What's new
Van's Air Force

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

Fuel flow oddities

kaweeka

Well Known Member
I've been struggling with aberrant fuel flow readings for the past 2 years. Tom Swearengen and I have batted around multiple ideas without nailing down the cause so I'm throwing this out to the brain trust.

Engine: IO 320 B1A about 800 hours
Injection: AFP FM 100 (older model with the purge valve) plumbed between the servo and flow divider
Ignition: Lightspeed Plasma III
Transducer EI FT-60 (Red Cube)

Fuel flow fluctuates randomly and quite drastically between 0 and 5 gph during taxi.
Fuel flow is ROCK SOLID during full power climb and high power settings running ROP

If I run LOP, I'll get to about 20* LOP 62% power and see about 6.7 gph which is about right for my plane historically. After a few minutes, the ff indication will drift down to 4.8 gph without any loss of poser, change in ANY engine parameters, TAS or MP. (I'm not really burning 4.8 goh at 147 kts.) If I leave it alone, it will correct itself after about 1 hour to 1:10 back to 6.7 gph and remain there.

I rewired the run from the GEA 24 to the transducer, replaced the transducer, tried several iterations of hose configurations (thank you Tom) but nothing changes this odd behavior. I cannot figure out the problem. Here is a link to all my G3x data courtesy of Savvy folks. I'd sure appreciate some insight to this phenomenon.

https://apps.savvyaviation.com/my-flights/4123/f520fec6-e341-4f88-9871-c9a31209f356


I'll answer anything I can to help the diagnosis along. Poor Tom has been along for this ride for too long. He is a saint!

thanks all,
David
 
Last edited:
My part in all of this. Dave was going to replace is old FLoScan from the firewall location, with a FT60. He contacted us about our flow divider mount. Installed, worked just fine, as advertised, then after some hours of flight time, this anomaly started popping up. WE moved the transducer to several other points all with new hoses, and get virtually the same thing. Just when you think you have it solved, the anomaly pops up.
Dave sent the transducer back to EI for testing. Ei passed it and returned it to him. Reinstalled, re did all the electrical. Worked fine for some period of time, then the anomaly started again. Made no sense, no matter where we put the transducer, the anomaly would rear its ugly head.
Dave purchased another FT60, installed it and ist was good for X number of hours, and it started EXACTLY the same thing as the first one. WE went back through all the different install locations and retested, same results with the anomaly showing up.
Dave contacted Don at AFP, and he an Kyle went back through the servo, purge valve, and flow divider. Returned, reinstalled. Similar period of time, and here we go again. I had 2 other things I suggested. 1) bypass the mechanical pump and use the AFP boost pump to force the fuel at a higher constant pressure to the servo. Fuel pressure doesnt seem to be an issue as its pretty consistant. Try that and see the results. 2) Reinstall the original FloScan and see if is an electronics issue, or an actual fuel system issue. Dave has checked and rechecked all the plumbing and it's correct, tight, and no suction leaks.

It seems to Dave that for some mysterious reason, the FT60 does a 'reset' on it own, and works perfectly, then after X period of time starts misbehaving. It does appear that it does this at LOP settings at around 65-62% power. But the other engine parameters are in normal ranges. Guess its like alittle heart fluxuation.

So---anyone with any ideas, feel free to chime in. No stone will be left unturned. He's been fighting this for 2+ years-want to find the root cause and eraticate it!
Tom
 
Just for grins, can you try a different Fuel Flow XCDR supply and input on the GEA 24?
 
Brian, I'm not sure what that is (XCDR supply). Sorry but can you explain a little more for me? What other input can I use on the GEA 24?
Thanks.
 
Brian, I'm not sure what that is (XCDR supply). Sorry but can you explain a little more for me? What other input can I use on the GEA 24?
Thanks.

The GEA 24 has 4 channels available for a Fuel Flow transducer -- See pages 26-30 and 26 -31, 26-34 of the G3X Installation manual - Revision AN.

You'll also need to change the configuration when you move the transducer to a different channel.
 
Last edited:
Years ago I talked to the manufacturer of the cube and they told me the "best practices" for mounting it.

1. Don't hard mount to the engine. (I mounted mine on a 2" slightly angled spring steel support).

2. Mount with the intake slightly lower than the outflow.

3. Inflow and outflow need to be straight in/out and not angled or 90* fittings.

I did all that and haven't had an issue in 800hours.
 
Myron,
I have checked all of your noted restrictions. Mine is mounted below the 1/3 inter cylinder baffle in a vertical manner with straight connections above and below. I also have heat wrap around the cube and lines.
 
