One down side it that the opto sensors cost about $2.50 I think, and the opto mouse measurement would cost about $5. At what point do you simply make it a servo system.

hehehehe

That ones easy

If your either a priest or supplicant then you refute any technology that is'nt your one true faith, The rest is heresy and must be cast out by any means possible.

If you are an engineer then you choose the established technology that best suits the application within the specified cost constraints and that choice can be backed up by math etc that stands up to verification by any other engineer.

If you are a scientist then you choose a novel technology and method of application that is as close as possible to an engineering solution whilst never having been done quite that way before. (Preferably starting from first principles, and can be verified by another scientist)

If you are an accountant the cheapest technology that you can buy irrespective of whether it can do a job that you don't understand and have no intention of trying.

If your a salesperson the most profitable technology that contributes towards your commission that you can coerce a potential customer into buying irrespective of whether it can do a job that you don't understand and have no intention of trying.

Everyone else tends to choose what ever they have to hand.

Which are you ??

I am a combination of scientist,engineer and everyone else.



aka47

Edited 1 time(s). Last edit at 12/12/2007 03:13PM by Andy Kirby.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

aka47 Wrote:
-------------------------------------------------------
> In reality a more sensible solution is to place
> finely slotted opto pickups at the center of
> travel for all axes and center all jobs on the
> same point.

The problem with this approach is that if something goes out of control and you lose track of where you are (especially if you don't know which side of the center pickup you are on) you can end up running too far and the software won't know when it's sending the extruder head out of the build volume the hard way! The opto endstops on the ends of the axis serve as a safety feature as well as an alignment feature. I don't think you can eliminate them unless you can find something else to fulfill that role.

Kyle

You could use three slots in the opto sensor or three posts. What ever you use as a sensing edge in the middle could be use for both limits as well. That way if you trigger and don't expect to trigger, you know something has gone wrong. The chances of hitting a limit while expecting to cross over are quite small. So it would likely prove to be reliable enough.

... in my Tripod i have small sized stepper-motors with a magnet atached to the backside-shaft and some extra coils for tachometer-sensing ( [bp3.blogger.com] ) but i didn't actually use them - so you can easily control your actual position in coarse ...

Viktor

Edited 2 time(s). Last edit at 12/18/2007 08:53AM by Viktor Dirks.

If you mount your opto sensor on the moving platform (Whichever axis) you only need one for the axis. And an interrupter (Post/flag/sticky out bit) for each of the three stations ie either limit and the middle.

I guess if you want to be really sure and don't want to construct clever rules to determine state you could use two opto sensors mounted on the moving platform and arrange them so that the limit interruptors trigger one only and the middle interrupter (or position checker) triggers only when centered.

All in all the object of the game is error reduction.

If you start off assuming 100% error you initially minimize this by over engineering and then further by having some method of checking that what you want to happen is happening.

100% error checking is a servo system. (However you motivate it, Stepper, Voice Coil, Brushed DC Motor is completely irrelevant)

I have been thinking long and hard about servo systems for a while and have come to the conclusion that more feedback is always good. Getting what you haven't is always more problematic than ignoring what you have.

Most servo systems seem to lack a method for feeding back how much energy it is taking to maintain or attain a given position. My initial design ideas include measuring forward current to give this sort of indication.

It's a bit like weighing things or testing things by hand based on how your muscles feel.

IE force feedback from current consumption.

This is doable with servos because they are proportionally driven against position attaining and maintaining.

I don't think this is doable with steppers as they are not proportionally driven against a point in real space.

If this were doable with a stepper you would be able to tell when you had hit a hard stop or maybe even were loosing steps due to loading/resonance.

Although an opto interrupter (however fabricated) is arguably cheaper and simpler.

Speaking of which opto interrupters and opto interrupter arrays are eminently reprapable......... just add light source and photo diode/transistor.

aka47

Edited 2 time(s). Last edit at 12/13/2007 03:53AM by Andy Kirby.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Hi Andy,

... for another project (the "c't-bot") i manufactured index-plates for position-encoding of the bot which is driven by high-quality DC-gearbox-motors per PWM.

