What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Diagnosing fuel pressure rapid continuous variation

rjcthree

Well Known Member
I’m constructing a plan to diagnose a change in fuel pressure behavior in my most recent flight. Previous flights have had stable and predictable fuel pressures depending on the phase of flight (MAP/throttle), with the most recent flight showing wide and rapid variation (needle continuously jumping +/- 2 psi around the mean, mean pressure being also down from typical by 0.5psi with or without the boost pump on) of fuel pressure seen on/by the EFIS. I will add a mech gauge , put a scope on the sensor output, you name it. it’s a D180 with a three wire kavlico 0-15 psi sensor that’s been on the plane for 80 hours. Note no engine running abnormalities were seen in flight or in the data.

After studying the data and doing modest inspection and diagnostics ( not able to run the engine due to rain) I do not have any reason to doubt the data. Boost pump on or off in flight changed the mean by 0.5 psi when on as normal, but did not change the sensed pressure fluctuations recorded and seen. Pressure is steady on the boost pump only when not running.

Even though it’s an Ellison, it’s a typical system, with tanks to selector to boost pump to collator to engine driven pump to carb, which is a pass through to the sensor and primer solenoid.

The question: is it possible the outlet check valve of the engine driven fuel pump, if leaking, could cause the pump to ‘breathe’ from the output side, thereby causing variation?
 
The question: is it possible the outlet check valve of the engine driven fuel pump, if leaking, could cause the pump to ‘breathe’ from the output side, thereby causing variation?

There is a diaphragm with to check valves, inlet and outlet. When the diaphragm relaxes, the outlet closes and the inlet opens. As it compresses, the inlet closes and the outlet opens. The pumps are volume based in design and can only produce pressure when there is a restriction and the consumption is below the max flow capacity of the pump. Lets say it has a 45GPH capacity. At 1 GPH, it may develop 5 PSI and at 45 GPH it may develop 0 PSI. There is a somewhat linear change from 0 to 5 that ends somewhere in the middle of that range and then the max 5 is maintained due to springs in the diaphragm that reduce the compression stroke.

Check valves are unlikely to cause intermittent problems like you are seeing IMHO. When they go bad, I would possibly expect to see volume and pressures varying, but would not expect them to be normal in some flights and not others. I would expect a bad check valve to have low volume and pressure across the board. I would expect these kinds of problems to deliver relatively consistent results.

A leaking outlet check valve would prevent the diagphragm from pulling in the full amount of fuel and therefore would reduce volume delivered on the compression stroke and if that volume is low relative to consumption demand, pressure will drop. A leaking inlet would have a similar effect, but would be due to some of the fuel leaking back to the inlet pipes on the compression stroke.
 
Last edited:
More data

I was able to get the data off my other machine and get some comments in for clarity. The first is completely normal, from 9/9. The second is the very next flight, 9/18. Clearly different. The scale is is psi, and is the same for both.

A working hypothesis that drives the mech pump outlet check valve question is a reduction in observed (calculated) mean pressure. If the pumping efficiency is less because the intake stroke is drawing from the inlet and a small amount from the leaking outlet check valve. The pump cycle would repeat. Since all I have at the moment is 5 sec data intervals, so what is logged is essentially random points, but it shows the working variability. My working assumption related to the pump is it’s likely on cam lobe frequency, something I can get with an external tool.

So yes, something changed, maybe the pump is compromised, maybe it’s a dislodged pc of debris from whenever making its way, but something isn’t right, and I’m grounding it until I can root cause it and replicate. Maybe this hypothesis is trash. We’ll see.
 

Attachments

  • 6E6B8225-62D0-459F-8D97-25B0C91559F6.jpeg
    6E6B8225-62D0-459F-8D97-25B0C91559F6.jpeg
    155.7 KB · Views: 100
  • 14518D3A-32ED-4D75-9FF3-1917F95DA21B.jpeg
    14518D3A-32ED-4D75-9FF3-1917F95DA21B.jpeg
    204.4 KB · Views: 126
Last edited:
Rick,

I would look very carefully at the wiring to and/or the sensor itself. Got a spare?

B
 
Looked at the wiring….

Oil pressure and fuel pressure share power and ground from the Dynon, per the Dynon installation/ retrofit from VDO instructions. All of that looks intact, manipulation does not drive any changes in output. Tapping and the like does give any sensor response. Oil pressure is solid, no variation. When I get a scope on it and can run it, I think much will be learned.

I don’t have a spare kavlico of the right range, for testing purposes I have plenty though at 0-150 psi. I still want to put the mechanical sensor in parallel before replacing the sensor.

