What's new
Van's Air Force

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

New SDS Closed Loop Testing

rv6ejguy

Well Known Member
With the news that unleaded avgas is starting to be distributed in the US now, we've decided to get started early on perfecting closed loop feedback control software for aviation applications.

We'll be shipping off a modified ECU to one of our test partners in Colorado tomorrow for flight testing later this month. We have several strategies in mind for O2 sensor fault detection and correction, building off our 20+ years experience with this feature in our automotive ECUs.

Initial goals for production ECUs would be to have user programmable AFR targets for several rpm and load ranges. The O2 sensor will command auto fuel trim based on these parameters.

The eventual goal would be something closer to full FADECs where all fueling and timing is automatic. I'm not sure how well that can be implemented on multiple engine types and configurations in the Experimental world but we can explore heading in that direction at least and see how we end up.

We'll also be acquiring a test engine we can bench run under load with a prop later this month to validate code and performance.

The hardware is already in place on present EM-5 ECUs to implement closed loop control so only a reflash would be involved to have this capability.

Always interesting and challenging to work on new projects.:)
 
The hardware is already in place on present EM-5 ECUs to implement closed loop control so only a reflash would be involved to have this capability.

Always interesting and challenging to work on new projects.:)

Thanks Ross... I look forward to receiving my EM-5 system from you shortly.

I am not the "test partner". Mine will be open loop at this time.
 
Last edited:
If this system uses an O2 sensor, are you limited to no lead only fuel?

We probably won't release this software until there is a more widespread supply of unleaded avgas. It takes about 10 seconds to turn off the closed loop function in the programmer and revert to the normal open loop mode.

Most people are already using a wideband O2 sensor to monitor AFRs with their EFI systems. We've seen sensor life vary from about 10 hours to over 300 on 100LL so far to date.
 
Last edited:
Looking forward to seeing this develop. Presumably the impact of leaded Avagas on o2 sensors make closed loop impractical in today's market?
 
I have played with turning closed loop on in my aircraft now below 18" and 1100-2100 RPM. Is there an ultimate limit of allowable % change from closed loop?

My tune is close, but having closed loop to clean up small changes is nice.
Only reason I cant use it below 1100 is my lycoming requires a 12-13:1 mixture to idle nicely.

Caleb
 
Ross,

For those with split exhaust systems, as in no crossover, do you still plan to use a single sensor, and assume the other side is running the same mixture?

Or, is there a way to incorporate the output from a sensor on each side of the system?

Randall
 
I have played with turning closed loop on in my aircraft now below 18" and 1100-2100 RPM. Is there an ultimate limit of allowable % change from closed loop?

My tune is close, but having closed loop to clean up small changes is nice.
Only reason I cant use it below 1100 is my lycoming requires a 12-13:1 mixture to idle nicely.

Caleb

The present closed loop software targets stoichiometric (peak EGT)and can trim mixture +/- 25% maximum.

Once we introduce the aircraft closed loop software, you'll be able to target several different load ranges at different AFRs.
 
Ross,

For those with split exhaust systems, as in no crossover, do you still plan to use a single sensor, and assume the other side is running the same mixture?

Or, is there a way to incorporate the output from a sensor on each side of the system?

Randall

Since we can trim all cylinders, having a single O2 sensor on one or two primary tubes should work fine. You'll still be able to balance all cylinders with the trim function.
 
Ross,

Is O2 sensor lead poisoning cumulative, or is it somewhat self cleaning if you predominately run unleaded? I hope to run regular old auto gas in the Rocket locally, but anticipate a primary diet of 100LL going cross country. You say that you have seen up to 300 hours with 100LL... Wondering if my situation will extend that time any noteworthy amount.

Also, is the closed loop feature selectable on the fly? Meaning- I have an established open loop tune loaded as a fallback in case the O2 sensor dies while Im in the middle of nowhere?
 
Ross,

Is O2 sensor lead poisoning cumulative, or is it somewhat self cleaning if you predominately run unleaded? I hope to run regular old auto gas in the Rocket locally, but anticipate a primary diet of 100LL going cross country. You say that you have seen up to 300 hours with 100LL... Wondering if my situation will extend that time any noteworthy amount.