Here some images:
Milling and lasercutting the index-plate:
[www.ctbot.de]

Attached to the wheel:
[www.ctbot.de]

Assembled with the sensor-PCB's:
[www.ctbot.de]

With the same principle you can measure/control your motors too (either servos or steppers)

The c't-bot has 60 pulses per revolution which are measured with a reflex-optocoupler - but with a finer adjusted self-made opto-sensor i can measure much more.

So it shouldn't be a real problem to get 100 or even 200 pulses per revolution with a home-brewed two-LED-two-Phototransistors-sensor or salvage better ones from old printers or optomechanical PC-mices (or by the sensors somewhere in bulk)

For the c't-bot comunity i first made the index-plates with my CNC-mill, but when it went to much work (i mailed over 100 milled index-pairs! to them), then i tried with a small CO2-cutting-laser (in the first image) and then give the job to the laser-cutter completelly ...

Viktor

Viktor

Looks cool. I like bots.

Just for fun here's an idea for 2 opto interrupters in 1. length of the interrupter decides which sensors are interrupted.

Separation distance for the two sensors and angle is determined by the angle of emission of the LED.

This is so obvious I think someone must have done it before somewhere.

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

aka47 Wrote:
-------------------------------------------------------
> If you mount your opto sensor on the moving
> platform (Whichever axis) you only need one for
> the axis. And an interrupter (Post/flag/sticky out
> bit) for each of the three stations ie either
> limit and the middle.

I see what you're saying now - that is a good idea. That way you could install more positional checks wherever you felt you needed them along the axis just by putting up more flags, which is simple and dirt cheap. And by having only one opto sensor per axis you can more easily fit the whole thing on one arduino, which is desirable. Nice! This change should be incorporated into Darwin and hopefully the McWire Repstrap as well - although I don't recall what the official McWire endstop layout is right now.

Kyle

Thanks kyle

Correctly speaking the limit switches should really be something that cut the drive for anything but the reverse direction. (Saves burnout and smash) Ideally without requiring the motion controller to think about it (A hard wired reflex if you like, that saves burnout and smash caused by software crash).

This is very easy with micro switches and servo driven DC motors and correctly speaking should be equally so if you mess with the direction and step signals to the stepper driver IC.

The center position thing is better as a photo interrupter and is in reality a soft interrupt. ie the motion controller needs to take note that something has occurred and will use it where it can but it isn't the end of the world if it is ignored till next time.

I guess a final addendum is worth mentioning here and that is that the center position photo interrupter thing (the sticky out bit) has a measurable and well defined width.

Where the width is known this can be used to auto calibrate the steps per mm without human intervention.

Ie the width of the interrupter sticky out bit can be optically measured in steps (cos it's in the middle) and this figure divided by its width in millimeters (Measured with a micrometer) to give a repeatable calibration reference. In this repsect having a relatively wide 10mm to 20mm + interupter is going to be preferable over a narrow one. The validity of the result is a function of the rounding error from the division etc over the distance measured. ergo the longer the distance measured the less is the error per mm.

If this is followed by a run along the axis from limit to limit (Remember the limit switches/sensors) the length of travel in steps and millimeters (to fractions of) will be discovered along with the exact position of the centering flag relative to everything else.

Again correctly speaking you would want to calibrate limit to limit as mathematically and statistically this reduces likelihood of error (as already mentioned above) but for what we are doing the question is would calibrating on the center flag be sufficient for our needs ???

All of this of course supposes that the axis can initially be driven slow enough to almost guarantee that no steps are lost (in a stepper system) If steps are lost the results are arguably compromised.

With a servo system the above calibration techniques are going to be pretty much spot on.

Yet more musings for what they are worth, sorry guys.

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

I suggested running to the limit switches as that hardware is already part of the design, so no more hardware to solve the problem. Otherwise, I'm with you, put a sensor on the head, and let it watch markings on the guiderail rather than using the endstops. You COULD get away with one sensor, if you were willing to mark the ends, middle, and reference points along the length, with three different values. Say, a single dot for center, (easiest to calibrate against,) a dash every fifth to the edge, (less accurate, but it still tells you if you reach it prematurely,) and perhaps a double dash at either end to indicate the terminus, or even a warning to slow down so you can barely touch the pre-planned end-stops for calibration.

