Logo
Builders' Plan Gallery  |  Hip Pocket Web Site  |  Contact Forum Admin  |  Contact Global Moderator
November 19, 2017, 05:20:42 AM *
Welcome, Guest. Please login or register.
Did you miss your activation email?

Login with email, password and session length
 
Home Help Search Login Register
Pages: 1 2 [3]   Go Down
Print
Author Topic: Rubber comparative test.  (Read 6388 times)
0 Members and 1 Guest are viewing this topic.
tomv
Bronze Member
***

Kudos: 0
Offline Offline

United States United States

Posts: 18



Ignore
« Reply #50 on: October 01, 2012, 08:50:14 AM »

Tapio,
I believe that at lower stress levels all of the new rubber is fairly similar with respect to energy return. When you push this stuff to its limit (stress just under the breaking point of the rubber) you see some significant differences. The fact that Leo's results vary only .5% verifies this.
In my testing the total range of results for 2009-2012 samples runs from 1620 meters to 1800 meters, about 11%. This is based on about 300 samples tested so far from most of the batches produced since 2009.
The result is expressed in meters because it is the total energy in the sample divided by the sample weight. Theoretically that is the height to which the energy in the sample will lift the sample with 100% efficiency. Actual still air flight results (launch altitude) are highly correlated to these results.
Tom
Logged
Tapio Linkosalo
Platinum Member
******

Kudos: 27
Offline Offline

Finland Finland

Posts: 1,089



Ignore
« Reply #51 on: October 01, 2012, 09:13:19 AM »

Tom, the "common wisdom" with indoor fliers claim that the whole energy return curve of the motor depends on the maximum strain that it is exposed to. Your results seem to oppose that, when you suggest that with a lower strain all the samples give the same output, but only when stressed close to the breaking load will there be differences. Unless it is, after all, a matter of variation in the maximum strain? I think you described the method by stating that the relaxed lenght and weight of the sample dictates the maximum pull force? Shouldn't the stretched length be used to make sure that each sample is subjected to the same strain?

Logged
tomv
Bronze Member
***

Kudos: 0
Offline Offline

United States United States

Posts: 18



Ignore
« Reply #52 on: October 01, 2012, 09:47:33 AM »

Tapio,
If you just pull all the samples to the same length, the variation in sample thickness and width will cause the cross section to be different  and therefore, the stress will vary. Also pulling to the same length assumes that you can ster wit samples that are exactly the same length before stretching. In practice, this is impossible to achieve.
I agree somewhat with your comments with respect to indoor motors. However, in that case it's not just maximum stress that's important. The properties of the actual rubber used in the motor is as important as winding skill. One can't get same energy return from any batch of rubber stressed to the same level as from 5/99. That's why the F1D guys still use 5/99 exclusively. I also believe that those guys wind their motors to much higher stress levels than we can in F1B.
Tom
Logged
Tapio Linkosalo
Platinum Member
******

Kudos: 27
Offline Offline

Finland Finland

Posts: 1,089



Ignore
« Reply #53 on: October 01, 2012, 11:39:31 AM »

Not to the same length, but to the same strain. As in the Pearce-method.

Logged
tomv
Bronze Member
***

Kudos: 0
Offline Offline

United States United States

Posts: 18



Ignore
« Reply #54 on: October 01, 2012, 12:24:20 PM »

That's what I'm doing. All samples pulled to the same stress.
Logged
Tmat
Titanium Member
*******

Kudos: 64
Offline Offline

Canada Canada

Posts: 2,884




Ignore
« Reply #55 on: October 01, 2012, 12:53:23 PM »

Not to the same length, but to the same strain.
Strain is relative change in size or shape due to an applied stress. Stress in an internal force (per unit area etc) associated with a strain. So I assume you mean stress Tapio not strain?

