[ckelloug]

Pete,

Anybody who can read specific gravity of epoxy by color is the kind of guy we need on this thread! The index thread (a sticky in the forum) provides a good summary of the first two thirds of the thread. I haven't updated it in a while due to time constraints.

I think you are right about the amount of epoxy being a key factor in how it settles. Given no epoxy, the aggregate design that John is following (the min S distribution from Concrete Mixture Proportioning, a Scientific Approach by Francois de Larrard), has a tendency to trap the small particles between the larger ones preventing their mobility. Adding more epoxy than the amount of empty space between the packed particles will add degrees of freedom for the smaller particles allowing them to move and to get non-uniform settling.

My calculations with the packing model show that the amount of epoxy that just fills the voids is going to be about 25% by volume in this case. (John had previously posted his aggregate distributions allowing me to run the compressible packing model code I've written on it). Whether this is enough epoxy to get a usable composite is debatable.

I suspect in this case that a combination of hydrogen bonding between the epoxy and the aggregate and viscosity of the epoxy provide the ultimate amount of epoxy left between the particles when the molds are spun.

So Pete, what's your background? You seem to know quite a bit about this stuff.

Regards all,

Cameron

[greybeard]

Hi Pete. Very happy to have your ideas, and I think you are spot on with your current thoughts of the way my mix is behaving. Cameron, I'm sure, would concur, as he has suggested to me that making my mix much stiffer may well reduce if not remove the effect that I'm seeing.
I'll repeat here that this may only be a problem while I'm trying to produce the test samples. Once I move on to produce the machine castings, with a much bigger cross section, I can stiffen up the mix without having the difficulty of getting it into the 1/4" mold opening I have at the moment.
My current resin has an sg of 0.8 and when the two components are mixed has a viscosity of about 600cps(I think, if I remember the spec correctly).
I shall keep with the spinning method, as this is what I'm trying to develop as an alternative to open mold casting.

My own idea right from the beginning was that building the machine structure from separately molded parts would enable the builder to use smaller components which would then be assembled with a similar mix.
(Experience gained with a successful production of a small cnc machine would then allow scaling up without much difficulty.)
In addition, the high G forces involved will replace the need in open casting for vibration to "liquefy" the mix to get it into all parts of the mold; will increase the buoyancy of the air bubbles, replacing the need for vacuum; and will bring the packing density in the outer layers of each component to the best possible for the particular recipe being used, with any excess resin, or even unburst air bubbles congregating on the inner surface.
Regards
John

[greybeard]

Hi Cameron.
Seems I need to check the posting update before I send mine out
John

Edit
Pete, on re reading your post, I'm not sure I agree with you on the point of difference in density have no effect on particles in a viscous material.
As my gel time is in excess of 1 hour, probably nearer 2 at the temperature of the workshop, I think the effect of 200G for that time would overcome the drag of the viscous material quite easily.
John

[probinson]

Hi Cameron,

Actually, knowing that the epoxy pours means that the resin is a fairly short molecule and the fact that you can see the aggregate through it means it doesn’t have heavy oxide fillers so it will be reasonably close to the density of water. Close enough to be sure that rocks won’t float anyway.

As for my background, I have just always had a need to know how and why everything works. Electronics, physics, chemistry, biology, all the physical sciences are one in the same to me. I live to experiment and build things. Unfortunately, both because my interests are so varied and because I simply have to build everything from scratch, I rarely finish anything. A situation which my wife is always keen to point out. I keep telling her that if we moved out to the desert, where I could have 15 acres of workshops, I’d have room to build all the tools I need. She’s not buying it though.

Anyway, I currently work at the Medical Collage of Virginia designing and building equipment for medical research, which is where most of my experience with plastics and polymers comes from. It is certainly nice when your job and your hobby are one in the same!
I am not an expert by any means but, having dabbled in just about every topic in the CNCZone, I find this quite an interesting place to poke around, when I have the time.

So how about you guys? Care to share your interests and how you get caught up in this project?

Hey John,

Got your post so I’ll hit that too while I’m here. )

It’s not the drag or buoyancy that prevents the fine material from returning to the bottom. The fine particles build up as they attempt to migrate back down blocking off access to the lower areas.

An hour glass gives a good illustration of the flow problem. The sand flows nicely as long as the waist is large enough and the sand falls completely away from the opening. Make the waist too small, try to make it turn a corner, toss in a few large grains, or add even a small amount of moisture and the grains interlock, blocking off the flow. All of these conditions are present in the case of your mix which is what I believe is preventing even distribution.

