Designs on software v/s practical stuffs

[Marvel]

Hello,
Its been a long time I was wondering; how can I actually come to a conclusion (economically*)that the stuffs I am doing on software and the ones I am doing while actually manufacturing are close enough(may be not exactly...manufacturing errors!)
For example,
how can I conclude that the position of suspension brackets while actually mounting it on the chassis is slightly the same that was decided on software. There may be some deviation but this will deviate many other parameters. For instance, there may be some deviation from initially decided static camber suppose -2degree to somewhere -2.1 or even -3degrees or 0degree?
This will further change some other characteristics like the camber again, contact patch...etc...etc...
I understand this all depends on manufacturing processes and proper care is to be taken but a slight deviation of 2-5mm can change many parameters.
economically* - without any sensors etc...
(I am myself confused...but I hope my words are clear enough to understand what I want to say... http://fsae.com/groupee_common/emoticons/icon_rolleyes.gif )

[Marvel]

Hello,
Its been a long time I was wondering; how can I actually come to a conclusion (economically*)that the stuffs I am doing on software and the ones I am doing while actually manufacturing are close enough(may be not exactly...manufacturing errors!)
For example,
how can I conclude that the position of suspension brackets while actually mounting it on the chassis is slightly the same that was decided on software. There may be some deviation but this will deviate many other parameters. For instance, there may be some deviation from initially decided static camber suppose -2degree to somewhere -2.1 or even -3degrees or 0degree?
This will further change some other characteristics like the camber again, contact patch...etc...etc...
I understand this all depends on manufacturing processes and proper care is to be taken but a slight deviation of 2-5mm can change many parameters.
economically* - without any sensors etc...
(I am myself confused...but I hope my words are clear enough to understand what I want to say... http://fsae.com/groupee_common/emoticons/icon_rolleyes.gif )

[Tilman]

Heyho,

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Marvel:
economically* - without any sensors etc... </div></BLOCKQUOTE>

Take care of your manufacturing processes. But the only proper way to tell wether you positioned your parts/points/whatever correctly is to measure them. You can put lots of effort and money in precise manufacturing, but measuring may be easier and cheaper.

Here is a photograph of how the Delft team positioned their suspension brackets, and I believe I once saw a similar photo on the Stuttgart website.

http://dutracing.nl/wp-content...s-dut10/img_6803.jpg (http://dutracing.nl/wp-content/gallery/chassis-dut10/img_6803.jpg)

[RollingCamel]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Tilman:
Heyho,

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Marvel:
economically* - without any sensors etc... </div></BLOCKQUOTE>

Take care of your manufacturing processes. But the only proper way to tell wether you positioned your parts/points/whatever correctly is to measure them. You can put lots of effort and money in precise manufacturing, but measuring may be easier and cheaper.

Here is a photograph of how the Delft team positioned their suspension brackets, and I believe I once saw a similar photo on the Stuttgart website.

http://dutracing.nl/wp-content...s-dut10/img_6803.jpg (http://dutracing.nl/wp-content/gallery/chassis-dut10/img_6803.jpg) </div></BLOCKQUOTE>


Ehem......kinda close to what we used..."it looks ok to me".

[Z]

Marvel,

You have discovered what 90% of real engineering is. Control of tolerances.

This includes dimensional tolerances, but also quality of materials ("so this is an aerospec bolt, eh?"), costs ("but you quoted...???"), and timelines ("two months ago you said a couple of days!!!").

All those 8+ digit numbers coming out of the computer look quite silly when you struggle to get the part positioned to the first digit.

To improve your chances I suggest:

1. A largish (say 1+ metre x 2+ metre) "build" table with a horizontal surface flat to within +/-1mm.

2. A "bubble level", preferably 2m long. Maybe two of these. They are accurate to less than 1mm over their length, and not too expensive.

3. Some reasonably accurate right-angle "squares", the bigger ones with about 0.5m sides, preferably clearly dimensioned on the outer edges.

