What's new
Van's Air Force

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

Variable ignition timing (100LL no longer available at RHV).

SDS to the rescue

Go with SDS fuel and ignition, EM-5F system, and your worries are over.
I have an 0-360 8.5 compression ratio and I run Costco premium.
Yes I have the 6 port Andair valve for return lines to tanks and all SS/teflon lines.
Starts same cold or hot.
Problem solved.
My three cents worth Art
 
Dynamic spark advance.

By ignition timing, I mean advancing or retarding the spark to compensate for lower octane fuel(s).
 
Last edited:
If you mean flex fuel type capability (to adjust automatically based on onset of detonation) I am not sure anyone has that capability for air cooled engines. First of all the heads are too hot to mount regular knock sensors (designed for water cooled engines). Secondly I doubt that suitable locations for the knock sensors would be available.

I think any electronic ignition system would be set up for a particular fuel and thats it. For tuning, sensors and signal conditioning systems are available to be able to detect detonation very early (sending a signal out to a pair of headphones) with great sensitivity and the timing tables would then be adjusted accordingly and then locked.
 
With SDS CPI, CPI-2 or EM-5 systems, ignition timing is user programmable.

When burning 91 Mogas and using std compression ratios we normally retard timing at high MAP around 3-5 degrees so total timing might be around 20 degrees at 30 inches and then as MAP drops, we slowly bring more timing in so the full 25 deg is in at 25 inches and below. This is all automatic once programmed.

There will be a slight drop in takeoff power (around 6-7 hp on a 360) but cruise power and economy are unaffected above about 5000 MSL.
 
Last edited:
I'd like to do some testing with various fuels in the 7, but there was another post mentioning it's almost impossible to find a knock sensor that works in lycomings. Anyone have any info on why knock sensors don't work, or suggestions on how to make them work?
 
Knock frequency varies between engines so a sensor optimized for one engine may not be ideal on a different engine.

Std Lyc heads don't have good places to mount knock sensors as they were never designed for them. Placement is important for useful knock detection.

You have to make the engine knock first to detect it. How many times do you want to subject your engine to this? Light knock for short periods probably isn't damaging but continuous medium knock is.

Lycs have individual cylinders so you may need individual sensors to properly detect knock.

Lycs have a lot of mechanical noise which must be filtered out so it's not falsely interpreted as knock. You need a way to filter out the unwanted false frequencies and just display when actual knock is occurring. Can and has been done on air cooled Corvair and Porsche engines with integrated heads at least. These are also mechanically noisy engines.

Lycoming's IE2 engine has different cylinders and multiple knock sensors so it can be done with a bunch of money and engineering thrown at it. Standard cylinders may work with enough experimentation and I read an account online from someone who'd done this on a big Conti or Lyc a few years back. They believed they were able to detect knock in flight.

You then need a programmable EI to effectively map around knock and probably full EFI/EI to optimize as mixture also makes a big difference in knock propensity.

If you decide to experiment, please let us know what you discover. It's an interesting area of study.
 
Last edited:
I'd like to do some testing with various fuels in the 7, but there was another post mentioning it's almost impossible to find a knock sensor that works in lycomings. Anyone have any info on why knock sensors don't work, or suggestions on how to make them work?

I am not convinced that a lyc is any noisier than any other engine and the volume of noise has no bearing on electronically observing knocks. In most auto engines, they use a piezo electric sensor to listen for a specific frequency and that sensor is mounted in a strategic position on the engine block. Each sensor is built to only listen for that specific frequency, though it can be done with broader sensors if you know the frequency that is present during detonation. In a lycoming the cylinders and heads are much thinner with very little contact with the block, meaning that it is likely challenging to get a good resonance in the block, possibly requiring a sensor for each cylinder, with few good places to mount them. Auto egnines have everything cast into a block and heads. Therefore all noises can be observed resonating in that block. Not so much with a boxer engine. Also, engineers spend weeks on the dyno with forced detonation, figuring out which frequency to listen for and which spot on the block will best resonate that frequency during detonation. To my knowledge, no one outside of Ada, OK has done that for a Lyc and I don't think they are sharing. I am sure there are others who do this, but don't know. Clearly it is done, as Dan has posted several dyno reports that show detonation as part of the plotted data. These are likely engine lab type settings with instrumentation well outside the reach of guys like us.

