Building the full sized winder and chamber.
11'th April 2009.
After a few months doing other stuff I've started the full scale filament winder project. If you haven't seen it before have a look at the 3/8 scale model first.
This winder is being built for one winding job. It is expected to run for under twenty hours total time. It doesn't have to look good or be durable - after the job is does it will be disassembles because I want my garage back.
The job to be wound is a vessel approx 2.5 metres long and 750mm wide.
The stepper.
The stepper I used for winder-2 turned out to be under-powered for the way it was set up. By using a bigger motor and a smaller pulley I hope to get around three times the linear force onto the carriage. I also suspect I hit a resonance running the small stepper at 100 steps/second. I need over twice the linear speed on this winder and I expect to be running at about 400 steps/second.
Arjan gave me a motor to use. It was off a large printer. The big motor had nine wires to sort out but I managed to figure it out. I am running this unipolar stepper as a bipolar one by running two pairs on windings in series. This is a very low voltage motor so series connection helps raise the drive voltage and lower the current. Each winding is 0.4 ohm and rated at 4.7 amp. V=IR says 1.88 Volts. In series we still only need 3.7 volts to get the same current but you usually drop it a bit because you have a higher duty cycle when using a bipolar drive.
The driver I'm using is a L6208 - it should be able to handle the current but I need to cool it better before I try. The driver is a chopper and is power by the 24V supply.
The motor spins fine at 1 amp but doesn't have much torque. Because of the low voltage it has less torque at 1amp than the small motor. The motor has less power to weight that the smaller one. It is probably four times the weigh but it won't produce four time to power.
I ran into problems straight away when I tried to get the original pulley off. It ended up taking me hours and I had the destroy it to remove it. I had to use a fair bit of brute force and after I'd finished the motor didn't work. I feared the worst and decided to open it up to have a look. My past experience with disassembling steppers said you'll never get it back together properly but it came apart and went together again quite easily. This one had very little magnetic force to work against. When I put it back together it kind of worked but didn't rotate as freely as it should have.
Eventually I pulled it apart a second time and found it works much better with one of the thrust springs omitted.
I lathed a new pulley out of hard epoxy. It is pretty rough but should do the job.
Instead of kite string I'm going to try hi-tech fishing line as the cable. This gel spun UHDPE is about as thin as sewing thread but very strong. The main negative is you have to keep it fairly cool (<60C?) or it looses strength.
The large stepper pulled apart. Note the rust. |
The large stepper with the back off. Without the back the rotor is not aligned. |
The spindle.
The spindle setup will be much the same as winder-2 expect the power source is a cheap five speed drill press. The second fastest speed (1650RPM) is close to perfect. After the 153:1 planetary gears I should have around 11 RPM. If for some reason we can't wind that fast we have an easy fallback to slower speeds. We're expecting a linear fibre speed of around 500 to 600 mm/sec max. This is over twice what I've done before so the ability for the resin dipper to keep up is untested. I think it will be fine.
The gear box will be driven via a right angle drive to make the winder more compact. I have a space problem trying to wind a >2.5 meter tank in a 3 metre wide garage.
Later.
I put a pole through the middle like I did on the scale test. I wasn't happy with the aluminum tip though. It was scored from the scale test and the full size chamber is a lot heavier and will be spinning a lot longer. I made a steel tip instead.
The spindle drive is now working - not sure for how long though. |
The first attempt at rotating the pattern using drill power was short lived. The nut attaching the wooden disk to the bench grinder shaft is a left hand thread and promptly unscrewed itself. It did the same after I disassembled, tightened and reassembled. On the third go I used exopy glue and lock tight and this has held up so far.
I also found the pattern is out of balance because I add an extra layer of glass to one side.
I've got it all working now but it sounds terrible - I'm not sure if it will last the distance.
The pattern.
The pattern will be made in a similar manner to this one.
It will be a cylinder with a cone at one end and a dome at the other.
The bag was first made without the cone etc so I could check the size. I'm not sure how much the plastic stretches and I want to be sure this will go through a 800mm doorway. |
I made a silly mistake and ended up with the cone seam on the wrong side. Taping it down will have to do.
The first glassing sessions didn't go well. I began to loose pressure. I swapped from fish tank pump to mini-compressor but still had marginal inflation. I glassed over one leak and had to apply a pressure bandage to stop the leak - this caused a small depression in the glass. There is also a small depression near the cone seam close to where it joins the cylinder - I suspect this was due to poor pressure. At the end of the day I found a leak in the back seam. I taped it up and was able to go back to the smaller pump again. Most of the seams are now glassed over so the leaks should be less of a problem.
Some of the resin hasn't set properly. I'm not sure if this is sloppy mixing on my part or some sort of contamination on the plastic. It is savable.
Time to plug the plug and deflate the ball. |
The ball deflation caused a temporary cave in of the wall. Once the air seal was broken it popped back into shape. |
You can see the colour change as the fibreglass separates from the ball. |
A close up of the ball being peeled off the fibre glass. |
A rough trim of the opening and the first view inside. |
Inspecting our handy work. |
The opening was tidied up and made oval. A couple of pieces of 1mm polycarbonate were used as a former to make a fibreglass neck.