4. Some good steel rulers and tape measures.

5. Work in millimetres, and ignore most of those digits from the computer... http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

[Warpspeed]

This is why we always build some adjustability into the suspension.

Even if you think you know what is going to work best, some actual track testing may prove your pet theories a little bit wrong.
That is when the spanners come out, and you get to try a few different ideas.

Cumulative tolerances are the curse of practical engineering. But a few shims, some strategic threaded adjustments, and the odd alternative hole here and there can work wonders.

[Tilman]

Heyho,

there is a way to easily measure points in 3d space: With appropriate software, you can determine points in 3d space from some good photographs. I know there are companies who sell such software and objects to mount to your vehicle which will be recognised on the photos. And you need a good camera.

The theory behind all this is quite simple, however, you need to know a bit on the camera settings like focal length to determine the camera matrix (function to map a 3d point to the 2d picture).

I do not know whether there is free open source software available to do this.

[mk e]

One last point is to spend some time on the software doing a tolerance study so you actually understand how close is close enough.

It's much easier that you'd think during your optimization work to find a design that work properly at one and only 1 condition. You avoid that trap by running cases with the dimensions of each part altered. Excel has an add-on to do what's called a Monte Carlo analysis. Here you give all the components a reasonable tolerance, assume a distribution (pick normal) and then the software will build assemblies and test there performance against an equation and spit out the the performance profile distribution....so you know what could happen if you do what you're planning to do.

It's boring work and we didn't do it when I was building a car, we barely had time to actually build the car, but in industry this is the kind of work I expect to see if they expect to have engineering approval on the design or change because it's pretty important.

[bminus]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Tilman:
Heyho,

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Marvel:
economically* - without any sensors etc... </div></BLOCKQUOTE>

Take care of your manufacturing processes. But the only proper way to tell wether you positioned your parts/points/whatever correctly is to measure them. You can put lots of effort and money in precise manufacturing, but measuring may be easier and cheaper.

Here is a photograph of how the Delft team positioned their suspension brackets, and I believe I once saw a similar photo on the Stuttgart website.

http://dutracing.nl/wp-content...s-dut10/img_6803.jpg (http://dutracing.nl/wp-content/gallery/chassis-dut10/img_6803.jpg) </div></BLOCKQUOTE>
calling measuring easy and cheap... and showing a picture of a $50,000 Faro arm.

for the intents and purposes of this project, I agree with everything Z said.

[OspreysGoSWOOP]

I understand your pain.

You basically need to mimic the X,Y,Z coords from CAD over to real life as closely as possible.

Here is how our first year team tackled the issue. There are better methods but this is the best we could do given our tools & equipment.

I'm going to make 4 assumptions:
Your chassis is square and accurately built.
You have a flat, level building surface.
You have access to a mill.
Your a-arms are built accurately and you know your chassis mounting solution.

We created a coordinate system in CAD that we could transfer to real life.
1. CAD your build surface then mount the chassis securely at ride height.
2. CAD brackets to mount your outboard suspension points.
3. Mark the origin of your table's coordinate system. We used the first lower bar of the chassis front tube. Using a tape measure we marked the centerline then hung a plumb off the bar and marked the 0,0,0 point. This gave us a reference in both real life and CAD.
4. CAD jigs to reference your outboard points.
5. Manufacture jigs. NOTE: If you weld the jigs like we did you must machine them so the bottom and the front are a true 90deg to each other.
6. Using a mill, we zeroed off a corner of our jig and drilled the suspension holes. Your jigs are now complete and must be fastened to the table.
7. In CAD, measure from your origin to your JIG location. Use these numbers to locate your jig and fasten to the table.
8. Bolt A-arms to the jig.
9. Locate on chassis.

Here are some pictures

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-ash4/s720x720/391925_10150565948795767_333774555766_11462483_101 0970099_n.jpg

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-ash4/s720x720/417992_10150621675375767_333774555766_11652944_130 6516776_n.jpg

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-snc7/s720x720/380906_10150521126025767_415945619_n.jpg

[Marvel]

Thank you all for help...
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Z:
Marvel,

You have discovered what 90% of real engineering is. Control of tolerances.