That said, it can be done. I installed a custom EFII system on an older, 80's vintage porsche 911, which is an air cooled boxer engine very similar to the Lyc. While I never was able to rig the knock detection into the system (never attempted due to the complexity), I was able to rig a system to hear the knock frequency through headphones and used that when tuning.

Larry
 
Last edited:
To the OP

According to Santa Clara County:
The unleaded fuel available at Reid-Hillview and San Martin is 94 octane and is FAA-certified as safe for use in approximately 68% of the piston-powered aircraft of the type that use the Reid-Hillview and San Martin airports. Larger jet aircraft, such as those operating out of Mineta San José International Airport, use a Kerosene-based fuel that does not contain lead.

With 94UL available at your home field, why don't you just determine if you fall within the 68% of (FAA certified as safe) and use it before redesigning your ignition system?
 
Most of you guys are focussing on microphone type knock sensors, and I agree those come with challenges. Not knowing which frequency to listen for can be overcome by using a large range sensor and real-time FFT data analysis. While it varies a little bit based on combustion chamber design, I remember my internal combustion textbook saying its focussed around 3,500Hz for automotive engines. If I run the motor at 2100 rpm, that should give me a mostly noise free zone between 2100 and 4200hz, until the 2/rev vibrations come back into play. As far as mounting the sensor I see 3 options, the primer port which is unused with an SDS system, the CHT port, or one of the 1/4-20 baffle bolts that go through the fins in the head. I would start with a leak-down test to see which cylinder has the best compression, and lean it out a few percent over the other 3 to make it knock first. That would mean I would only have to focus on that 1 cylinder instead of all 4. Using the LOP button with ignition retard instead of advance would allow me to pull massive amounts of timing almost instantly when I do encounter knock.

I think the easiest way to do this would be to remove a spark plug and mount a pressure transducer in its place. We work with this company for a few of our ground test stands, so I might be able to work a deal on a high-temp, 5000psi pressure transducer and amplifier. (https://www.kistler.com/en/solution...n-analysis-and-cylinder-pressure-measurement/) No noise to worry about and I could plot the trace on a battery powered o-scope in the cockpit. The downsides to this approach would be if I did end up losing the other ignition, I would only have 3 cylinders to fall back on. I would be ok with this as I would do all these tests in gliding distance to the runway, so my only massive risk would be losing the other ignition between 1 and 1000 ft. The biggest issues would be 1 vs 2 plug detonation variables.

This would be a ton easier on a dyno stand, but I think I can make it work on the airplane. I'll think about it more after Oshkosk. If the pressure transducers are out of reach price-wise, I can always hook up a good old stethoscope style system and listen to the cylinder through headphones.
 
Last edited:
Most of you guys are focussing on microphone type knock sensors, and I agree those come with challenges. Not knowing which frequency to listen for can be overcome by using a large range sensor and real-time FFT data analysis. While it varies a little bit based on combustion chamber design, I remember my internal combustion textbook saying its focussed around 3,500Hz for automotive engines. If I run the motor at 2100 rpm, that should give me a mostly noise free zone between 2100 and 4200hz, until the 2/rev vibrations come back into play. As far as mounting the sensor I see 3 options, the primer port which is unused with an SDS system, the CHT port, or one of the 1/4-20 baffle bolts that go through the fins in the head. I would start with a leak-down test to see which cylinder has the best compression, and lean it out a few percent over the other 3 to make it knock first. That would mean I would only have to focus on that 1 cylinder instead of all 4. Using the LOP button with ignition retard instead of advance would allow me to pull massive amounts of timing almost instantly when I do encounter knock.

I think the easiest way to do this would be to remove a spark plug and mount a pressure transducer in its place. We work with this company for a few of our ground test stands, so I might be able to work a deal on a high-temp, 5000psi pressure transducer and amplifier. (https://www.kistler.com/en/solution...n-analysis-and-cylinder-pressure-measurement/) No noise to worry about and I could plot the trace on a battery powered o-scope in the cockpit. The downsides to this approach would be if I did end up losing the other ignition, I would only have 3 cylinders to fall back on. I would be ok with this as I would do all these tests in gliding distance to the runway, so my only massive risk would be losing the other ignition between 1 and 1000 ft. The biggest issues would be 1 vs 2 plug detonation variables.

This would be a ton easier on a dyno stand, but I think I can make it work on the airplane. I'll think about it more after Oshkosk. If the pressure transducers are out of reach price-wise, I can always hook up a good old stethoscope style system and listen to the cylinder through headphones.

