Shop OBEX P1 Docs P2 Docs Learn Events
LVDT (Linear Variable Differential Transformer) — Parallax Forums

LVDT (Linear Variable Differential Transformer)

parskoparsko Posts: 501
edited 2008-10-15 22:19 in Propeller 1
Hi all!

I thought I would start a new post for this subject specifically rather than hijacking other threads. I will dedicate this to the homebrew LVDT makers out there that enjoy doing fun, but somewhat unpractical things.

First, a list of links:
I'm doing this so I can measure the linear displacement of a vehicle suspension. Plus, I need to be distracted by the fun proposed by inductors. Inductors have fascinated me for a couple years now, and this seems like another good project to play around and learn them.

The style I'm going for is as follows:

lvdt_how.gif

First, I'm going to start small, with a proto. My primary and secondaries will be 1" long, measuring a pseudo displacement of around 2 inches.

I'm going to leave this first post at that, and write a follow-up to ask my actual question, so this post isn't wicked long.

-Parsko

Edit 1: Added link to Beau's Joystick

Post Edited (parsko) : 9/29/2008 4:51:40 PM GMT

Comments

  • parskoparsko Posts: 501
    edited 2008-09-29 15:51
    I don't want to go into too much detail in this specific post. But, I would like to ask Beau a follow-up question from another forum post from the link list above.
    Beau Schwabe (Parallax) said...
    "Does this AC wave go positive and negative (aka, -10V to +10V)" - Yes

    "(All values chosen arbitrarily)" - For L and C yes, but the 100K resistors were chosen to limit the current to within a safe range for the internal I/O diodes to safely handle.
    "You push the output through a diode prior to any further circuitry. If the output at B is (+/-10V), does this diode chop off the (0V to -10V) part of the output?" - Yes

    "You then "clip" the circuit with the voltage divider..." - The internal diodes that are built into the I/O actually do the clipping.

    "...When you "read" this signal, are you "reading" the 1.65V threshold?" - Yes, but through the function of a Sigma delta ADC.<SPAN style="mso-spacerun: yes"> But remember, since we are cutting [noparse][[/noparse]/color]half of the wave out (the 0V to -10V) not only does the ADC measure magnitude, but now because of the nature of the Sigma-Delta ADC, the width of the pulse also becomes a factoring component that just happens to be beneficial to this application.
    parsko said...
    Thanks Beau! That gave me much to think about. I still need to think about the clipping thing more, but I think experimenting will yield me answers.

    One more question:

    What kind of capacitor is safe to use for an LC circuit?

    The circuit I'll be building will look for a low resonant frequency (~1-10kHz). Given the dimensions of the coil I'll be winding, the math calls for a 1uF capacitor. These seem to be harder to find in the non-polarized (metal-film/ceramic disc/etc...) style, but easy to find in a polarized (electrolytic) style.

    My instinct tells me that I need a non-polarized capacitor, is this true?

    I was planning on using:
    THIS ONE

    Thanks again,

    -Luke
    Beau Schwabe (Parallax) said...
    "What kind of capacitor is safe to use for an LC circuit?" - safe? I think this would depend on voltages involved... something Non-polar for sure.


    That RS cap should be fine.... remember though your 'Q' factor goes way down with higher L and C values and especially if your Coil has an Iron core. With the lower Q, you probably will not get a great deal of gain. Do you have a specific frequency in mind that you are looking at?


    A 250uH coil is not that large for 10kHz, but 25mH coils can start to get bulky for 1kHz. Is there a way that you can hetero-dyne and use a higher detection frequency with a smaller Coil and Capacitor? This would also give you a higher 'Q'.

    Beau,

    From what I have read, LVDT's are normally excited by a 1-10kHz wave. After thinking about your replies over the weekend, I have concluded that it shouldn't make a difference. It's relative. I think I should be able to use any frequency I want, since I'm measuring the amplitude of the Voltage output in the secondaries.

    With that said, I should be able to go with a value for L and C that are lower so as to get a better Q factor. Though, I still wonder wheather that actually matters in a hardware setup such as this??? I don't thing it does due to the distance being so small. From my reading, Q only matters when you are trying to transmit data over larger distances (aka RFID). Can you confirm this?