I really think you should vibrate the mix for several reasons:

1) It is very easy to do. You don’t need to vibrate the entire mold. All you need to do is push a vibrating rod or paddle into the mix. Extracting the vibrating tool slowly will heal the wound as the tool is withdrawn.

2) Vibration allows the particles to jostle their way completely to the bottom. As long as everything is dense enough to sink, it will continue to displace epoxy until all areas are fully packed. This allows you to choose from a wider variety of materials. Differences in the relative density of similar sized particles can produce minor layering but it won’t prevent finer particles from packing all areas of the mold.

3) Vibration aligns the particles so their irregularities fit together better. This both reduces the epoxy content and produces an interlocking structure that greatly enhances the shear strength.

Pete

[probinson]

I just took a look at the index thread you pointed me to. VERY NICE.

Of course now I feel kind of stupid for posting things that have already been said.

Pete

[ckelloug]

Pete,

I am a research engineer for a large not-for-profit in California. I mainly do software but I work on anything that needs fixing. I got involved here because I woke up one day on sabbatical last year and decided I wanted to make concrete machine tools. I found this thread with google and the folks here convinced me that epoxy granite was the way to go. I then concluded I would use my engineering skills to find the optimum material for this and would share and collaborate with those here. I've got a small commercial building which I am equipping as a laboratory as well as a small materials test machine for flexural testing.

John's method is equivalent to setting the mixture under pressure which is desirable according to the 1975 paper by B.W. Staynes.

I have formed the opinion from reading some of the academic literature that interlock is of little importance in epoxy granite. As I understand it, the bond between epoxy and aggregate is so good that interlock doesn't increase the properties significantly.

Vibration is considered a necessity by most of the industrial folks doing E/G. John's method of setting the epoxy while spinning it produces some nice effects and I am not convinced that the segregation he is describing is significant. The true test will be when we I get his samples tested. I suppose this means I need to get out to the shop and get busy machining the improved top for the diamond saw so I can trim his samples accurately.

In general, the mix design in use has been shown by academic researchers to be the least prone to segregate under vibration of any mixture of the given aggregate sizes.

Regards all,

Cameron

[sigma relief]

John

Just a few ideas on the separation issue.

Try changing the speeds. I know this sounds silly, but if 200G for several hours is too much, 200G for a few minutes may be enough to remove air, then a low speed cure to hold everything in the mold could be used. Another more outside the box idea would be use the 200G full speed mode to remove air bubbles, then let the mold sit still, If I remember correctly, you have a multi cavity so all of the top cavities will slump down slowly, but probably won't entrap much air. A slow continual rotation would allow the resin to slump, not fall and possibly entrap air. A few sets of this followed by a slow spin to actually set the resin in the parts could help.

You could also try letting the resin begin to set off, or at least get close before you go to a high speed de-airing spin. This will limit the travel of the particles, but timing may be difficult on small molds and impossible on larger ones. If you have a slow resin and are quick at mixing small batches, just letting the batch sit a while before starting the process may help.

With 200G, you may be able to try some additional additives that would never work with vibration alone. I am thinking mainly of fibrous additives such as short (<1 inch) or long glass fiber (preferably s-glass, but e-glass would be fine), or more exotic carbon roving. In a slightly unrelated issue, I would be interested in seeing what steel based additive such as nails or screws would do to modulus. I know steel is bad for damping, but I don't think a bucket of nails suspended in resin and rocks will transfer vibration very well, besides, cutting damping by a factor of 4 still gives us better properties than steel ignoring the fact that the cross sections will be many times lager to get stiffness up. Using assorted lengths and diameters of nails and screws would help spread out the resonant frequency and reduce peaks.

I will check by the metrology lab tomorrow at work, our 200x stereo scope with digi-cam was possibly going out for upgrade to 400x, but if it is around, I may be able to grab some images over lunch some time. PM me if you are interested.

John K

[greybeard]

I don't know how many of you are familiar with the nursery rhyme which has the refrain "...and this is the house that Jack built". It's a line that I mutter to myself quite often when I have to build a jig to make a tool to get inside a piece of equipment that I need to fix to....... but I'm sure you get the picture.
Thanks for the offer JK, I'll be in touch.
Lots of ideas to think about.
John

EDIT
What has made me try to get a better view of the last sample I made is that the 'milkiness' that I have labeled excess flint continues down amongst the larger particles, and be just that, the flint well distributed.
The alternative is that the opacity of the resin in the lower part of the picture may be the shadow of other larger particles deeper in the mix.
Without a higher magnification, I can't distinguish between the two possibilities.