The problem with home brew testing of this sort is getting all the parameters in place. With a permanent knock sensor tied to EFII all is good. If you are testing to determine an advance threshhold that is safe, that is hard. Detonation risk increases with heat in several areas. Getting OT, IAT and CHT into the high ranges will be problematic on the ramp, even orbitting the field.

Larry
 
We also need to understand that knock sensors in cars are required because they are trying to optimize timing and mixture to a very, very tight standard to meet fleet emissions standards. Our standard is much easier - do not knock. I have shown that the typical Lycoming running typical take off power mixture is very insensitive to timing retard. Put another way, you can back timing off far enough that you will never have to worry about detonation and retain essentially all of your take off performance - and you can do this open loop with a simple lookup table. This capability is available right now, today. Just give SDS a call.

OTOH, if you are really looking for that absolute corner case where you want to ride right on the edge of detonation reliably with a closed loop system, then please let us know what you find out. We will all benefit.
 
We also need to understand that knock sensors in cars are required because they are trying to optimize timing and mixture to a very, very tight standard to meet fleet emissions standards. Our standard is much easier - do not knock. I have shown that the typical Lycoming running typical take off power mixture is very insensitive to timing retard. Put another way, you can back timing off far enough that you will never have to worry about detonation and retain essentially all of your take off performance - and you can do this open loop with a simple lookup table. This capability is available right now, today. Just give SDS a call.

OTOH, if you are really looking for that absolute corner case where you want to ride right on the edge of detonation reliably with a closed loop system, then please let us know what you find out. We will all benefit.

My main goal is to find out how 94UL affects detonation margin on an engine with 9:1 CR. Lycoming has only approved 8.5 or 8.7 CR engines for use on the new unleaded. I don't want to push the limits and run it on the edge. If I run back to back tests with 100LL and 94UL and keep advancing the timing until I hit initial detonation on each, I'll know how much extra timing to pull when running 94UL to have the same amount of margin. If 100LL starts to knock during the takeoff roll at 28 degrees advance, and 94UL starts to knock at 26, then I know I have to pull 2 degrees of timing out whenever I fill up with 94UL.

Larry, as far as getting parameters correct, I don't see it being that difficult. Fill one tank with 94 and the other with 100LL. Spend 30 minutes in the pattern to get everything at operating temps and do back to back takeoffs with the 2 fuels. I'm not trying to determine absolute knock margin, which would require having all the temps be super high. I just want to know the relative difference between the 2, which means the parameters only have to be the similar.
 
Detonation on takeoff is a "thing" when you have fixed timing with magnetos. Remember that with fixed timing you are trying to satisfy two masters - a large enough advance to give reasonable high altitude and lean performance, but not so much advance that you detonate at take off power. It just so happens that those two requirements are in almost perfect opposition. So in the certified world Lycoming has to play a very careful dance. With variable timing we can completely eliminate this sometimes razor edge compromise.

You dont need to find out what the margin is at high advance at takeoff because you should not be at high advance at take off. Thats a problem for the magneto guys.
 
Larry, as far as getting parameters correct, I don't see it being that difficult. Fill one tank with 94 and the other with 100LL. Spend 30 minutes in the pattern to get everything at operating temps and do back to back takeoffs with the 2 fuels. I'm not trying to determine absolute knock margin, which would require having all the temps be super high. I just want to know the relative difference between the 2, which means the parameters only have to be the similar.

That assumes that heat increases detonation at a comparable rate for both fuels. Not a combustion engineer, so cant answer that. I, however, would not assume they are the same.

I agree with Mike. Just get a variable ignition system and drop the high MAP timing levels. I only run 21* at 29". Don't believe I am losing anything over 25*, as it is not optimum for ROP WOT; It is a compromise for fixed ignition timing. I run close to 33* at LOP cruise in the 540.

Larry
 
Last edited:
My main goal is to find out how 94UL affects detonation margin on an engine....

....If 100LL starts to knock during the takeoff roll at 28 degrees advance, and 94UL starts to knock at 26, then I know I have to pull 2 degrees of timing out whenever I fill up with 94UL....

Just re read this part. Why on earth would you want timing set anywhere near this on takeoff? My 8.5 engine is set to 17 degrees on TO. It's a set and forget issue - 100LL or car gas, the engine is not going to detonate.
 
Back
Top