And I know this has been tossed about earlier. I recall the difficulty of printing the ruler strip with a first generation reprap being a prime concern.

I still think steppers would be easier to work with than servos. I think there'd be less fiddly calibration with the things, but I've not worked with either, so I could easily be wrong.

Sean

The dash dot idea is a good one purely for position sensing by a controller that is compos mentis. Using the minimum of Hardware. It does rely rather heavily on the controller monitoring the duration of interruption versus travel distance instructed. (Tweaks to software)

If the controller isn't with it (accumulated error or crashed software) you may smash the assembly into the end stop and or burnout drivers/coils if left. Whether the Darwin & Mendel are over engineered sufficient to withstand this sort of system failure if left unattended for an extended (Long print runs) I don't know.

Some thoughts for what they are worth....

If you are using a servo track (ruled strip, encoded disk on motor) however constructed, you are running a servo system. What ever the method of locomotion (Stepper, DC brush less motor, Universal motor, Voice coil, Hydraulic Ram etc etc etc).

Sorry if I am laboring a point here.

A hard drive head poitioner built using voice coil technology (ie Like a loud speaker) and positioning itself via a written magnetic servo track on one of the head patters is still a servo system. (The clue is in name servo track)

An Oki Microline printer that uses a graduated positioning strip but has a carriage driven by a carriage mounted "Space Motor" Oki speak for a printed circuit stepper motor is still a servo system.

An RC servo which uses a Potentiometer but is driven by a PWM fed DC Brushed motor (Also confusingly enough called a servo motor when used in this type of system) is still a servo system.

A CNC Crane which uses Tachometric output from a 3 Phase AC electric motor for positioning is again still a servo system.

The servo bit sort of indicates that it has continuous real world feedback as to it's actual position therefore also change of position and the positioning device what ever it is is driven against this.


Servo v not Servo is all in the positioning and feedback.....

Servo = Pessimisticaly poitioned ie "I am going to put the tool head here and as I am not going to take it on blind faith that it will get here, I will monitor it every step of the way and either make it so or quit with an error."

Non Servo = Optimisticaly positioned ie "I am gong to put the tool head here and as I don't believe that it is not going to make it here, I won't worry about whether it makes it or not."


So I guess the discussion becomes one of DC Brushless Motor v Stepper Motor.


Interestingly enough a DC brushed/brushless motor, cannot be optimisticaly positioned with any accuracy, precision or resolution at all. It has to be driven in a Servo System manner to work for us at all. Stepper motors can be positioned either way.

The drive electronics for a Stepper motor are more complex than for a DC Brushed Motor (I am purposefully not calling it a servo motor to keep the point already made quite clear)

The drive electronics for a DC brushed motor allow you to drive the motor with just enough power to achieve the move or hold station. The drive for a Stepper does not.

The drive electronics for a DC brushed motor allow you to "Feel" or feedback how much power is required to achieve a move or hold station. A Stepper does not.

The processing (ie cleverness) required to use A DC brushed motor in a servo motor systems is greater than that required to use a Stepper Motor.


Hmm nothing more major springs to mind, anyone got any more comparisons. I guess the more we list the more informed our judgment.decisions can be.


It is worthy of note that Stepper driving is much more simple than it otherwise might be as a not insignificant quantity of the drive cleverness has been moved into hardware ie the controller chip.

If the same were done for a DC Motor and servo track design it would be just as easy to drive yet by virtue of the fact that you can't not servo drive it yield arguably potentially less error.

Question???

Do we have a case here for a programmable logic design for a DC Brushless Servo System controller (or even a pic sub controller) that makes signals equivalent to the Stepper driver controller available to one of the existing axis controllers with the Stepper Controler chip removed.

Remember folk it's your vote that counts, opportunity knocks...

(God that makes me feel old)

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

... it's interesting to look at the development of the driving-methods of printheads.