    Should that be true, I will certainly redesign my coil to accomodate different, more readily available capacitors. Right now, I'm putting together a coil winder, so I can think about the coil design a bit longer.

    BTW, my coil winder will be similar to one Paul made a while back that will use a servo and hall sensor to count turns. I'm making the assembly out of an old floppy drive (recycled). It has forced me to learn servo's and hall effect, which so far aren't too tough to learn...

    -Luke
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2008-09-29 16:16
    parsko,
    ·
    "From what I have read, LVDT's are normally excited by a 1-10kHz wave. After thinking about your replies over the weekend, I have concluded that it shouldn't make a difference. It's relative." - This is true if your LVDT does not exploit it's own self resonant frequency.· At the resonant frequency you will have a heightened sensitivity if for no other·reason than your Slew rate dramatically changes.
    ·
    "With that said, I should be able to go with a value for L and C that are lower so as to get a better Q factor. Though, I still wonder wheather that actually matters in a hardware setup such as this??? I don't thing it does due to the distance being so small. From my reading, Q only matters when you are trying to transmit data over larger distances (aka RFID). Can you confirm this?" -·Since you are not using a particular resonant frequency, the Q really doesn't apply here.· The term 'Q' refers to the quality factor.· When referenced to transmitting or receiving, the Higher the 'Q' the better the selectivity and the greater the discrimination between side-band signals.· This can be applied to both transmitting and receiving and is not confined to RF or radio... it can be IR, ultrasonic, etc.
    ·
    "Should that be true, I will certainly redesign my coil to accommodate different, more readily available capacitors. Right now, I'm putting together a coil winder, so I can think about the coil design a bit longer." -Again, since you are not using a particular resonant frequency, the coil itself will have its own parasitic capacitance that will act in parallel with the coil·(combined with the L value this is what is referred to as the self-resonance) no need to use a capacitor in this case.
    ·
    ·
    ·
    ·
    ·
    Not to distract you from your current LVDT design, you might find this link below interesting.· This application on the other hand does exploit·self resonant properties of the sensor coil to produce an LVDT without the need for a primary coil.

    http://forums.parallax.com/showthread.php?p=598919




    ·

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.

    Post Edited (Beau Schwabe (Parallax)) : 9/29/2008 4:22:38 PM GMT
  • parskoparsko Posts: 501
    edited 2008-09-29 16:47
    Beau,

    Again, thanks for the reply. And, yes, I forgot to include that link in my list. I'll add it, as I reference it regularly.

    So, to summarize, there is no need for a capacitor?

    Would it hurt to use it anyway?

    -Luke
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2008-09-30 03:38
    Luke,

    "So, to summarize, there is no need for a capacitor?" - "Would it hurt to use it anyway?" - in answer to both questions, probably not. It could depend on what the rest of your circuit looks like.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.
  • parskoparsko Posts: 501
    edited 2008-09-30 15:37
    I thought you would say something like that. I have thought it would be a good idea to include a capacitor anyway. I think I'll design it with a resonant frequency of around 100kHz. This makes the design a bit more friendly to what we might have in our box of parts.

    If I were to tune my primary to 100KHz, would it be a good idea to also design the secondaries (which are two inductors in series, whose inductance will ADD together) to also resonate at that frequency? Would this help, or just be overkill?

    I figure, by doing this, I might get a better response (I think you called it slew rate, or gain) which should allow me to get more accuracy out of my displacement measurements. Tuning the whole thing to one frequency should also help me avoid any issues that I might have with noise, right?

    Anyway, I made good progress last night with the coil winder (modified my servo to be continuous, figured out that code, and got the hall sensor working). I'm hoping to wind my first coil before I go to bed tonight. I tell ya, I haven't done much in SPIN lately, and I forgot how quick and easy it is!!!

    -Luke
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2008-09-30 17:47
    parsko,

    Tune your secondaries to 100kHz, but adjust the frequency going to the primary so that the output of each secondary is about half of where it would normally peak.· Make sure you do this with your "metal·slug" in the center position.· This way you can capture the addition or subtraction of metal from the influence of the "metal slug" ... see the link that you provided to·"Beau's Inductive Joystick" for an explanation as to why you want a midway "de-tuned" frequency to the primary coils.

