153 replies to this topic

[Arthur Brown]

IMO the debate and (possible but unlikely!) agreement on the algorithm will take LOTS longer than the CAD and fabrication.

[PyroSam]

LOL! Yeah, I agree with that! BUT . . . by using a controller that simply needs ACII characters sent to the serial port and a high level programming language it should be easy for about anyone to implement whatever taping algorithm they want . . . and to share it with others. It's not like it's a hardware part or anything, just dl a file and give it a shot.

[digger]

I noted that you are going down the VB route.

For this App if you have excel you have all of the tools you need. I run my firing system from excel! (as do a few others I know).

If you are any good at maths (which I assume you are using CAD therefore an engineer). I reckon it should take less than 20 hours to go through a few iterations to get somewhere close.

If Phill does not dig out a flowchart for the math "as it is easy", then I will put one up mid August, this is when I will get a chance to unpack my notes as I have a few shows to do over the next few weeks whilst getting my old house ready for renting out.

Cheers

G

Edited by digger, 16 July 2012 - 11:23 PM.

[PyroSam]

Hi Digger,

THANKS! I use Excel every day but I had no idea it had VB capabilities! Too deep in the woods to see the trees I guess! LOL!

I'll probably go that route just to get going as it's the easiest path. I was looking at Visual Basic simply because I could compile something down to share with others and I'll probably continue that path once I get something working. I'm thinking the requirement of just sending ASCII out the serial port would even lend it's self to the use of FreeBasic, in that case others could experiment and share without expensive software. Eventually I'm thinking it may even make a nice BasicStamp project.

I'm really trying to make this a project most anyone can do - it's not about making a WASP for me, I can do that I'm sure - I have access to a fairly well equipped shop, but I'd like to come up with something that most anyone could do without issue - including the software.

At some point would you have access to your WASPs? I'm drawing up a Stinger simply because it's the best documented with all the pics in Ned's operator's manual - do you have a Stinger, if so once I've got things drawn could I impose on you to verify the measurements?

Thanks again - I'm off to see how VB works in Excel!

[PyroSam]

Since Arthur brought it up - here's a couple pics to get the grey matter working on. The first shows the top pole of a WASP pasted shell - note the offset in the line as it turns. The second is a pic of the bottom of a WASP pasted shell - note they are straight lines so the motor speed offsets are only occurring at the top pole.

Edited by PyroSam, 17 July 2012 - 03:47 AM.

[Vic]

PyroSam

I have been following this projected with great interest, I am an old die hard and will probably always paste by hand as all my shell's are one-off.

But I do like the open source ethos that you are clearly advocating.

Brilliant to see you all working together.

[martyn]

^ yep, exactly what the society / forum should be about in my opinion.
sometimes it's difficult or foolhardy to share what we are doing as it may not be legal, but as far as I know it's not illegal to build a taping machine though! yet!

[PyroSam]

Hi Vic & Martyn, yes - I'm really wanting an open source project. When I discovered the pasting machines I was really surprised that no D-I-Y projects seemed to be out there, at least any with any substance.

Work is ramping up for me so I'm going to disappear for a couple days - don't worry, I haven't went anywhere and while I'm gone I'll be thinking about how to build this thing! 8->

The CADs are going slow - the camera creates a lot of distortion so i'm having to guess at a lot of measurements - we need to find someone with a WASP that can take some measurements for us!

[parabolic]

To be honest your best bet is to see what material you can get locally, I can't see that it matters having the same dimensions as a wasp. With my engineer background I would automatically look up what stock is available online or local, ask my self does this project need to be light weight or not? Does this need to be compact or can it be designed to fold away or be easy to dismantle for storage?

If weight is not an issue and you want it cheap then black bar angle iron or box section would be good for frame work, just like many Chinese machine frames are built on. Cheap and easily weldable or bolt together then paint.

If Lightweight then aluminium is an idea but then you start adding £££

I guess some could be made from wood but only used as a prototype because wood is subject to shrinkage and warping.

