Some Seam Weld Photos
 

Ugly welds but you can see some of the zones you need to hit. I saw these at an S2000 engine swap thread.

Another one...

BTW why are the factory welds weaker than these?

http://www.lightweightmiata.com/v8/seam/

This site tells it pretty well, basically Mazda just placed all those sheets of metal together and tack welded them, often not even getting all the sheets in each weld.

I don't see how smearing a few boogers in your door jamb is going to tighten everything up

Isn't it already known that stitch welding helps?

real stitch welding, sure, but that looks like it's only warping sheetmetal

+1...

those welds would help, but judging by how they look there was not much penetration of the parent and filler (cold weld)

This is just helping your car rust faster on the back side of the welds that you cant access.

if you want to really want to tighten everything up, just weld in a cage like a boss would

Like I said, ugly. In racing…results are mostly due to the person and not the equipment. Same thing in welding. Seam welding does beef up a unibody. But it has to seem like welding haha. ;-)

Most seam welding is actually pretty forgiving. The glob and grind crowd can succeed here. It can be a decent way to improve your welding skills. Even those ugly welds probably help a great deal and of course they can be cleaned up. There is certainly room for improvement.

You really want to lay a bead of weld to really get some strength. You can basically assume the first and last .25" of a weld aren't really holding anything together. IIRC it is a safe to assume that about 60% weld coverage (weld area vs. total length of area welded) is considered to be a "solid weld".

So when Mazda tightened up the 01, did they improve the welds?
Or was it cheaper for them to do what they did instead? (add the spiderweb bracing under the prop shaft)

Do other car manufs, on more expensive cars, use better welds?

the purpose of seam welding in most cases is getting rid of the the seam "glue" that is used instead of welding on 90% of the unibody. i hope you guys are joking. you guys sound like newbs from those "other" miata forums.

and no, the manufacturer will never get close to the same chassis stiffness adding braces, compared to what a raceshop would get from seam welding.

yes, other maufacturers use more welds or a stiffer chassis design (or tub) to account for the seam filler thats used.

So why don't manufs do similar welding instead of adding weight and cost with braces?

What is this seam filler glue stuff?

All I know is that I did a similar job to my door/window frame seams. Now when I jack up one corner, the door does not rub against the sill like it used to. Could be all in my head, but something is not flexing as it once was.

I don't know about the '01s, but in general NBs have stiffer unibodies than NAs. If you look at the interior floor pan sans carpet you can see part of how they tweaked the deal. A bit more metal or bracing where it would help.

I've heard claims that the early EVOs were factory seam welded. Might be cool to pop off the door opening seals or look elsewhere to check it out. Oh yeah, certain 911s like the GT3 too, supposedly. Drool.

The cost to weld up a chassis like what we are talking about would add a ton of cost and complexity to the design. I imagine that they have to use special robots and jigs on exotic cars to keep them from warping during welding.

BTW…there are different strategies or philosophies on seam welding. Some folks like ending up with one solid weld like you see at the lightweightmiata site, above. Others suggest doing some pattern such as 1 inch welds with some spacing between each weld.

Another aspect is that the sheets of metal can have gaps between them. You can see this on most any Miata by popping off the door opening trim by hand. Like here:
https://www.miataturbo.net/attachmen...ine=1312514817

My thought there is that any gap found after you clean and prep before welding should be clamped shut. Use vice grips or something like that. Others may feel differently and just want to fill the gaps with more weld material.

Because monocoque

The reason for the spacing is to limit crack propagation. If you have one solid weld and a crack starts, it will almost inevitably spread through the entire weld, ruining everything you just did. By making numerous smaller welds, a crack will ruin just one weld. The "pattern" or spaced out method would be the right way to do it.

It's not just that. When you do normal welding, the weld and vicinity are harder and more brittle than the rest of the metal. Hence more prone to crack in the first place. By stitching your seam welds, the loads are more spread out. Even if this could be avoided, the brittleness, it would not help the car, since it is always a matter of what is the next weak link in the chain. Any extra strength from a full 100% welding of the seams can't express itself usefully since the next weak link elsewhere will limit you. I think it is best to stitch.