    Note: adjustable caps on the secondary coils·is a must here.


    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.
  • parskoparsko Posts: 501
    edited 2008-09-30 20:04
    Beau Schwabe (Parallax) said...
    Note: adjustable caps on the secondary coils is a must here.

    Grumble Grumble!!! <==== Meaning, no gots!!!

    I'd like to get away with putting a header there, since I have a box of caps that I can swap in/out pretty quick. I assume that would work too.

    Otherwise, I understand, for the most part, what you're talking about. I certainly understand what you mean by tuning it to fall in the middle. I just have never had experience in doing so.

    Also, I have to apologize. Today was the first time I actually opened the Joystick object. It has much more info in it than your forum post does. It cleared a bunch of stuff up for me. Specifically, the Vdd/2 suggestion to keep the signal above Vss. I've been wondering how that was done!

    ======================
    Beau Schwabe (Parallax) said...
    Tune your secondaries to 100kHz, but adjust the frequency going to the primary so that the output of each secondary is about half of where it would normally peak.

    By "tune your secondaries", I assume you mean, the L and C values such that they resonate at 100kHz (for instance, L=293uH C=10,000pF => ~92948Hz).

    Or do you suggest tuning them all to work at 100kHz (where they would all theoretically peak), then DE-tune the secondary via adjusting the secondary's capacitor value (in your case the adjustable cap)? For instance, if the secondary's peak resonance worked with a 10,000pF cap, try swapping out capacitor's with values higher (or lower) than 10,000pF, maybe 9,000pF.

    Thanks again,

    -Parsko
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2008-10-01 03:28
    "...Or do you suggest tuning them all to work at 100kHz (where they would all theoretically peak)" - Yes, that's correct , but just for the secondary coils

    "...then DE-tune the secondary via adjusting the secondary's capacitor value (in your case the adjustable cap)? For instance, if the secondary's peak resonance worked with a 10,000pF cap, try swapping out capacitor's with values higher (or lower) than 10,000pF, maybe 9,000pF..." - No, DE-tune by altering the input frequency to the primary coil until the signal no longer peaks, or peaks to about half AFTER both coils have been tuned to peak at the same frequency. <-Make sure the metal-slug is in the center position when you do this.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.

    Post Edited (Beau Schwabe (Parallax)) : 10/1/2008 3:38:29 AM GMT
  • parskoparsko Posts: 501
    edited 2008-10-07 14:35
    Finally turned my first coils last night using my coil winder (which needs a bit of tweaking).

    I've thought about this before, but never bothered to ask...

    Does it matter which direction to wind the coils?

    My instinct lights a big sign saying "YES!!!". I will assume I should wind all coils in the same direction around my tube. Meaning, I will wind them all away from me, over the top. If you were looking down the tube (holding it in my left, theoretically winding with my right), the direction would be clockwise.

    So, all would be wound in the correct direction. But that leaves me with another question:

    Should I wind them all from left to right?

    My instinct tells me that I should wind the middle (primary) and the right (secondary) from left to right. The left (secondary) should be wound from right to left (aka, from the middle towards the end)???

    I'm pretty sure this will affect the resultant voltage output from that particular winding. Each secondary will have two wires dangling off, and will produce a voltage across them. The two secondaries should be tied to one another, as seen in the picture at the top of this post. If I find that I would it incorrectly, I think I could simply swap which wires are hooked up between the two secondaries, right?

    Thanks,

    -Parsko
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2008-10-07 21:31
    parsko,

    Think about your two statements...

    "...Does it matter which direction to wind the coils?..."

    "...If I find that I·wound it incorrectly, I think I could simply swap which wires are hooked up between the two secondaries, right?..."

    The two statements go hand in hand but if you do one, you can't do the other.

    You could either alter the winding direction OR swap the wires.


    Probably the most important thing is even winding distribution.· If you have a "hot" spot by spending too much time winding in one place, then the results can be thrown off.· Speaking from experience, this·can easily happen if you are winding coils by hand.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.
  • parskoparsko Posts: 501
    edited 2008-10-08 15:39
    Below, you can see my first two attempts. Granted, the first needed a serious code change, and addition of a keyboard to provide input for starting and stopping the servo. The second was okay, but was me guiding the wire onto the form. I thought that I need to add another servo attached to a threaded rod and nut to guide the wire more accurately so that I don't get grooves in the winding. I figured that the more perfect the winding (aka your comment about winding distribution), the better my performance (and repeatability!!!)

