diy cordless drill servo

[domonic_cnc_guy]

hi all, recently i have managed to get the x axis on my hm-30 moving back and forth in 0.1mm increments, i have used a cordless drill motor and gearbox
geared down 3:1 with a 10 count encoder disk cut out of aluminium,
(see atached file)


1, has anyone else done anything like this before? or am i the only crazy one to even think of it or try it

[Xerxes]

Drill DC motors are not very good as servos. The topic has been discussed several times here and you may find explanations by doing some search work.

Also encoder should be coupled directly to motor shaft for stability and should have higher resolution for decent performance (at least few hundred CPR).

If you find a good DC motor and a real encoder, then diy-servo can be very successful.

[domonic_cnc_guy]

i did the searches on the topic, and there isn't much out there as far as other people trying it out

so far now i have both the x and y axis's moving , when the ir photo diodes aren't being flooded by IR flooding in through the window in my garage movement is actaully quite good however yes i could probably use a higher resolution encoder i think i might make a pair of those first

my other problem is money i don't have the hundreds upon hundreds of dollars to spend on decent servo and off the shelf encoders

[vger]

If there is a computer scrap yard in your area, check em out for big old dot matrix printers. Some of them have servo motors with high count encoders attached for head movement.

Steve

[domonic_cnc_guy]

so far though i've had some good success getting the desired 0.1mm resolution on both axis's i've already been able to millout holels for bearings , the problem now is that we're not talking a high feedrate 0.2 to 0.3mm/sec which equates to 12mm a minute of less than 1/2 an inch per minute, any faster and i start loosing repeatability,

i've got a few high count encoders scavenged from various printers and
trouble is with the cordless drill motors there is a large amount of slop in the gearbox so when you move the table 1 step the tension in the belt and slop in the gearbox will allow the encoder to actaully go backwards enough to give you a false reading, the higher count the encoder the worse it is, hence why i'm using crude 10 count encoders and not the high res ones, i actaully have to have the motors turning for about 10ms after the encoders pass the transition from 0 to 1 just to make up for it,

i do have another plan to get around this and get a higher more reasonable feedrate out of it, and that is to mount a dro and interface them to my controller which shouldn't take too long, that will also eliminaate the need to software compensation of backlash in the leadscrews,

the question is whats a reasonable price for a 800mm dro and 250mm dro and 200mm dro?

[Larken]

Actually a cordless screwdriver motor and gearhead makes a great cheap robotic arm actuator.
With an encoder on the end as long as you have weight on it to get rid of the backlash, it will work fine with a dc servo drive.
I do a lot of work with tech in schools and have seen some great projects using them, from robotic arms to battle bots.

[domonic_cnc_guy]

im not quite happy with the performance of my optical encoders, while they are fairly reliable they aren't really reliable enough,
and i've stil got the problem of backlash which because of the nature of these encoders and being only 10 cpr i still have to deal with,

so i'm now looking at alternatives and more in the direction of a digital readout mounted to the x y table and the quil

so far though and looking at the quotes i'm not entirely happy payingout over $600 for a capacitive digital readout that is made in china for peanuts and imported, the same can be said for any of the locally availble lvdt units and glass scale readouts,

and the $600 odd is only for 1 axis still got 2 more,

so it's going to be a diy job and i'm looking at a diy lvdt unit,
after speaking to a few engineers about it while getting these quotes the diy LVDT option doesn't actaully seem all that un-realistic, really when it boils down to all that is needed is an exciting coil ,a secondary coil and up to a 16bit adc plus a few other little components to smooth things out


so far by virtue of google i can only find 1 example of a diy lvdt and the ressults were pretty good with that one

[Al_The_Man]

If you ever come up with, or come across a good diy LVDT transducer, I would be interested in the details.
Al.

[vger]

Al,
Here are a couple of links to datasheets for some chips that might be of help.
http://www.analog.com/static/importe...eets/AD598.pdf
http://www.analog.com/static/importe...eets/AD698.pdf

Somewhere in storage I have a couple of lvdt's that have about 4" stroke, and I have about a dozen "Swiss Precision" LVDT's that have 1/4" stroke. How long a stroke length are you thinking of making?

Steve

[Al_The_Man]

I was trying to come up with a cheaper alternative similar to Heidenhain ERA series, they use what they call a AURODUR steel scale tape, fastened to the outside circumference of hollow lathe spindle etc.
Al.

[Zoidberg]

What about a DIY version of this?

[domonic_cnc_guy]

looking at the design and basic operation of an lvdt, it's a remarkably simple device i've tested ,a simple miniture one based on some winding wire and a pen barrel, i used a piece of threaded rod as the core, so far produced a poor result though, not that i could play with it too long before the wife went crook on me about dismantling her pen!

i energisied it with 5v pulsed at 1khz , 100 winds on the primary coil with another 100 on the secondary, even with the core fully inserted at most i got 1v out of the secondary, i think this might as been atributed to the huge distance between the coils and the core, and probably my power supply

so i'm off to the hardware place, to get the proper tubing and core of the proper size and if i can more winding wire