Raising the Roll Center and Assessing Bump-Steer
It’ll either be glorious or a rolling dumpster fire
Measuring bump-steer on my GTI
If you’ve followed any of my recent Trip Reports, then you’ve seen me mention several times that what the car needs is some more front-end “bite”… so I finally decided to take care of that by ordering some 034 lower ball joints (shout out to Sean at www.itsnotstock.com for the crazy fast shipping - got them in less than 24hrs!). Yes - even though my car is at stock ride height. The idea was to add some needed camber fast and easy.
The other thing being messed with of course is the roll center (as they are roll center adjusting ball joints). Rather than go on about that here, I’ll link to this excellent Moto-IQ Article for further reading.
So why did I take this gamble knowing that it was likely going to be a “bad idea” or at least just not ideal? While I don’t “need” the roll center raised, all of the extended ball joints on the market do it. Ball joint cost is less than half that of extended lower control arms, and install is pretty quick and easy. I ordered these with about 10 days to go before Fastivus 2023 at Summit Point, and wanted them in beforehand… I figured I’d try them and if I don’t like it, I’ll sell them or set aside until I can use them more effectively. Also when it comes to track cars - adding grip (via camber keeping the contact patch flatter mid-corner) almost always wins out as long as other “less than ideal” things aren’t too glaringly terrible.
All of the MK7 Golf/R/GTI/Alltrack/etc all use identical subframe and control arm and knuckle geometry, but with wildly different ride heights. My hope was that since VW has such a huge variance in ride height with the same components, that the chances are good that my roll center wouldn’t end up too ridiculously high in relation to the center of gravity (Alltracks aren’t exactly known for flipping or having a “jacking” effect of the suspension). I also have a friend who has been using them on his stock height Golf R successfully on track (yes a bit lower than mine, but close enough).
In a nut shell: raising the roll center (to a point) can benefit you in the form of less body roll. There is more of what I like to call “geometric roll stiffness” because the roll couple (distance between COG and roll center) is shorter. Scroll back up and read the linked article if that went over your head.
It is important to note that my car is not 100% STOCK suspension-geometry-wise to begin with, despite being at stock ride height:
034 Camber upper mounts
Powerflex offset LCA bushings (originally set for max camber outwards)
I have the stock springs, but with Koni Special Active shocks - an OE replacement equivalent dimension-wise. The car prior to this was sitting at approximately -2.4 to -2.5 deg of camber per side. I wanted to get to at least -3.0, if not a bit more.
The problem is that when it comes to adjusting the roll center, you are also going to be affecting bump-steer. If you’re not moving the steering rack geometry or the tie rod location on the knuckle while messing with the roll center, chances are there is some bump-steer being added. You can read more about what bump-steer is and how it can be tuned out in this other Moto IQ article here, as they will cover it a lot more coherently than I can (once again).
Bump-steer is something very few people in the MK7 world talk about. I knew that I’d probably be adding a bit - but how much was a question. ANY time that you play with the lower control arm pivot point, the upper strut location, the ride height, any knuckle pickup points, or LCA length, you are going to be affecting bump-steer.
So I took this as an opportunity to attempt to measure it as found, and see how badly the ball joints affected it. Strut suspension by design does not have an upper control arm, so bump steer can only be optimized in one spot/small range of travel. As it deviates further from that height, bump steer will increase (whether positive or negative depends on the geometry). Graphed out along the entire range of travel, it will appear as a sine wave.
Generally speaking, having a bit of toe-out on compression is OK if it’s within reason, while moving towards toe-in on extension. Having the opposite (toe-in on compression) means that the harder you lean into a turn, the more the car will “self-steer” (at both front and rear, though we are only discussing the front in this article). If the amount of toe change over a typical bump mid-corner is excessive, it can make the car very darty and unpredictable. This is very common to happen when a strut car is lowered excessively also. Old Nissan B13 chassis cars have HORRIBLE bump steer, at BOTH ends of the car… upwards of 10mm worth, and they are known to be absolutely diabolical to drive - particularly when lowered.
There isn’t necessarily a “you must have no more than X amount of bump steer” rule when it comes to struts… but less is better. Minimizing bump-steer can help lower tire temperatures as they won’t be scrubbing back and forth with every little bump mid-corner, and improve predictability when at the limit. Keeping bump steer under control can make a car easier to drive fast.
I was originally planning on possibly installing these ball joints with the LCA bushings still offset to their max position - but I found that this was dangerously close to the axle shafts’ max extension… and I did not want to be breaking a CV axle on track (potentially wrecking the trans or engine block)… So I decided to measure the amount of bump steer with the bushings clocked:
Stock ball joints with Powerflex at max camber (for reference)
034 ball joints with Powerflex LCA bushings at max camber (for science - even though I wasn’t comfortable with the axles being extended as far out as they were)
034 ball joints with Powerflex LCA bushings fully clocked upwards raising the inner pivot
034 ball joints with Powerflex LCA bushings fully clocked downwards lowering the inner pivot
Here is a graph of all the values plotted to visually see just HOW MUCH any single change can have on the amount of bump-steer:
Now it IS important to note that this gauge wasn’t really meant to fit the 5x112 bolt pattern, and required shimming outwards to clear the caliper (as I was feeling lazy and didn’t want to remove the caliper bracket). But I essentially jacked the vehicle up, disconnected the sway bar, and then jacked the RF corner up until the hub to fender measurement was identical to the same as was measured on the ground at ride height (366mm in my case).
