal and mount to the end cover then drill, ream and thread the end of the shaft to hold an extension for the encoder.

All i really need to know is will the average servo drivers work with such a creation or have i completely missed something?

[Al_The_Man]

There has been a few threads here on the differences between DC servo motors and non servo type, this was one of the latest. http://cnczone.com/forums/showthread.php?t=28447
Although you can transfer the encoder to another motor the results may not be up to servo performance.
Al.

[max_imum2000]

i think it will work fine, except for stoping and starting the motor, you could face a gradual stop and not a sudden braking of the motors.

also the current requirment of these motors in my opinon is very hight

it is rated at about 19amp for 1 motor thats about 50amp for the 3 motors.
not considering that it will take much more than that if stalled or when begining its motion.

i think you will need an extra powerful PS.

what you will save on motors you will end up paying for other things.

consider ebay

[Al_The_Man]

The other thing to consider is to check wether or not they are P.M. motors, often, motors made for traction modes of operation are series wound motors for high torque, they should never be operated off-load.
Al.

[mcpltd]

As Al said, make sure they are PM motors, up untill recently, traction motors were field wound motors.
18.3A is about right for a 500W motor rated at 36V, so you are going to need a driver that can give you that current continuous (Not Peak)
As long as the motor is well balanced, has good bearings (Unlikely) you will have no problems running this motor in a servo application.

[higgrobot]

Thanks for all your help, The supplier has smaller versions of these motors that are definately PM as they are being used as wind generators:
http://www.oatleyelectronics.com/motors.html

I may end up trying one of the 100W ones first to evaluate it's performance and if all works out then i may try a motor of a higher wattage.

I found a link where a guy has converted a large 1/3HP PM DC motor to a servo motor:
http://www.truetex.com/servomod.htm
He notes that because his motors have plenty of torque he can run them on the feed screw @ 1:1, this helps the lack of acceleration due to high inertia.

[NC Cams]

Servo motors are pre-designed to be small in diameter (low intertia) so that they can spool up and stop quickly as that's their intended useage environment.

Similarly, they are high torque and high speed potential, again because of their intended useage potential and/or needs.

It is not difficult to get high torque and high rpm BUT to do that with LOW INERTIA/low mass/small diamter is NOT easy nor at minimal cost especially if you want smooth and accurate running. Sort of mutually exclusive goals.

It is not coincidental that servo's are sized and shaped as they are - nor priced accordingly. High speed, high torque, low intertia motors tend to be small in diameter and long in length and not low in cost of a PMDC configuration so that they can be high efficiency, bidirectional rotation and low cost (unless if you live in Australia).

[higgrobot]

I would have concluded that a PMDC motor with it's larger diameter rotor would infact have a greater torque rating per watt simply because of it's mechanical advantage between the magnets and the windings? I'm sure there exist a formula for working all this stuff out but it may be easier to just convert a small PMDC motor into a servo and see how it compares to the real thing. Maybe a project for next year though, I will order some of these motors and keep you guys posted. If the larger motor doesn't turn out to accelerate quick enough for an axis then i can always use it on a spindle or a beer fetching robot

[bobs bots]

Hi Higgs,
These look like Oatley Electronics type SC500 motors (www.oatleye.com) are you sure they are brushless? Looks more like they are permanent magnet type (PM motors have two wires red and black usually) (BLDC motors have 3thick wires and 5 or more thin wires).

I have a couple of the Oatley SC250G (250W geared) motors as drive motors on my robot project. I will be using hall effect pickups on the sprocket teeth as a crude encoder, one pulse ~ 6mm of travel, they will be in a fairly simple closed loop servo control. I'm running them at 12v instead of 24, but easily more speed and torque than I need, (I'm powering them with a 12v/15AH gelcell harvested from a $39.99 Supercheap jumpstarter and I get a free charger), the 10" pneumatic tyres are harvested from a Supercheap $19.99 two wheel trolley.
The disadvantages of these traction motors is the really high inertia which would make them unsuitable for most servo work (as another member has pointed out earlier). However you can still close a positional control loop with them but it will be much slower, perfectly adequate for your robot to fetch your beer and not spill it.
:cheers:

On the plus side , though, these motors have a much tighter coupling than your average DC non-servo motor, This means you will still get rotation and torque out of the motor even as low as 1 volt.

