Jump to content


Photo

amateur buliders and three dimensional space


  • Please log in to reply
21 replies to this topic

#1 mariner

mariner
  • Member

  • 1,381 posts
  • Joined: January 07

Posted 04 February 2012 - 10:57

No , I am not thinking of building the Starship Enterprise, this post is about how an amateur builder ( in which I would include some race shops) can achieve an accuracy level for chassis build close to the big OEM's.

When I built my toy car I cheated by buying a Corvette C4 frame and first bolting it to the floor on adjustable stands , then cutting most of it away. All that was left were the truncated side rails, the " horns" to the rear axle and the roll hoop/rear bulkhead. However that gave me a square and rigid basis to add all the new bits. I did it that way because my welding skills are not up to starting from scratch and I was worried about keeping everything a accurate as possible. even so my real tolerance was +/- 5mm across the full wheelbase of 2.4M.

Now I have seen spaceframes welded up by top fabricators on a flat bed and although they can beat my accuracy I don't think they are THAT accurate unless a full jig with the suspension holes rigidly located is used. Even then you still have the problem of fixing the jig points in free 3D space.

All I read these days is how modern OEM designs hold the suspension mounts to an accuracy of under 1 mm. If you are Red Bull etc you have a CFD/CAD/CAM system that can make the whole tub to tiny tolerances.

So any suggestions/thoughts on how an " amateur" bulder can acheive +/_ 1mm tolerance and do you need some sort of 3D free space measurement system external to the jig to do it.

One thought I had was to make your jig adjustable then trailer it down to a four wheel alignment fixture at a tyre shop to set up the key points on a jig which you place onto the 4 wheel alignment rig to set it up with adjuster screws then weld them solid and take the now precise jig back on a trailer to where you are fabricating. I believe those 4 wheel alignment systems are very accurate and most towns have one.

OR ( Greg?) maybe you don't need +/- 1mm on suspension points in real life whatever OEM's calim??




Advertisement

#2 cheapracer

cheapracer
  • Member

  • 10,388 posts
  • Joined: May 07

Posted 04 February 2012 - 11:15

The shock mounts are always the very last thing I fully weld after final chassis welding and checking/alignments.

A well clamped down chassis doesn't move too much and you kind of get the feel for where to add a bit more weld to get some compensating pull after a while.

#3 Greg Locock

Greg Locock
  • Member

  • 4,543 posts
  • Joined: March 03

Posted 04 February 2012 - 11:30

The only way you could achieve a true (3-5 sigma) tolerance of 1mm in production would be to machine the suspension points AFTER assembly and welding.

1mm sounds too little, in general. You'd probably manage that on the spindle, between the notional wheel centre and each of the 3 ball joints, but even then it is on the tight side. A typical problem is how much pull-down you get in tapers on ball joints, in aluminium spindles.

On crucial relationships in spotwelded assemblies, such as steering rack height vs lower arm height the usual approach is to make sure that the two parts are located on the same piece of metal. Even then some holes are postpierced.

Note that some dimensions don't matter all that much, for instance the width of the car doesn't really matter. The tolerances inside each corner are far more important.

Most OEMs shim the mounting brackets for say upper arms and trendset them (based on the previous days production) for castor and camber. However I know of a Japanese manufacturer that measures the castor and shims every vehicle in one plant. I'd get lynched for proposing that.

Incidentally the salesman will tell you that a CMM will measure all these things to better than 1mm. Set up a realistic demo and let him measure the same car twice. CMM needs a lot of experience before it is reliable. Spanning introduces very large errors.


#4 mariner

mariner
  • Member

  • 1,381 posts
  • Joined: January 07

Posted 04 February 2012 - 17:50

Thanks guys , you have made me feel better about my +/- 5mm accuracy, which I had to correct by shimming things. My main problem was over the full wheelbase which is where the error got to about 10mm side to side. The 4 wheel alignment got the thrust lines etc set up square as the rear toe was easily adjusted.

Greg's comments remind me of the crude but quite clever technique used on the awful mid 1980's Ford Fairmont to set up the front strut geometry.

The top mounts had the usual oversized holes and the struts were moved in these to get the right camber/castor. The top plate had small pre-drilled holes and when things were aligned ( hopefully) a quick drilling down into the body panel was followed by inserting a pop rivet. Result - setting held whilst top bolts tightened and a locating hole for any future strut swaps.

At the other end of the scale the Pontiac Fiero ( remember that?) had a steel internal frame on which they hung plastic outer panels. I think the idea was ease of re-skinning but that didn't ever happen . Anyway square sockets in the inner frame were filled with resin and the whole car went into a huge cage with drills aligned where each panel bolt was located.