Years back I used to lay layers of composites into some beater car floors, rails and tunnels. Epoxy glue and fiberglass or kevlar, whatever was handy. You know what? It helped a ton. Even before a cage was added. I bet it would help Miatas too. It is labor intensive and there are some tricks to it, but the material costs are not bad at all.

Then there is the old structural foam strategy. You can do it yourself, and many new cars already come foamed. Besides stiffening, it helps with NVH.

An old trick from a buddy at Koni was to strip a car and then drive it hard with that white correctype paint on suspect areas. The stuff they used to use back in the typewriter days. It is opaque but very weak paint and it dries instantly. After your drive you examine and photograph whereever the paint cracked. That is where your unibody is flexing excessively. Fix those areas and your car will wind up faster.

The goal of these structural mods is to make the car stiffer both in bending and torsion. This helps both ride and handling and makes the car easier and more predicable to control.

Here is a picture from a local car I'm seam welding. Seats out, carpet rolled back. Ugh, look at this crummy brace. It is not even achieving stock benefits. This is the driver's side brace that lives behind the lower seat belt mount on the rocker. It will be seam welded of course. I may first cut it off and start again so that it fits flush. Or leave it and tap the gaps flush. This is a very important brace and as you can see it cannot be doing much since it is barely attached to the rocker. Hmm…looking it over again I see a couple more "fails" I should have added. For shame Mazda!!

See, part of doing things like seam welding, corner balancing, and engine blueprinting involves more completely achieving the design intention. In the real world things don't always turn out quite so perfect in cheap car mass production.

Here is a setup weld, getting the settings close on a HF welder using Lincoln wire. Not that bad for the price really. This is just wire brushed for like 5 seconds after the weld cooled. It's a serviceable unit if your budget is tight. Just use better wire than they provide is my tip. You can see the factory spot welds to the left and right of it.

Okay I have the car I just interior seam welded for just a bit longer. So here is an anecdote.

It is an older high mileage Miata. That has trouble with opening a door if the car is jacked at a corner. Jack it and then the doors balk at opening or closing.

It also showed flexy flier signs entering NC driveways. In this area many of the streets have these combination curb and water drainage feature. So they are like a wide but steep curb. On lowered cars you MUST enter them at a steep diagonal. Well this twists the car torsionally along the F/R axis. This older Miata would flash it's interior light when entering or leaving my driveway.

Now the light stays out!

The automatic switches for the doors live at the lower rearmost door jams. That way they activate when you open a door. Well this car would twist enough that the switch would close when you do a driveway entrance or exit. Now that the unibody is stiffer, the interior light no longer flickers. Definitely the car is more solid.

I will see if the guy wants to do those frog arms or maybe I will make him a set. That should further help the car.

Here are some rocker cross section photos I came across. The one with extra sheet metal is from an NB, the basic one is an NA. Hmm I didn't know this! It seems like there are some extra zones that could benefit from seam welding on an NB, the outboard layers to one another. I wonder if the NA cars varied over the years. I think that Miatas got more unibody gussets over the years. FWIW my 97 has reinforcement at the rocker behind the seat and at the corners of the foot wells under the dash at the tranny tunnel and at the outboard kick panel.

Fuck me, if you're going to go through all this shit, why not just install a cage? Are you that much of a pleb?

I've noticed that Mazda did a few things over the years to improve the structure of our cars. Nothing wrong with pointing out some of this, and those clues are very important. Having looked into the matter a bit and also being an experienced Engineer, I can tell you that the commonly suggested seam weld locations are not really 100% there. There are some key zones that should also be done IMO. And some that are present on those diagrams are not high impact.

On cages, many folks are not interested in them. In my case I can't even seam weld my car, never mind caging it as it is not allowed in my autocross class. However I do it on other cars and just seam welded part of a local street Miata that was getting to be too much of a flexible flyer. I had a chance to look it over pretty carefully while it was apart. The guy wants some more things done so I will further firm up his car in a month or so. You can stiffen up the unibody quite a bit for low cost and without giving up any space.