[ad_bfl]

I don't know how many of you are familiar with the nursery rhyme which has the refrain "...and this is the house that Jack built". It's a line that I mutter to myself quite often when I have to build a jig to make a tool to get inside a piece of equipment that I need to fix to....... but I'm sure you get the picture.

VERY VERY Scary, you just described my shop and "projects" perfectly.

Thanks for capturing that simple insight into my life

[greybeard]

John
.......Try changing the speeds. ........use the 200G full speed mode to remove air bubbles, then let the mold sit still....... A slow continual rotation would allow the resin to slump, not fall and possibly entrap air. A few sets of this followed by a slow spin to actually set the resin in the parts could help.

.........a high speed de-airing spin. This will limit the travel of the particles, but timing may be difficult on small molds and impossible on larger ones. If you have a slow resin and are quick at mixing small batches, just letting the batch sit a while before starting the process may help.......
John K

So it looks like a programmable, variable speed motor..........Hmm. Now, do I go mechanical or electronic. I've built potter's wheels with variable speed, but never actually built a speed control circuit. Brushed motors are noisy, but if I use an induction motor/cone drive I've got to solve the programming issue.
Could use a cam, but having to make a new one each time I wanted to try something else would be a pain......... Wonder if I could get enough torque out of one of my steppers to use that ? I've also got a servo motor not doing anything, but I don't know anything about servos yet. Perhaps if I find that site on the web where they discuss cnc design I might find something.... :drowning:

Cheers JK, I do think your right. I'm going to need control over the speed for the main casting, so I might as well get it sorted out now.
John

[sigma relief]

John,

I don't remember what you were using for a motor, if it is a simple AC line voltage motor, you can do what we tell our customers not to do and use a lighting dimmer. Anything will do, but the use of a dimmer specificaly for Magnetic Low Voltage or MLV would be prefered as they are balanced to limit the amount of DC current thru the motor which results in heat and damage. Most motors respond to phase controll dimming, but some like ceiling fan motors do not (at least not with a decent amount of noise). Don't worry about power rating on the dimmer, A little ingenuity improving the heat sinking of the triac and removal (or relocation farther form the triac for thermal reasons) of the anti hum RF supression inductor should allow you to safely work at double or tripple the rated limit.

John K

[greybeard]

JK- funny enough I've been using a 400watt dimmer on an old B&D 250watt drill quite successfully, but the recent spill didn't do the bearings any good, and the noise level is now too much.
I've already re-built the machine to use a near silent split phase motor, but unfortunately that didn't respond to the dimmer.

Having given it some thought over the last 24 hours, I'll go with the idea of a short, high speed spin to compact the aggregate and de-air the resin, then the slow speed for the duration of the gel time to avoid the mix slumping in the mold.
This will be in addition to stiffening up the mix with a lower resin content.

A thorough search of my various 'scrap bins' has produced a 250rpm motor, and with simple manual changeover, this will allow me continue in the workroom without fear for my eardrums, or any other part of my anatomy.

What I'm doing is to rig up the slow motor as a friction drive onto the face plate of the faster one, dropping it into engagement while turning off the power.
This produces very little additional load, so the slower motor can drive both motor and mold quite happily for the next couple of hours.
Photo to follow on completion of build.
John

[probinson]

Hey John,

Split phase-motors us a capacitor to delay the AC line which creates what's called a shadow phase. The time delayed phase is needed to produce a rotating field for the rotor to follow, otherwise, the rotor lines up with the single field which just switches back and forth so the motor just buzzes. The shadow phase is applied to a secondary winding usually referred to as a run or a start winding. The secondary winding isn't really needed once the motor gets up to speed, hence the term "start winding", but "start winding" can also used to refer to a high power winding connected to a centripetal switch that increases the starting torque on higher power motors so "run winding" is often used instead. OK, more than you need to know.:nono:

A standard light dimmer is basically a solid state relay but, instead of just connecting the power to the load, it turns off when the voltage goes away (which happens every half cycle) and waits until part way through the each voltage swing before it turns back on. The power to the load is governed by how much voltage is left when the switch tuns on.

The upper waveforms show the AC line voltage and how the capacitor delays the voltage to produce a shadow phase. The lower waveforms show the output of a dimmer and what happens to the shadow phase if you try run it off the dimmer.

You can use a dimmer with a split-phase motor if you keep the shadow phase intact. It doesn't work great at low speeds but it does provide some degree of speed control.