    I guess my instincts are pretty good. I'll follow them. Thanks for the guidance thus far. I'll update this when I get something better wound.

    BTW, the second attempt has two layers, then things went awry around where the hot glue holds things from unwinding.

    -Luke

    PS - Is there a way to resize pictures in this forum, so the width narrower, by simply adding some additional code to the "img] /img]" stuff?

    EDIT: Resized the picture to be more friendly to my browser...

    DSC00263c.jpg

    Post Edited (parsko) : 10/9/2008 2:43:02 PM GMT
  • Brian LBrian L Posts: 60
    edited 2008-10-09 00:12
    I also have an interest in making LVDT's and have been trying to understand this thread, but I'm a bit confused by the last three comments about the direction of the windings. If the windings are to have more than one layer, don't you end up with windings going in both directions no matter what? Maybe I'm wrong, but I was under the impression that LVDT's can be made with windings several layers deep. I assumed that what would matter was the CW or CCW direction of the windings, and not whether you go left to right, or right to left when you start since multiple layers gives you windings going in both directions anyway. And am I understanding correctly that the CW or CCW direction needs to be the same for all three coils?

    I'm also working on the mechanical design for a winding machine of my own and would be very interested in seeing pictures of yours parsko. I used to be a tool & die maker (which pretty much means I'm an expert machinist) and I have access to all the machine shop equipment you could ask for. So for me, doing a really professional job on the mechanical end is the easy part. (Now I design industrial tools for a living.) Perhaps we could collaborate on a really nice machine. I was planning to use the my PPDB as the controller for mine.

    I was also planning to use a lath to machine my bobbins out of plastic rods & make them very precise.

    Post Edited (Brian L) : 10/9/2008 12:17:57 AM GMT
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2008-10-09 05:32
    Brian L,

    "If the windings are to have more than one layer, don't you end up with windings going in both directions no matter what?" - Yes, that is correct. What we are referring to is the Phase or polarity of the coils with relation to one another. There are two ways to align the proper phase polarity... 1) reverse the winding direction with respect to each secondary coil ...or... 2) Swap the output terminals of one of the coils.

    It is also very important to have an even winding distribution throughout the entire coil.


    I don't think that you need to re-invent the wheel though as far as a winder..... A Filament winding technique used in many fishing reel mechanisms would be perfectly fine producing a pattern very similar to what you see on spools of twine...

    www.kprogroup.com/gifs/twins.jpg
    www.automation.siemens.com/mc/mediadb/inplace/textiles/en/b3be444b-15b7-4209-9668-c2585fcca3bb/banner-filamentwickler.gif

    ...or even a straight winder would work as well...

    www.tesla-coil.com/images/Coil2.JPG

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.

    Post Edited (Beau Schwabe (Parallax)) : 10/9/2008 5:40:34 AM GMT
  • parskoparsko Posts: 501
    edited 2008-10-09 14:42
    Brian,

    Thanks for your enthusiasm! Yes, we can certainly collaborate. I'm in this for fun, not money (doing this stuff blows away watching "must see TV").

    Last night I made the executive decision to disassemble my winder, and start fresh. Today, when I get home, I'm going to grab my son and head off to the Depot for some quality time in the hardware section to grab a few more things. I have decided to move the operation to a 1/4" plate of aluminum I have laying around. I'm using aluminum angle stock from the depot as well to hold everything together. My problem with my previous setup was that the whole thing was too compliant. I need something more solid.

    I found a nice linear bearing that will give me about 7inches of travel for the wire guide. I wish I could have found something that would give me 12" of travel, but we'll work our way up to that in time.

    Regarding the direction of windings. I think you and Beau have it right. Not sure what I was thinking when I asked. It boils down to wanting the windings to all go in the same direction around your form. The left to right thing shouldn't matter, since the inductor will create a field regardless. The field direction is a matter of rxf (or the same formula as torque). You can follow the "right hand rule", if you are familiar.

