Raleighcopter

i spent some time cleaning up the new s.bus controller code and making sure it reads channel pulses accurately. it does and can be used at this point. the benefit of switching to s.bus is that reading channels no longer requires an interrupt routine. i can now use the interrupt routine to read the RPM pulses without concern of an interrupt happening while insude another interrupt routine (which would prevent the interrupt from being recognized and the rpm pulse would be lost).

the s.bus signal needs to be inverted before going to pin 7 of the seeeduino xiao and that can be easily accomplished using a transistor and 2 resistors or, even better, one channel of a uln2003a and a single resistor. i'm leaning towards the uln2003a solution because it can also be used to buffer the solenoid signal without any added parts which reduces the total part count bcause it eliminates the mosfet and resistor we currently use. it also reduces cost as the uln2003 is about a quarter the price of the mosfet we've been using.

i'll be adding RPM sensing to the controller and using another channel of the uln3002a plus a resistor for that. i also plan to add the air pump control (using yet another channel of the uln2003).

'll use 2 channels of the uln2003a to control a high current power supply (such as to power the ignition which would eliminate the cost of the opti-kill switch and it's required channel input). i plan to thoroughally test this to make sure there's no interference from the ignition but we already connect the RPM signal to our telemetry devices/radio without any isolation when we use openxsensor devices without issue.

there's one more channel on the uln2003a available for future use and i'll not connect it at this point.

Rcplanedan

I would choose a decathlon, to do such a conversion would involve shortening the crank shaft of the rear engine to reduce total length..and yes the over slung prop shaft would add a little extra length over a pegusus..i know maybe i shouldn't take on such a modification, but cant help but wonder

Raleighcopter

The code is updated and can now read multiple channels over s.bus and control multiple outputs. There's the solenoid pulse plus 3 more digital/pwm pins (currently 8, 9, and 10). I've breadboarded and tested the circuit...

and now it's time to build a more permanent version. I think this one goes together easier than the prior version but I need to build a few first. The code has spaces for defining how to control those 3 new pins plus rpm measurement coming soon.

Rcplanedan

My crankcase vent is rerouted via a T-connector to both rocker covers. No more valve lash adjustments
Where does the engine now breathe to?

1967brutus

You need drains on the rocker covers... The oil has to go somewhere...

The pic is a bit fuzzy but the drains can be clearly seen. It makes no noticable difference whether you have the oil supply aimed at the exhaust valve (forward connection) or intake valve (rear connection),, just do wat makes for the visually neatest solution.

You can leave the drains open and a little bit of oil will soil the sides of your plane, but there is absolutely no issue with rerouting those drains back to the middle exitting at the centre of the firewall, or from each cylinder to a main gear leg if you want to keep the plane oil free.
The only criteria is to use equal; lenghts of tubing so the flow will be as equal as possible. The backpressure does not create any problems.

It is advisable to cut cardboard gaskets for the rockercovers and wrap a bit of teflon tape around the heads of the rocker cover bolts to keep the cylinder heads clean. No biggie, that's a one time affair.
Allow for an approximate 10~20 hours to let the valve lash stabilize, depending on state of break-in.
Mine from new required a 5 hourly check/adjustment up till about 20 hours, then valve lash simply stopped changing, and I basically kept checking every 5 hours until 40, then I decided enough is enough. I check when I change the plugs, which is about every 35~50 hours.
Can't judge your flying intensity, but for me, 25 hours is about a year of use (I have progged a runhour counter that ONLY counts ACTUAL running time based on "ignition ON+RPM<500". So prep-time with RX on, or even ignition on, or anything where the engine is NOT actually running, is not counted)

Rcplanedan

Thanks..well i figured thats what you have done, , great idea to utilise the waste from the crankcase, well i have put a good 4hrs on this plane and only one exhaust valve needed rocker clearance set, and done another 30mins this morning, keeps running without fault basically..except the single exhaust pressure nipple i found blocked up at 3.5hrs. But this time i have pressure off both pipes..i found this problem when refuelling, resistance when filling, rerouted the tank lines as per your instructions, im happy with all that now..instead of making rocker gaskets i used automotive silicone and let it set on the rocker covers..that works fine..

1967brutus

Yup, the silicone also works fine.
I am a bit oldfashioned in that respect, I decided for gaskets

I have done this rocker lube mod on ALL my fourstrokes except for the radial (due to "logistical reasons", the tubing and the baffling result in a nightmare) and I am extremely happy with the results: soaking wet and brand sparkling clean rocker box and once the running surface of the cams has settled (that is where the initial change in valve lash comes from) no valve adjustments anymore.