Aside from trying a different input on the GEA24, (a good idea) since the data you are seeing is as interpreted by the GEA24, what you really need is a solid confirmation of what the red cube is actually providing to the GEA24. That said, what I would do is tie into the GEA24 input from the cube and send the signal to a portable (battery operated) oscilloscope, preferably with frequency counting capability, and take the setup for a taxi and look at the cube signal both for signal integrity and frequency (with a second person). You can easily convert flow numbers to frequency and vice versa using the cube K factor. If you get one of those flow fluctuations as per the display, see if it is there at the source, coming from the cube. Or, you may find an unexpected other signal anomaly. This may sound like a complicated setup, but you need better insight which side of the GEA24/cube interface the root cause is.

While there is a huge amount of talk surrounding do's and dont's of mounting the cube, the conversation seems to end up making the cube appear far more sensitive than it actually is. It just spins a little wheel with fuel as it goes by and doesn't really recognize much of the outside world, other than if it directly sees a very pulsing flow. So other than a faulty one (or even faulty replacement it does happen), I wouldn't obsess over cube location.
 
If it's mounted between the engine driven fuel pump and servo I would probably suspect the fuel pump.
I always mount the xdcr between the servo and spider.
 
Last edited:
fuel flow location

I initially mounted my red cube on the firewall before the engine driven fuel pump and fuel flow fluctuated. I then moved it on top of the engine before going into the spider. Rock solid and never any problems. Try to have both the in and out as straight as possible, at least 3-4" and have a nice curve on both ends.
 
Dave's is mounted between the servo and flowdivider, in a lower location near the sump. Its been moved around to verify the readings are NOT related to any mounting orientation, horizontal, vertical, upside down, with straight AN nipples, and hose lengths----just to satisfy the 'requirements ' of EI's recommended install. Nothing changed due to the location.
Brian may be on to something.

Tom
 
I would be looking for a leak. A small leak would allow air in under suction before the mech pump. IF the Cube is mounted on the spider inlet (high point of the system) that air could collect in or around the cube and mess with the readings, as air in the cube internals could mess with the spinning of the little paddle wheel. The significant ramp down to the 1/8" SS lines, along with the injectors being below the high point of the system would mean that the air is not impacting flow to the cylinders. It is likely just a pocket of air staying in the cube.

You could prove/disprove this theory by moving the cube to a lower location, but still between the servo and spider.

Larry
 
Good thought, Larry. I was thinking along those lines as well but there were also excellent suggestions about the signal quality from the GEA 24. I'll be giving attention to both.
 
Good thought, Larry. I was thinking along those lines as well but there were also excellent suggestions about the signal quality from the GEA 24. I'll be giving attention to both.

I think connectivity is unlikely, as the red cube is a transducer, not a resistive sender. These types send a voltage pulse from an I/C, based upon wheel revolutoin (like a modem sends 0's and 1's via voltage or no voltage) and are far less sensitive to wiring issues than resistive senders, which rely upon a very low current signal that is sensitive to line resistance. The GEA count's pulses against a set time window and uses math with the k factor offset to compute FF. With a transducer, there is no signal from the GEA. The transducer runs off 8-12 VDC and the red cube I/C generates a set voltage pulse (0 volts - 5 volts - 0 volts) for each paddle on the wheel that moves past the optical sensor. The GEA only needs to count the pulses, just like it counts pulses from your EI or mag to compute RPM. Wiring would have to be really wonky for it to show a low reading vs no reading. An intermittently failing wire or connection would not reduce the flow a specific percentage and hold it there for an hour and then come back to normal.
 
Last edited:
Larry----Dave has moved the cube to several different locations, 2 of which were between the servo and flow divider. Several were upstream of the servo---just for testing purposes. All of these give the same anomaly, no matte where you put it. SAME with a NEW FT60. I believe that he has re wired everything using different wire, connectors, glue, tape, bailing wire, bubble gum, shoe goo, you name it. I think the only thing he hasnt changed is the EFIS itself.
I suggest that for giggles he reinstall the old FLoScan and see what it does.

Something about this plane and a FT60 ( or different FT60s) that it doesnt like.