In my oldest printer (a Brother-HR1-typewheel-printer) the head is moved by a linear-stepper with only a home-switch at the left side.

The next one (a NEC CP6 24-needles-printer) has a big and strong 1,8

Edited 1 time(s). Last edit at 12/18/2007 06:30AM by Viktor Dirks.

Viktor

In the interest of economy I am with you on using what is to hand "The everyone else" approach.

Unfortunately I don't have such stock to hand

Like yourself I have a history of looking inside printers etc and seeing how they worked (I did IT Component level repairs and Field Service for a while, as well working for an instrument manufacturer as a design technician) I guess this has contributed luckily/unluckily to my understanding of how things work.

Something else I just thought of to add to the discussion of optimistic v pessimistic positioning is the question of speed.......

Lets take the case of a system having one axis only and using a method that allows it to move a step at a time and compare optimistic v pessimistic.

In an optimistic drive scenario you don't actually now when the next step has been completed (if at all) and therefore have to wait a worst case epoch (Time period) between each step. Irrespective of how long it actually took to achieve the step.

In a pessimistic drive scenario you can sequence the next step on immediately the first has been completed and you will know exactly when that is.

In comparison then the optimistic system has a maximum speed governed by the worst case condition even when working at light loading, something that is arguably always going to be the case in a system over engineered to avoid loosing steps. the pessimistic system has a maximum speed governed by it's maximum capabilities.

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Hi Andy,

... i'm in a sort of cooperation with ISEL ( [www.iselautomation.de]〈=en&oldlang=de ), a manufacturer of CNC-systems - they actually prefer the use of servos instead of steppers too

And i have a big old 3D-CNC-stepper-system (something like this [www.proma-systro.de] but much bigger, stronger and older) from this company too, which i actually rework for higher accuracy and for milling and 3D-fabbing with different tool-heads ...

With the actual controller i have the possibility to drive steppers until 70Volts/6Amps with step-rates until 10.000 Hz and it have lots of I/O-Pins, so it's very good as CNC-mill, but heavy overdefined for 'only' 3D-printing - but it's no real problem, as i can reduce the output-current and all signals as CLOCK, CW/CCW, Switches and so on are on a common bus and i have all the documentation and much experience with this stuff ...

As this is a perfect controller for cartesian robots as the normal Darwin/Mendel or the McWire too, i borrowed me a second Interface-card, so i can assemble on the fly a second setup with my microstep-drivers.

Now i'm experimenting with driving my Tripod and rethinking some concepts for String-Tripods - here i can play with other methods or drivers ...

In 2008 i think to try with the Arduino and some of my own microcontrollers too, so my way isn't predefined, but it's a bit time-comsumpting -- i can't go so fast, as my family claims most of my day-time and the reprap is my 'night-work' only

Viktor

***
Edit: they seems actually to change their company-structure, so some links may be misleading ...

Edited 2 time(s). Last edit at 12/18/2007 09:02AM by Viktor Dirks.

i just thought i'd jump into this thread.

personally, i agree that servos are the way forward. the reasons they are superior to stepper motors have been laid out numerous times above.

however, as also stated above, servo motors are more complicated in software and in general. the goal of this project is to build a working 3D printer. steppers allow us to solve the positioning and drive problems in a relatively easy fashion that works well.

steppers are great because they allow us to get something up and working with a minimum of effort, and get down to the real task at hand of building a great robot. of course there are better options out there, but the magic of open source is that anyone can take our design, switch the steppers for servo motors, and prove that such a system is better.

bottom line is that now is the time for getting stuff *working* and that polishing the design to be perfect comes after that. steppers work well, and obviously having positioning feedback is something we'll need in the future, but for now we're focusing on getting it out the door.

anyway, this is a nice, lively discussion. cheers!

I'm kind of surprised that we didn't see someone bring up the servo option from the future plans page found here [www.reprap.org]

Hears a direct link to the servo page [reprap.org]

Fairly cheap and reasonably easy way to get the servo option. Also appears someone has made some form of software to make it work.