    I will gladly share pictures of my mechanism, once it matures a bit. I noted to myself that I didn't take any pictures of the first Proto winder, as it was that bad. This next one will be many times better, and worthy of publishing to this thread.

    I like your idea about turning precise bobbins. But, I don't know if it is necessary. My final design will likely use a tube of teflon that I will wind the coils around, and will allow the ferrous rod a nice frictionless surface to slide in. The killer thing is the spacers between the windings. I made my proto ones out of old credit cards. It worked quite well, but needs a bit of super glue to hold them in place (on my list for today). I used a 1/4" section of my brass rod (which is 0.014" wall thickness) to punch a hole through the card, making a nice perfect sized hole. But cutting them out of the card wasn't so easy. This is where a lathe would have helped. A 1/16" to 1/8" spacer made of Delrin or teflon with the correct ID (0.25" in my case) turned on a lather would be awesome. I'm sure they exist in McMaster Carr, but I haven't really put much effort into finding them.

    This post is long enough as it is, so I'll leave it at that.

    -Parsko
  • Graham StablerGraham Stabler Posts: 2,510
    edited 2008-10-09 16:01
    "perfect wound" coils might be best but you need to lay alternate layers in the grooves as it were so the cross-section almost looks honeycombed. I've done quite a bit of coil winding with home made winders for micro RC actuators, compared to winding with 0.02mm wire you have things easy so good look.

    Cheers,

    Graham
  • Brian LBrian L Posts: 60
    edited 2008-10-09 20:25
    Parsko, sounds good. I have a lot of work to do right now, but will get back with more information later tonight or tomorrow. For me, if I set up a milling machine or a lath to make one part of a winding machine, going ahead and making two of each part it really doesn't change the effort much at all. I have good sources for metal and bearings and such. Programming a prop board (I also have a Basic Stamp PDB too if you were going that way) to control the servo and the motor is the end of it I was hoping you might be better at. Maybe we could both end up with a nice machine and some good controls for it. I can provide nice CAD drawings in both 2D and 3D to publish here for all the mechanical stuff. I'm wanting to have a machine good enough to make many copies of identical professional quality LVDT's for a work related project we have going on here, so I was planning to go for a really well built machine, but what you're doing right now is probably good preliminary work.

    PM me if you like.


    Graham,
    You're right. The coils I want to make are going to very tiny, about like the actuators you mentioned, so the machine will have to operate very precisely.
  • Beau SchwabeBeau Schwabe Posts: 6,566
    edited 2008-10-09 21:44
    Just for a winder idea, why wouldn't a couple of hobby servo's work?· One continuous rotation servo, and another standard servo.· The continuous rotation servo would also need some sort of feedback line to indicate·position and or # of rotations, while the standard servo provides what ever winding pattern you program it to do.· I was just thinking that using servo's would minimize the rest of your hardware·requirements.

    http://www.parallax.com/Store/Accessories/MotorServos/tabid/163/CategoryID/57/List/0/Level/a/ProductID/102/Default.aspx?SortField=ProductName%2cProductName

    http://www.parallax.com/Store/Accessories/MotorServos/tabid/163/CategoryID/57/List/0/Level/a/ProductID/101/Default.aspx?SortField=ProductName%2cProductName



    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Beau Schwabe

    IC Layout Engineer
    Parallax, Inc.
  • parskoparsko Posts: 501
    edited 2008-10-10 13:25
    Beau,

    That's exactly what I have going on. Except that I'm using two continuous servos. One is hooked up (hot glued) to the form, and has a Hall Affect sensor looking at 4 screws screwed into the horn. The second will turn an 8-32 brass threaded rod (no feedback) which will move the wire "guide" so that any pattern can be obtained. My ratio will be about 3:1 (which I will experiment with to get just right). So, the form will turn 3 times to the brass rod's 1 (my wire is about 0.25mm). I will simply start both servos at the same time to keep the timing.

    I obtained a really nice linear bearing (which could be substituted with a Home Depot/Lowes drawer guide bearing) for the guide to mount to. On this linear bearing will also be my pretension mechanism so that the wire source can be loosely mounted (my proto design pretensioned the source spool itself, which has some disadvantages). I'm thinking of running the wire through some sort of foam which would be compressed between two plates. I'd love to hear an idea how I could pretension the wire so that it's tight around the form. Also, I'd eventually like to add a glue sponge to this guide, so that the coil will be glued together once wound.