Tom
 
I think connectivity is unlikely, as the red cube is a transducer, not a resistive sender. These types send a voltage pulse from an I/C, based upon wheel revolutoin (like a modem sends 0's and 1's via voltage or no voltage) and are far less sensitive to wiring issues than resistive senders, which rely upon a very low current signal that is sensitive to line resistance. The GEA count's pulses against a set time window and uses math with the k factor offset to compute FF. With a transducer, there is no signal from the GEA. The transducer runs off 8-12 VDC and the red cube I/C generates a set voltage pulse (0 volts - 5 volts - 0 volts) for each paddle on the wheel that moves past the optical sensor. The GEA only needs to count the pulses, just like it counts pulses from your EI or mag to compute RPM. Wiring would have to be really wonky for it to show a low reading vs no reading. An intermittently failing wire or connection would not reduce the flow a specific percentage and hold it there for an hour and then come back to normal.

While the connectivity is unlikely the root cause, you never know what kind of goofy wiring error, perhaps even repeated in re-wiring, is there. But the replacement of the cube with another "presumably good" unit "appears" to vindicate the cube, it all adds up to "you don't know who's messing up", and moving stuff around and replacing parts isn't yielding an answer. So, I suggest it's back to - you need to know for sure what the GEA is seeing to point the finger in one direction or the other. And that means putting instrumentation at the GEA input. A frequency counter at a minimum, a portable oscilloscope is optimal to see the actual signal. That will give you an answer beyond all the guesswork so far. Hopefully the suggestion of an oscilloscope isn't too off-putting, but in a situation like this, unless you get lucky with the guesswork, that's what will get to the bottom of it.
 
I like your approach and process. This task is beyond my simple knowledge base but it can't hurt to look around for someone who might be able and willing to help in this regard. I'm still going to re-double check all my connections in the fuel system.
 
While the connectivity is unlikely the root cause, you never know what kind of goofy wiring error, perhaps even repeated in re-wiring, is there. But the replacement of the cube with another "presumably good" unit "appears" to vindicate the cube, it all adds up to "you don't know who's messing up", and moving stuff around and replacing parts isn't yielding an answer. So, I suggest it's back to - you need to know for sure what the GEA is seeing to point the finger in one direction or the other. And that means putting instrumentation at the GEA input. A frequency counter at a minimum, a portable oscilloscope is optimal to see the actual signal. That will give you an answer beyond all the guesswork so far. Hopefully the suggestion of an oscilloscope isn't too off-putting, but in a situation like this, unless you get lucky with the guesswork, that's what will get to the bottom of it.

I don't disagree. However, all that is happening here is the cube sending voltage pulses and the gea counting them. The fact that it never stops working would imply that the voltage and ground feeds to the cube are solid, otherwise it would intermittently stop sending data. The odd part here is that the pulse rate drops to a specific new rate and stays there for an hour and then returns magically to the correct rate. Failing wiring or other intermittent electrical issues should show MUCH greater variability. That just doesn't sound like a wiring issue. I simply cant imagine a scenario where wiring to a transducer can create this scenario. Seems far more likely that either the OP got two bad cubes, the G3X s/w is taking a nap or it's clock is messed up, or something is causing the wheel to occasionally spin slower than the flow should spin it (air or some mfg defect in the bearing that is causing rubbing of the wheel and case). The cube measure flow via both mechanical and electrical components. The moving fuel mechanically spins the wheel and the electronics observe the revolutions and report the result via on/off electrical signals. If the G3X is working correctly, I don't see how a scope helps here, as it likely just confirms what the G3X is saying. Now, if the G3X has a H/W issue and is not correctly sensing on it's port, then the scope will prove that. However, that same clock is likely used for the RPM sensing and the OP is not seeing issues there. Problem is, not a lot of folks around the airport with scopes.

That said, I have been wrong before. I agree with Tom. If it were me, I would but the floscan back in and see if the problem remains.
 
Last edited:
Larry I dont disagree either. Out of over 100 installs of FT60s, I can only think of 4 with anomalies, 2 of those were traceable to a wire failure at the cube body. The other 2, one of which is Dave's, has just been one of the great mysteries of our time. Both showed 'normal' readings as compared to other similarily equipped aircraft. Then for what ever reason, begain to show some anomalous readings, then went back to reading normal, either after period of flight time, or after landing and engine shutdown, Both seemed to be fine after the next engine start.

In both of those instances, as a matter of customer service and curiosity, we moved the transducer and provided hoses and fittings for differnt locations. In Dave's case SEVERAL locations, both upstream and downstream from the servo. Even though we have many of these installs flying with no issues, we wanted to make sure we didnt miss anything. Both transducers, or at least the electronics interface part with the EFIS---seems to have a mind of its own. Reads normal, then acts flakey for a while, then acts normal.