+1 I wouldn't want a cage in my street miata. Additionally, I rode in a caged spec miata that somehow had more chassis shimmy than mine.

sjmarcy,

Do you think the door sills will have a significant effect? (I've got door bars already-huge effect) It's probably the easiest to prep and do:
https://www.miataturbo.net/attachmen...ine=1313260157
I have the opportunity to get this done for free, but I don't want to burn through a favor if it won't help much.

I did pretty much those red dots in the pic you posted.

So far in my research and ongoing experience on this subject…the bang for the buck impact and easy to access areas with the most impact are below. Do nearby seams leading to these zones if possible:

A) Door sill seam like in the photo…suggest doing those in a stitch fashion rather than one continuous weld for various reasons. This is the zone that the pull-off weather stripping and scuff plates cover. I would not be too worried about the front seat mount brace but while you're there it can be done.

B) All three of the braces Mazda added compared to the earlier cars. The one joining the rocker and rear panel behind the seat. It is near the rocker mount for the seat belts and I included a picture of one in this thread. In that case that brace was not even active as delivered by Mazda since some of the factory welds were not performed at all. I guess the guy had seen too many pictures of Hustler's Mom, so his vision was burnt out for a few days.

The other two braces live above where your feet go in the footwells. One connects the footwell and the other side of the panel Frog arms reinforce. The other connects the footwell panel and the tranny tunnel.

Below is the outboard brace..center of picture…not the extra metal you see added, that was for a wheel well revision on a racecar. The inboard one is very similar and can be seen in the Honda motor swap thread pics from last week. Mazda put those there for a reason.

C) Finally…the upper structure that is covered by the front fenders. And the area below it and behind the wheelwell. Again, the area covered by Frog Arms.

I am not too worried about the underhood welds as that structure is pretty beefy compared to most non-convertibles. Besides, to do those well it would be easiest with the motor yanked. Especially since there is a lot of glue and seam sealer there, making prep extra important.

Note that later RX8s had their front subframes filled with two part structural foam to increase stiffness. That might help Miatas too if you are not rules limited.

Those are my present thoughts on low cost partial seam welding that is fairly accessible. I'm sure that this will change a bit after I do a few more cars and try some things out.

With your seats and plastic trim out along with carpeting rolled back access is decent. Fenders are not hard to remove either. If you show up at your friends with one seat and no fenders this can all be done pretty quickly. The areas to be welded need to be prepped. Alternate where the welds are done to control local heat. Have the car on level ground. A welder's blanket should be adequate for most of your car protection needs.

Watch out for fires…the carpeting, under-carpet insulation and those white plastic clips for the door opening zone of the carpet burn easily too.

Here are some pics showing important structure. To an extent they are reachable for welding with the dash installed but with nearby stuff removed or protected.

Getting ready to do the front of a car…guess it is a good idea to clean out the Crap-Trap from time to time:

So I had a friend do the door openings. Grinding took more time than the actual welding. He's a race fabricator, and suggested seam welding with about 30% coverage, e.g. 1" welds with 2" gaps. I ground the seams flat and marked where he was to weld, with a sharpie. I tried to more or less hit the seams between the existing factory spot welds, and then put a bit more coverage in the corners where the factory had extra layers of sheet metal (gussets) and extra spot welds. He suggested not going all the way to the top of the windshield but only to about the top of the dash. In the tight areas where the grinder wouldn't fit, I used his mini air-powered belt sander. Very cool tool. We used a sheet of aluminum to insert behind the seams to prevent damaging interior bits.

Ran out of time to clean, repaint, and reinstall interior bits.

Result:

Butt-o-meter says there was a noticeable improvement in the stiffness. The way I noticed it, was there was less shimmy in relation to the "single pulses" from tire impacts. So the tire's impacts seem to come through "clearer".