Hurro Zach

I agree, there are effectively three threads in one, one discussing the where we are now, one discussing technology comparatively and a third to discuss what next.

I struggle to agree that any technology is "Superior" to any other merely more/less optimal a solution to a given problem.

On effort something to consider.

For a software oriented individual with the tools to hand more software over less mechanics and electronics, is the least effort route.

For a mechanically oriented individual with the tools to hand more mechanics over less software and electronics, is the least effort route.

For an electronically oriented individual with the tools to hand more electronics over less mechanics and software, is the least effort route.

I cannot comment for anyone else here but I struggle for the mechanics (no tooling, space or scrap bin to draw from, plus there is no McMaster & Carr in the UK and every engineering supplier considers themselves a specialist with prices to match)

The reprap project and RM/RP in particular addresses a chunk of that shortfall. If I can get there.

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Hey Andy,

Good to hear.

I think the idea solution is one that uses sophisticated electronics + software. The reason being that it only takes 1 or 2 smart people to create the electronic schematics / boards, as well as the software to run them. After that, any number of people can easily replicate what they did.

Mechanical solutions are a bit trickier, since as of right now, everyone has to build it themselves.

Anywho, I'd love to see a side project focused on taking a chip like the ATtiny, a L298, an encoder (like.. austria microsystems 5040), and providing a step/direction interface to a DC motor (with the possibility for serial comms too). that would be fantastic! if someone comes up with a design, breadboards it to test it, and then lays it out as a PCB, the RRRF will make a prototype run for you and send you free copies of the PCB to test and write firmware for.

cheers!

Already part way there zack,

I have put up a worked servo motor example just for fun on one of the other forum threads.

What I am working on at the moment is a universal 1 per axis servo motor design.

The H-bridge power stage will be built from discreet N type MOSFETs and is likely to be driven by a pair of high/low side Gate drivers. (LM5100A/B/C, IR2011, HIP2100 or some such)

The gate drivers do clever things necessary for logic to High sided N type driving and do the necessary voltage translation to enable servo motors up to maybe 48V+ to be driven.

The power output stages will benefit from all 4 transistors being the same type (bulk purchase deals) and potentially mounted on a separate board so that it can be sized to the power of motor to be driven. Mounted on a heat sink, using agiain whatever folk have to hand etc.

What I want to achieve is the ability to drive brushed DC motors (cheap and plentiful) at a range of voltages and powers from 12V upwards. So that folk can use whatever they can get their paws on.

The servo axis board will have limit switch inputs (open contact) together with the ability to accept quadrature positional input (shaft mounted encoder) form a motor mounted encoder either optical or magnetic which ever you fancy.

Oh and last but not least an opto sensor input for centre axis and auto calibration.

Just starting to think about control logic now. Probably going to be programmable. I need to think about this one a bit and work out quite how complex I want it to be. Ideally at least enough to continue using the same microcontrolers as are already used on the per axis boards maybe a C18 (Cos I have the C18 compiler and ICD etc)

I think it should have a base subset of lines/controls that make it look to an existing controller like the stepper controller chip. Maybe with some extra cleverness built in possibly later (32bit 2s compliment signed position counters, centre being 0)

S'about where I am at the nou.

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

cool, well i'll take my suggestions over to that thread. keep in mind that the more compatible it is with the current arduino stepper driver board, the more likely it is that we'll adopt it. if you provide those signals + advanced features, that would be very nice.

For design process, I think more folk will use it if they have participated in the design and implementation.

What's the form with the Wiki for this project (bunch of projects ?)

I think the design work etc would best be done as Wiki Documentation alongside forum discussion. This gives more folk the ability to participate and winds up with a formalized set of doc/s that describe the design such that anyone can do some of it/all of it as part of the development group.

So thread plus Wiki. I don't mind taking ownership of the Wiki bits and working with folk via the forum discussion. (My web site includes Twiki so I am familiar with it already)

What do I need to use the groups Wiki (I don't want to cut across what other folk are working on etc) if anything other than a valid login......(do I have one already, do I just need to go get one, does someone need to grant me one ????)