I'm beyond puzzled--

Tom
 
Larry I dont disagree either. Out of over 100 installs of FT60s, I can only think of 4 with anomalies, 2 of those were traceable to a wire failure at the cube body. The other 2, one of which is Dave's, has just been one of the great mysteries of our time. Both showed 'normal' readings as compared to other similarily equipped aircraft. Then for what ever reason, begain to show some anomalous readings, then went back to reading normal, either after period of flight time, or after landing and engine shutdown, Both seemed to be fine after the next engine start.

In both of those instances, as a matter of customer service and curiosity, we moved the transducer and provided hoses and fittings for differnt locations. In Dave's case SEVERAL locations, both upstream and downstream from the servo. Even though we have many of these installs flying with no issues, we wanted to make sure we didnt miss anything. Both transducers, or at least the electronics interface part with the EFIS---seems to have a mind of its own. Reads normal, then acts flakey for a while, then acts normal.

I'm beyond puzzled--

Tom

I have had 2 red cubes fail. In both cases, it would work fine for a while, then it gave erratic readings for a while (never consistent), then it would read 0 for a while, then back to normal. It was never as consistent as the OP indicated; normal and then a precise error/offset for a while, followed by normal again. It just doesn't seem to fit the normal failure mode.

In both of my cases, the plug on the wire at the body had worked itself out and I suspect internally the wires were pulling off the circuit board. I also speculated that acrylic window on the hall sensor had clouded over.
 
Last edited:
I don't disagree. However, all that is happening here is the cube sending voltage pulses and the gea counting them. The fact that it never stops working would imply that the voltage and ground feeds to the cube are solid, otherwise it would intermittently stop sending data. The odd part here is that the pulse rate drops to a specific new rate and stays there for an hour and then returns magically to the correct rate. Failing wiring or other intermittent electrical issues should show MUCH greater variability. That just doesn't sound like a wiring issue. I simply cant imagine a scenario where wiring to a transducer can create this scenario. Seems far more likely that either the OP got two bad cubes, the G3X s/w is taking a nap or it's clock is messed up, or something is causing the wheel to occasionally spin slower than the flow should spin it (air or some mfg defect in the bearing that is causing rubbing of the wheel and case). The cube measure flow via both mechanical and electrical components. The moving fuel mechanically spins the wheel and the electronics observe the revolutions and report the result via on/off electrical signals. If the G3X is working correctly, I don't see how a scope helps here, as it likely just confirms what the G3X is saying. Now, if the G3X has a H/W issue and is not correctly sensing on it's port, then the scope will prove that. However, that same clock is likely used for the RPM sensing and the OP is not seeing issues there. Problem is, not a lot of folks around the airport with scopes.

That said, I have been wrong before. I agree with Tom. If it were me, I would but the floscan back in and see if the problem remains.

OK, once again, a conversation full of "ifs", "seems", "unlikely", etc., but you and Tom both keep agreeing that you don't really know. And that is the problem, you need to do something where you will know something! Not sure why you are resisting the idea of looking at the signal right into the GEA, because then 1) If you see nice pulses that actually change in sync with the displayed numbers, then you can firmly condemn the cube, get another one yet, or put the FloScan back in; 2) If the pulses look good and don't go wrong when the display shows a radical change, then you have a GEA problem; and 3) If for some reason the signal isn't the neat clean opto-coupler generated stuff you are assuming and expecting, then it's wiring/signal problems - say for example somehow the voltage being sent to the cube isn't the required 12V.

Now, if you are reluctant to tap/cut into the GEA input - here's how to do it the easy way: Pull the signal pin out with a pin remover tool - If you've built this thing, you most likely have one or can get one. Then, make a short T-jumper with a pin on one end and a female pin on the other, put the pin into the GEA connector, and the signal from the cube into the other end of the jumper. The T goes to the scope, ground can be clipped onto metal near the GEA. It's a bit of effort, but it can get you to real information.

And if access to a scope is too big a problem, many digital voltmeters have frequency counting capability. With that you can check the pulse frequency and look for the big changes, and do a rough signal integrity check - since the pulses are 50% duty you can look for a voltage around half the power voltage, which per the G3X manual is half of 12V.

If this isolates to the cube, then you still have to figure out why. Two bad ones? Possible. Cavitation, captive air inside, maybe, but located just before the flow divider, I think pretty much everyone with that location has no problem.

Sorry to sound a bit strident. It just seems this mystery is begging for an approach that will get some hard data. Too bad I am 2k miles away in Chicago, or I would be right over there with my battery op scope!
 