Tmat
-interesting discussion by the way... Grin
Logged

F1B guy...
But don't hold that against me!
Tapio Linkosalo
Platinum Member
******

Kudos: 27
Offline Offline

Finland Finland

Posts: 1,089



Ignore
« Reply #56 on: October 01, 2012, 01:07:57 PM »

Tony, right, stress. Thanks for the correction.

Logged
lbodnar
Bronze Member
***

Kudos: 1
Offline Offline

United Kingdom United Kingdom

Posts: 19



Ignore
« Reply #57 on: October 01, 2012, 03:54:25 PM »

What is your the average sample weight? What is the max tension and max length to which you are pulling the sample?
Tom
What equipment are you using the measure the pull force?
Positioning:
I am using a 16Nm 2kW servo motor driving a timing (toothed) belt via a matching pulley with 250mm circumference.
Maximum pulling force on the belt provided by servosystem is about 40kgf.  
4m long timing belt is made from steel reinforced polyurethane and has 4kN tensile strength.
The servomotor can be positioned with high precision (40,000 steps / revolution) translating to linear step resolution of 6.25um. This obviously far exceeds sensible requirements but I just happen to have the equipment at hand.
One end of timing belt has a rigid steel hook attached to it.

Force sensing:
Current system contains a 2kg load cell attached to a load cell amplifier and 12-bit ADC converter.  This provides a force resolution of about 0.5gf when calibrated.

Control:
I have written a simple application that controls the positioning of the servo system while simultaneously reading data from ADC.

Sample size is a rubber loop of about 120mm long when folded in half and having weight of 0.48g.
Maximum force applied to the rubber sample is in the order of 2000gf but can vary.  I don't use force now as a limiting factor for pull tests.
I am using either differential stiffness (curve ramp) or total specific energy injected into sample.

I am thinking along the lines that it is not pulling force that makes particular piece of rubber break.
It is either excess of energy or pushing sample beyond limits of structural change in the material.
That's why I don't see much use in pulling samples to fixed force (or level of stress) when comparing batches.
I am still at the very early stage of trying to understand behaviour of rubber so I am very interested in any opinions!
I definitely don't understand how ambient temperature affects the tests so my strategy is to try and keep room temperature at controlled level.

Cheers
Leo
Logged
PeeTee
Titanium Member
*******

Kudos: 48
Offline Offline

United Kingdom United Kingdom

Posts: 2,163



Ignore
« Reply #58 on: October 01, 2012, 04:14:29 PM »

Leo

Many thanks for your efforts. I'm a strict amateur when it comes to testing or conditioning rubber, but I am interested in what others are doing. Are the results of your testing likely to be published in say the FF Forum report, or the NFFS Sympo (or elsewhere perhaps). Are you able to tell us who you are collaborating with?

Cheers

Peter
Logged
Tapio Linkosalo
Platinum Member
******

Kudos: 27
Offline Offline

Finland Finland

Posts: 1,089



Ignore
« Reply #59 on: October 02, 2012, 12:37:10 AM »

I am thinking along the lines that it is not pulling force that makes particular piece of rubber break.
It is either excess of energy or pushing sample beyond limits of structural change in the material.
That's why I don't see much use in pulling samples to fixed force (or level of stress) when comparing batches.

This in one detail that I have given lots of thought during the 10+ years that I have tested my F1B motors on a computerized setup. In principle I agree with you, however for practical application you cannot tell (beforehand) how much force a full motor could take, and therefore testing to a fixed stress emulates winding the motor to a fixed target combination of torque&turns.