This includes dimensional tolerances, but also quality of materials ("so this is an aerospec bolt, eh?"), costs ("but you quoted...???"), and timelines ("two months ago you said a couple of days!!!").

All those 8+ digit numbers coming out of the computer look quite silly when you struggle to get the part positioned to the first digit.

To improve your chances I suggest:

1. A largish (say 1+ metre x 2+ metre) "build" table with a horizontal surface flat to within +/-1mm.

2. A "bubble level", preferably 2m long. Maybe two of these. They are accurate to less than 1mm over their length, and not too expensive.

3. Some reasonably accurate right-angle "squares", the bigger ones with about 0.5m sides, preferably clearly dimensioned on the outer edges.

4. Some good steel rulers and tape measures.

5. Work in millimetres, and ignore most of those digits from the computer... http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z </div></BLOCKQUOTE>
Sir, those were some really reasonable solutions..(most of which were actually used...)
but while manufacturing, when I was asked why am I so worried even for 1mm??How am I going to check whether my "1mm" is correct??
Basically, how close is close enough..
Nonetheless..thank you for your valuable advice...

[Marvel]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by OspreysGoSWOOP:
I understand your pain.

You basically need to mimic the X,Y,Z coords from CAD over to real life as closely as possible.

Here is how our first year team tackled the issue. There are better methods but this is the best we could do given our tools & equipment.

I'm going to make 4 assumptions:
Your chassis is square and accurately built.
You have a flat, level building surface.
You have access to a mill.
Your a-arms are built accurately and you know your chassis mounting solution.

We created a coordinate system in CAD that we could transfer to real life.
1. CAD your build surface then mount the chassis securely at ride height.
2. CAD brackets to mount your outboard suspension points.
3. Mark the origin of your table's coordinate system. We used the first lower bar of the chassis front tube. Using a tape measure we marked the centerline then hung a plumb off the bar and marked the 0,0,0 point. This gave us a reference in both real life and CAD.
4. CAD jigs to reference your outboard points.
5. Manufacture jigs. NOTE: If you weld the jigs like we did you must machine them so the bottom and the front are a true 90deg to each other.
6. Using a mill, we zeroed off a corner of our jig and drilled the suspension holes. Your jigs are now complete and must be fastened to the table.
7. In CAD, measure from your origin to your JIG location. Use these numbers to locate your jig and fasten to the table.
8. Bolt A-arms to the jig.
9. Locate on chassis.

Here are some pictures

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-ash4/s720x720/391925_10150565948795767_333774555766_11462483_101 0970099_n.jpg

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-ash4/s720x720/417992_10150621675375767_333774555766_11652944_130 6516776_n.jpg

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-snc7/s720x720/380906_10150521126025767_415945619_n.jpg </div></BLOCKQUOTE>

Looks very similar to what I have been thinking...I have not yet tried but just wanted to share..
1. Make a complete 3 dimension co-ordinate system(table as the reference) with the jig having a hole/bore all over its length (to adjust the height)to fasten the members.
2. Then make an adjustment to move your jig along say, x and z co-ordinate(may be a long horizontal groove in the table to which a jig is fastened at the bottom..so that it can be moved..)
3. Get the desired height of member(y co-ordinate) with the help of "multi-slot" jig to which member can be fastened.
4.Use most of the devices/instruments(as said by Z)..
(I think the only issue that may arise is while adjusting the angles..)
I will try to make some CAD diagrams to get a better visualization of what I am trying to convey.

[Z]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Marvel:
Basically, how close is close enough... </div></BLOCKQUOTE>
Marvel,

The biggest part of "control of tolerances" is knowing how accurate your tolerances NEED to be. A big disadvantage of all the CNC machine tools available today is that many engineers have forgotten (never learnt?) what this means. Today most parts are automatically made more than accurate enough, even when they don't need to be.

In the "olden" days the blacksmiths would only spend time making things accurate, that really needed to be accurate. The parts were mostly made from rough and ready forgings and castings, and only the critical surfaces were made accurate, usually by hand scraping or filing, because that was hard work!