Also, is the closed loop feature selectable on the fly? Meaning- I have an established open loop tune loaded as a fallback in case the O2 sensor dies while Im in the middle of nowhere?

I am wondering the exact same thing...
 
Ross,

Is O2 sensor lead poisoning cumulative, or is it somewhat self cleaning if you predominately run unleaded? I hope to run regular old auto gas in the Rocket locally, but anticipate a primary diet of 100LL going cross country. You say that you have seen up to 300 hours with 100LL... Wondering if my situation will extend that time any noteworthy amount.

Also, is the closed loop feature selectable on the fly? Meaning- I have an established open loop tune loaded as a fallback in case the O2 sensor dies while Im in the middle of nowhere?

We've seen O2 sensor life from 10 hours to 300 hrs. on the Bosch 4.9 wideband sensors when using 100LL. 10 isn't acceptable, 100 + maybe is.

We're having some new sensor bosses made which are longer, so that most of the sensor element is shielded from direct impact of the lead. Other people have noted that this tripled their sensor life on leaded fuels. We're not sure yet how this impacts response time but we'll be testing it this month on our O-200.

Closed loop is easily selectable in a window called Closed Loop, toggle on/off with the +1 button when in the window.

Just be aware right now that the software in place only targets stoich- about 15.2 AFR on 100LL. Not the place to run the engine since this corresponds to peak EGT.

In 2007 I tested a retargeted value of 13.0 on an RV10 in Florida and it worked fantastically. Move the throttle anywhere and within less than .75 second, the AFR went right back to 13.

As mentioned previously, we plan to expand the capability so that the user can define what AFR they wish to target at 3-4 manifold pressure ranges and possibly a couple rpm ranges. Right now, you define closed loop operation with a high and low rpm range and a high and low MAP range. Whenever closed loop is on and you are inside those 4 limits, the ECU automatically jumps into closed loop. It immediately jumps out if any of the 4 limits are exceeded.

We also need to have diagnostics to detect an invalid sensor output, jump the ECU back into open loop when that happens and flash a warning on the programmer letting you know. That code is likely to take much longer to develop and validate than the actual closed loop control which is just a variation on what we've been doing for 20 years already in the automotive field.

This is all planned but we have 4 other development projects to finish off first. I'd hope that we can have beta versions to test sometime in early 2018. Older EM-5 ECUs will be able to be reflashed to have this feature.
 
This discussion seems to have prompted some European builders to ask about the future closed loop capabilities. Unleaded avgas, in various forms, has been available in certain countries there for over 20 years and they are keen to be able to use closed loop EFI systems.

In light of this, we will try to accelerate the development of this aspect for our aviation EFI systems.

Some interesting links:

http://www.total.co.uk/en/aviation/generalaviation/wheretobuy.html

http://www.hjelmco.com/pages.asp?r_id=14663

https://generalaviationnews.com/201...nleaded-avgas-approvals-revealed-by-lycoming/
 
Would it be possible too run 2 wide band sensors, and when a certain discrepancy exists between the two trigger a disconnect of the closed loop?
 
Would it be possible too run 2 wide band sensors, and when a certain discrepancy exists between the two trigger a disconnect of the closed loop?

This is what the auto OEMs do and if we were doing a clean sheet ECU design, we might adopt that strategy. With only one O2 input though, it's probably easier to monitor MAP, rpm and TP and jump it out if closed loop trim exceeds 5% where none of these parameters change more than 2% within a certain time span.

We'll have to test the strategy on a running engine of course
 
This is what the auto OEMs do and if we were doing a clean sheet ECU design, we might adopt that strategy. With only one O2 input though, it's probably easier to monitor MAP, rpm and TP and jump it out if closed loop trim exceeds 5% where none of these parameters change more than 2% within a certain time span.

We'll have to test the strategy on a running engine of course

Sounds like a good tactic to me. If you are implying that you need a volunteer, I'm your man!

Caleb
 
Thanks Caleb. When we get this done, we'll reflash your ECU and get your feedback. Appreciate the offer.
 
Back
Top