Steering Rack Tear-Down / De-Power / EPAS

[Justin Daniels]

BLUF: I am de-powering an M3 rack so I can use an EPAS column from a Toyota Prius to give the assist.

Disclaimer: Let me start by saying that I am not pushing this as a solution to any problem... or even as a good idea. I'm doing this because I want to. This thread is purely informative and nothing else. The car will never be driven on the street; as such, the context/goals of this "project" may be far different from yours.

Preamble: I'm building a racecar out of a dirt-cheap (ended up free after selling some parts) 325i that I acquired. I picked up a wrecked M3 to pick some parts off of (driveshaft, diff, axles, rear subframe, uprights, arms, sway bars, etc.) and decided to use the rack on the racecar since it's a quicker ratio. Unfortunately, the damage to the front suspension was worse than I originally expected so I'm going to have to do a lot of checking to make sure things aren't bent/broken, including the rack. Since I can't confirm the condition of the rack, I wouldn't have felt comfortable sending it in as a core or selling it to some poor sap; and certainly wouldn't feel comfortable putting it on the car. So I decided to pull it apart to be able to measure the components for trueness, mainly the rack shaft itself.

On Servotronic: Originally, I wanted to tinker with the Servotronic system to be able to dial in the amount of assist from the cockpit. I quickly realized that without starting with a working system, it was going to be hard to understand how to spoof whatever signal the servo gets. Others who have swapped an M3 rack into their non-M cars have reported that the level of assist is acceptable with the servo disconnected. I was curious if the M3 pinion was the same as a non-M pinion, and it just had a stiffer torsion spring and different valve body slid over it. that way I could just swap the valve body and run the non-M valve-body cover. Turns out I was wrong, the pinions are actually completely different (other than the teeth) due to the complexities of the Servotronic system.

On The Big Picture: After getting expensive Pentosin fluid everywhere while disassembling the steering rack, I started to think about everything upstream of the rack; the pump, the lines, the cooler, etc. I thought, maybe incorrectly, that I should probably get an M3 power-steering pump due to the flow characteristics... but then I thought maybe the pump is built to handle Servotronic that I won't be using? Do I need to run the power steering cooler? I'll need to adapt the fittings on to custom lines since I'll be running the pump on a non BMW engine in a non-stock location. Where will I put the reservoir? It leaks... so maybe I'll make my own? Pentosin is expensive. This M3 rack swap quickly began to get more expensive/complicated than I wanted it to be... and I was still bummed that I couldn't dial in the assist.

On De-Powering: After having a flashback to Miata/RX7 days, where guys frequently de-power their stock racks for various reasons, I decided that I would explore this option for the M3 rack. De-powering the rack would make for a cleaner engine bay, less points of failure, and less cost. Seems like a win win. Unfortunately, it would also leave me with a very uncomfortable driving experience even with a 350mm steering wheel. I'll be running 255/45/17 Trofeo R among other R-Compounds, in a 3000lb car, on tight technical tracks mostly... I did NOT want to have zero assist. So there was only one solution that gave me everything I wanted... de-power the steering rack and then add power assist back!

To properly de-power the rack, I will have to remove the rack shaft, grind off (or otherwise render useless) the seal between the two chambers, drill and weld the valve-body, pack everything with grease, and plug all of the ports. I will also be replacing any important seals along the way. The result would be a rack that now has no resistance (removed piston), slop (welded valve-body) or assist (plugged lines).

To add assist back, I decided to take a page from many home-built/track/off-road vehicles I've seen and re-purpose an OEM electric power steering column. The two popular ones seem to be the Toyota Prius (since it, in the absence of any VSS/CAN input, defaults to an assist level needed at 40MPH) or the Saturn Vue/Chevy Equinox (since there is a module that can spoof whatever signal the control unit needs and allow the driver to dial in the assist). I haven't decided which to use yet, but I'll keep this thread up to date with the process.