Some examples in FSAE;

1. The nose of the car can probably be 30+mm out of position with no noticeable effect. Likewise other "cosmetic" body parts.

2. The chassis engine mounting holes have to be accurate to within +/- ~1mm, otherwise the engine won't fit! This can be done relatively easily by using a good jig, perhaps a spare engine case bolted to the mounts during chassis build.

3. Wheel toe angles should be "right" to +/- ~0.1 degrees, which might mean +/- ~0.1mm at the toe-base. This can be done with very thin shims, or with the threaded adjustment of standard tie-rods.

4. So how accurate do your wheel camber angles need to be? You can do a "sensitivity analysis", as suggested by mke above, but a quick look at typical Fy vs Camber curves, and some common sense, suggests...?
~~~o0o~~~

With regard to my previous post, the most important bit of equipment is the "build table". Make this both solid and accurate enough and it can serve for both construction, as the jig, and also for final measuring, etc.

Something like a cast-iron bed from a very large mill, with its built-in T-slots, is ideal. The base should have screw adjustments to get the top surface as close to flat and horizontal as possible. With the table top horizontal you can then check whether points have the same z-coord by using the bubble level. (BTW, usually x = longitudinal, y = lateral, z = vertical).

Other options are a billiards/snooker/pool table with its slate surface. Or I have used an 8' x 4' x 5/8" (2.4m x 1.2m x 16mm) "compressed cement" sheet on a suitable base. I currently have a 1/2" (13mm) thick steel table top, but this has a bow in it, and it picks up weld splatter. Or kitchen suppliers make reasonably accurate MDF counter tops with laminex finishes, or, top-of-the-line, granite! http://fsae.com/groupee_common/emoticons/icon_smile.gif

Z

[rrobb]

The well funded teams could always use something like this:

http://www.bluco.com/applicati...cles1/GO%20Cart.html (http://www.bluco.com/applications/vehicles1/GO%20Cart.html)

[safetyfactorof1]

1) You may be surprised at some of the equipment (think Faro arm) your department has that isn't readily advertised or widely known about. Time to start talking to professors. FWIW, we do not use the one we have access to for our chassis.

2) Take your time to manufacture and plan out jigging. Do it right, and it can be a great way to teach your freshman how to machine some parts early in the semester. You don't NEED a bluco table either. For under 1k and some resourcefulness, you can get a decent size steel plate, have a standardized grid water jet/laser cut/CNC plasma cut into it and do all of your jigging off of that. After the initial investment, it might be ~200/yr to make parts that cant be carried over/adapted. Money well spent. Last year the major structure of our chassis was completed in a single 24 hour push, and we only have 1 TIG machine.

3) Establish some sort of reference posts at your corners with set height markets and you can use them like toe bars for double checking your positioning with a tape measure for sanity checks.

[Mazur]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Marvel:

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-ash4/s720x720/391925_10150565948795767_333774555766_11462483_101 0970099_n.jpg

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-ash4/s720x720/417992_10150621675375767_333774555766_11652944_130 6516776_n.jpg

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-snc7/s720x720/380906_10150521126025767_415945619_n.jpg </div></BLOCKQUOTE>

I'm confused on what you're trying to achieve here. Is this how you located the pickup points on the chassis? If so, I would consider that a very bad way of doing it. You need a fixture that locates these points in direct relationship to your chassis.

[ed_pratt]

Hi,

Going slightly further than Z's method..

How about taking that basic flat table and starting at your suspension points?

With a little thought you could easily hold the hardpoints in 3d space. Then put in the roll hoops and bulkheads and work inwards. This way you would at least know that each corner was in the correct place relative to all the others, even if some of the chassis members were off by a mm or two.

On another note, one of our electrical engineering guys has been fiddling with an MS kinect for pc and has found that it can be used as a 3d scanner with a +/-1mm level of accuracy within a reasonable range.
This may be a good way for a team with very limited budget to get their hands on a 3d scanner....

Ed