I don’t imagine that my results are dead on absolute - there is probably a +/- 0.5mm wiggle room, but based on getting pretty consistent sine waves from averaging multiple cycles of the corner from +1in of compression down to 3 inches of extension, they are at least reasonably accurate in relation to each other. I would like to repeat this test over winter, but with the spring removed so I can assess the full range of bump travel (or at least +3in down to -3in). I couldn’t go any higher than 1in of compression before the car started rising off the Quick Jacks with the spring installed.
Notice the [relative] flatness of the curve originally with the OEM lower ball joints? There is far less variance from +1in to -1in of suspension travel vs any with the ball joints installed. This does not necessarily reflect on the ball joints being bad - as stated, I am running this on a stock ride height car.
I was able to utilize the offset LCA bushings to at least minimize the unwanted effects of the lower ball joints on my car by pointing them in the full-downward position. I ended up adding only 1mm of bump-steer in the +1/-1in range, still keeping below 3mm total change within that range.
I am measuring out a hair over 3 deg of negative camber now, so mission accomplished there. I reset my front toe to approx 0.5mm of toe-in. I’m really not terribly picky and will run anything from +/-1mm from neutral.
As for on the street? I haven’t noticed any ill-effects, and the car definitely bites a lot harder on turn-in. I’ll have to take some careful measurements and run the numbers at some point, but back-of-the-napkin-math says that lowering of the lower ball joint pivot should be good for somewhere between 0.1 to maybe 0.2 deg of negative camber gain (vs stock) dynamically.
For reference, these ball joints added approx 12mm to my overall track width over the previously maxed out offset LCA bushings. I believe the overall width added from an otherwise stock vehicle may be in the 22-24mm range.
I’ll be figuring out my impressions of them on track at Fastivus this coming weekend, and later in October at VIR again (lots of laps there - better for assessing corner speeds and overall lap time improvements).
UPDATE 10/1: The verdict is rolling dumpster fire. Read the Trip Report HERE.
These ball joints added enough bump steer that it made the car nearly undriveable at the limit. I am confident this is NOT a ride height issue.
I will add that of course the raising of roll center in a non-lowered car is likely not the ideal option. I have since removed the ball joints and will set them aside until the car IS lowered to test further next year. I don’t expect bump steer to get any better, but might be able to answer the question of if it is the roll center or the bump steer that is the cause of the issues.
The control arm geometry is the control arm geometry regardless of ride height. It will just be in a different part of the curve depending on static ride height. If you look at the curve above - you’ll notice that as ride height gets lower (negative in the graph) the bump steer induced deviates further from 0 at a faster rate than with stock ball joints (even with the LCA bushings clocked to minimize effects as much as possible - which most people do not have installed.
I was unable to test beyond ~1in of compression due to the spring still being installed when checking, though due to the sine wave that the suspension geometry follows it’s reasonable to assume it’s only going to be worse. Re-testing with the spring removed would be able to pinpoint exactly how much bump steer is being introduced.
Running super stiff springs and not having lots of wheel travel will lessen these negative effects, potentially to the point they’re less noticeable. You can’t have bump steer if you don’t have any bump travel - which most cheap coilovers lack anyway.
This winter I intend to pull the springs and do a FULL sweep of the suspension travel with the bare shocks to confirm or deny my suspicion that bump steer will only get worse as the car is lowered more to at least answer that part of the question.
Currently the only bump steer correction available is Verkline outer tie rod ends, but only can move downwards 2mm before the knuckle is the limiting factor. I’m pretty confident this would not be enough, though would be a move in the right direction.
By my estimation, the outer tie rod end pivot needs to occupy where our steering knuckle is placed to get bump steer to an acceptable level with these ball joints.
To their credit: I didn’t feel like the car was lacking grip in the lower speed corners, and the car bites and turns in well - though I attribute that 100% to the bit of extra camber and not from moving the roll center around. The higher speed corners were where the car felt far more “edgy” (primarily entering the “big bend” left-hander leading onto the front straight. There was more grip, but it was very different feeling. Likely some from the roll center effects + bump steer mid-turn. I suspect the ill-effects are far less noticeable for autocross.
Lastly - just to be clear: These same issues will apply to ALL “roll center correcting” ball joints. This isn’t strictly an 034 issue. Whiteline/Superpro/Hardrace/etc will all have the same issue. Until there is a viable bump-steer correction kit (that will require more movement than the Verkline kit offers - it appears to be better for fine-tuning more-so than making a massive adjustment needed with roll center correction).
It’s not an inherently bad product for it’s intended use: raising the roll center on lowered cars. The issue is it introduces another problem that isn’t as easily fixed. Realistically speaking, most production based strut cars should only be slightly lowered anyway to preserve enough useable bump-travel to not be upset hitting curbs at speed.
These will be getting looked at again in the future when lowered before the 2024 season before writing off completely. I would LIKE them to work out well… as I still really need more camber.