Going Forward: In this thread I'll do my best to measure/catalog any seals, gaskets, or o-rings that I come across. If anyone wants any specific measurements, let me know.

Cheers

[Richbot]

Excellent thread and you're an ultra-nerd for using BLUF in a forum post

[Justin Daniels]

Originally Posted by Richbot

Excellent thread and you're an ultra-nerd for using BLUF in a forum post

Thanks, and I suppose I should get with the times and start using "TL;DR"!

[Justin Daniels]

Took a closer look at the Lots of slop in the driver side (impact side) bushing; the inner tie rod is bent on that side so I'm not surprised the bushing took a beating. Passenger side is tight. At this point, I think I may not use this rack and just dissect it, for science/measuring.

Everything I am doing is on a 2008 E93 M3 steering rack from and M3 with the VIN ending in P330463. Real OEM shows the part number as 32102283633.

Start out with inner and outer tie-rods removed (do this ON the car, inner tie rods are a PITA with the rack on the bench). I also have the black alignment cap on top of the pinion housing removed. I'm not sure how to do this "properly" since mine was damaged; As such, I just cut the thing off. If you're basically re-designing the steering system, I don't think it really matters much.

Since the rack's major components come apart fairly intuitively, I won't provide many pictures of this process.

Initial Teardown

• Remove the two bolts securing the Servotronic servo to the pinion housing; remove the servo.
• Remove the two lines that run from the pinion housing to the rack housing. There is a fair amount of fluid in the system, don't cycle the rack quickly unless you want a Pentosin geyser (ask me how I know).
• Remove the two bolts securing the pinion housing to the rack housing; remove the pinion housing. If I recall, there is an o-ring between the pinion and rack housings.
• Remove the "nut" (image below) at the front of the rack (I used an impact gun) to expose the pre-load spring and the rack shaft "holder/slug"
• Remove the o-ring that is preventing removal of the "holder/slug"
• Remove the rack shaft "holder/slug", I used a pair of needle nose pliers and twisted while gently pulling this piece out. You can also likely just point this at the bench and tab the back of the rack with a soft mallet and it will pop out.
• Remove the pinion. This will require some force, but it does just pop out. Grab it tight and gift it a good yank.
• Remove retaining ring from passenger end of the rack housing. This sucks, I tap the ring until one end is in the recessed area (you'll see) and then use a pick to lift it up. Once you have a pick under it, you can fit a screwdriver and pop it out; cover it up with a rag when you're doing this so the spring/ring/clip doesn't go flying.
• Tap the driver-side end of the rack shaft; the seal on the passenger side should pop out now. Slide it off over the end of the shaft.
• Remove the rack-shaft from the rack housing.

Listing of o-rings/seals/gaskets in this specific post:

• Top of Pinion Housing - Oil Seal, this is the one that leaks and causes expensive puddles in your garage/driveway. You can remove this with a socket that fits the hole at the top of the housing, 3/4" worked for me.
• Pre-Load Spring + Holder Cover - O-Ring
• Lower Pinion Shaft - Oil Seal (will likely stay in the rack body when the pinion is removed)
• Pinion Housing to Rack Body - O-Ring
• Servotronic Servo - O-Ring + Screen
• Hydraulic Lines - O-Ring (x4)
• Passenger Side Rack Shaft Seal/Housing - Oil Seal/Bushing

That's it for the initial dismantling of the rack, if anyone wants specific pictures feel free to ask.

Cheers.

[Justin Daniels]

The shaft has a piston (First Attached Picture) that the fluid from the valve body/pinion housing acts on to provide the assist to the rack. The harder you turn the wheel, the more the outer pinion portion rotates about the inner, the more fluid gets let through to the rack housing.

Step One: Bust out your death wheel and cut the rack shaft piston on opposite sides of the shaft (Second Attached Picture). Don't worry about nicking the shaft, or do worry... I didn't, perhaps I'll die. If you do nick it, just don't make any cuts deeper than the reliefs you see in the third attached picture, that's my pretengineering logic.