Also the inductance of your windings will probably be higher than a true servo motor.
If you were controlling the motor with a microcontroller, you would probably find it easier than a true servo motor as the PWM switching frequency could be lower, and your control loop response would be lower (e.g. 10Hz).
So for example if you wanted to start from dead stop, rotate 360degrees, then come to a dead stop with 1 degree, all within a 1 second timeframe, then yes this motor could do it. (The start and stopping would be burning way more power than real servo motors too)
Or you could drive a ballscrew directly with this motor, (they have TORQUE!), and implement a reasonable CNC router or plasma cutter, (0.5mm accuracy?).
You might even drive a lathe or milling machine (with a bit of gearing to enhance accuracy), but it would be slower than a true Servo, and not quite as accurate, But 0.1mm may be all you need, and hey if it takes 15mins instead of 1min to do a job you spent 8hrs programming, what the heck!. (Or if it finishes the job before you finish the beer whats the point?). $109 Australian instead of $3000US is a lot of beer in the fridge!.

Cheers, Bob
BTW I'm at Nelson Bay

[higgrobot]

Thanks for your reply bob, I look forward to getting these motors and having a bit of a play with them. I'm convinced the smaller 100 watt motors from oatley with it's smaller diameter rotor will make a good little diy servo motor. I'm planning on trying one of these with my 5A CNCTeknix mini drivers and comparing it to a 48W real servo motor. The larger motor will take a lot more work to drive? I have this strange idea about using a small servo driver to switch some paralleled mosfets, does that sound crazy?

[bobs bots]

Hi Higgs.
Your existing driver may well work with the cheaper motors,

But note there is more than one way to make a 100w motor!
you have a motor rated at 100v and 1amp (e.g. a typical servomotor)
or you could have one rated at 10v and 10amp. (e.g. electric bike motor)

The difference is one has a few turns of thick wire on the rotor , while the other has many turns of thin wire. The parameter that defines this is the "emf constant" i.e how many volts per rpm you get. You also have a current constant in amps/nm or nm/amp? (someone will correct me here no doubt). I.e. this is exactly like a gear ratio, in fact you can scale your effective motor constants by installing a reduction gear. So you need to do a bit of homework here and dig out the data. Otherwise it will be like trying to pedal a bike with the sprockets the wrong way around!.

So you need roughly similar emf constants for the swap to work (somewhere between double and half would be close enough).
And obviously the torque and speed limiting values need to be appropriate.
You also need to be able to superglue your encoder somehow on the back of the shaft. (this may not be as simple as you make it sound) And allow for the inevitable runout i.e by mounting your encoder on some shim stock. (n.b. these motors have incredibly strong magnets, be careful not to chop the tip of your finger off getting it back together).

Having done all that your cheap motor will probably work on your Teknix minimill, although you may have to lower the gain somewhat.
Of course if your emf constants don't match you may need to modify your drive for more amps or more volts. The cheap motor will use more power accelerating and decelerating, but at constant speed not a real lot more than a real servo.
The brushes on a real servo motor are built better, and designed for reversing (unlike your motor), the Oatley motors have 4 brushes just like real servo motors, and have quite low cogging (another desirable feature)

Another aspect you may not have considered is the stiffness of your servo motor combination, i.e. how springy it is. This is critically important in a milling machine where you need to do climb milling (backlash is important too!) , a real servo motor will have lower inertia, which means a higher closed loop gain can be used, which means a higher stiffness will result.
Yet another aspect is the closed loop resonant frequency , which will be higher for a real servo (with its lower inertia and higher stiffness). You may get unlucky and find your resonant frequency is low enough to align with the tooth passing frequency of your cutter -- bad vibrations-- .

Oh and did anyone mention about how servo motors have all the windings glued together so the windings don't break due to fatigue? Probably not a concern as a hobby user wouldn't notice the reduction in operating life.

Cheers BobT
PS Please excuse my misunderstanding about BLDC motors, I jumped into this topic from a BLDC topic and my brain hadn't caught up.

[higgrobot]

The Mini servo drivers are maxed out at 30V 5A, The 48W Servos are plated 22V 2.2A ~2500rpm and the Diy 100W servo motor is rated 24V 6A 2300rpm. I'm unsure why they call it a 100W motor when it would draw 24x6=144W, Hmm, maybe it's it's 69% efficient? Anyway to run this with my mini drivers i'd either have to run it at 20V so it would only draw 5 Amps max or make myself a 5A limited current supply and run it at full 30V. I could use a resistor but that would be wastefull not to mention hot. I'd probably see a higher rpm ~2875 but i'd lack the full torque.

