Steering assist goes away, comes back

[Hyper]

{ed, July 2017: This thread documents the diagnosis of my hydraulic system, searching for the reason for loss of steering assist. Ultimately, the root cause was traced to the steering box itself, and the failure of seals or sealing therein, possibly caused be contamination. Even further along, and on another diagnostic endeavor not covered here, it was found that contamination in the hydraulic system had also impacted the performance of the brake boost.
Back to the past...}

*****
About a month ago I brought the car out of garaged status. It was overdue for a pentosin swap.
An independent engine issue had me on a wild chase into the hydraulic system, where I had the Luk68 pump apart.
There was nothing in there I didn't like. It was merely a loose accessory belt which couldn't get a grip on the pulley.
You feel every belt slip, like right now. Makes me wonder why there is not a bomb in the steering circuit like the one in the brake circuit.


Now the engine works fine, no problem with belts, and the steering is smooth...up until assist suddenly goes away.
This typically happens after 30 minutes of driving so its got me thinking its heat related.
After cool down, I go out and the steering is back. Other than that, no other symptoms.
Pump is quiet, belt is not slipping, no brake assist msg on the MID. Nothing.
Note: I already tried the lock-to-lock procedure, which cleared it once but not the last time.
Next time it happens I'll investigate whether I can also deplete the brake bomb.

Looking ahead though, I'm also considering a more scientific plan.
The question is whether there is any factory sanctioned procedures and equipment for checking pressures around the regulator.
I might go it alone except finding Tee fittings would be difficult and guessing "normal" would be less than useful.

[ByMW]

When the assist "goes away", does it come back if you rev the engine up?

When I had a steering assist problem, the pump had weakened to the point that steering boost would drop out at low RPM's. Had to get a new pump - problem solved.

Your issue sounds different, however. Have you cleaned/changed the pentosin filter?

[Hyper]

The assist does work perfectly up until the second that it goes away. Last time, I was driving straight at a constant 25.

Not rpm sensitive, one of the reasons I think its down stream. New filter, which is on the return side anyway. And new fluid.

The inside of a vane pump is rather simple, just a spinning head with those radial vanes, plus the piston/spring valve for limiting. It looked shockingly pristine to me and therefore I'm reluctant to throw down for another.

Rather than getting into the guessing mode, I'd like to get some instruments onto the regulator. Was hoping any service techies out there could say whether hydraulic diagnostics and tools are mentioned in factory shop manuals. Could be that the existing sensors could reveal something, either directly or through a diag port.

[DanMarshall]

Makes me wonder why there is not a bomb in the steering circuit like the one in the brake circuit.

We have seen the enemy and he is us...
That's my way of saying that there is indeed a "bomb" for the power steering, it is the same one that the brakes use. The whole Pentosin system is for both the brakes and steering. The idea being that if the engine quits for some reason you will still have some semblance of power steering and brakes to get you safely off the road.

Cheers,
Dan M.

[Ahmed303]

I would venture a guess that the problem is the Bomb as well.

[m10n]

http://www.bmwtechinfo.com/repair/main/721en/index.htm

then go to 32-41.005

[Ahmed303]

I wonder how much that Special tool cost.

[Hyper]

...there is indeed a "bomb" for the power steering, it is the same one that the brakes use...

Uh, no.
On my '97 840, the brakes come directly off the bomb fed by the regulator, and the steering comes directly off a different port on the regulator. I can pull the belt giving immediately no steering assist but still have brakes until I deplete the brake bomb. So the two circuits are separated by at least a check valve, possibly also running different pressures.
Thanks anyway.

***
M10n, I'll look it up.

[DanMarshall]

Uh, no.
On my '97 840, the brakes come directly off the bomb fed by the regulator, and the steering comes directly off a different port on the regulator. I can pull the belt giving immediately no steering assist but still have brakes until I deplete the brake bomb. So the two circuits are separated by at least a check valve, possibly also running different pressures.
Thanks anyway.

***
M10n, I'll look it up.

I don't know if the Brake Bomb is your problem but both systems are definately connected through the regulator. See pipe # 2 in the first diagram and hose #12 in the second. Could be a bad regulator.

[m10n]

http://www.bmwtechinfo.com/repair/main/721en/index.htm

also check 34-33

[Hyper]

She blinded me with science.
-Thomas Dolby
*******

So here is the system schematic and I've added two sensors at point B ("brake") and P ("pump").
I would have added sensor S but I haven't figured out how to undo the far end of the steering hose.
The ball valves used to shut down the subsystems were not installed and instead, return lines were clamped.



Brake sensor on the left, pump sensor on the right. Both of these are 1000psi max. Therefore the pump can go out of range until I get a 2K sensor retrofitted.


The sensor outputs are sampled by a microcontroller and results formatted onto a character display.
Fyi, the controller is hanging around for an entirely different purpose and we are temporarily borrowing 5% of its input-output capabilities.
It can also kick out to USB if I were so ambitious. Software by me.


Here we see the pump maxing out the 1000psi sensor.
The regulator scales this back to 52bar for use by the brake booster.
Steering sensor is not connected, reading is bogus.