The first schematic shows how the split-phase motor is normally wired and the second shows how to wire in a dimmer so it doesn't affect the shadow phase.

If you are not sure which winding is which you can measure them with an ohm meter. The shadow winding is much lower power and will have a higher resistance.

Reversing the polarity of the shadow winding will also reverse the rotation of a split-phase motor. Although that little tidbit really isn't useful here, knowing how to add a reverse to an AC motor might come in handy at some point. Like when you want to add a tapping attachment to the CNC that you made with the mill that you made from the EG castings that you made.........

Pete

[greybeard]

Hi Pete.
I should have added that the centrifugal switch inside gave me the clue to it being a split phase motor. Never having tried a dimmer on this particular type, I gave it a go, assuming that if I had the dimmer full on when I switched on the power, the start winding would operate, trip out, then I might see if the dimmer would have some effect.
Unfortunately no, it didn't. I did notice the motor quietened down a bit, so under a decent load it might have produced some speed variation, but I decided to move on to the idea of two separate motors as being the simpler way forward for me in this instance.

By the way I do have some years experience as an electronic technician. You had earlier asked what we did. Your not going to believe it, so I suggest a quick trip to www.fanmaker.co.uk for a resume ! But looking after the British Army's cipher equipment in Malaysia on my own for several years, was a useful experience in "making do", as for some reason I was expected to be able to repair anything that came my way.

I've just tried out my latest aquisition, a set of five test sieves. :wee:
Like a kid with a new toy, I was amazed to be able to pour a beaker of sand in the top, a couple of minutes shaking, and there it was, a complete separation and information on the make up of the sand. This must be the simplest analytical machine possible - no moving parts, and a result in minutes.

[ckelloug]

Hi John,

I haven't forgotten about your sample set. I'm working on a new saw table for my tile saw to make it work well as a sample cutter. The first attempt with the tig welder at tig'ing up a tile saw table wasn't good so I made it into a jig for making a better table. As soon as I complete the jig then I can make the table.

One thought that's been bugging me lately however is the fact that the modulus and strength posted by the manufacturers are likely the compressive modulus and strength while my test machine can only get the flexural modulus and flexural strength. The compressive modulus and strength are much higher for E/G than the flexural one so I'm thinking we may not be as far from spec as we thought. I suppose it will be necessary to send out some samples to a lab to correlate the data.

Regards all,

Cameron

[greybeard]

Cameron, I've now collected together a range of sieves, some good quality, some not so good, but in total 12 ranging from 2.5mm to 44 microns, so I'm now trying to sort out the grading I shall need to do of my local sand/quartz sources.
Referring back to an earlier posting where I asked if from a range of sieves, a set of portion sizes could be determined for the optimal mix, I've now decided to try a 7 part mix, based on the following sieves sizes :-
2470/1560/1200/850/600/420/300/210 microns
If I reject the >2470 and the <210, can you tell me what the individual portions should be, given they will all be quartz ?

Nearly finished the 2 speed spinner, somewhat delayed by needing to cut the 150' long hedge. There's a project crying out for a solution - a cnc hedge trimmer

Regards

John

[ckelloug]

John,

I'll see if I can't run the sim on that for you tomorrow. I'm up to my eyeballs trying to get day job work done along with working with my contractor to straighten out the last house fixing before we finish up on the huge maintenance project at my house. . .

I'm still working on getting your samples cut for testing. I am however busily destroying tig welder parts making the jig to make the table to repurpose the tile saw to cut the samples to do the test to build the house that Jack built. . . Why can't anything ever be easy?

Regards all,

Cameron

[ckelloug]

John,

The cpm simulator doesn't like your mixture: it won't converge. Could be human error on my part or it could be a simulator bug that I'll have to debug. I don't know yet.

According to the simple appolonian model here is the optimum distribution for your aggregate since your sieve sizes match the optimum size distribution nicely:

mm percentage
1.560000 0.337500
1.116796 0.223594
0.799508 0.148131
0.572364 0.098137
0.409752 0.065016
0.293339 0.043073
0.210000 0.028536

--Cameron

[greybeard]

The percentage column seems to be short by 0.056013. Could this relate to the other problem ?
John

[ckelloug]

John,

That came from the simplified version of the model. I hadn't looked closely enough at it. The output there is the percentage of space filled by the components incuding empty space. The missing 5% is the model's view of the amount of leftover empty space. I guess I need to figure out why the big model is acting sick.

I think this means is that dividing each of the components in my last post by 1- the missing percentage (.95 or so) ought to give the correct percentage values of the components rather than the percentage values of the total.

--Cameron