Black & Tan 332iT

[Bry5on]

Also while we're on the subject of setting things right for the record, my 1:10 tapered reamer showed up. After fitting it into the stock e46 M3 tie rod and control arm balljoint locations, I can confirm without a doubt that the BMW taper is a 1:10 Diameter:Length, rather than a 1:8 or a 7 degree, cited elsewhere on the internet. This means that for every ten millimeters of length in the taper, the diameter grows 1mm. Yes, diameter, not radius. Always measure yourself folks!!

Also, based on the torques BMW is calling out for the control arm and tie rod M14 and M12 nuts, respectively, this is likely around a steel equivalent to a Grade 5.8 fastener. Plugging these numbers into the simulation yields the following results - the tie rod and knuckle will fail about the same time:




All this had me thinking some more. Once I get a control arm here to measure, I can press out the outer ball joint, and if it happens to be a common diameter (perhaps the same as the rears) then I can press in a bushing designed for double-shear. And instead of doing the factory taper joint, I can design this thing for a much stronger and stiffer double shear joint at the control arm. Porsche did this for the 991.2+ generation GT 911 and I always thought this was a really clever way to stiffen the knuckle up and provide better wheel control, see below:


And last, a fellow member has asked me to consider making a knuckle that is designed to:
1) Correct the geometry of lowered track cars (this would have the effect of making more stable geometry and also increase front roll stiffness, so you can soften the front sway bar and still get good roll control)
2) Move the strut mount more inboard so you can fit more front tire by going inward

I've been thinking more about this, specifically #2, and how I think we might be able to mitigate the downside of the increased bending moment on the strut tube from relocating it inboard. If we play our cards right, we can achieve another 20mm or so of inner tire clearance AND stiffen the strut up to make it respond even better to mid-corner damping and displacement changes. And also last longer, because it will be bending less. So this is intriguing and I may pursue it. Again, see below from Porsche for inspiration, noting the two clamps on the strut, one of which is placed very high up:

[Bry5on]

Alright, so a couple people were skeptical that the e60 hubs would be compatible with the e46 DSC, enough so that it was starting to make me nervous. So let's address that in two ways:

First, let's break out the oscilloscope and check the output signals directly. These sensors are pretty clever, they are *active* and not passive like MK20. They output a square wave with a very short pull down from 12V to ~6V when they detect a trigger on the hub (this is very different from the e9x and e60 M5 still).


Here's what the e46/e6x non-M looks like on the active 2-wire system when spinning.


Ok, so we're looking good so far, let's make 100% sure we're right here by plugging the e60 sensor into the e46(it has the exact same keying on the connector) and firing up INPA to read the individual wheel speeds.



And would you look at that, it reads the sensor just perfectly

[Bry5on]

Hi Bryson... as for load cases I can think off...

* Pot hole, tough to estimate. I can ask around here to see if there is any directional input, like shock g's and damper speed.
* hitting a curb, although the week link here is the tie rod.

since we have more unknowns than not, one idea is to scan the OE design, and run the same modes, then you can have two data points you can use to compare under the same conditions.

Yeah, good idea. I was waiting for you to pop in here. thank you.

I’ve got the scan of the OE parts, just need to close the holes and make it a solid body.

I’ve also done a quick model of the tapered joints on the control arm and tie rod and placed the appropriate loads that put the tapers in bending. At first blush it looks like the control arm taper will fail before the knuckle by a small amount. I’ll need to disassemble a control arm to get good dimensions of the taper to make sure, I just estimated length of the taper based on an old iPhone scan.

[maupineda]

Hi Bryson... as for load cases I can think off...

* Pot hole, tough to estimate. I can ask around here to see if there is any directional input, like shock g's and damper speed.
* hitting a curb, although the week link here is the tie rod.

since we have more unknowns than not, one idea is to scan the OE design, and run the same modes, then you can have two data points you can use to compare under the same conditions.

[Bry5on]

Analysis run with the appropriate boundary conditions at the tie rods and strut tube. Interestingly it wasn't possible to use these during generative design. Looks more accurate now, this thing should be extremely beefy. Low stress and low displacement.

Next up I will put the loads into the strut tube, tie rods and hub to see what that does to the simulation, instead of applying the loads directly to the part surfaces. The simulation can then factor in displacements of the steel parts as well as the aluminum parts.