Rcplanedan

absolutely a brilliant idea. And will be followed on with the top cylinders on my 3 cylinder radials, .
i could surely imagine the mess of pipe work on the asp 5 cylinder..damn catastrophe

1967brutus

One of the reasons I did not incorporate it on the radial, was that the radials have a (near) constant crankcase volume. Singles and boxers have a strong pulsating crankvent, meaning that the oil gets "atomized" in the rockerbox(es).
The radial has a steady flow, so I expect issues with too much standing liquid level in the rockerboxes. It is no biggie, the radial makes very few running hours, so....

John_M_

Well redirecting the oil vapor from the crankcase through the rocker boxes is a damned sight better way than oil migration through the valve guides, and what gets past the cam followers and up the push rod tubes, it works to some extent... rather that, than the os pcv system, or the Moki radial's with their greased cam followers, which is an asinine way to lubricate a valve train.

Raleighcopter

How's this sound? Read rpm, 2 engine temp, controls the solenoid and an air pump (for whatever reason you want) s.port telemetry for everything and we'll figure out an air pump algorithm at some point or repurpose the control point for a future use. I could also make it take 5 engine temperature measurements if I try.

This makes it a single device for controlling our engines and keeping track of how they are running through telemetry and it's own integral display. Depending on final dimensions I may use a 64 row display but only if it doesn't make the package bigger.

Glowgeek

Sounds great to me, Dave.

Raleighcopter

I still need to attach the wires for the sensor and display and then finish programming it. It's almost built but not as easy to build as I had hoped. I'm not sure it's worth building unless we have circuit boards made.

Raleighcopter

...but what I am sure of is that an s.bus capable solenoid controller is worth it. Switching to s.bus requires an inverter in order to use the serial port on pin 7. There's a software serial library that may let me receive inverted serial on pin 2 which is the same pin we currently use to read the rc channel. If it works we could flash new software to our Xiao controllers and make them work with s.bus. also, I've figured out how to calculate the required additional fuel for acceleration from one speed to another so a new firmware with acceleration and deceleration pumps is hopefully on the way.

Raleighcopter

Needs a case and some software. I'll need to work harder to get softserial to work on the original controller because it doesn't play nice with the sbus library I'm using.

Glowgeek

Looking good, Dave. A lot of power in a small package. I wouldn't mind investing in a PCB if it helps clean things up and simplifies assembly.

Raleighcopter

I might do that, order some circuit boards.

The controller has minimal parts: 4 resistors, 2 diodes, 1 uln2003a and a metric crapload of wires.

I worked on software some last night. None of the thermistor libraries I found work with 2 thermistors which means I'll probably need to write my own code. Rpm should be pretty easy, count 4 or 5 revolutions and convert the elapsed time to RPM.

Controlling the air pump will definitely be pretty easy and I already can turn it on and off from my transmitter.

Adding an acceleration pump is a bit more complicated. Since we tune our mixture curve at static rpm, the extra fuel needed to accelerate the propeller to a new rpm is the increase in rotational energy: I x omega squared. Omega is the angular velocity and I is the moment of inertia. If we know the specific rpm vs throttle value we can create a rpm/throttle position table and use that to calculate the rotational energy going from one throttle position to another throttle position. Since we know the throttle position (since it comes on the sbus data stream, we know how much extra fuel to include. The only unknown here is the moment of inertia but we can just use a channel to provide that constant and tune it for the snappiest acceleration.

Raleighcopter

Pcb way says I can get 100 1"x2" boards made for $50. I guess this is happening. Let me get the prototype working completely before I design and order boards.

Glowgeek

Are you planning to standardize on JR/Futaba connectors for I/O?

Raleighcopter

I would use pin headers for all the wire connections to the controller. I might make the bmp280 integral and mount the controller on the firewall with an optional remote display. This would require 4 conductors back to the receiver and a plug for 4 conductors to the optional display. It would be a 2x8 or 3x6 pin header, probably.

Glowgeek

3x6 90 degree pin header sounds perfect. Most everyone has, or can borrow, a crimp tool for rc connectors.

I won't use the display if telemetry is available. If I did use one it would be a temporary thing so yes, remote connection is my preference. Components on the PCB would simplify wiring in a big way.

Raleighcopter

So if we move the controller to the firewall, a remote display makes sense. just 4 wires go back to the receiver, everything else is under the cowl except possibly a display mounted in the airplane's dash board. I may want to move the air pump to pin 2 as i can do some more timer stuff and future proof a custom board. The rpm input would likely move to pin 1. If we use a remote display the controller winds up about 1"x1.5"x1/2" overall size.

If anyone has any other ideas they think we should implement, now it the time to bring it up.

Glowgeek

Does the new design include the voltage dropping diode for use with Life batts? Was thinking 2 batt inputs would be nice, one for 4.8v-6.0v and one for 6.6v-?v.

Raleighcopter

If I build a board, it probably includes a voltage regulator

Glowgeek

Even better. A regulator would allow for 4.8v-8.2v?