Interesting observation: I’m surrounded by some very talented electrical, instrumentation, and mechanical engineering people. When shown the data you see, the electrical/instrumentation people swear it has to be electrical or the sensor, the mechanical people are confident in a physical fault. I expected the opposite. It’s a head scratcher.
 
Last edited:
I was able to get the data off my other machine and get some comments in for clarity. The first is completely normal, from 9/9. The second is the very next flight, 9/18. Clearly different. The scale is is psi, and is the same for both.

A working hypothesis that drives the mech pump outlet check valve question is a reduction in observed (calculated) mean pressure. If the pumping efficiency is less because the intake stroke is drawing from the inlet and a small amount from the leaking outlet check valve. The pump cycle would repeat. Since all I have at the moment is 5 sec data intervals, so what is logged is essentially random points, but it shows the working variability. My working assumption related to the pump is it’s likely on cam lobe frequency, something I can get with an external tool.

So yes, something changed, maybe the pump is compromised, maybe it’s a dislodged pc of debris from whenever making its way, but something isn’t right, and I’m grounding it until I can root cause it and replicate. Maybe this hypothesis is trash. We’ll see.

While not an expert in pump design theory or diagnostics, based upon my understanding of the mechanical principles involved with these pumps, I really struggle to see how any mechanical issue in the pump can cause such wild swings up and down over such a long period, excluding possibly issues related to ongoing introduction of air on the intake side. I am with the EE's here, My vote is on instrumentation issues; WAY too consistent of a frequency to be mechanical in nature. Just one man's opinion though.

Debris in the check valve would deliver a reduced flow, and commensurate pressure, that was consistent for any given RPM and fuel flow. It wouldn't bounce up and down wildly IMHO.

The lobe for the pump is on an idler gear between the crank and cam, NOT the cam itself. Though that idler gear does turn at cam speeds (1:2 from the crank). At cruise revolutions, that lobe is activating the pump 20 times per second. WAY TOO fast to observe pump cycle / frequency related issues on your instrument. Similar to why you can't see AC electric lights flicker eventhough they are reducing to 0 volts 60 times per second. At some point, the on off cycling becomes irrelevant.
 
Last edited:
Interesting observation: I’m surrounded by some very talented electrical, instrumentation, and mechanical engineering people. When shown the data you see, the electrical/instrumentation people swear it has to be electrical or the sensor, the mechanical people are confident in a physical fault. I expected the opposite. It’s a head scratcher.

https://en.wikipedia.org/wiki/Blind_men_and_an_elephant

...have to agree with Larry though, the mechanics of the pump and the physics of fluid dynamics preclude any mechanical issue yielding the data that you've captured.

OTOH VDO sensors do fail...
 
H2AD pump ahead of #2

H2AD fuel pump arm is on a separate cam lobe, left front top of the engine. It’s one of those things about an H2. It’s also already a fairly fast kavlico, not VDO. If it’s real, the sensor should have the speed, and the oscope should record it. I’m most worried that the mech gauge won’t give enough visibility. But that’s a problem for another day.
 

Attachments

  • FCCA318B-E6E1-436A-958A-88E2F3D1B1A1.jpg
    FCCA318B-E6E1-436A-958A-88E2F3D1B1A1.jpg
    384.7 KB · Views: 106
Last edited:
H2AD fuel pump arm is on a separate cam lobe, left front top of the engine. It’s one of those things about an H2. It’s also already a fairly fast kavlico, not VDO. If it’s real, the sensor should have the speed, and the oscope should record it. I’m most worried that the mech gauge won’t give enough visibility. But that’s a problem for another day.

Sorry, missed that in your post.
 
A root cause, not sure it’s THE root cause

I confirmed the variation is real electrically and with a mech gauge. I confirmed my previous hypothesis of leaking fuel pump output check valve is not the singular root cause, as swapping pumps with a tested sound pump did not correct the issue. What I have found is the Ellison has the dreaded death drool. Since the Ellison is no longer supported, I’m forced to go a different direction, which will be to an MS carb. The Ellison gave me 250 good hours, but the lack of support brings me to the end.

Once I have the MS carb on, plumbed, bracketed, FAB’d and a million other little things, we’ll see! Until then….grounded. :(.
 
If there is a hose from the pump outlet to the sensor or from the carb to the sensor, try disconnecting that hose at both ends and allowing it to drain completely. Then reattach to the sensor and pressure source. The air in the line will act as a snubber and possibly dampen out the fluctuations you are seeing.
Good Luck,
Mahlon
 
Back
Top