I had a few theories about why the DTM5 was running "popcorn fart lean," a known problem with the car when I bought it, and verified on a local chassis dyno. Could my fuel pump be dying, not able to supply enough flow to the injectors? Could the injectors themselves be clogged or dirty, or just too small for the amount of air this (modified) 3.5L can flow at WOT?
Despite not updating you, dear reader, on a regular basis I have been thoroughly enjoying the DTM5 after sorting out some "pressing" issues. This is the 3rd installment in that vein. This feature was originally written by Cflo for Daily Turismo Project Car series.
Might the tuning be utter crap, and not allowing the injectors stay open long enough at higher engine speeds? I've kept you guys waiting long enough, so in this episode of DTM5 Troubleshooting, we get to the bottom of it.
After seeing the lean-spiking air/fuel ratio curve from the dyno test, several people smarter than myself suggested rigging up a fuel pressure gauge, taping it to the windshield, and driving around. This seemed like a logical first step to diagnose a lean running issue. The real issue is lack of fuel flow, but why? Ruling out supply pressure as the culprit would point to either the injectors or tuning. Inversely, finding an issue with pressure would point to the remainder of the system - the in-tank pre filter, in-tank pump (and its electrical power circuit), main filter, supply lines, fuel rail and pressure regulator.
So I grabbed a decent analog fuel pressure gauge from my local NAPA and rigged it up using some scraps of old hose and brass hosebarb tee fittings into the supply line. Ferdinand BMW placed a fuel pressure damper just prior to the rail when he designed this system, so I tee'd in between the damper and rail with a long dead-headed hose up to the gauge. This way, I'd be looking at a value as close as possible to the actual fuel rail pressure. I could now watch the fuel pressure in real time as I drove.
Simultaneously monitoring a fuel pressure gauge off to the right of my field of vision while running a high-strung S38 from 4000 rpm to up near 7000 rpm at full power in So-Cal traffic was a bit of a chore, so I enlisted some help from friends and family. "Want a ride in the M5?" I'd say. After the inevitable "sure!" and once they were firmly belted in, I would spring the assignment on them. "Try to stare intently at this tiny needle as it vibrates around inside a gauge that's taped to the windshield wiper while I mat the throttle in this stiffly sprung sedan on potholed LA streets. Then tell me the lowest number you saw." It was like having my own semi-reliable memory recall fuel pressure gauge with voice readout - just heavier and smellier.
What repeated rounds of this "pressure testing" (really just hooning with data) showed was a consistent drop after 4000 rpm at wide open throttle. This happened...every...time. It was the opposite of what you want from a manifold-referenced, regulated EFI system on a naturally aspirated car, which is low pressure at low throttle (e.g. high manifold vacuum) and then an quick jump to full pressure at wide open throttle as manifold pressure approaches ambient. My fuel system seemed fine at idle, with the fuel pressure regulator (FPR) maintaining about 40 psi. At part throttle, in the middle of the engine's rev range, fuel pressure was up to 25% lower - but OK, that's fine because manifold vacuum increases as you rev up at a given low throttle position, meaning the manifold absolute pressure is lower, which reduces the reference air pressure to the FPR, in turn reducing fuel pressure.
A little background - all of this variable pressure malarkey is necessary in a port injected pulsed EFI system to keep the flow rate consistent. The injector tips poke out and spray fuel into the intake manifold just prior to the intake ports, after the throttle plate. Since manifold air pressure changes with engine speed and throttle position, the fuel pressure on the "back side" of the injector needs to follow along in order to keep the pressure differential across the injector (∆p) constant. Base fuel pressure is usually 43.5 psi on your average Bosch multi-port EFI system. So if manifold pressure is at ambient - meaning the throttle is wide open - there's no need to vary fuel pressure. The other extreme is light load or idling; in this case a mostly-closed throttle lowers intake manifold pressure, so the fuel pressure has to be adjusted by the same amount. This is the job of the FPR.
I already verified the FPR was working fine by removing the manifold pressure reference line with the engine idling, and watching fuel pressure on my ghetto-rigged gauge. With the FPR open to atmosphere, fuel pressure was 43.5 psi at idle. With my MityVac pulling down the FPR air reference by 5 psi (for example), fuel pressure was also reduced by 5 to 38.5 psi. With the reference line disconnected and the fuel rail's return clamped off temporarily, I watched as fuel pressure rose to 80 psi or higher, meaning that the fuel pump was pretty strong when deadheaded. "Ok, so the pump is probably not the culprit," thought I.
Back to driving the car with the FPR and gauge connected. As soon as I'd whack the throttle open at low engine speeds, pressure would jump up to around 43.5 psi - that's good - but then start falling pretty linearly as the engine revved higher - that's bad, mmmkay. The lowest we saw was around 32 psi near redline. Definitely a delivery issue. The drop-off in fuel pressure vs. rpm aligned neatly with AFR curve from my dyno test, leading me to turn away from tuning and injectors for the moment.