The accelerometer data of the drive to and from the shop shows that the shimmy often appears as a (musical) "chord" of 3 or 4 "notes" which are closely spaced. When there's a doublet or a triplet, before the seam welding the lowest frequency was 16.2-16.3 Hz. Now it's at 16.7 Hz. This means an increase in stiffness.

Unfortunately I took a different route going to and coming from the shop, so the roads were different, so it's hard to say if the amplitude is smaller. I will try and get a datalog on the speed bumps I did STB testing on.

So in decreasing order of effect:
- door bars
- butterfly
- shock tower bar
- door opening seam welding

I'm reasonably happy with the results, for the effort.

I think it stands to reason that taller sills (door bars) will help the most on cars whose center section strength is provided mostly by the sills. But it also depends on what you are measuring. Damping, torsional rigidity, bending strength.

I wish I had measured flex on the beater Miata I just door opening and fender seam welded. Its doors would not work properly if you jacked it at the front spot on the pinch weld. But now it works fine and does not creak nearly as much over rough roads. Are you saying that a front shock tower braces helps torsional rigidity more than seam welding the weak center portion of a Miata? Or is it more the damping post speed bump? I think I understand your point but request that you clarify it a bit.

The earlier non-US Elises had stiffer frames as their sills were higher. This made ingress/egress too difficult for the American market so they were cut down quite a bit to help out there. About 1 1/2 inches IIRC. So in effect the version for the "fat" Americans had door bars removed more or less.

For handling…what you want is high torsional rigidity along the longitudinal axis. You can measure that if you apply a twist to the car via weight or by jacking a corner and then measuring reference points.

Another method which is pretty simple involves tape. Tape a big "X" across the car from top of windshield frame to the opposite side of the car roll bar top or trunk lid zone. And do it the other way too so that you form an X. Do this on level ground so there is no flex, slight tension on the tape to avoid droop.

Now jack the car in various places. See the tape droop or break? The droop could be measured before/after a change. Less droop = more torsional rigidity. They used to do this in the 1960s when creating convertibles. I mean all they had were slide rules. The moon mission used the computer later utilized by Pong arcade games once they greatly enhanced it haha.

Another roadster design concept is used on the Honda S2000. In that instance they use an X concept before and after a central backbone. On that car, most of the strength comes from the tranny tunnel region. So the rocker/sill region is less important. They use a massive tunnel with more vertical sides. Loads are fed into it via unibody load paths from wheel corner to tunnel, diagonally.

The STB had a significant effect subjectively. In my accelerometer thread I hypothesized that while it did not increase the torsional rigidity (no increase in resonant frequency), the engine bay (which the STB should stiffen), vibrates less (and at a higher frequency), and thus the main torsional mode isn't sympathetically excited as often.

I'll have to play with STBs some more. My car came with one and I did not notice much from it. Tower to tower braces are legal in my autocross class (no three point mounts). But I removed it cuz it was ugly and got in the way of working on things.

I think the main thing for handling is the torque needed to twist the chassis along the F/R axis. If something damps vibes that is a good thing too, but won't do much for at-the-limit cornering if torsional rigidity is not changed much.

I came across some torsional rigidity figures for Miatas. Bear in mind that the required amount will change with car weight. And track width. A two ton car will need around twice that of a Miata. So…the heavier/later Miatas are stiffer, yes, but the extra mass offsets the benefit a bit. To me an NA has a vintage toy-car sort of nature…jump into a decent NB and they have more of a "car" feel. Part of that is the more solid and refined nature of the structure. Or just jump into a Porsche to feel that solid chassis characteristic. The Elise is around 10,000 IIRC [check units] but note that it is lite and has front wheel weights in the 300s - they feel subjectively solid and rigid.

For Mazdas, the RX8 has one of the more torsionally rigid chassis around, especially after they foam filled the front subframe. If rules issues allow this option, it might help Miatas too. RX8s handle wonderfully and their chassis has around double the torsional stiffness of the RX7.