Edited by parabolic, 19 July 2012 - 05:17 AM.

[Arthur Brown]

The Chinese will have down engineered something to the point when it has minimum cost, The Americans will re-engineer upwards til it looks good and costs a lot.

We have the opportunity to optimise function and cost, or form a committee and design a camel!

[PyroSam]

Hi Parabolic,

I agree that ultimately one is going to use what one has available to them at a reasonable cost. The distances between the motor rollers, the roller angles, the angle/distance between the right and left side uprights (which affects bearing block placement for the various sized shells), etc. are important though so I think it's prudent to have an accurate set of reference plans from a proven working design, then working from them one could see how their locally obtained materials could be used to mimic the original design as closely as possible. Just buying some material and blindly building is just going to waste material and probably end up with a less than optimal machine.

Steel certainly could be used but I think part of the reasoning for using aluminum was to try and stay as non-sparking as possible even though the shell should be completely sealed before it's brought anywhere near the machine. I also think part of the reasoning was due to the ease of use of the t-slots, it just makes a very nice way to adjust the bearing blocks, motor distances, etc.. I'm sure weight and portability also played a role and should be considered too.

In the grand scope of things the frame is really the smaller cost of the project, the money will be in the stepper motors, power supply and related components. In the end, I think the frame material will come down to what's available locally and the builder's preference - but he's still going to need to know what makes a good functioning machine before he starts the build.

Arthur, in general I agree with you but in this case we must also factor in that this device started as a hobbyist machine and though many have been marketed and sold to commercial manufacturers, the bulk of the users are hobbyists (the pyro hobby in the US is MUCH larger than over here). This machine is also under US patent so there is no competition for sales over there so there's no reason to overbuild the thing as a selling point.

I don't think we should discount the fact that this is a hobbyist used machine. The developer is an amateur pyro and he not only sells machines to his fellow pyros but he routinely interacts with them at the PGI Convention, club get togethers, on the online forums and via personal communication. No doubt many have offered their insights and suggestions about how to improve the machine - in fact Ned Gorski outright states this in the operator's manual he wrote.

From Ned's Manual:

-Snip-
"You and I, as users of the machine, are part of the design-and-build team. We will make our WASP our own, personalize it, help it evolve over time, and our feedback will help Jim refine the design so that it's even more streamlined in the future.

Jim, as an owner-manufacturer, is readily available to offer suggestions if questions or problems arise for a WASP owner. He's also willing to hear suggested changes or improvements from those same folks."
-End of Snip-

Many, many shells have been produced with these machines and the operators of these machines for the most part are very skilled fireworkers and D-I-Yers and they HAVE offered considerable input and that input HAS been listened to - the machine has evolved from a plywood mounted behemouth that took a cargo van to haul to a very compact machine. That's a LOT of research and development and I don't think we should overlook that fact.

Quite simply, I think the shortest path is to not re-invent the wheel, just replicate what works and get moving.

I think the key is the software, if we can get that working and get a working machine to be a relatively simple project than I'm sure many will build them and as time goes on I'm sure many will experiment with different frame designs and they certainly may come up with a better design - if that's the case then it will be an easy retro-fit as most of the components could no doubt be re-used.

If we can get a working design into CAD there are many that have 3D design software that may chime in and help with an improved design too - and they'll be able to do it without having to build several prototypes.

Do I think the design could be refined? YES I do! I'm pretty sure a machine could be made to do small through large shells, it just makes more business sense for him to sell two models. I'm sure we can combine the two designs into one, it's just going to take time and a couple prototypes (or a few runs through a 3D CAD program). Personally I just want to get a machine going and replicating a working design as closely as I can I think is the easiest route to that goal and is the one I'm persueing.

I'm not the final word on any of this though, and I apologize if I sound like I am, this is just what I'm doing and others are free to do whatever they want AND I HOPE THEY DO! And I hope they share what they've done with the rest of us! Let's make this a living, breathing, evolving project!!!