The drills went into the resin pads and ( in theory) every panel was perfectly ailgned regardless of inner structure welding tolerances. I guess you could modify the approach to do all the suspension mounts in one super accurate go but the frame drilling rig was HUGE and must of cost GM a fortune it never recovered in Fiero sales.

#5 Greg Locock

Greg Locock
  • Member

  • 4,543 posts
  • Joined: March 03

Posted 04 February 2012 - 20:59

The original Renault Espace, designed and built by Matra, used much the same technique as the Fiero. Lotus was quite interested in that at one point, as a method for building low volume cars with reasonable panel gaps.

I wouldn't feel too bad about 10mm in wheelbase, that is probably about the same as production. As you say thrust angle and crabbing are easily accounted for. One common production problem is stagger between the chassis rails, where say the LH rail is 5mm rearward of the right hand one. So your subframe mounting holes have to allow for that, so when you set the subframe square to one rail then you've automatically got 5mm in wheelbase, from just ONE tolerance.

For many years one company used floorpans that could be welded to account for the possible 25 mm error in width between the chassis rails, width wise.





#6 bigleagueslider

bigleagueslider
  • Member

  • 862 posts
  • Joined: March 11

Posted 05 February 2012 - 03:13

mariner,

As Greg Locock alluded to, rather than ask what can be achieved in terms of dimensional tolerance, you should first ask what dimensional tolerance is really necessary for your application. If you get hold of a good factory chassis service manual, it will give you dimensional limits for various points on the chassis. This will give you an idea of what the factory works to, as a starting point

As for your project, holding tight tolerances (+/- 1mm) in large "as welded" aluminum structures is next to impossible. Welding aluminum, with its high thermal conductivity, requires large amounts of heat input due to the high heat transfer rate away from the weld zone. Plus, aluminum's high CTE means lots of local thermal growth during welding. Thus welded aluminum structures tend to have lots of post weld distortion and residual strain. The dimensional distortion problem in welded aluminum structures is compounded by the fact that the post weld residual strains will naturally relieve themselves over time, which will change the shape of the weldment.

If it is necessary to hold tight dimensional tolerances in your welded aluminum chassis for things like suspension mounts, it can be done. You would simply need to make the chassis suspension mounting points as separate bolt-on parts. This way, the mounting points can be accurately located, then match drilled and shimmed. Using this method with a good surface table as a baseline, you should easily be able to achieve +/-0.5mm positional tolerances.

slider

#7 NeilR

NeilR
  • Member

  • 472 posts
  • Joined: October 09

Posted 05 February 2012 - 04:20

I have just welded my centre section and the movement is noticeable at around 2-3mm even with significant care taken. Since all suspension mounts are bolt on and shim-able I am not too worried.

#8 mariner

mariner
  • Member

  • 1,381 posts
  • Joined: January 07

Posted 05 February 2012 - 14:18

Bigleague, this all sort of started because I was reading a book on body repair which has a diagram from the Rover group showing the datum points for crash damage checking on a Rover 800.. They specified +/- 3mm. So my +/-5mm seemed poor given that is a mass production car etc.

I then checked the GM service manual for my CTS and the chassis datums are given in mm with NO tolerances!

However, after Greg's comments I looked again at the Rover diagram and , interestingly, they do not specify any datums across the whole wheelbase so there is no evidence that they hold +/- 3mm all over the chassis.

What you do about a GM diagram with zero tolerance I don't know, I will read it in more detail

#9 Charles E Taylor

Charles E Taylor
  • Member

  • 175 posts
  • Joined: December 07

Posted 05 February 2012 - 16:07

No , I am not thinking of building the Starship Enterprise, this post is about how an amateur builder ( in which I would include some race shops) can achieve an accuracy level for chassis build close to the big OEM's.


Holding tight tolerances on one-off engineering structures is difficult, especially if you have limited funds. You will find that with a little application you can achieve some remarkable results.

In the modern race car world, composite structures and metal fabrications are treated as if they were castings, most are CNC machined after fabrication. This style of manufacture is not generally achievable for a one off amateur build project.

Nevertheless very good accuracy can be achieved by careful design. Many mistakes are made by referencing external co-ordinate systems.

You may find that you can gain much more accuracy by using the Article (Vehicle) itself as a reference system. This ability needs to be designed in from the start. Reference holes are a good (Cheap) way of doing this. Plumb lines (Fishing Line) can be positioned through these holes to assure correct alignment.