After patching up all the pin holes in the lines, I was ready for a road test. Never left the driveway due to the original symptoms already present.
The pump suddenly hovered in the low teens while the brake system still showed 52bar due to the healthy accumulator.
Brakes dropped to 43bar when the pedal was operated, but never recovered back up to 52.
Couldn't get the pump out the teens, even when rev'ing.


One final piece of logic was to clamp off the return lines in case these were creating a massive pressure loss that even a healthy pump couldn't support.


$700 for a new pump. Hmmmm.
I'm still a bit skeptical about the pump being bad because
1) they are really too simple to fail, and 2) I've been inside this one and didn't see any thing unusual or worn.
I will place an order, but also inspect this one again.

[Max Lumens]

Nice bit of diagnosis, thanks. IIRC there's a relief valve in the back section of the pump, may have failed open. No clue where the fluid goes if that happens, maybe internally back to the suction side?

[Timm]

We have seen a couple of similar problems that were eventually solved by replacing the pump.

[sleepy81]

Hyper, You're right about the bomb. I have had the engine die & the power steering going immediately mid way through a corner.
Do you have any noise when the steering fails? My noisy pump related problem was with the level in the Pentosin canister being too low although I didn't have steering failure related to that. I'm inclined to agree with Max, although when I dismantled my pump, I agree, very simple & I didn't notice any relief/bypass valve. More likely a faulty pressure regulator IMO.

[Hyper]

... the level in the Pentosin canister...

Yes, I left out a step or two in the write up above.
After the the pump refused to pressurize the input to the regulator, I DID check the reservoir level.
...only because I've been burned in similar situations. (Car doesn't start...oh, no gas.)



From the overlooked writeup: HowTo->Rebuild the 840 brake booster...

The H31 brake booster on a '97 840Ci is rebuilt using the kit supplied by BF member Max Lumen...This rebuild was motivated a rash speculation that the booster hydraulic path may be stuck open, leading to low system pressure, leading to loss of steering assist. The symptom list later grew to include brake pedal floor tap. The only doubt was whether the booster was stuck open because of internal issues or whether it was, say, due to the master cylinder failing to return. Fast forward: The booster was rebuilt and there was no change to the steering symptoms. However, after further swapping in a new MC and fluid and bleeding, it more apparent now that the pedal floor tap was rather a bottoming of the booster travel before getting into MC travel. Probable cause: low hydraulic pressure. Since both the steering and brakes are affected, probable source is the pump.

I followed the booster rebuild with a new MC. This had the supreme benefit of flushing years-old brake fluid (the brown kind).


With the resulting system (fresh MC hydraulics and very bad boost hydraulics) it was finally revealed to me that the brake pedal stroke had two phases: the (non) assisted travel, then the abrupt tap as you come up against the MC hydraulics. This two phase stroke had been merged previously because my MC hydraulics were so bad. Previously I thought I was tapping the floor, and probably was.
So, Tip-of-the-Day: even if your boost assist is bad, don't ever regret also throwing maintenance into the MC side of the hydraulics.

This added to my epiphany about how hydraulic boost can mask bad mechanicals and/or bad maintenance.
On the steering side of the system, I traveled 1000+ miles cross country, unassisted.
Without boost the center dead-zone widened significantly and I not only had to provide the missing muscle but also had to use much larger, more frequent corrections.

***********
Here is the relief valve on the LUK LF68 2105975. Output pressure builds against the spring until (at 130bar) a hole is exposed and fluid is diverted back to the input.
Also too simple to fail, but its what's left. That and stuck vanes.

[Hyper]

New crime scene, new suspects, another smoking gun. All for your entertainment on a busy Garberville weekend.
In the last episode our main character, the hydraulic pump, was failing to pressurize the line to the regulator.



Even though I was skeptical that a too-simple-to-fail pump had failed, I sincerely attempted to buy a new one.
Disty insisted on not only $700, but a core with shipping on me, plus 5 weeks delivery, plus up-front non-refundable payment.
Me: what do you people actually do?

The eBay gods were kind and someone was offloading an e38-740 pump for $45 delivered.

A special note on compatibility....
I could find NO difference between a Luk LF68 2105975 (aka bmw 32411141568) for the e31 and the Luk LF68 LF68 2106384 (bmw 32411092015) for the e38.


After the pump swap, my super-secret box was still saying "no good" on pump pressure.
So we resurrected a theory mentioned earlier, that there might be a massive flow back through one of the three return lines that was overwhelming the pump.

I drained the reservoir enough to reveal the return inlets from both the regulator and the H31 brake booster. Nothing.
I then removed the filter at the bottom.

Note that in the case of the 840, the steering return is in the bottom center and the drain to the pump outside the filter on the reservoir wall.
Got something big. Real big. A geiser if it weren't contained by the filter lid.


So, it looks like the pump is flowing significant amounts, just not pressurizing the system due to a big return.
The question is, if not for this massive return, would the pump then build pressure?

Going back to the system diagram, we DID anticipate the need for shutting off the subsystems with ball valves.
Ball valves never made it in.