Step Two: Take a chisel (or whatever) and separate the two halves if they are still on the shaft. (Third Attached Picture).

Step Three: Go have a drink, you just rendered your $400 (used) or $150 (core) steering rack completely useless to anyone but yourself! You're committed now, congrats (or my apologies).

That's it for the shaft. Next I'll dismantle the pinion into its constituent components.

Cheers.

[Justin Daniels]

I'll keep this specific post up to date with progress on the steering column.
After doing some research, I decided to use a steering column from a second generation Toyota Prius. The control will be handled by a steering module from a first generation (2006-2011) Toyota Yaris without ABS.

The non-ABS Yaris steering module requires a wheel speed input (standard 0-5V pulse) for adjusting the assist and engine RPM (over CAN) to "wake up" the ECU. In the absence of the VSS or CAN RPM signal, it will "wake up" once the wheels roll a bit, albeit in fail-safe mode. The nice thing about the Toyota steering computer is that it will fail safe (for a few different reasons) to 40MPH of assist; so it really just needs 12V and you're good to go. This is what I will be doing until I can figure out the CAN messages with the Megasquirt.

It's important to note that, from my quick research, the GM columns/computers don't appear to fail safe. Which made me go this route.

Steering Column: 2004 to 2009 Toyota Prius (PN: 80960-47051)
Steering Computer: 2006 to 2011 Toyota Yaris without ABS (PN: 89650-52070)

Attached is a before (as pulled) and after (cleaned up) of the steering column; and a comparison with the E90 325i rack.

[///Mobbin]

Nice project and write-up, had no idea this was even a thing, very cool.

[Justin Daniels]

Originally Posted by ///Mobbin

Nice project and write-up, had no idea this was even a thing, very cool.

Thanks!
I'll have some more updates soon. Put a 4.5" cutoff wheel into my forearm last week so I've been focusing on the design of other bits (adjustable control arms and BBK) for the war effort.

Cheers.

[Bartledoo]

Originally Posted by Justin Daniels

Thanks!
I'll have some more updates soon. Put a 4.5" cutoff wheel into my forearm last week so I've been focusing on the design of other bits (adjustable control arms and BBK) for the war effort.

Cheers.

Sorry to hear, but let's see the gore!

[Justin Daniels]

Originally Posted by Bartledoo

Sorry to hear, but let's see the gore!

It's hard to tell in the picture how deep it is, I won't attach the image since it may put some people off. However, here's a link to a high-res picture that the resident took for me. Cool thing I learned is that they cannot stitch a cauterized wound since there won't be sufficient blood flow; so they have to cut back the cauterized flesh before they can stitch you up. The picture is from before the cute resident augered it out some more with a scalpel.

I'll have more steering updates soon, I hope!

Cheers.

[TristanTellone14]

Not trying to thread jack but need to reply to a thread to be granted access to post. However, I just recently purchased a 2011 e93 m3. It was running fine when I test drove it, had to get it shipped from Florida to NY. Upon delivery the car has about 3-4 lights on the dash. I get ABS "Brake" traction control light as well. When IK start the car it says "Increased battery discharge" as well as "Brake/Chassis Malfunction". Anyone know whats up with it? Delivery took around a week, I have been reading it coulde be battery related as it wasnt started for about a week or driven. I have driven it to work and back home twice just to get some mileage on it and possibly to charge it. Still getting all the same errors, anyone know what it could be? I will be charging the battery overnight to see if I get the same error tomorrow morning. Once again sorry to thread jack a little.

[///M3 Gusta]

Any updates on this project?

[Justin Daniels]

Originally Posted by ///M3 Gusta

Any updates on this project?

I'm not sure why I didn't receive any notification of your post... very strange.

To answer your question; yes.

Steering Rack
Last week I welded up the rotary valve (pinion); effectively locking the inner/outer valve bodies. I also filled all of the ports in the valve housing with JB weld, and blocked off the ports on the rack body. I re-installed the heavily greased steering rack, welded pinion, and lash adjuster; and then re-installed the fully sealed valve housing.