In reality land being as I am "resting between laptops" and a touch "hand to mouth" whilst I can and will do theoretical Electronic Design Actual Software/Firmware/Programmable Logic etc etc anything that requires expenditure of your earth Pounds, Dollars, Euros & _snip_and_substitute_your_inflicted_currency is problematic.

But I anticipate a lot of progress can be made without resorting to such base pastimes as spending money.

Hope this helps

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Ok guys

[forums.reprap.org]

AKA

A worked Servomotor example

Under the Motors area is the place for contributing to a worked Servo Motor one per Axis Design.

As Linus's Biography suggested.......Just for fun.....

(Is standing on the shoulders of giants fun ? Maybe it's an extreme sport.)

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

we hid wiki registrations due to spammers. i'll message you the link and then once you register, i'll add you to the authors group.

lets put all the docs here: [www.reprap.org]

as for money... this is exactly the type of thing that the RRRF was founded for. if you're willing to do the work on designing PCBs and such, then the RRRF will have prototypes manufactured and if we have enough money at that time, we could also probably buy components as well.

if you're willing to bring the brainpower, the RRRF will bring the $$$. (even if it is not very much)

Here is some interesting reading about PID controllers. Offers pseudo code.

[en.wikipedia.org]

I'm still looking for a source of those magnetic strips, I feel they would be easier to implement then the optical mouse idea. Harder to fake out with dust and such as well.

Hot dang, it was their under my nose all along. I've wanted to put a linear encoder on the platform pushed by the motor for a variety of reason, including making it a servo-ed system. From [en.wikipedia.org]

"Digital calipers contain a linear encoder. A pattern of bars is etched directly on the Printed circuit board in the slider. Under the scale of the caliper another printed circuit board also contains an etched pattern of lines. The combination of these printed circuit boards forms two variable Capacitors. As the slider moves the capacitance changes in a linear fashion and in a repeating pattern. The two capacitances are out of phase. The circuitry built into the slider counts the bars as the slider moves and does a linear interpolation based on the magnitudes of the capacitors to find the precise position of the slider."

I thought they were using embedded magnets and hall sensors not variable caps. This makes the setup much easier to produce. I was barking up the wrong tree all along. Any one know where to find more info about how to make these variable cap on a PCB?

Looks like I should have read the references at the end of the article. Lots of detailed info there about how they work. Woo Who. I think I just found my next project.

Edited 1 time(s). Last edit at 12/21/2007 06:27AM by Jared Harvey.

either that, or we make caliper mounts and use them directly

I've been tempted to use the calipers directly, but this can get a bit expensive. Cheapest I've found with 12" is about $50, If creating the PCB's and your own circuits it should be more like $5.

This is an interesting open source DRO project

[www.shumatech.com]

Basically it's the addition of the digital calipers as used by a digital read out display.

Guys

Sorry to sugest something a little simpler but I just found this whilst having a rummage through Cypress Semi's web site

They do a nearly all in one laser navigation chip aimed at the mouse market (Read cheep) umm
it doesnt need a mouse mat etc either.

It does do X Y but I guess we could use it for one axis anyway and if you wanted use the spare axis to check up on backlash and end float etc.

It is the cyons1001 the datasheet is cyons1001_8.pdf and makes for interesting reading.

In summary this device should be cheep and used correctly can read position straight of the axis rail/slide using non contact laser specle. No printed/etched or any track needed (So no alignment issues).

All the benefits of a directly read linear strip with no linear strip.

Too good to be true huh...

I need to sit down with the spec sheet and work up a comparative spreadsheet converting their quoted resolutions into the metric that we use in Europe but it looks extremely promising.

I have summited a request for a sample from Cypress and am waiting to see if they come back with one.

Using this method of servo positioning should also make any variances in the drive train irrelevant and improve positioning whilst allowing cheaper drive trains to be used. (Taught chain and axis mounted drive sprocket for example).

cheers

aka47

PS Zach I just tried to create the doc you wanted us to do this in but don;t seem to have the relevant perms.

---

Necessity hopefully becomes the absentee parent of successfully invented children.