Early Miatas ?,??? [Less]

(97-99) = 5,130

(99-01) = 6,412 [17 Hz IIRC was claimed?]

(01+ NBs) = ?,??? [Stiffer Rockers: more metal added]

(05-06) = 9,426 ftlb per degree of twist

(Latest NCs) ?,??? [more]

https://www.miataturbo.net/attachmen...1&d=1313710656

Speed bump test. Before (top), After (below). I hit the speedbumps faster this time around, yet the worst peak vibrations (red intensity) is less intense. Also if I look at the center of the worst red spot, the new data shows about 0.2 Hz higher frequency.

I'm pretty convinced it helped, but again, not as much as the STB (based on data) and not as much as the door bars or butterfly (butt-o-meter)

In a fit of industriousness, I may get a round tuit and install my Frog Arms.

Doesn't the butterfly's X shaped sorta do the same thing?

Sort of, if I try to be charitable. ;-)

The X is too small. And does not connect to the front and rear of the unibody as well as it could. But that would take more work and expense than most can justify.

I'd be careful about placing too much credence to the spectrographs. It might make sense to cross check things using actual rigidity tests.

For one thing I'd like to look into how a shock tower brace connecting each side to one another in a strong zone is so clearly superior to seam welding a weak area of the unibody. Can you jack the car and measure twist in some manner, or try the simple tape test?

I have not seen test data on this, but it is claimed that the hard top is mostly an aid to NVH as in dampening things down more quickly and less so a rigidity mod although it does help there too.

Below is an image showing the primary structure for an S2000 marked in yellow. See the diagonals (part of the X) tying into the special backbone zone? If Honda did not send diagonal structure right to each corner of the car and thence to a designed backbone, the car would not be as torsionally rigid. I'm sure you've experienced S2000s, so you know they are quite solid. Of course Honda had quite a bit more money to spend since they started out at 35k.

Why would an X be more torsionally rigid than a box?

The spectrograph measures vibration, which is my primary concern. My goal is more of ride quality. The shimmy/vibration cheapens the car's feel. Not stiffness directly, nor handling - I don't run a track biased super stiff suspension. My miata is a street car / canyon carver.

Your goal is primarily ride, and you are ranking mods in the race-prep section?

;-)

The X concept is much more efficient at creating torsional rigidity since structure is placed directly along the load path. After all, cars have tires at their four corners which are the only places the entire assembly touches the ground. So that is where road forces are applied to the structure. You can increase torsional strength a number of ways but some are inherently more efficient. When you twist a frame, one diagonal gets stretched (tension) and the other gets compressed.

Just try out my tape example on your car and you'll see what I mean. The one where you tape an X across the car. Roof down, top of windshield diagonally across the car. And the other way too so that an X is formed. Jack the car and see the twist primarily hit one diagonal as exhibited by droop.

Drop by a book store and examine the Herb Adams book Chassis Engineering. In it he shows example car frame made of wooden sticks. He mounts them and then hangs a weight to one corner. This applies a twisting force to the model frame. Which version twists the least? The one with an X.

Edit: Here you can see the model frames being twisted, along with text from the book Chassis Engineering: http://books.google.com/books?id=rY2...page&q&f=false

Okay here are a few screen grab samples to ponder. They show some of the concepts fairly well.

LOL I just decided to post results in an existing thread instead of starting a new one.

Once in a great while on backroads with rippled surfaces while accelerating I hit a ripple at a frequency close to the chassis torsional resonant frequency. I feel the tire skip. Here's a case where the lack of rigidity affects grip.

I think the spectrograph is very useful for measuring the effect of rigidity mods.

I suspect that with a roof a box is superior. But in the absence of a roof, the miata's structure concentrates the stresses at the 4 corners (behind the seats, and in the footwells - which you pointed out needs help). The X structure, in your S2000 diagram, if the backbone is strong enough, doesn't concentrate stress nearly as badly where the "arms" enter the backbone, as the miata does on the corners.

What is the Boxster's structure like? The NC?

