LPG Discussion Forum

Looks interesting anyway they seem to cover a wide range in vehicles, have you got a simple strategy of how you mite initially trial and test it, looking at a petrol/Lpg advance graph I don't know whether it's universal, approx 3-3500 rpm the plots merge whereby advance is the same for both fuels, above that it looks like retarding would be optimal.
So I suppose up to around 2500 rpm + high degrees of over advance could be effective.
I suppose knowing the rpm s you usually operate at and tweaking those especially if that falls into the low/mid range could be effective, and yeah they look easy to fit.

I always make a point to read LPGC's posts. It's not often on forums that I get to learn too much. And that's not because I know all that much. More because, sadly... most times, I'm found reading mindless mis-information?

LPGC... thank you.

Back in the day, or in my case the present... I've gone thru' AEB Joker and AEB Wolf TAPs. But running a 1960s Land-rover on CNG, we're talking old skool stuff here. A TAP is essential!

He's certainly proved himself time and again. Lots of good info, so I find I'm having to reading it later on again.

I've heard the TAP makes much more difference on CNG, but I figured it was still worth trying on LPG.

Just managed to fill with lpg again to check consumption. It's down to 14.3l/100km now, which I'm much happier with (12.5 on petrol, so under 15% difference). Will probably get the TAP running next week and see how it goes for me. Will post results

Cheers @CNG and @disco stu

Yes a TAP will make a lot more difference on CNG, necessary for CNG.

Look forwards to your results Stu.

disco stu wrote: Thu Mar 25, 2021 6:48 am I recall reading in the manual for my car that ignition timing is based on the crank sensor angle and injection based on the cam sensor, or something to that effect. I still can't understand why, as they both do the same thing so why not just base it all off the one sensor (cam would be logical).

LPGC's comments re VVT are valid but it's more basic than that I think.
The cam sensor is the only sensor that knows 'where' the engine is in the cycle. The crank sensor can't tell a compression from an exhaust stroke so the cam sensor has to be there for start up (unless like my smart (the only engine I've ever heard of that is so configured) it employs another method). Once running, cam sensors aren't really needed as the crank sensor can provide enough info for the ECU - once it's been primed by the cam sensor. Some engines however may continue to use cam sensor data for sequential fuel injection. Cam sensors also pinpoint misfires to relevant cylinder(s).
The crank sensor though will be more accurate for ignition timing which will eventually stray with wear in the cam drive - which is why I think, it is the primary source of data for ignition timing.

What might be useful for those messing around with timing is data logging software. I'm looking into TunerPro software (free downloads available from their website) that runs on a laptop via an appropriate interface to the OBD port.
One of the parameters it can measure is 'knock counts''. The guys on my Chevrolet truck forum adjust (via TunerPro) their timing in particular speed/load bands then subsequently monitor the knock counts. If overly high, they then retard a touch.

I agree with LPGC that advance isn't everything. The aim is to burn as much of the fuel at the most favourable piston position relative to TDC. 50% of mass fuel fraction burned by 10 degrees ATDC seems to be the norm. Over advancing will promote detonation and the contained (un-expanded) heat will exacerbate the tendency in the extreme but negative work can accrue long before that. A good indication would be exhaust gas temperatures if they were easily monitored. When they cease to cool with advanced timing - that's probably optimum timing.

We broadly agree mate but I would repeat that some engines with hydraulically controlled VVT do compare cam position sensor pulses to crank position sensor pulses to know how far cam timing is advanced/retarded. Might have an electrically controlled hydraulic master solenoid that feeds oil to a remote slave hydraulic cam timing adjuster, cam timing adjustable by controlling the duty cycle of electric pulses to the master... But with this type of setup the ECU cannot assume that (say) a 50% duty cycle pulse on the master will give X degrees cam advance, it has to measure the cam advance and control duty cycle to achieve the cam advance that is aimed for (or otherwise cam timing might change when viscosity of oil changes). Like you say the crank sensor cannot tell the difference between a compression and exhaust stroke but plenty crank sensors have gaps in teeth so do have reference points for crank position which can be compared to cam position data. Plenty engines will give 'cam timing too advanced' type errors if a cambelt/chain is fitted a tooth out.

LPGC wrote: Thu Apr 15, 2021 6:33 pm We broadly agree mate but I would repeat that some engines with hydraulically controlled VVT do compare cam position sensor pulses to crank position sensor pulses to know how far cam timing is advanced/retarded. Might have an electrically controlled hydraulic master solenoid that feeds oil to a remote slave hydraulic cam timing adjuster, cam timing adjustable by controlling the duty cycle of electric pulses to the master... But with this type of setup the ECU cannot assume that (say) a 50% duty cycle pulse on the master will give X degrees cam advance, it has to measure the cam advance and control duty cycle to achieve the cam advance that is aimed for (or otherwise cam timing might change when viscosity of oil changes). Like you say the crank sensor cannot tell the difference between a compression and exhaust stroke but plenty crank sensors have gaps in teeth so do have reference points for crank position which can be compared to cam position data. Plenty engines will give 'cam timing too advanced' type errors if a cambelt/chain is fitted a tooth out.