    What I found in my proto was that my frame was compromised bigtime. So, I moved it to a plate. I got most of it done last night, and should have all the hardware done in my next session. Then it will be a matter of programming. I'd like to add some smarts to it, where I ask for the number of turns, number of layers, and wire OD. I'd also like to add a "pause" button so that I can inspect should a problem occur.

    I'll get a picture or two later tonight/this weekend once the hardware is done. We are on the same page, but Brian and I are trying to (I think) do a little better than what the average hobbiest might do.

    Also, my setup change is such that I will be able to wind just about any length coil less than 7 inches. That was the major motivation for change between the old frame (3 inch coil length max), and the new frame (7 inch coil length max). I have enough space to make coils that are 12 inches, but I'd need to get a longer linear bearing and threaded rod. Baby steps...

    So, don't worry, Beau, I've got some Parallax parts incorporated into my winder (two servos and a malexis)!!! No need for the sales pitch. tongue.giftongue.giftongue.gif

    -Parsko
  • Brian LBrian L Posts: 60
    edited 2008-10-10 19:18
    Parsko,

    I was also thinking of a continuous servo driving a lead screw set up to feed the pretension mechanism back and forth. I want to do as you said and program the software to let you set parameters for the feed per revolution (wire diameter) number of coils, and the exact start and end position of the bobbins.

    I was also going to design it so that lead screws of different pitch could be swapped out to provide for at least two different levels of precision and feed speeds, since that way the machine could be used with a wide variety of wire diameters, and still operate at decent speeds. Some of the coils I want to do will have wire of only about .003" diameter and if you set up with a fine pitch lead screw to provide that kind of feed precision, it would take too many turns of the screw to feed for larger wire diameters, (which I suspect would slow down the operation of the machine quite a bit when using larger wires - I'll have to calculate on that and see what I think). I'm also considering using a zero backlash drive on the lead screw. Meaning that instead of having only one nut riding on the screw, there will be two nuts with a light spring between them, which completely removes all free play between the screw and the mechanism being driven by the nuts. That would provide better precision of the feed movement when you wind coils with multiple layers and need to reverse direction at each end of the coil. With the clearance there is between the threads in normal screws and nuts, that slack would cause a loss of positioning accuracy whenever your pretension mechanism changes directions.

    I like the hall effect sensor for counting revolutions of the spindle. I'm thinking of something similar for the positioning of the servo driving the wire feed lead screw. A small round plate with about a dozen or more holes in it could be fastened to the existing servo horn, and an optical sensor or hall effect sensor used to give pulses every 10 or 15 degrees of rotation on the lead screw that way.

    I noticed that the maximum speed of the Parallax servo is about 60 revolutions per minute and was wondering if we can get a cheap stepper motor that would give the option of greater speeds for the main spindle. (But that complicates the software by requiring that the speed be ramped up and down nice and easy when starting and stopping the spindle.) I was thinking about using a cheap stepper motor for the main spindle because even in open loop mode we should still be able to keep track of it well enough. I think it's really just the positioning accuracy of the wire feed mechanism on the lead screw, and keeping it in sync with the spindle that we are concerned with. And if the stepper motor is large enough, it's own spindle could be used as the main spindle for the machine, holding the bobbins on the same shaft. Another option I like for the main spindle is to use a cheap DC motor geared down with plastic gears or tiny flat timing belts. Then a hall effect sensor could count whole revolutions of the motor and that would translate into steps of just a few degrees rotation on the spindle.

    A good place to browse for parts for a machine like this is Berg Components at www.wmberg.com, but there are cheaper places to buy than from Berg. Another is McMaster Carr at www.mcmaster.com. You might be surprised how cheap is is to use tiny, moderately precise ball bearings, plastic gears, timing belts and other stuff that you can buy off the shelf from industrial component suppliers. Since I was planning all along to mill my own aluminum plates to make the frame of the machine, it will be easy to include gears or pulleys and the shafts they ride on. If using a Parallax servo to drive the main spindle at 60 RPM, I assume you were going to have a spindle shaft separate from the servo that drives it anyway. A servo shaft does not really give you much way to mount bobbins and spin them very accurately.