[Peret]

I've followed this thread with interest, wondering whether to jump in. I probably should, since I had an idea last year to make a small compact paster for shells up to 4 or 5 inches. I took it as far as a prototype, but then I dropped it because I live in the US and US patent law doesn't make any distinction between commercial and non-commercial use. British law does, however, so even if the WASP is covered by a British patent (pretty sure it isn't) or some cross-border treaty, there's nothing to stop you guys making them for your own personal use. I worked out the algorithm and put it in an integrated controller with an LCD display based on the Atmel AVR processor, so no PC or external boxes were needed. Here's an explanation of the algorithm.

First, the angle between the drive rollers should be 120 degrees. That means the contact points on the ball are one ball radius apart. Imagine the cross section of the ball as a hexagon with a flat face running between the contact points, then the rest of the geometry follows easily.

The basic unit of measurement for the algorithm is one stepper motor step. It can all be done in whole numbers (integers) - no fractions or floating point is needed.

If the X axis runs horizontally through the ball center, and the Y axis runs vertically, then with both rollers turning at the same speed in the same direction, the ball rotates around the X axis. If both rollers turn at the same speed but in opposite directions, it rotates around the Y axis. If the rollers turn at different speeds it rotates around some other axis and tends to ride up and off the drive rollers, causing slipped steps. That means you can't skew the ball by slowing down or stopping one drive roller, you have to reverse one of them.

With the ball rotating about the X axis and the rollers at 120 degrees so the contact points are one ball radius R apart, the diameter of the vertical section through the ball above the contact point is (root 3)R, so the path length traced out by the drive roller for one complete turn is pi(root 3)R. If the diameter of the drive roller is D, then the number of drive roller rotations for one ball rotation is (root 3)R/D, and since you know how many steps it takes for one drive roller rotation - usually 200, or some multiple thereof - you can work out how many steps it takes for a complete ball rotation. Round up to the nearest whole number. This has to be worked out once every time the ball diameter changes, but then it becomes a constant in the algorithm.

Another variable is the tape width, You need to know how many complete wraps it takes to make a complete layer, and this is the ball circumference divided by the tape width. You are more interested in the amount of Y axis rotation needed to lay parallel wraps, and this is the reciprocal - the tape width, divided by the ball diameter, times the number of steps for a complete revolution. Again, work it out in advance and then it becomes a constant. Every complete revolution, reverse one motor for this number of steps, and the ball turns about the Y axis just far enough that the tape wraps kiss at the equator.

To place a layer at right angles to the previous, rotate the ball one quarter revolution in X, then reverse one motor and rotate it one quarter turn in Y, then proceed as before.

[PyroSam]

Hi Peret,

Sure, you could do it scientifically! ;->

I was actually going to do it imperically - my thoughts were to build
the machine and build a pencil holder for it, then setup a program to
slowly run the motors and watch as the line was traced - when they
just touched hit a button to pause the program and read out the steps
taken.

Along these lines, I bought a copy of Visual Basic off of eBay and I
got it reasonably enough, but wouldn't you know it - the seller sent
me a note and a disc is lost so they voided the transaction (wouldn't
be they didn't want to sell it that cheap, would it?). If we use the
serial controller then it's just the simple act of sending ASCII
commands out the serial port but it seems Visual Basic is
expensive/hard to find. Excel can be made to work but not everyone
has it. I tried FreeBasic but it hangs my system. I'm thinking I
may need to learn Python - it's a higher level language so hopefully
it won't take too much of my already short time to learn, it seems to
have some i/o libraries so serial communication shouldn't be that
hard, and it's free. Not only is it free, it's also cross platform so
someone could setup an old computer with Linux and still make it
work. I'm still kicking around the options and would appreciate any
readers inputs.

Your PIC solution sounds interesting and we'd be grateful for any
further specifics on it you wish to share.