My comment was based on simpler engines such as my Vortec that has no VVT but still times the ignition from the crank and not its cam sensor.
I'd say driving the ignition from the crank began there when the cam sensor being used was theoretically possible - but impossible on the engines you describe for the reasons you give.

The discussion has got me thinking about tackling timing from the other end - EGT. If setting up from scratch, measuring EGT with known to be retarded timing and then advancing the timing until the EGT ceased to drop that would represent an ideal as combustion was early enough not to miss the expansion event and wastefully through heat down the exhaust pipe. And without having to push the timing into negative work territory.

That's a good point regarding timing belt stretch etc and ignition timing, and now you say that it does seem much more logical. I was thinking too simplistic about that sort of thing (for probably the first time ever-chronic over thinker!)

EGT is another really good point also, and not sure if I have come across anyone discussing monitoring that for optimising timing. That also lines up with things that I mentioned in another thread, where the Russians kept mentioning advancing the timing to save exhaust valves, as apparently the lpg mixture is often still combusting on the exhaust stroke leading the valve/seat issues. I'm not certain of the correctness of that info, but it seems to line up with info from Mitsubishi Australia.

Problem with my car is its pre obd2 for Australia (2005), and also lacks knock sensors. Fairly poor I thought for a 99 model

I did try some different timing yesterday and testing acceleration times with the software. I saw no real difference with up to 9deg extra advance. The car did feel a bit better with some advance, not as jumpy, but feel is very subjective. I emailed the seller and he gave me some things to try which I'll look at today

On some engines, those that use a wasted spark ignition system, cam sensor won't be needed as the sparks fire on both the compression and exhaust strokes. So as long as the crank sensor is working the timing can be calculated from that alone.

On my last post I meant to include a bit at the end to say there are other ways some ECUs can detect incorrect cam timing besides direct comparison between cam and crank sensors but direct comparison is one way.

I think it's the norm for ignition and even point at which each injection pulse begins to be crank sensor referenced (as opposed to cam sensor referenced), the cam sensor only purpose on a less advanced engine (without VVT) is so the ECU knows which part of the 4 stroke cycle each cylinder is on, but we approached this subject with TAPs in mind so I raised the point about some VVT engines using cam sensor info to read cam timing and how on some engines fitting a TAP to just the crank sensor could make the ECU incorrectly read cam advance. For sure there are simpler engines where fitting a TAP to just the crank sensor will affect ignition timing without messing any other aspect up.

I like the theory of measuring EGT to set ignition timing, not sure if it will work but would be interesting to see how such tests go. Slightly different subject but I know some piston engine'd aircraft pilots set mixture using EGT. They may all watch EGT to make sure the temp isn't too high but some take things a bit further - to save fuel instead of backing off the throttle in a relatively low speed cruise they keep the throttle relatively wide open but lean the mixture, which at first sees EGT soar, but then leaning the mixture further sees EGT fall again. They reckon this method helps save fuel partly because the engine isn't losing as much to intake throttling losses (inlet manifold pressure still relatively high) and partly because the leaner mixture burns slower in the cylinder giving a more average pressure push on the piston on it's power stroke (more like a diesel) rather than have a big push closer to TDC, which at lower RPMs helps give more torque for less fuel use. There is argument between the minority who use the mixture method to control engine power / airspeed / EGT and those who more conventionally use the throttle, those that use the throttle reckon it's risky to use the mixture method for the very small gains and that a lean exhaust mixture is more likely to cause valve damage at any given EGT because there will be a lot more unburned oxygen in lean mixture exhaust gas.