I suppose in theory, with a sensitive enough system you could calculate the derivative of the motor stress and from that conclude when you are reaching the breaking stress. But then in practical solution, when you wind the motor instead of pulling, I'd suspect that the internal friction (and the variation in winding technique between individual motors) would mask the fine details of motor torque development...
Logged
lbodnar
Bronze Member
***

Kudos: 1
Offline Offline

United Kingdom United Kingdom

Posts: 19



Ignore
« Reply #60 on: October 02, 2012, 04:08:43 AM »

Many thanks for your efforts. I'm a strict amateur when it comes to testing or conditioning rubber, but I am interested in what others are doing. Are the results of your testing likely to be published in say the FF Forum report, or the NFFS Sympo (or elsewhere perhaps). Are you able to tell us who you are collaborating with?
Hi Peter,

I am working on this together with Pete Brown.  For now we have a tool but no definitive idea of how to use it properly!
If we get anything interesting to share of course it would be great to show the results.

I suppose in theory, with a sensitive enough system you could calculate the derivative of the motor stress and from that conclude when you are reaching the breaking stress. But then in practical solution, when you wind the motor instead of pulling, I'd suspect that the internal friction (and the variation in winding technique between individual motors) would mask the fine details of motor torque development...
We are testing individual strands so the friction is not an issue - unless you mean internal rubber friction?  

The force derivative is measured as "stiffness" - it is reliable enough to use as one of the criteria for stopping the pull.  It increases very rapidly towards the end (by a factor of 20!)

Leo
Logged
Tapio Linkosalo
Platinum Member
******

Kudos: 27
Offline Offline

Finland Finland

Posts: 1,089



Ignore
« Reply #61 on: October 02, 2012, 05:09:33 AM »

I suppose in theory, with a sensitive enough system you could calculate the derivative of the motor stress and from that conclude when you are reaching the breaking stress. But then in practical solution, when you wind the motor instead of pulling, I'd suspect that the internal friction (and the variation in winding technique between individual motors) would mask the fine details of motor torque development...
We are testing individual strands so the friction is not an issue - unless you mean internal rubber friction?  

No, I mean that too refined testing is purely academic, as the actual use of our motors, by winding, involves the friction between the strands, and this probably adds more variation in the energy output than is the (potential) difference between samples...
Logged
Modelace
Gold Member
*****

Kudos: 34
Offline Offline

United States United States

Posts: 763



Ignore
« Reply #62 on: October 08, 2012, 01:36:29 PM »

Tapio:
Both Fred Pearce  and Chris Matsuno  developed rubber testing techniques in the 60's. Fred was the "tester" for Ed Dolby (Later FAI) for many years. Both Fred and Chris tested rubber using techniques that involved friction, Fred's using twisted (wound) rubber and Chris usng stretched only. Both of these gentlemen published definitive papers on rubber energy storage (Fred's published in The Freeflight Symposium and AMA'S magazine and Chris in a private article).
I suggest a Google Search for those documents which may add some background to your efforts.

Bob
Logged
Tapio Linkosalo
Platinum Member
******

Kudos: 27
Offline Offline

Finland Finland

Posts: 1,089



Ignore
« Reply #63 on: October 08, 2012, 11:52:02 PM »

In message #34 of this thread there is a link to the Pearce articles in Model Aviation, describing his stretch method; essentially the same one that Eimar has used as basis of his method, and I've based mine on Eimar's. In addition, I found a paper by B. Hunt in INAV, that also relies on the Pearce stretch method. But could not find online anything else, not from Pearce and nothing from Matsuno (just references to some Sympo articles).

 
Logged
giladmark
Copper Member
**

Kudos: 0
Offline Offline

Israel Israel

Posts: 4



Ignore
« Reply #64 on: April 30, 2013, 03:59:53 AM »

Hi everybody,
I find this topic very interesting.
As been said here before, the deviation in energy between splices (or batches) a long the past 10 years was a bout 10%-12% max, and probably less in the past 3 years. If so, the testing method resulting error is very important and must be at least one order of magnitude less then the actual result (am I wrong?).
I'm struggling for some time trying to understand the error magnitude.
Lets say I don't have a hi tech machine, I test manually a loop specimen of 125mm in length (250mm/2), weighing about  0.700g.
I use a ruler with 1.0mm accuracy and a scale with 0.02kg accuracy.
can someone clarify the error calculation?
Logged
lbodnar
Bronze Member
***

