8 march 2020
There are few things more alarming than barrelling up a road at speed and seeing engine oil pressure failure on the instrument panel. The first incident was in june 2018, 2500 miles from home, driving to South Carolina to visit a friend while on the Hot Rod Power Tour. After a moment's panic, lifting my foot off the gas pedal, the oil pressure rose. Oil volume (and pressure) varies with engine RPM, so this was odd. Long story short, I diagnosed this as sender failure, rigged up a parts store fix with a conventional oil pressure switch and continued the drive.
The second was April 2019, trying to catch up to a group of sports cars, climbing a steep grade at 80 mph. 35 psi instead of expected 60, after another moment of terror, I lifted my foot and declutched, oil pressure rose as before.
When i got home google-research (sic) uncovered an epidemic of failed oil pressure sensors, in all sorts of cars and applications. So I disassembled the sender that had failed in South Carolina (it was in my junkbox, the current bad one still installed) to get to the bottom of this failure -- though by now I knew exactly what the cause was before i even looked.
The cause of failure is high ohmic resistance at the constant-wear spot, for me, 56 - 60 psi. In both instances I had warning of impending failure in the form of erratic readings at constant rpm. When new the reading at a constant rpm was steady (or plus/minus one digit, quantization error). As the months wore on, the steady reading would vary more and more, eg. 58 plus or minus 1, to +/- 2, then 3, 4, 5, changing multiple times per second. Final "failure" was readings varying erratically from nominal 60 to 40, 30, 20 psi, then back up again with no change in engine speed.
Dropping rpms appeared to increase oil pressure readings, because the wiper move to a non-worn part of the card. Vibration at idle is quite low and doesn't seem to wear the card out.
Oil pressure senders are a piston or diaphragm exposed to oil pressure which pushes a bar against a spring. The bar rotates a wiper across a resistance-wire card. The wiper is at electrical ground and causes the resistance at the (single) electrical terminal of the sensor to vary from "high" (usually 230 ohms) at zero pressure to "low" (33 ohms at 80 or 100 psi). This is a fine design; the problem is in the crappy construction.
The sender is screwed into the engine block, usually sticking straight out from the block. Orientation matters here -- nearly all vibration in otto cycle engines is in the plane parallel to connecting rod movement, directly in line with the bar with its soft spring.
At highway speeds, engine rpm/pump volume is enough to hold the relief valve open, therefore oil pressure is fairly constant. This means that the wiper sits at the same spot on the card for most of it's operating life, and right when engine vibration is highest (high rpm).
After years/decades of cost cutting the wiper and resistance wire and insulating card are about as cheap and thin as possible. The materials used are not wear nor corrosion resistant, and when the vibration is in the worst-case direction, sensors last months, not years.
One astute customer at Summit Racing noted that when mounted parallel to the engine the sensor lasted longer, which jives perfectly with this failure mode. My solution is to remote mount the sensor on the inner fender, and feed it with a braided hose; see photo below.
The cost difference between a cheap sensor and one-that-might-be-reliable is significant: $40 vs. $150 or more, and a more complex (incompatible) interface. so we all stick with the cheap ones. OEM sensors don't seem much better, anecdotally.
Below are photos of my previous sender (failure #1), chopped up; it lasted half a year maybe. I got it from Summit Racing, screwed it into the main oil gallery, up near #1 cylinder, sticking straight out of the block. Maximum exposure to vibration, which you'd think a part like this would be designed for. I hacksawed through the crimp to take the top off.
Here's a 5-second video showing bar and wiper operation.
The problem was immediately visible. The bar rotates between two pivots, and is fairly massy, compared to the wiper and spring. Also the wiper tip's displacement, being the end of relatively long arm, amplifies any small bar vibration.
here the wear is abundantly visible. the wiper is tin plated brass, which i would think would be fine, but there is black corrosion or something around the shiny brass spot. the sensor was sealed and not contaminated with liquid. The wires over the white plastic card are visibly mashed, though wiper pressure seems minimal. Some dissimilar metals issue?
My installation/application passes 5 milliamps (.005 amps) through the wiper (1K ohm pullup to 5V, input to a computer analog in) so it's certainly not burned up by current, and 5mA is a sufficient "sealing current" (if you're familiar with telephone technology) to overcome galvanic stuff.
here's my current solution. ask me in a year or two how this worked out.
March 2020: It's working fine. No failure, no jitter, solid readings. SHIP IT.