So instead we rip out the steering box (connections) and cap the regulator with a 1000 psi pressure sensor.
Hose to reservoir was decoupled and plugged but, in retrospect, that was probably unnecessary.


Showing off: Roll-your-own M16x1.5 male to 1/8-27 NPT female. Beats waiting 5 days for the postal truck.


Also time to swap in a 2000 psi sensor on the pump line.
Electronics distributor Digikey wanted a minimum order of 5 copies on my SSI brand sensor but they had a Honeywell in single units.


The Packard connector style was new to me and took me several attempts to build correctly. Note: Pull-to-seat pins, watch the orientation.


Now we have three sensors on line, and the plot becomes more fascinating.
You're expecting a commercial interruption but it doesn't happen.



As predicted, with the steering removed the pump could now pressurize the regulator input, and with enough engine revs, even blow past my new 2000 psi sensor.
(Didn't quite like that, seeing its on the far side of 130bar.)

What I really didn't like though was seeing the steering sensor exactly track the pump sensor.
Had me checking the software, looking for some amateur's mistake.

280 psi on pump & steering. Brake line rock steady at 48 bar.


980 psi on pump & steering


Blowing past the steering sensor's 1000 psi range. Pump at 1520 psi but can easily top 2K now.


Hey, doesn't the steering hydraulics pass thru the regulator to perhaps get regulated?
And didn't I read somewhere that steering pressure was similar to brake pressure?

Is this my smoking gun, that over-pressure from an evil regulator blew out the seals on an innocent steering box?

[IcemanBHE]

Are your at 8Fest? We need to meet.

[Hyper]

Update, for the benefit of anyone in the ZF organization that may have someway/somehow gotten hold of this link.
My car has been on jacks for over a month now, waiting for a favorable action from ZF regarding steering box rebuild kits.



Plan B to be revealed soon.

[Fatandre]

How about a better solution than those boxes?

[Hyper]

How about a better solution than those boxes?

My super-secret boxes? Almost weekly I build a better solution.



Seriously, I'll guess that the ZF steering box can be designed out.
The ratio is not correct for the car anyway unless you are CSi.
Pretty much it a stand-alone subsystem, just sitting there on the sub-frame, held down with two bolts and fairly easy to get to.


"A better solution" would be a common catalog item from the majors, with similar dims between steering input and pitman, correct ratio, cheaper AND serviceable.
Time, money & luck. Plan Z.

[Fatandre]

Hyper, have you seen this?

http://forums.bimmerforums.com/forum...ng-in-your-E31

[Hyper]

Thanks, didn't see it before. That's taking a full swing.

******** My Plan B ******
A new regulator was installed and seemed to show the same response as the original.
iow, the steering port was following the pump input up to the 1Kpsi sensor limit.

A junk-yard steering box was installed and very brief readings seemed to indicate no change from original.
iow, with the steering box installed (and the steering sensor off line) the pump was reading way low.

Note that I have now replaced every item in the entire hydraulic system except the radiator and bomb.
And also keep in mind that a double fault on the regulator and/or steering box is improbable (but could happen).
That leaves open the possibilty that the pressure readings are normal, that I just haven't tested properly.

The JY box DID briefly make it out on the street and back without any noticeable problem. A longer road test to come.
I also need to hack a sensor directly into the steering line so that we get readings while its operating rather than readings off of a plugged regulator port.
That should reveal the mystery of the last paragraph and further nail down the fault. A 2K psi sensor is inbound to replace the 1K unit I have been overrunning.

The original steering box goes onto the bench to be torn down, now that a temporary replacement has been swapped in and we are mobile.
The two repair kits out of the bmw cat are inbound, but I'm skeptical that these cover the problem.
You may expect a writeup on the original fault (if any), wrenching procedures and what exactly is in the bmw kits.

ZF inc is still on the other line, but there is a bit of static.




**** update ****
A motley crew is organizing a bench top party for around mid December.

[tvjake2]

I'll keep listening.....

[toomanyparts]

Hi Hyper

Thanks for pointing out the interchange between the E38 and E31 pumps. My pump went recently and before I order one, did you have any long term issues with the use of the E38 pump? They appear identical in physical form, but the LUK part number leads me to believe their might be a slight difference in flow rate.

A new LUK box is not possible for an 840, but a 740 box is around $300. I will likely source a used box for under $100 and take the gamble. Rebuilds seem to be garbage by most accounts and the ones with any rep approach the price of a new box.

Your input is appreciated.

[Hyper]

Just for you my friend, I put down my morning coffee, jacked up the car and verified I'm still running the green label 2106384 salvaged from the e38.
So that would be nearly 4 years, and perhaps 15K miles without incident.
***
Re: PN differences.
Don't know specifically but in general PN's could also capture...
- Pressure limit. However, another label field says they are both at 130bar.
- Mounting hardware attached. True according to the pic above.
- Destination app, ie e31 v e38
- Sub contract sourcing
********
While I'm called back here after 4 years,
I can report I've had no further trouble with the overhauled steering box.