I've also noticed, like Cat, that if I completely seal the air bleed valve, some fuel vapors rise in the hose and tend to severely perturb engine operation at idle.

Nope, won't work... In theory, it should, but in reality, it is two throttlevalves in series, where the servo controlled valve will ONLY have a noticable effect when its passage approaches that of the main needle, leading to a very abrupt control and a very narrow usable range.

The trick with the modulating solenoid, is that it is an on/off switch, basically, and the actual metering is still done with the manual needle valve itself. The solenoid opening and closing imposes a "dutycycle" on the flow through that main needle, allowing for a very linear and precise control of the actually metered mount of fuel.

Say your fuel tube allows for a full flow of 100 ml per minute, and you trim the main needle back to the 10 ml/min the engine actually needs at full throttle.
A servo-operated throttlevalve in the fuel line will only have a noticable effect on the flow if its passage closes all the way from full flow down to 10 ml/min or less. Meaning a lot of the actual throw is not effective, and the actual control all needs to be done in that last 10% of servo travel. The slightest amount of friction or play in the mechanical linkage will destry any remaining accuracy and repeatability.

The solenoid allows you to set the main needle at 11 ml/min (a touch rich, providing some headroom for the atmospheric compensation) and a 91% dutycycle will then interrupt that flow of 11 ml/min to the exact value needed. Meaning you still have approx 90% of the full linear range availlable to control down when the throttle is reduced.

A fairly decent indication on whether the engine can run properly on its original not modified carb, is to lean the engine to its most optimal setting, and check (with a piece of cloth or filterpaper behind the exhaust) how "dirty" the exhaust residue is. For this you need a stable setting and test after at least 30 seconds of running at that constant RPM.

Then reduce the throttle to wherever you think mixture is worst, leave the engine running for a short while at that setting. Check the exhaust residue again, and check for differences in sootiness of the residue,
Then at that same throttle setting, check how far you can close the main needle and whether running improves.
That will give a rough indication of behaviour in the air.

But I am not very optimistic...


felt pretty good on the ground, but turned out to be barely flyable in the air.

Got the system in a plane so we are closer. I’m 4 for 4 on system repeatability. Same results with a stock ST 2300. I did make sure that I have a full 3 mm minimum round cross section for the whole air track and feel I could use a bit more as the “low-mid setting” is basically wide open. But it runs very well and does not run well if I close the valve off (or open it up). Might get a chance to get in the air this weekend to see if this effort will pan out.


a few pics of the quick setup. Valve is behind firewall and injector is again in the post throttle space.

Cool! Sounds good. I'll be following closely.

Ok I understand what you mean Bert.
So on the fuel line, apart completely revised the carb design, the only option is a solenoid to impose a duty cycle to the fuel flow.

I will do the test.
I noticed this FS61 has a very clean running compared to others engine (I have Rossi 61 2 stroke, and even if there is good engine setting, it's like a coal fired locomotive 🤮 )

"completely revised" is a bit too strong a wording. It IS possible, with some careful grinding, to modify the linearity to such an extent that the result will be pretty usable.



That engine of the Calmato a few posts earlier? This is basically the ONLY real modification I did to that engine. As a result I had to double the spring under the throttle barrel to push out the increased lateral clearance, but this worked very well, and that engine to this day is in my posession and flies as good as it did back then. That is now almost 10 years ago (the pic dates 15 June 2016) and the engine behaviour was VERY acceptable and constant, both on the ground as well as in the air.
It did allready have a servo-slowdown function applied, and the remote fuel needle was heat-decoupled from the engine.

No electronics and just a fully "accoustic" carb.

The grinding of the carb is relatively easy and initially a matter of taking off a little bit, and retesting, Once you have done two or three, I could do them basically freehand in one go.
Of importance is, that the lower part of that groove (which is the part that handles 1/2 to full throttle) remains UNchanged, and the lower half, which handles the 0 to 1/2 range, needs to be gradually rounded.


If you look in my YT account, any gasoline conversion vid that predates the Spacewalker vid I posted a few days ago, by definition was modified the above way, and most of them (about 30 of them even if not every one is on my channel) was modified this way (except for the SuperTigres, which offer sufficient carb adjustability to allow direct conversion) and virtually all of them still run like that to date.

Its interesting that the mechanical mod on a Carb groove end up looking like that, where as the "air injection mod" seems so far to favour the low and high ends untouched and mid range being the area focused on.

Not really when you think about it...
If you consider an unmodified twin needle carb, calibrated for glow but used for gasoline, both needles can be set accurately for idle mixture and full throttle mixture, but when opening the throttle coming off idle, the idle needle lifts off it's seat at a too rapid rate. In original design, that rate of lift is constant (the slanted groove is a straight line).
Now regardless of whether the carb is modified "airbleed-style" or "grindstone style", in both situations, at idle, the idleneedle is in the same position relative to the spraybar: the position needed to pass the required amount of fuel for idle RPM.

My mathematical theory is very rusty, and my algebraic skills are even barely existent, but what IS interesting, is that the top of the curve you found empirically, more or less coincides with the inflection point (where the ground part of the groove transitions into the straight original shape of the groove), basically eactly how the top of a mathematical function coincides with the inflection point of its integral.

I have noticed this correlation of "derivative" and "integral" quite often between process control values, and control element values.

Got out for a quick test today. Forgot my sunglasses and had to fight of snow blindness.

I can report success. The ST 2300 flew as good as it ever has on any other type of system. Still some tuning to be done but midrange and throttle response were good for a quick tune. I did have a dead stick at the end of this video. The HS needle turned out to be a couple clicks lean and I failed to do a proper “nose up” before flying. The landing hurt a stay cable attach ring on the skis so we were done for the day ,but I got enough flying in to determine that this system is worth perusing and testing further.