OK, on to trying to wrap my head around the algorithm, let's put it
in a real world application - let's say we want to paste a 4" shell
(I'll stick with inches since you're in the US).

Shell OD = 3.4"
Radius = 1.7"
(root 3) = 1.73
Pi = 3.14
Drive Roller Diameter = 4" (I'm guessing this is what the WASPs use?)
Motor Steps per Revolution = 200
Shell Circumference = 10.68"
Tape Width = .75"

> With the ball rotating about the X axis and the rollers
> at 120 degrees so the contact points are one ball radius
> R apart,

The contact points in our example are 1.7" apart.

> the diameter of the vertical section through the ball
> above the contact point is (root 3)R,

1.73 * 1.7 = 2.94"

> so the path length traced out by the drive roller for
> one complete turn is pi(root 3)R.

3.14 * (1.73 * 1.7) = 9.23"

> If the diameter of the drive roller is D, then the
> number of drive roller rotations for one ball rotation
> is (root 3)R/D,

(1.73 * 1.7) / 4 = .74 rotations

> and since you know how many steps it takes for one
> drive roller rotation - usually 200, or some multiple
> thereof - you can work out how many steps it takes for
> a complete ball rotation. Round up to the nearest whole
> number.

200 *.74 = 148 steps

> This has to be worked out once every time the ball
> diameter changes, but then it becomes a constant in the
> algorithm.

Is this worked out at the end of every full tape layer before
the next is began, or is the few thousands difference
insignificant?

> Another variable is the tape width, You need to know
> how many complete wraps it takes to make a complete
> layer, and this is the ball circumference divided by
> the tape width.

10.68 / .75 = 14.24

It seems to me this should be halved? It would take
14.24 strips but a wrap is really two strips - front &
back?

> You are more interested in the amount
> of Y axis rotation needed to lay parallel wraps, and
> this is the reciprocal - the tape width, divided by the
> ball diameter, times the number of steps for a complete
> revolution.

(.75 / 3.4) * 148 = 32.65 steps, so 33 steps.

> Again, work it out in advance and then it becomes a
> constant. Every complete revolution, reverse one motor
> for this number of steps, and the ball turns about the
> Y axis just far enough that the tape wraps kiss at the
> equator.

> To place a layer at right angles to the previous,
> rotate the ball one quarter revolution in X, then
> reverse one motor and rotate it one quarter turn in Y,
> then proceed as before.

So 37 steps in x and then 37 steps in y and proceed on?

[Peret]

I think you've got it right, congratulations!

VB6 has been discontinued by Microsoft for some years, and you can only get bootleg copies now. Microsoft gives Visual Basic 10 Express away for free, just download it. It is, however, dot.NET, so it's similar but different in syntax from VB6. It's easy to make a super pretty application that can display movies or calculate a spreadsheet, but to get it to work properly with a serial port, now that's another matter. Microsoft never got serial ports right. If you're starting from scratch it might be better to design a controller with a HID class USB interface instead of serial. It would be a lot of work but probably less painful and more reliable overall.

Not everyone has Excel, but everyone can get Open Office for free, and its spreadsheet is superior to Excel in nearly every way. I've never used its scripting language but I bet it's good.

Python is a good choice and easy to learn, though it has an unfriendly user interface. I use it with Eclipse as an IDE. The only reason not to use it for an application is that you can't make a one-file setup to give to other people - they need to be geeks and nerds to install it.

[PyroSam]

Hi Peret,

Glad to see I was following along correctly!

I'll have to look into the VB freebie deeper. I thought serial communications were OK in VB, it was parallel that windblows had issues with.

I took a quick look at Open Office Star Basic and it seems it's pretty difficult to get it happy with serial communication.

The Python install went easily, as did the Eclipse install which I'm also using. The Python install went so easily I don't think it really requires a geek. I was sure I'd read also that it is possible to make an executible from Python code, I'll have to check again.

Things are getting busy again but I'll try to keep an eye on the forum - I love all the ideas being floated!