Kudos: 1
Offline Offline

United Kingdom United Kingdom

Posts: 19



Ignore
« Reply #65 on: April 30, 2013, 06:18:34 AM »

Hi everybody,
I find this topic very interesting.
As been said here before, the deviation in energy between splices (or batches) a long the past 10 years was a bout 10%-12% max, and probably less in the past 3 years. If so, the testing method resulting error is very important and must be at least one order of magnitude less then the actual result (am I wrong?).
I'm struggling for some time trying to understand the error magnitude.
Lets say I don't have a hi tech machine, I test manually a loop specimen of 125mm in length (250mm/2), weighing about  0.700g.
I use a ruler with 1.0mm accuracy and a scale with 0.02kg accuracy.
can someone clarify the error calculation?
If you are measuring independent values X with relative error Ex and Y with relative error Ey then the result is usually of relative error Ex+Ey.
It does not matter whether you are measuring X*Y or X/Y.  This of course applies to small Ex Ey (few percent or less.)

The other source of errors is integration process.  The larger your integration steps the bigger is the potential difference from actual energy.

Leo
Logged
Tapio Linkosalo
Platinum Member
******

Kudos: 27
Offline Offline

Finland Finland

Posts: 1,089



Ignore
« Reply #66 on: April 30, 2013, 07:20:13 AM »

I think the biggest potential source of error is the variability in the maximum stress.
Logged
giladmark
Copper Member
**

Kudos: 0
Offline Offline

Israel Israel

Posts: 4



Ignore
« Reply #67 on: April 30, 2013, 02:17:36 PM »

So if I understand correctly, we must sum the errors of each measurement (lets say we are making 10 measurements starting at a bout 2kg force and 1.2 meter elongation, 0.02/2=1% plus the errors in elongation 1/1200= 0.08% we get more then 1% error in each measurement). also the error percentage getting higher for each measurement because the accuracy stays the same while measurement magnitude getting low, but in the other hand, as we go along the measurements percentage from total energy getting lower.... confusing.... Huh
Did anyone ever calculated the error percentage?
Logged
Tmat
Titanium Member
*******

Kudos: 64
Offline Offline

Canada Canada

Posts: 2,884




Ignore
« Reply #68 on: April 30, 2013, 02:31:17 PM »

Gil,
Paul Rossiter from Australia looked at this problem in either the NFFA Sympo or Free Flight Quarterly (I'll look it up). He calculates both the sources of error and the required accuracy to get meaningful results.

Tmat
Logged

F1B guy...
But don't hold that against me!
lbodnar
Bronze Member
***

Kudos: 1
Offline Offline

United Kingdom United Kingdom

Posts: 19



Ignore
« Reply #69 on: April 30, 2013, 02:39:26 PM »

So if I understand correctly, we must sum the errors of each measurement (lets say we are making 10 measurements starting at a bout 2kg force and 1.2 meter elongation, 0.02/2=1% plus the errors in elongation 1/1200= 0.08% we get more then 1% error in each measurement). also the error percentage getting higher for each measurement because the accuracy stays the same while measurement magnitude getting low, but in the other hand, as we go along the measurements percentage from total energy getting lower.... confusing.... Huh
Did anyone ever calculated the error percentage?

I might have unwillingly confused you!

The accurate and possibly intuitive way of calculating the error is to do the calculation twice - one for all the measurements that err on the side that reduces your final figure and the second for all the values that push the result higher.
Say you want to calculate the error spread of A*B + C*D where A = 1.00 +- 0.02kg, B = 1200 +- 1mm, C = 2.00 +- 0.02kg, D= 1500 +- 1mm.
It does not matter whether this is correct formula or not, we assume it is.
Calculate the lower end result as 0.98 * 1199 + 1.98 * 1499 = 4143.04
The top end result is 1.02 * 1201 + 2.02 * 1501 = 4257.04
You can then write the result as average of two results +- error interval width / 2
or 4200 +- 57 kg*mm in our case.