FYI. Under the snow is our “new” flying site. Grass strip went in last fall and will be relocating here next year. I’m in heaven as it’s less than 5 minutes from my house!!!

It appears the YT link is not working? Didn't open, copied the link from message to separate tab and gave error message "page not availlable" (while YT in itself is responding normally on the connection I am currently on)

Edit: don't know what changed but the video is playing... COOL!

That link does look a bit off - this one captured my normal way...

Yes this does make sense after your explanation and my thinking on it a bit more... Thanks Bert.

Howdy Gents from Houston Texas,

I have looked over your projects with quite a bit of interest. I enjoy the hobbies of electronics, engines, and rc aircraft. I feel as though I have the skill set to complete and fly your system. As I read over the RCU, RCG, and Github I find myself having a
few questions as the design seems to be in evolution and the GitHub info contains separate paths when it comes to microcontrollers.

My Equipment:
1. I have a Spektrum 8 Gen 2 with 7 point throttle mix and can handle other programmable mixes as well.
2. My Saito 72 which is currently running on glow ignition is my primary target but I also have a Super Tigre 45, GMS2000 76 ring, Thunder Tiger F-91S, and a Magnum AR180 all of which hold futures in gasoline hopefully.
3. I own a 3D Printer
4. I'm already familiar and invested in Adafruit products.

My Questions:
1. The M-Tronic solenoid has different part numbers depending on where I read the information (RCG & GitHub). Which one do I need?
2. I would like to purchase 3 of the housings that fit the solenoid. Which one of you guys is the man for that?
3. Which microcontroller is currently recommended of the 2 options?

Moving Ahead:

I'm going to guess that based on my not very demanding single cylinder engine range that the option for me is the BME280 compensated solenoid. Unless I'm understanding it incorrectly the air pump design wont be required in my case.
Looking forward to everyone's input.

Hey Texas!!

I will Jump in an answer a few of your questions but I know the others will add to the chorus..

Not familiar with the 8 gen 2 - I have an NX8 that I just checked but I haven't used that radio for my "gas conversion" stuff yet. Many ways to skin this cat but my way is a separate channel (6/7/8 etc) that uses the throttle stick as its input. Then the output from this channel is "curved" to meet the needs of the fuel adjustment output. The Spectrum 8's don't seem to have a way to do this so you will have to use a "mix" (throttle to mixture channel) and set the curve of the Mix. You will be limited to 7 points it seems.

I will have to look into the solenoid part numbers as I can't recall the best one or the one currently available.

I'm currently working through a "much simplified" solution to the "glow to gas" situation - info in the last few posts - It is simply a manifold "air bleed" that is able to lean the off idle to mid range part of the curve that is usually troublesome when trying to get a glow carb to work on gas. It eliminates the Microprocessor and the solenoid and replaces it with a simple air control valve (servo) that plugs directly into the receiver. I have had great success but still in the testing stage. The valve can be reliably 3D printed as the control medium is air...

I hope this get you a few answers - will look up and confirm a few of the answers..

The housings were available from Bert (he was recently out but not sure of the status) but I'm working on a CNC'd solution here too. Just need a but more time to perfect the file to get a good fit for the solenoid. I have tried a printed version but have really struggled to get a unit to seal well and print reliably. I have given up and moved to a machined version.

I have only built with the Xiao version.

Part number on the solenoid I prefer is 00001205104 (white one) I have used both this one and the green version and both work just fine, but I like the sealing method of the white one better.

At the moment I am indeed more or less out of housings, with no real insight in whether I will get them again.
I do not make money on this, and the design is out there for anyone to copy, but if demand justifies, I might have another run made.
The hurdle for me is that I have to take a minimum order of about 50 pieces, and pay in advance.

For what it is worth: that SuperTigre of yours, its carb should allow for satisfactory adjustments on gasoline, no mods needed.

I can still get the drivers relatively easily (and in smaller batches) so that won't really be a problem, but shipping to the US might be a bit prohibitive WRT cost.

Ok gents.

Thank you both for the information.

I've dug an Adafruit Feather ESP32-S3 with TFT screen out my pile of junk as well as an I2C BME280. In further review I've also noticed the Adafruit Feather comes with a provision
to install a SMT BME280. This would yield an all in one compact package of Controller/Display/BME280.

Adafruit Feather w/ onboard TFT

Onboard provision for BME280

I'm far from a Arduino MPU expert but I do know the code that is run has a "timing sequence" that is "ported" for the Xiao board in the original sketches. I remember trying to load it on another board and didn't have luck. Probably not an issue for someone with some MPU skills but this was above my "pay grade".

I will also see if I can get set up to get a few housings made - The set up is 90% of the work and once the setup is "calibrated" I should be able to punch out a few for "inventory". I am north of the 49th - so shipping to the US might have some extras added as it crosses - not sure.

I’m surely no expert of anything that’s for sure!!
While the BME280 provision straight on the Feather board is very interesting I’ll leave that bit for later if this works out in the end. In the mean time I wired up a separate break out with the simple i2c stems QT connectors. The serial interface is giving out the information as intended.

Howdy,

I've had a little more time to read through some of the threads and have a better picture now.

I was wondering your thoughts of a hybrid solution between air bleed and weather condition compensated solenoid.

The short idea is a high speed continuous rotation servo that contains a feedback line would operate a standard remote needle valve to do the air injection handling but
in overview the same idea could be applied to the solenoid solution and perform fuel limiting. This would eliminate the complexities of building a solenoid housing and eliminate
the solenoid driver. The servo would be controlled with the Feather TFT to allow for weather compensation.