Reinhard-----ANY suggestions, good, bad, ugly, etc for Dave would be appreciated, thats why he posted to begine with. I'm NOT and electrical engineer, nor profess to be one. I know the plumbing of these systems.
Dave even enlisted the help of EI, the manufacturer of the unit, sent back the first offending one for testing. They gave it a clean bill of health, he reinstalled it, and it had a brain flux not long afterwards.
Tell you what I can give you his cell number----I'm sure he would be GLAD to hear from you, or better yet, he likes to fly----bet he'd love to fly to Illinois and have you look at things. Bet we all could learn something.

Tom
 
Thanks Tom!

I responded to Dave PM including my phone and email.

Yes, this sounds confounding. My plane had a VM1000 which just last year got replaced by a GEA24 and red cube, so I just recently installed this same setup, including the removal of a Floscan. Cube just ahead of the spider. Luckily, mine works just fine.
 
OK, once again, a conversation full of "ifs", "seems", "unlikely", etc., but you and Tom both keep agreeing that you don't really know. And that is the problem, you need to do something where you will know something! Not sure why you are resisting the idea of looking at the signal right into the GEA, because then 1) If you see nice pulses that actually change in sync with the displayed numbers, then you can firmly condemn the cube, get another one yet, or put the FloScan back in; 2) If the pulses look good and don't go wrong when the display shows a radical change, then you have a GEA problem; and 3) If for some reason the signal isn't the neat clean opto-coupler generated stuff you are assuming and expecting, then it's wiring/signal problems - say for example somehow the voltage being sent to the cube isn't the required 12V.

Now, if you are reluctant to tap/cut into the GEA input - here's how to do it the easy way: Pull the signal pin out with a pin remover tool - If you've built this thing, you most likely have one or can get one. Then, make a short T-jumper with a pin on one end and a female pin on the other, put the pin into the GEA connector, and the signal from the cube into the other end of the jumper. The T goes to the scope, ground can be clipped onto metal near the GEA. It's a bit of effort, but it can get you to real information.

And if access to a scope is too big a problem, many digital voltmeters have frequency counting capability. With that you can check the pulse frequency and look for the big changes, and do a rough signal integrity check - since the pulses are 50% duty you can look for a voltage around half the power voltage, which per the G3X manual is half of 12V.

If this isolates to the cube, then you still have to figure out why. Two bad ones? Possible. Cavitation, captive air inside, maybe, but located just before the flow divider, I think pretty much everyone with that location has no problem.

Sorry to sound a bit strident. It just seems this mystery is begging for an approach that will get some hard data. Too bad I am 2k miles away in Chicago, or I would be right over there with my battery op scope!

From EI:

The input (red power lead) will operate from 8 to 30 Volts. Typical supply current is 14 mA over the entire operating range. Power to the FT-60 should be regulated and supplied by the instrument interfacing with the FT-60.

The FT-60 incorporates an open collector output (white lead). The output can operate from 0 to 30 Volts. Saturation voltage at a given sink current is 0.25 Volts (typical) at 4 mA and 0.7 Volts (max) at 10 mA. The output should be limited to 15 mA max. Recommended pull-up current is 1 mA. The output incorporates a two-stage comparator that keeps the output waveform square, even at low flow rates.

Don't really see how varying voltage from the gea24 can create serious problems. At least not problems with stable readings that are only 10-15% less than normal. your advice is certainly sound; I was only trying to create options that are easier for the OP, given that access to a scope is quite difficult for many. I am sure you will help him figure it out. Just thought that going back to the floscan to confirm or deny the red cube is much easier. These things are out there in the thousands and problems like the OP's are very uncommon and routinely addressed without the use of a scope.

The fact that the operation of the red cube has been confirmed by EI and the readings are very stable (just not always accurate) would point to a mechanical issue with the spinning wheel, causing it to spin at a rate slowly than the design point for a given flow, or something internal to the GEA 24, as opposed to wiring or frequency anamolies.
 
Last edited:
Just a question to help with troubleshooting.
The data I reviewed would indicate the FF is probably at a higher number than reported and not that the FF has truly dropped.
However an easy test or check is to see if the calculated total fuel used is equal to what you put back in the tanks.
I have my tanks calibrated to be within .2 of a gal after consuming 30 gallons.
If your calibration was good before the problem, it would be easy to tell if this is a reporting issue i.e. FT60 or air or if it is truly a FF issue.
 
Back
Top