Unrelated to original topic but, what are you using to record & view this data? Are you logging a single axis accelerometer? thx.

The thread describing my test setup:
https://www.miataturbo.net/suspension-brakes-drivetrain-49/spectrographic-analysis-chassis-shimmy-59764/

Here's an interesting piece of work I found on prodracing.com, "chassis stiffness rule of thumb" thread. Yes, a Miata is not a NASCAR "Cup" car, but this SAE Technical paper is VERY enlightening.

Since adding my EP/D-Prepared cage, there is zero noticeable twist when jacking up my car, but I'm still Jones'n some more cage/tubing from the front hoop to the front strut towers.

Cheers,

Why not use high density foam in the voids. That is what I'm about to do, not just because making the chassis more stiff, but to keep the water out that keeps getting in there some way. Here is an article I found on the foam by Eric Hsu

Chassis Stiffening - Question It - Import Tuner Magazine

Alfasud... :)

Funny how this got bumped today. Look what I just did!

Also pictured, proper penetration.

https://www.miataturbo.net/attachmen...ine=1362962492

Please report back if you notice much of a difference. I'm about to buy some FM rails, but always open to cheap alternatives.

BTW, to answer some questions posted earlier in this thread - if you look at the bare chassis of various Miatas over the years, you can see the attention Mazda paid to the ends of the transmission tunnel. The flange between the rear bulkhead and the tunnel got bigger and bigger. On a 2001+ car, it's significantly larger than earlier cars. I read somewhere (somewhere legitimate, but I don't recall the exact source) that they actually went to thinner sheetmetal on that rear bulkhead in 2001 thanks to the extra stiffness from elsewhere. There's a photo of the differences in my latest book, I don't have them handy in digital form.

Various braces at the corners of the cockpit got beefed up as well.

About the butt-o-meter - we can really feel the difference in seam-welded cars when we're building the V8s for customers. Never tried to quantify it, but it sure feels good.

Thanks for the update Kieth. I am doing some seam welding myself. I have not done any welding in the driver compartment because I am doing a cage. I did to the frame rail project however.

I am spending some time on the front of the car where frog arms are often installed. Any input where the welding should be on the front would be helpful from anyone in the know.

You can use the diagrams that have been on the FM site for years or the clearer ones in my book.

Flyin' Miata : Tech: Seam welding

A good friend of mine seam welded his 2000 and it made a SIGNIFICANT difference over my unwelded 2000. Our setups are almost exactly the same, except I have the under-fender frog arms. Driving our cars back-to-back mine feels super limpy whereas his is planted and happy.

I WILL be seam welding my car.

It's pretty boss when you can respond to someone's "how should I do _____" with "I covered that in the book I wrote".

-Ryan

Thank you very much. The site has so much info, it is hard to know what all is there.

My thoughts exactly.

So my car is a little different than most here, it is missing the windshield header.

BIG increase in stiffness around the firewall and front of the cockpit came from seam welding. Before you could push down on the area were the wiper arms used to be and feel it move and hear creaking. After the seam welding, there is no more noticeable flex and the creak disappeared.

This was welding the section where the dash bar attaches to the body. I have a straight tube welded instead of the factory curvey tube. I also stitched the braces from the firewall to the tunnel, the firewall to body on the outside and the seams along the bottom of the firewall. *More was done but this section made the most immediately noticeable difference.

Our first experiment with seam welding involved the big seam around the door opening and windshield frame. Before, you could flex the frame up and down by hand. After, not so much.

Anybody else have issues with the goo Mazda used to hold on the rubber trim piece? When I got around to welding that part, my welder would splatter and spit horribly.

Clean up of paint, seam sealer, and adhesive has been the biggest hurdle in any welding I have done on the body. It is like 3:1 in time investment over welding. Maybe more.

I have had good luck with a wire wheel on a drill. A 90* angle die grinder with a scotch bright pad works pretty good in some areas too. I used various size wood chisles for the globs of seam sealer or undercoating in many spots.