Steering Column
Since this is just going to be a lapping day car; I am going to do an experiment and see how bad it is to drive on 255/40/17 Pirelli Trofeo R's with a fully manual setup. I have put the Prius column on the shelf for now and will be running the stock E90 column initially. I'm thinking I may need to dial out some caster to make things a bit more manageable; but time will tell. The 17x9.5 ET35 Apex ARC-8 I'll be using calls for a spacer (assuming to clear suspension) but I'll have to see if I can do without since any increase to the scrub radius is just going to make turning the thing even harder without any form of assist. At the end of the day it's going to spend 95% of its life at speed on the track, the rest being spend loading it on a trailer or transiting in the pits/paddock. Since it's not going to be used competitively; if I get tired, I can just pull off and take a break. If it's unbearable then I'll have to rig up the Prius column. I need to do some CANBUS sniffing on an non-ABS Toyota Yaris to figure out how to send the EPS computer the parameters it needs for standalone operation (sounds like some Arduino work is in order).

All that to say; no real-world feedback yet. I'll update this post with some pictures when I get a chance.

Thanks for the interest!

[Richbot]

Depending on the suspension and tire you might not need much spacer. My 275/30-19 PS4S on OEM 220M rear square setup (ET23) has quite a bit of room to my (sachs performance coilover) damper/spring perch, you might be able to get away with a 5mm.

That's a pretty quick rack to not have power steering on though. You'll be feeling it unless you compromise performance with alignment. Hopefully you've got a nice big steering wheel lol

[Justin Daniels]

Originally Posted by Richbot

Depending on the suspension and tire you might not need much spacer. My 275/30-19 PS4S on OEM 220M rear square setup (ET23) has quite a bit of room to my (sachs performance coilover) damper/spring perch, you might be able to get away with a 5mm.

That's a pretty quick rack to not have power steering on though. You'll be feeling it unless you compromise performance with alignment. Hopefully you've got a nice big steering wheel lol

I'll be running some JRZ 2-Way Motorsport dampers; probably a Trofeo-R... but maybe a Firestone Indy 500 for dicking around if I don't run a super aggressive pad compound. Only one way to find out if it will fit!

As for the rack, I definitely agree with you... it's going to be a workout. I'm hoping the fully adjustable front arms will help me set it up on some way that will work better than if it was just a stock car. I don't think dialing out some caster (new top plates needed, mine are camber only) would hurt performance, just make it a bit more "darty" in a straight line... maybe a lose a bit of self-centering too; or am I missing something?

I'm using a stock steering wheel for the time being, and hoping to have the car a fair amount under 3000lb. We'll see, I guess!

[Bartledoo]

Dialing out caster will decrease the dynamic camber change no?

[Justin Daniels]

Originally Posted by Bartledoo

Dialing out caster will decrease the dynamic camber change no?

Ah balls, yes you're absolutely right. Camber gain under bump will decrease. Thanks for the reminder. How does the E9x fair in this regard? Barely any gain through bump or enough to spare a little?

I guess I'll see how much I suffer with OEM caster and go from there; getting the EPAS column sorted may be easier than farting about with suspension settings to dial out a self-imposed hurdle.

[Bartledoo]

Originally Posted by Justin Daniels

Ah balls, yes you're absolutely right. Camber gain under bump will decrease. Thanks for the reminder. How does the E9x fair in this regard? Barely any gain through bump or enough to spare a little?

I guess I'll see how much I suffer with OEM caster and go from there; getting the EPAS column sorted may be easier than farting about with suspension settings to dial out a self-imposed hurdle.

Maybe Malek@MRF would know what kind of camber change this thing goes through under compression since he has a lot of experience aligning these and messing with the settings.

As for tracking without PS, my PS went out on track a couple months back and I would've still finished the day out but it was coming on and off midturn which was basically undriveable and dangerous. It definitely seemed difficult but driveable without PS, but would've been a lot slower without the confidence/ability to quickly correct steering.