I should be able to fit the encoder without too many hassles, i'm not going to superglue it but thread an extension shaft into the end of the motor shaft, if i then machine the extension afterwards i will reduce runout. Obviously i have to get the motors first before i decide how i'm going to do it, next year i will.

BTW, I'm in Perth where it's Hot and sticky.

[Al_The_Man]

Anyway to run this with my mini drivers i'd either have to run it at 20V so it would only draw 5 Amps max or make myself a 5A limited current supply and run it at full 30V. I could use a resistor but that would be wastefull not to mention hot. I'd probably see a higher rpm ~2875 but i'd lack the full torque.

Torque is a function of current, period.
With PWM amplifiers, the voltage could be between 20 and 30 volts as the mean current will always be limited by the max. or the set current limit of the amplifier.
Al.

[higgrobot]

Yep, I agree, not too sure if the mini drivers actually have a current setting, think they just get to 6.5A then overheat, i have no datasheet for them!

What i should have made clear was if i ran the 24V 6A motors on a current limited power supply of 5A then i would lack the same torque as when they were run at 6A. However because i can also run them at a higher voltage i'd gain a few rpm. Probably feed the mini drivers with something like this:

[bobs bots]

Hi Higgs,
As both the 48W and the 100W have nearly identical emf constants ~100rpm/volt they will be essentially interchangeable.
As Al mentioned above the current required is purely a function of torque required. If your emf constants are identical then your current/torque constants will also be identical. So the 100w motors will draw ~2amps when driving the same torque as the 48w motors would at ~2amps.
(Not forgetting more torque is required to accelerate your heavier motor armature, so less is available to accelerate your load).
Nice circuit diagram, looks like its copied straight out of the natsemi red linear databook!, but this type of supply is not required at all, just use what you have. Current limited supplies are not particularly helpful in servo systems due to the large surge current requirements. Most likely your 5amp driver has a 10amp surge capacity anyway, and If I designed a 5A driver I would use 50Amp MOSFETs, to avoid surprises of the smoky sort.
If your drivers are halfway decent designs, they will shut themselves down at overheat, look to see if you have a sensor of any sort attached to the heatsink. If it doesn't have one, and it really bothers you, just screw a thermal cutout (90C NC) e.g. Farnell 1006844 onto your heatsink.

Best of luck attaching your shaft encoder!, if your motor shaft is alloy rather than low carbon steel, you won't be able to drill and tap it.
Cheers , Bobt

[FPV_GTp]

hi

where in ossie are you ?

nice convo


cheers

[higgrobot]

I'm hiding here in Perth, Still planning on making my own bigger servos just waiting on some stuff to arrive. I have an assortment from 100w to 1000w motors and i also have various types of encoders. I only have some little Servo drivers at the moment so i will be playing with the 100W motors first up, should be fun.

[FPV_GTp]

hi higgrobot

have a look at this website

" hi

has anyone had a look at doing something like what this guy did ??? Converting an Ordinary DC Motor to a Servomotor ??? have a look at his website http://truetex.com/servomod.htm "


post #15 http://www.cnczone.com/forums/showthread.php?t=22723


I posed the same question some where on the forum about making your own servo motors

Hepas of info on the forum just need 3 years to read all the threads

btw u and me im in Melbourne

cheers

[higgrobot]

What was that they say about great minds? Thanks for posting the thread and Good luck with what you are doing.

[bobs bots]

Hi Guys
This website is pretty impressive, thanks for the tip! ( http://truetex.com/servomod.htm )

And if you were a real cheapskate, you could always salvage the optical encoders out of computer mice, drill a hole in the motor the same size as the axle on the encoder, and push it in. Mounting the optical head may be little challenging though! and as bonus you get electronics to convert the counts to an I2C type interface.

Progress on the BudgetBot:
If anyone is interested I have installed quadrature optical encoders on the actual sprocket of one of the Oatley geared motors (i.e. using the sprocket teeth as the encoder wheel) , and ( just last night) closed the loop using an ATMEL AT90s8535 (On an ABC mini board), I've used H bridge drivers of my own design, and the servo loop updates at 1kHz. The resolution is woeful as expected (about 10degrees) , but as robot wheel motors, this equates to about 10mm of movement on the floor (or 1degree of pointing accuracy), which is all you need, (as I'm trying to keep the variables to a 16 bit fixed point format) . The servo response is "chunky" , but stable, in fact it was stable even with several combinations of dumb coding mistakes, and initially a very noisy 5v supply.

Cheers, BobT