    As for the pretension mechanism, I've seen how this is done on some commercially available machines and it's pretty simple. It involves putting your wire through two or three plastic pulleys, and one of them is spring loaded to hold an adjustable amount of tension on a loop of the wire right before it goes onto the coils you're winding. Your spool of bulk wire just sits on one end under the machine.

    What were you going to use for the microcontroller? I probably already have (or will get) whatever electronics from Parallax that this thing will need. I was thinking that maybe the Prop Proto board would be a good choice. With the VGA adapter installed it you could plug in a keyboard and a mouse to control the machine. Aside from a few mechanical parts that we can publish drawings for, (or I can get them made if enough people want them) this thing could almost become a Parallax kit.


    I'm just glad we have access to someone like Beau on this. I've always known exactly what I'm trying to accomplish with LVDT's as my end result, but I don't understand inductors nearly well enough to have made them work, let alone making it worth while to build a coil winding machine to make the LVDT's I need.

    Thanks for providing the light at the end of the tunnel on this problem Beau. Without you I would not have been motivated to follow through on making these LVDT's. (And I'm going to do some linear motors too.)

    Post Edited (Brian L) : 10/10/2008 7:36:07 PM GMT
  • parskoparsko Posts: 501
    edited 2008-10-10 19:53
    My experience so far is that the Parallax servo's can only get up to 45rpms. Maybe a few more, but that seemed to be the cap, without the servo sounding like it wants to explode. Others may have different experiences, or they may vary from servo to servo.

    The wire I'm using is 0.20mm, or ~0.0079". With coating, it comes out to around 0.24-0.25mm in diameter.

    You can control things with your servo RPM ratio (guide feed versus form rpm) pretty accurately. If you're winding 1000 turn coils, your speed will make more of a difference though (45 rpm will take 25 minutes!). I have the time to wait, so this does not concern me too much.

    I don't think you need any more accuracy on the rev counter than 1 pulse per rev. I'm using 4, just 'cause the servo horn has 4 legs, and I had 4 tiny screws to fit in there. So, don't kill yourself over this, IMHO.

    I don't think a stepper is needed either, unless you needed more torque or speed.

    I've thought about precise lead screws, but the cost/need prohibits it. If I eventually get to that route (which I doubt for my applications), I will get some. But, at a few bucks at home depot, I'm in good shape using brass threaded rod (but we'll see once I try it out!)

    Good idea on the zero backlash setup. I had thought about backlash, and decided it could be accomodated in software. But, if I double up on nuts, and put a wave washer between them, I'd have a zero backlash setup too. I'll see if this will be needed. For now, I'm not going to worry about it.

    My first proto pretension setup consisted of a string wrapped once around a shaft (that was hot glued to the wire source), with a spring on one end, and I simply tightened the string to adjust the preload. Simple and cheap. Although, it induced some axial translation into the wire source bobbin, which is why I'm planning on moving it to the moving guide.

    I'm setup to use the Propeller (aka this forum!). It's the only ucontroller I use. I would recommend a Proto board, as it already has the hook-ups for the servos built in. Plus, all the rest of the stuff needed for this circuit should easily fit on the the Proto board without needing any daughter boards. Plus, the Prop has TV out, which is a must for debugging and I/O feedback.

    -Parsko
  • Brian LBrian L Posts: 60
    edited 2008-10-10 22:16
    I've already got some Prop proto boards. Glad you're using it.

    I agree that this does not call for extreme precision lead screws. Regular threaded rods actually have pretty precise pitch control. One of my reasons for going with a zero-backlash nut setup was because the free play could also allow the wire feed mechanism to bounce back and forth within the play while it's in motion. For larger wire sizes, this would not cause a noticeable problem, but for very small wires it could.

    I'm going to do some coils that may have hundreds of wraps on them, and I want to make a lot of them. That's why I need to go with more speed than 45 RPM. If you're using a continuous servo for the spindle, were you just going to mount your bobbins right on the servo somehow? Or were you going to use the servo to drive a separate shaft as I was talking about?