[Bry5on]

Fresh off the printer, complete with a few tweaks from George's lessons learned. Dimensions are looking good, just needed to make the tapped hole a little deeper for the ABS sensor mount.

[Bry5on]

Would be cool if your brake duct supported brake cooling hoses.

Easy enough to design replacement bolt on backing plates that would adapt to a traditional hose - the backing plate should be easily replaceable with the rotor removed, no hub removal required.

[YoitsTmac]

Would be cool if your brake duct supported brake cooling hoses.

[Bry5on]

Some comparison shots against the factory steel parts that look pretty cool:


And some shots of the brake shield/duct hybrid in progress:

[Bry5on]

Thanks to George Hill (whom I trust greatly with my CAD files) who's already printed one up! Got some good feedback on socket clearance that I'm going to have to incorporate into the keep out space. And he's got so many spare parts that he's probably going to beat me to my own test fit cases

I also forgot to mention I started conceptual designs on a bolt in brake shield. I'll be tweaking it to incorporate a ducting feature that matches my scoop a bit better, and replaces the plastic part that zip-ties to the strut:

[Bry5on]

Actual weights from the bathroom scale, without heat shields or ABS sensors, repeatable 3x:
E46 total 14lb  = 6.35 kg
E60 total 13lb  = 5.897 kg
E60 aluminum knuckle 5.4lb = 2.5kg

so in theory the superknuckle is .15kg heavier than the e60 setup and .3kg lighter than the e46 setup. So no appreciable weight benefit, but definitely a stiffness increase along with a much larger (and stiffer) wheel bearing that happens to bolt in.

I’ve also decided that I’m going to reduce the steering ratio on the next iteration to match the non-m 13.7:1 steering ratio (vs the M3 14.5:1 with the same rack) that I find preferable. After adjusting for wheelbase differences, here are some baseline comparisons from other cars:

911 ST converted to e46: 13.5:1
911 GT3 converted to e46: 12.8:1
E90 M3 converted to e46: 13.3:1
Wife's Macan converted to e46: 13.9:1

So 13.7:1 has some good company, I guess that’s why it feels so nice. The MK60 is easy to program with a write-in steering ratio, so I’ll have to knock that out along with this change as well to keep DSC in its happy place.

[Bry5on]

What about much higher, peaky upwards loads (e.g. hitting a pothole)? I would think the wheel would see more than 1G upward acceleration in cases like that.

Worst case scenario sounds like steady state cornering and then a sudden upward force at the wheel (e.g. taking a corner and riding the kerbs on track).

Yeah, I thought about this. I had a hard time coming up with a real number to use here because the tire is a nice spring, and so is the bushing. So the load it sees is maybe 2x the worst case? Worst case being braking and hitting a pothole I think. The part has a safety factor of just under 3 to a 1.6G load of half the car's weight (0 load on rear tires) for this scenario, meaning that it's got to take a 5G+ load wallop to risk permanent deformation. Any experts on modeling suspension impacts here?

[heinzboehmer]

I mathed out and ran four load cases:

What about much higher, peaky upwards loads (e.g. hitting a pothole)? I would think the wheel would see more than 1G upward acceleration in cases like that.

Worst case scenario sounds like steady state cornering and then a sudden upward force at the wheel (e.g. taking a corner and riding the kerbs on track).

[Bry5on]

Okay, on to some results. I mathed out and ran four load cases:
1) 1.6G braking
2) 1.6G cornering, outside wheel
3) ~.8G cornering, inside wheel (more load on inside at lower Gs)
4) 1G braking + 1G cornering, outside wheel

I thought long and hard about these load cases and determined these to be the enveloping ones. Would appreciate any thoughts of other load cases that I may have missed or should consider.

All load cases included a lot of extra tie rod force to compensate for the fact that you can also be turning the steering wheel. You can see that the Fusion FEA spit out some real localized high sresses at sharp edges due to constraints. This is generally expected and not cause for concern as the bulk of the material has very low stress. I opted to run the solver for a stiffness-driven part, as I figured that would give the best steering feedback, and it means that generally the part ended up pretty low-stress.

The part, with just barely enough clearance to install fasteners and get an extension through the top bolts:


The stresses on the part in various conditions ("Load Case 1" is 1.6G Braking):


The displacements:

[heinzboehmer]

It has an MK60 and 48 tooth wheel speed sensor just like the e46 - I get to keep ABS and traction control.

And bolt in vs press in wheel bearing.