As I continued driving around "taking data" for several days, I opened the hood at one point only to find a huge bulge in the hose to my fuel pressure gauge. It turns out this wasn't fuel rated hose - probably hydraulic hose or something, and it started delaminating internally. This could have caused a major fuel leak and possibly a fire but I caught it in time (embarrassed smiley). So remember readers - always use the right hose for the job. Since I found this stuff in a box of random hoses in my garage it was no big loss; I just carefully bled the system pressure down by loosening a clamp, and then replaced the hose with EFI-rated stuff. Disclaimer: the non-fuel rated hose could have caused a major fire. Don't be a bonehead like me. Use proper EFI rated hose for this type of job.
Next I checked the fuel filter. This really should be the first part to change out whenever fuel system condition is called into question, since it's the cheapest and a regular maintenance item anyways. But without baselining the fuel pressure as-is, I wouldn't have known whether changing the filter would have any effects. So I jacked up the rear end and crawled under to see the filter tucked over one of the axles - and - nope, it was decidedly not fresh as the wanker shop had claimed to the PO. Changing the filter is super easy and dull on most cars so I won't bore you with the details. But once the new one was in, I cut open the old to have a look.
This Bosch filter is a simple steel can with a crimped-on lid not unlike a soda can, so I simply stuck it in my vise and went at the lid with a fine-toothed hacksaw. In retrospect, I probably could have used a simple can opener without much trouble. The actual filter element is a long strip of paper rolled up around a central plastic tube (both seen above right). Fuel would enter from the bottom center, force its way through the many layers of paper, and then exit through the nipple in the top of the can. Once unrolled and held up to the light it was obvious this filter was severely dirty and therefore restrictive - that paper element should be white! Little boring maintenance jobs like replacing filters may not be as cool as turning up the boost or chipping your ECU, but you can't hope to make consistent & reliable power without a "Stage Zero" or known good baseline.
The skanky fuel filter was the smoking gun I'd been looking for. With much higher restriction in the fuel supply line due to the dirty filter, the fuel flow rate simply couldn't keep up with what the engine demanded at WOT and high revs. But to be safe, I replaced another cheap and easy part simultaneously: the fuel pump relay. I decided to change this only as a precaution; I'd read sage advice from guru mechanics on the intardnetz that 20 year old BMW relays aren't the most reliable devices.
By prying the plastic cover off of an average Bosch-style SPST relay you can pretty readily see how it works. When terminal 85 is grounded (by a physical switch or a fairy in the ECU), a low current trigger signal flows from terminal 86 through a fat coil of small-diameter wire on to terminal 85 completing the low-side circuit. Current flowing through the coil induces an electromagnetic force that acts on a slug or rod of iron in the center of the coil. This moves the rod axially and completes the high-side circuit, allowing higher current to pass than the trigger switch (or ECU) could, thus activating the beefy component you want to turn on. In this case, that would be the fuel pump. But as a relay gets old, corrosion, fatigue, and wear can lead to higher internal resistance. The relay can't pass as much current on the high side, meaning the pump could be starved for power. I didn't want to chance it with the DTM5 and replaced the fuel pump relay with a brand new part even though the old one checked out OK.
After my two cheap fixes (which cost less than $40 combined) I went out for another round of testing.With multiple parties in the passenger seat eyeing the gauge, I could finally verify that fuel pressure stays rock solid to redline at wide open throttle. Approximately 43.5 psi all the way up means that the system isn't overly restricted any longer and that monster S38 can drink as much long-chain hydrocarbon juice as it wants. Great success!
Unrelated bonus: The E34 M5 has a fancy variable-length intake plenum; a resonance charging system that alters intake runner length within at WOT below 4120 rpm and also above 6720 rpm, giving a lot more "area under the curve torque" for not much effort. Have a look at this M5Board.com forum post and this repair guide for further details. During my fumbling around with the fuel pressure gauge, I found that this fancy variable intake system wasn't actually operating because one of the wires leading to the vacuum control solenoid was broken. It seemed fine at a casual glance - the wire was still inside the connector so nothing looked amiss. But when I was disconnecting my fuel pressure gauge from the supply line I brushed the wiring with my hand and noticed one wire had fallen out of the connector. A quick fix with my free soldering iron netted a max gain of 30 lb-ft of torque (40 N-m for you freedom hating un-Americans) at 3500 rpm. Flappy valve intake system: engage. Full fuel pressure: check. Commence hoonage.
Next stop: back to the chassis dyno to get a look at AFR, torque, and power after these small but effective fixes.
All of the Daily Turismo project car posts can be found here. That's right - the DTM5, Draken, Schmetterling and our old 242, all in one convenient link!
Image credits: Daily Turismo.