The pattern is now complete and ready to set up for winding. |
Winding it.
We have a problem. When the chamber is in our intended winding position a post is in the way of the fibre path. We have several options to hopefully get around this problem.
27'th April 09, a temporary halt to the project - OK I lied.
I had problem getting the carriage to work, I've run out of time and need to give it a rest for a few weeks. The main problem was that I crudely modified the pulley I used for the scale winder. I ground the groves to give a smaller diameter. Because the pulley is narrow 3 metres of string makes a big difference to the effective diameter.
This can probably be solved using a wider pulley but it was getting too late to fix it. What I would like to try is a ball-chain instead of a string.
28'th April,
I was given a long length of plastic ball chain. I also ordered a bunch of vertical blind spare parts which contain ball chain sprockets. http://www.blinds2go.com.au/ViewProductCategory.aspx?ID=38
I also have some blinds at home which can be gutted.
This is the sprocket from one of my vertical blinds. It is quite small and is part of a planetary drive unit. |
I pulled my blind apart. The sprocket may be a little too small - the ones I've ordered are larger.
The ball pitch is 6.5mm and so the chain moves 12*6.5mm every revolution. For a 200 step per rev motor this means 0.49mm per step - call it 0.5mm. So at 500 steps per second I'd get about 250 mm/sec travel - that might just do.
29'th April,
The sprocket is fitted onto the stepper motor and ready to go. |
The ball chain setup is running very well in the initial tests. It seem to be moving just fine at 600 steps/sec. There is a fair bit of tension but nothing bad happened when I stalled the motor.
30'th April, The first winding session.
Barry kindly came over to help me on the first session. I expected to need help because of the problem with having a post in the way. I figured I'd need to give it a helping hand but it worked ok without assistance.
The session went fairly well but not perfectly. There were several problems.
Tip problem.
Because of the way the winder was set up I could get quite enough travel to wind all the way out to the tip. I could have spent hours extending the travel but I'd already decided to do a hand lay up at the tip if I needed to. The last 60mm or so in not wound. This is a very low stress area and the existing fibre-glass would probably hold the pressure - I will strengthen it though.
The other tip problem is that the head should really be slow down near the tip. To make matter much worse I had the resin bath set up level with the other en of the chamber so the fibre would (almost) clear to post. What happens is the fibre goes on fine when the head it travelling towards to tip but when it reverses the small cone section can wind the fibre in fast enough to take up the slack.
This resulted in an uneven loose winding. I'm sure it is strong but it looks sloppy.
Spindle problem.
The left hand threaded nut unscrewed itself when I had a minor jam up. We drilled through the shaft and added a pin to stop it slipping. I'm not sure where the problem is but we ended up with a lot of slope somewhere and the rotation was very jerky. It wound very nicely considering how much it was jumping around.
Voids on the dome.
The fibres tended to bunch up and leave patches un-reinforced. The winding is much better than the scale model and I think it would hold pressure just fine but I will probably tidy it up by hand. Because the first winding layer is now set the next layer won't be able to fill the voids (ie holes will be bridged not filled).
I would have liked a little more fibre tension during winding but my tensioner applied too much drag.
For the next session the resin bath will be central and further back so some more tension may be possible.
Another minor thing was I had resin drip from the winder head. I think this is because the head was low and the fibre deflected up on the mandrel. I think when the fibre goes down after passing through the head it tends not to drip. I've never had this problem before even when the fibre was much wetter.
The chamber near the end of the first winding session. The first session had 313 head passes. This is about 40% complete. |
In this session there were 313 head passes in each direction. I didn't use nearly as much fibre as I expected. My initial guess was I'd use two and a half 18 kg rolls but now I think it will be well under two rolls.
A closer view showing the head being pulled by the ball chain. |
3'May - Minor disaster.
Things were going too well. Then with only 100 head passes to go my tip bearing failed.
I was upstairs at the time and heard the noises change.
The chamber had fallen onto the ground and traveled north, it stopped because the screws holding the gearbox sheared and the spline had come out.
To make matters worse I had a full batch of resin in the resin bath. I tried to improvise a fix so I could use the resin up before stopping to do repairs but gave up.
To complete to mood it started to rain.
There is no serious damage to the chamber. Hopefully I can repair the winder enough to make it work for another hour or so.
I'm calling it quits till the resin cures.
A few hours later - I repaired it and winding is completed.
May 2009,
The chamber lid is made. I have had a bunch of minor leaks in the nose cone. I've worked out a reasonably mess free way to apply patches from the inside so hopefully it will be air-tight before long.
June 2009,
I think I still have some leaks to find. There is definitely a leak but it may be it the hoses, regulator and so forth.
The good news it passed the 2 BARG pressure test without any dramas.
A set back.
I've over stressed the rim of the lid. I think the reason is my neoprene seal migrated towards the edge after a series of sessions due to the shear forces on the seal. Over time it became distorted and applied uneven force to the rim.
The lid didn't fail and I'm still using it but it will probably need to be replaced soon.