    Since it's normal in my job to build industrial quality machines and production tools that last forever, I'm shooting for a machine that can be used for serious production of large numbers of coils with pretty good speed. I know that sounds expensive to most people, but for me, I'm in a position to do it cheap. I realize this is not possible for most home builders, but I think with what we're on to here, the electronics and software could be designed to operate a basic machine design, and that machine can then be built to any level of speed or precision the builder wants to use, depending on the money they want to spend on the mechanical end if it. It just takes some basic decisions about the method of counting spindle revolutions and syncing that to the lead screw movement, and a screen interface that allows users to set the end points of the windings and so forth.

    I know you're going to drive the servos with a square wave pulse train, and a DC motor needs to be controlled differently, but maybe your software can be designed so that by replacing a small section of it, the same basic program can also be used in a second version that controls one DC motor and one servo.

    If you're willing to design the software so that a builder can have the option of using higher speed motors on the spindle, I'd be willing to send you a free copy of all my custom made frame parts and a kit of all the components you need to build it - all ready to assemble.
  • Brian LBrian L Posts: 60
    edited 2008-10-11 21:14
    Parsko,

    I was just looking through the junkyard of parts I keep in my closet, and discovered that I already have a brushless DC motor designed for RC model airplanes and a speed controller for it. The speed controller is made to be connected to an RC receiver and sets the motor speed proportional to the width of a 50Hz square pulse just like you'd feed to a standard servo.

    If I gear this down with tiny timing belts or plastic gears to a MAX RPM of about 500 on my main spindle and put the hall effect sensor on that spindle instead of on the motor shaft, I think that your software & electrical system - though I might have to make a few small revisions at most - should be able to operate my machine. I'm going to use a 1/4"-28 lead screw, two nuts with the anti-backlash spring, and couple the continuous rotation servo right to it in co-axial fashion through a standard misalignment coupling to avoid binding on the the servo shaft. With .002" diameter wire I should be able to run the spindle at 500 RPM and still easily turn the lead screw fast enough to keep up. With .010" diameter wire, I'd have to slow down the spindle RPM to keep the lead screw from falling behind, but that shouldn't be a problem since a coil with larger wire will usually have fewer wraps on it. I should be able to wrap a 1000 turn coil in about 2 minutes.
  • parskoparsko Posts: 501
    edited 2008-10-15 04:02
    Video of my winder in action

    So far so good. I'm pretty happy with my results. It will take me a little more time to get it right, though. I will try to wind a proto unit in my next session (which will likely be in a couple days).

    Location of a few pictures for you to check out, this one shows how tight the winding is...

    I'm sorry for the poor quality on the pictures.

    My next major step, after getting a proto working, is to work on the user interface.

    -Parsko

    Post Edited (parsko) : 10/15/2008 4:08:58 AM GMT
  • Graham StablerGraham Stabler Posts: 2,510
    edited 2008-10-15 15:07
    Very nice, have you got it doing multi layer coils as neatly?

    Graham
  • parskoparsko Posts: 501
    edited 2008-10-15 15:17
    Graham,

    Thanks. In my next session I plan to do just that.

    One issue is the transition from one derection to another. There is a bit of backlash that I need to accomodate for in software. That should be pretty easy to figure out. Once I do, I'm winding a 3-5 layer coil, 1 inch long, ~200 turns. Can't remember the exact numbers offhand. I'll post my results.

    Oh, another issue with the transition is that I don't have a "wall" to wind up against. Meaning, last night, when I was making my first transition from left to right, it took a few turns (~4) before the winding actually reversed direction (cause there was no wall to resist it, only the guide stage). Not sure if that description is clear... I thought I had a pic showing my multilayer winding from last night, but I didn't upload it. But, the second layer looks much like the first. Again, it's the transition that is the hard part right now. Oo, maybe saying that I didn't have a bobbin to wind this coil on will make my description more clear (a bobbin has ends! my form doesn't!)

    -Parsko
  • Brian LBrian L Posts: 60
    edited 2008-10-15 22:19
    Very cool Parsko. Can't wait to see the kinks worked out and the user interface.

    I had a feeling you'd need bobbins with end walls to get it to reverse, but maybe for now you can glue something together like you did on your first attempt pictures. When ever you get settled on exactly what you want for bobbin dimensions, I can make you a nice one on a lath because I need to spend a day making some for myself anyway.
Sign In or Register to comment.