Once familiarity with this is gained, you will find that this is a good, quick and very inexpensive method of making accurate structures, especially if they need to be symmetrical. This method, (alignment holes) together with gravity assistance, (Plumb Lines and Water Levels) has been in use for centuries; -do not let High-Tech cloud good practice. You might want to search how some small aircraft manufacturers achieve accurate geometry with wing assembly using nothing more that Plumb Lines.

Over 100 years ago many “hi-tech” people found out that “you should never underestimate a bicycle mechanic”.


Above all, Have Fun!


Charlie


#10 cheapracer

cheapracer
  • Member

  • 10,388 posts
  • Joined: May 07

Posted 05 February 2012 - 16:24

Plumb Lines and Water Levels .. have been in use for centuries


Indeed, plumb bobs, spirit levels and large squares are in constant use through my builds with the spirit level the most used, easiest and very accurate.

It's a joy sometimes seeing old world tools and techniques being used here on a daily basis by tradesmen.

Plan ahead for the inevitable as well, a corner clamped down to the floor, two bits of hardwood and a car to drive on to the other corner has straightened many a race car chassis.

Edited by cheapracer, 05 February 2012 - 16:27.


#11 mariner

mariner
  • Member

  • 1,381 posts
  • Joined: January 07

Posted 05 February 2012 - 16:52

"Indeed, plumb bobs, spirit levels and large squares are in constant use through my build with the spirit level the most used, easiest and very accurate."

That's more or less what I used too. As I mentioned I fitted the cut down corvette chassis rails with brackets bolted to screwjacks.These were bolted to the floor with concrete bolts to lock everything down.

Then I levelled the whole chassis using the screw jacks and worked from there.

I laid in centre line etc and fitted the front motor plate first as that holds the front suspension and could be marked off.

that got me the +/- 5mm accuracy end to end but I have often pondered how to do better hence my questions.

A big setsquare and a long level are certainly the key items!


#12 24gerrard

24gerrard
  • Member

  • 2,008 posts
  • Joined: March 11

Posted 05 February 2012 - 18:33

Perhaps you need a bigger set square mariner.
Joking apart the guys on here sure know how to build a chassis.

I think if I was to do one today without a factory and all the kit, I would try using lasers.
I know that sounds expensive but I dont think it would have to be.

#13 Wolf

Wolf
  • Member

  • 7,881 posts
  • Joined: June 00

Posted 05 February 2012 - 19:19

If I was building a spaceframe chassis, I think I would try to come up with some sort of jig that would allow all front/rear wishbone mounts to be positioned and tack welded at once... Would something like that even be possible or practical?

#14 carlt

carlt
  • Member

  • 1,064 posts
  • Joined: June 09

Posted 05 February 2012 - 19:59

some of the classic cars i've prep'd for competition have over 1" tolerance on suspension mounts over the wheelbase before we re set them
I guess with all the rubber bushes it wasn't so important - let alone with skinny crossply tyres

#15 Greg Locock

Greg Locock
  • Member

  • 4,543 posts
  • Joined: March 03

Posted 05 February 2012 - 21:55

I think if I was to do one today without a factory and all the kit, I would try using lasers.
I know that sounds expensive but I dont think it would have to be.

Yup, when working over the length of the car say for driveline alignment a mirror and a laser can both be mounted to turned fixtures which can be mounted into any accurately machined hole. This allows you to measure both axis misalignements and crucially angles in 2 dimensions, over large distances.



#16 NeilR

NeilR
  • Member

  • 472 posts
  • Joined: October 09

Posted 05 February 2012 - 23:50

I have been using a digital protractor which is 'accurate' to .1 degree. Has been very useful. Otherwise I have been using a lockable steel extension to check corner to corner measurements. The 2-3mm discrepancy so far is my error on the dash bar. The rest is square to 1mm and angles side to side of less than .5 degree. Again with rubber bushes and adjustable mounts I don't think it matters apart from pride in doing a good job.

#17 Lee Nicolle

Lee Nicolle
  • Member

  • 6,035 posts
  • Joined: July 08

Posted 06 February 2012 - 00:06

I have never built a racecar chassis, but have repaired quite a few. In my experience they are usually within about 1/2". Though really the only real concern is where the wheels and the weight is and most cars have some sort of suspension adjustability front and rear. If not get VERY critical where your suspension mounting points are as a car even 5mm short in wheelbase will be noticeable at high speeds and the balance may well be just a bit 'off'. But if the wheels are square and true with each other the pickup points are not as critical. But the squarer they start the easier it is to set the car up.
A LOT of production cars are way out, even now.A simple tape and string will tell you that. That is why some are lemons from new. All brands.
I had a front drive family hack here recently that has always wandered left, even with new tyres. It is nearly an inch shorter on the left than the right. And has never been involved in an accident. The corner weights are miles off too. And no real adjustment. Really it drives really well for being so far out!!
I have built several car trailers and have had no trouble getting those simple structures dead square, both to the frame and the suspension pickups.A racecar is a tad harder though !!