Notice that I did not write 4200.04 +- 57, it is considered poor taste to keep decimal digits that are rendered useless by error margin in the final figure.

Leo

P.S.  This is still basic error handling without going deep into statistical nature of error distribution, systematic, random ones, etc.
Logged
giladmark
Copper Member
**

Kudos: 0
Offline Offline

Israel Israel

Posts: 4



Ignore
« Reply #70 on: May 01, 2013, 10:43:10 AM »

Thanks Leo & Tony for your replies.
I've calculated the error on one example test and got about 2.4% error.
This doesn't include human error (its not easy for the eye to be accurate observing 1 mm intervals while stretching the rubber, meanwhile the tension reduces every second passing).
If the error is the same magnitude of the variance in test results it's no use performing it, am I wrong?
   
Logged
lbodnar
Bronze Member
***

Kudos: 1
Offline Offline

United Kingdom United Kingdom

Posts: 19



Ignore
« Reply #71 on: July 21, 2017, 01:37:07 PM »

Few pictures of rubber testing process as viewed with a thermal camera - for your entertainment.

Our pull/release speeds are quite low and yet we still see rubber temperature increasing and dropping by almost 10ÂșC against ambient.

The images are:
1 Before test
2 Pull
3 Release

Thanks
Leo
Attached files Thumbnail(s):
Re: Rubber comparative test.
Re: Rubber comparative test.
Re: Rubber comparative test.
Logged
Tmat
Titanium Member
*******

Kudos: 64
Offline Offline

Canada Canada

Posts: 2,884




Ignore
« Reply #72 on: July 22, 2017, 02:11:30 PM »

Interesting Leo and what I'd expect. When pulling, the rubber heats up above ambient, then when released it gets colder than ambient.
I had a discussion some years ago with Fred Pierce (USA) about the possibility of using a thermal imaging device as a way of telling if break-in has occurred. The idea was that you would look at the ambient temperature of the rubber before pulling. Then pull the rubber and maintain a constant stress (increasing distance as required until the motor stops relaxing) until the temperature returns back to ambient. At that point all of the break-in should have occurred. Then relax the rubber and let it rest.

I've never pursued this as I'm not convinced that it is all that important.

But very interesting work regardless!

Tmat
Logged

F1B guy...
But don't hold that against me!
lbodnar
Bronze Member
***

Kudos: 1
Offline Offline

United Kingdom United Kingdom

Posts: 19



Ignore
« Reply #73 on: July 22, 2017, 02:50:57 PM »

When pulling, the rubber heats up above ambient, then when released it gets colder than ambient.
I had a discussion some years ago with Fred Pierce (USA) about the possibility of using a thermal imaging device as a way of telling if break-in has occurred. The idea was that you would look at the ambient temperature of the rubber before pulling. Then pull the rubber and maintain a constant stress (increasing distance as required until the motor stops relaxing) until the temperature returns back to ambient. At that point all of the break-in should have occurred. Then relax the rubber and let it rest.
I've never pursued this as I'm not convinced that it is all that important.
Rubber plays few roles here - [reversible] heat engine, spring and a heat exchanger.
We are trying to do both pre-stretching and grading. Pre-stretching is relatively simple: constant stress at maximum extension is mostly defined by constant force, since cross-section does not change much (assuming ideal liquid rubber model.)
Grading on the other hand involves calculating injected and extracted energy ratio and heat exchange tends to messes it up.  To be quantitative, ideal pull/release cycle has to be done either very quickly (so that thermal energy remains inside rubber) or very slowly (so that heat dissipates and rubber remains at ambient temperature.)  Neither of them are practical so it's always an annoying compromise.
Leo
Logged
Pages: 1 2 [3]   Go Up
Print
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines Valid XHTML 1.0! Valid CSS!