[Richbot]

With the rigth front rates you might be ok. These cars gain a lot more camber in bump with wheel turned than the pre-dual pivot cars, but it's not magic and people are routinely running 3.5+ degrees static with caster maxed out to get the car to turn soooooo

But just from a pilot workload perspective, you're going to want to re-power the rack anyway, so it'll be a temporary problem. Life's too short and modern technology is too good to run without power steering in a sedan car

Just for perspective, I race a spec racer ford gen3. We went to a very high-caster setup (at 1600lb with driver and a full fuel load, and a 185 front tire on a 5.5" wheel). It is HOOKED UP now mid-corner to exit where it matters, but with a 280mm steering wheel, the damn thing was HARD to drive and tired you out. It was hard enough that my shoulders got involved in steering the car. With a larger steering wheel (effectively enough new leverage to amount to what you get from a light-assist PS setup) that all went away. My point is, caster matters a lot, and don't steer with your entire torso it's not a good time

[Justin Daniels]

Originally Posted by Richbot

With the rigth front rates you might be ok. These cars gain a lot more camber in bump with wheel turned than the pre-dual pivot cars, but it's not magic and people are routinely running 3.5+ degrees static with caster maxed out to get the car to turn soooooo

But just from a pilot workload perspective, you're going to want to re-power the rack anyway, so it'll be a temporary problem. Life's too short and modern technology is too good to run without power steering in a sedan car

Just for perspective, I race a spec racer ford gen3. We went to a very high-caster setup (at 1600lb with driver and a full fuel load, and a 185 front tire on a 5.5" wheel). It is HOOKED UP now mid-corner to exit where it matters, but with a 280mm steering wheel, the damn thing was HARD to drive and tired you out. It was hard enough that my shoulders got involved in steering the car. With a larger steering wheel (effectively enough new leverage to amount to what you get from a light-assist PS setup) that all went away. My point is, caster matters a lot, and don't steer with your entire torso it's not a good time

Oh I have no illusions about it being an enjoyable experience. But just how bad it will be is dependent on too many variables to predict... so I'll just go for it and see! Doesn't cost anything to run it as it sits as the de-powered rack was an iterative step towards the EPAS solution anyway. The end goal will be to run the car with the Prius column, I just need to figure out the CAN first (and how to perform an alignment on the column).

On a side note; I've spent time in a few open wheel cars - FF1600 ('89 Swift SE1), FF2000 ('85 Reynard), and FE ('08 Van Diemen) - and can only imagine the SRF being similar in seating position. The laid back seating position definitely forces you to use your shoulders/arms more than in a sedan; so HOPEFULLY there's a bit of mechanical advantage to be gained, especially paired with the OEM 375mm wheel!

Time will tell.

[Justin Daniels]

Originally Posted by Bartledoo

Maybe Malek@MRF would know what kind of camber change this thing goes through under compression since he has a lot of experience aligning these and messing with the settings.

As for tracking without PS, my PS went out on track a couple months back and I would've still finished the day out but it was coming on and off midturn which was basically undriveable and dangerous. It definitely seemed difficult but driveable without PS, but would've been a lot slower without the confidence/ability to quickly correct steering.

That's good to know; what tire are you running? Stock steering wheel?

From what I've experienced, a de-powered rack is noticeably easier to manhandle than a powered rack that has lost power. So hopefully that all reduces the effort a bit.

I think this is just going to be a "give it a shot and see" type deal.

[Bartledoo]

Originally Posted by Justin Daniels

That's good to know; what tire are you running? Stock steering wheel?

From what I've experienced, a de-powered rack is noticeably easier to manhandle than a powered rack that has lost power. So hopefully that all reduces the effort a bit.

I think this is just going to be a "give it a shot and see" type deal.

At the time, it was 275 NT01 and stock steering wheel.

I looked like Austin Powers trying to get into my garage in a tight alley later that evening though.