#18 kikiturbo2

kikiturbo2
  • Member

  • 638 posts
  • Joined: December 04

Posted 06 February 2012 - 00:31

I have been using a digital protractor which is 'accurate' to .1 degree. Has been very useful.



I use one too and it is very useful tool.. I also used a cheap laser to serve as a chassis center line during alignment of various parts of the frame..

#19 NeilR

NeilR
  • Member

  • 472 posts
  • Joined: October 09

Posted 06 February 2012 - 00:40

I tried a cheap laser as well but couldn't get it to do what I wanted. In the end I ran a 2mm wide groove up the centre of the build table as a datum line. Once the base chassis tubes are locked into place I have found little use for it.

Edited by NeilR, 06 February 2012 - 00:41.


Advertisement

#20 Tony Matthews

Tony Matthews
  • Member

  • 17,499 posts
  • Joined: September 08

Posted 06 February 2012 - 07:41

If anyone wants a cheap line laser I suggest checking building tool suppliers - I have one that generates a line of sufficient 'depth' to go over modest projections, and about 300mm up a perpendicular surface. Or wall, as we call them in the trade...

Edited to add that if cheapness is not a factor, you can get self-leveling cross-type lasers for not much more.

Edited by Tony Matthews, 06 February 2012 - 07:44.


#21 PhilG

PhilG
  • Member

  • 474 posts
  • Joined: September 06

Posted 08 February 2012 - 22:12

Being a professional measurist, its always interesting to see how different people approach stuff like this.

Having a big gantry type measuring machine(CMM) at my disposal, big enough to measure a full F1 tub on helps no end, but all of the important stuff is machined pretty much together, once the front nose and rear faces are done the rest is done off that, with the majority of the machining being on the bottom and sides , meaning the thing is done upsidedown, which is how i always checked them, so you could get to everything.

For WRC cars, its a totally different animal, a portable Arm type CMM is the kiddy for that, the cars are built up on a steel flatbed , with grooves every 250mm across and down , so you have an accurate grid to work from... first thing that goes in is the turrets, the jigs are made in sections , with risers etc to get the car in the air, and then assembled after inspection.. then inspected as an assembly on the grid to put it in exactley the right place, because it was adjustable to a small degree, bt design , there was no reason why it couldnt be perfect, but in reality 0.1mm was considered a result , and all four could be put in pretty much to that.

Then everything else came off the slots, with jigs for B post pins , crossmember mounts etc. Plus all the gearbox , engine mounts and roll cage stuff.

The clever bit at the end was inserts in the chassis for the crossmembers , where we got under the car , measured the true position of the mounts, and then fitted whatever insert was required to put it right to the strut tops.. it was a lot of work , but meant that everything was as good as it could be when the shell was fresh, plus you could always drop it back on and assess any damage if you had to.. we also had datum points in the floor that allowed a quick check back to the turrets, so you always had a good known reference.

As an amateur the best thing i can suggest is to build in enough adjustability so you can tune out any errors as best you can at the end.

#22 mariner

mariner
  • Member

  • 1,381 posts
  • Joined: January 07

Posted 09 February 2012 - 02:35

Thanks for all the comments, I guess it is a case of diminishing returns on accuracy terms of cost versus adjustablity. If you can adjust by say 2mm then why go below +/- 3mm on overall accuracy and +/- between what points - only some of them actually matter.

I believe it was Gordon Murray who started the trend of fully machined bulkheads. If it is a single seater then you can be very accurate on all the key mounting points in two planes if you have a machine capable of handling a piece say 800mm by 400mm. Then all you have to do is get those pieces set up vertical and parallel on a truly one plane base to be very accurate. I would think you can get the base plate flat by using screw jacks in the floor and levels etc.

I would guess that would let you hold things below +/- 2mm.

The key seems to be working out from the final suspension mounting points and letting the rest fit ( per PhilG above).
'
Interestingly there is a pic of the jig for the Lotus 25 in Ian Bamsleys book which shows the monocoque in a simple wood and steel plate jig. The jig base is 4" * 4" wooden beams with similar sized uprights held by screwed-on plywood triangles.The actual jig points are on thick steel plates across the wooden uprights at each end of the chassis. It looks very much like the front and rear cross members were made first then bolted to the jig by the suspension mounts and the paneling riveted up between the cross members. Esentailly per PhilG's description but using very simple materials.

The result wa certainly effective even if the jig was made of wood and steel plate!