We often think in terms of changing one aspect but there are some very complicated interactions between aspects - If we have a fuel injected car and advance timing by fitting a TAP we may also have advanced the point at which the fuel injection pulse starts.. which on the face of it should make less difference on LPG (which has no chance of wetting the manifold) but can make a difference to engine response particularly in transient (changing throttle / rpm) conditions if for example pipes between injectors and manifold are a bit long (and definitely on an engine which has VVT and runs a pseudo Atkinson cycle). Another difference between LPG and petrol is LPG is injected as vapour, petrol as liquid, they have slightly different stochiometric ratios anyway, petrol will take up less volume in the intake and have a cooling effect as it evaporates.. For any given MAF reading MAP reading will be very slightly higher on LPG than on petrol (because the vapour LPG takes up a bit more volume in the manifold raising it's pressure slightly) and this can affect throttling losses. I've recently converted a lot of the same type of engine and have noticed that at warm idle the MAF voltage reading seems to drop by 0.01Volts on LPG compared to running on petrol, which (if a true reading) means the engine idles using less air than it uses to idle on petrol. On the basis of the engine using less air to idle on LPG than on petrol it could be that the engine idles more efficiently on LPG than on petrol, which could be due to a tiny bit less throttling loss, could be something more complicated such as speed of the burn, or could just be a rogue reading (e.g. the petrol ECU has slightly different reference voltage for MAF due to not having the load of petrol injectors). Often discussion about making adjustments to timing and fuelling is about flat out power.. which is an easily repeatable condition for testing purposes (driver just puts his foot to the floor) making it easy to compare results between different tests, making it easy to find the ideal advance and fuelling... If we were to look at the idle situation I just mentioned should (or how would) we factor the difference in MAF reading into results we should aim for? Then it gets more difficult to find the ideals to dial in for all the various combinations of part throttle and mid RPMs?

disco stu wrote: Thu Apr 15, 2021 11:08 pm
EGT is another really good point also, and not sure if I have come across anyone discussing monitoring that for optimising timing. That also lines up with things that I mentioned in another thread, where the Russians kept mentioning advancing the timing to save exhaust valves, as apparently the lpg mixture is often still combusting on the exhaust stroke leading the valve/seat issues. I'm not certain of the correctness of that info, but it seems to line up with info from Mitsubishi Australia.

Thinking about it more, by the method I outlined the most retarded acceptable timing could be found. Then, by pursuing knock, the earliest acceptable timing could be found. Somewhere between those two points must lie optimum timing. To do that over the entire load/speed range is a massive undertaking though.

Late timing hammers exhaust valves - I have the burned ones from my smart to prove that!
EGT is ignored by 4T (4 stroke) guys (with the exception of aviation perhaps) but 2T guys employ to monitor in-cylinder temps not least because of the piston's susceptibility to heat. When looking for knock they observe cooler EGT - after an initial increase. What this tells them is the heat they expect to see in the exhaust is being transferred to the piston crown because detonation has scoured it's protective boundary layer. They then either drop the throttle or lose an engine.

disco stu wrote: Thu Apr 15, 2021 11:08 pm Problem with my car is its pre obd2 for Australia (2005), and also lacks knock sensors. Fairly poor I thought for a 99 model

Are you absolutely certain of that?
The generation of truck I run began in 1988 and even those with early OBD systems had knock sensors. They aren't always located in the head. On my engine there is one and it doubles up as a coolant drain plug on one side of the block.

LPGC wrote: Fri Apr 16, 2021 10:13 am

I like the theory of measuring EGT to set ignition timing, not sure if it will work but would be interesting to see how such tests go. Slightly different subject but I know some piston engine'd aircraft pilots set mixture using EGT. They may all watch EGT to make sure the temp isn't too high but some take things a bit further - to save fuel instead of backing off the throttle in a relatively low speed cruise they keep the throttle relatively wide open but lean the mixture, which at first sees EGT soar, but then leaning the mixture further sees EGT fall again. They reckon this method helps save fuel partly because the engine isn't losing as much to intake throttling losses (inlet manifold pressure still relatively high) and partly because the leaner mixture burns slower in the cylinder giving a more average pressure push on the piston on it's power stroke (more like a diesel) rather than have a big push closer to TDC, which at lower RPMs helps give more torque for less fuel use. There is argument between the minority who use the mixture method to control engine power / airspeed / EGT and those who more conventionally use the throttle, those that use the throttle reckon it's risky to use the mixture method for the very small gains and that a lean exhaust mixture is more likely to cause valve damage at any given EGT because there will be a lot more unburned oxygen in lean mixture exhaust gas.

Our posts were written at the same time and you have confirmed that aviators monitor EGT and given the reason why.
If they had the capability to advance the timing while leaning the mixture it would be beneficial as the closer an engine runs to the 'constant volume' model (vs the 'constant pressure' model) the more efficient it is as more of the expansion stroke can be utilised. (See diesel 'cut-off ratio' and how it hurts efficiency - probably the biggest advantage common rail brings to the party is its reduction).
In essence they are pursing 'lean burn' technology with the associated lowering of combustion temperature (and thus reduced heat loss in conjunction with the reduced throttling losses you allude to).

This is also what the guys (well one at least!) on the Chevrolet forum pursue. They call it HLC (Highway Lean Cruise) and can enable some very lean mixtures. They are however attending to timing requirements in every load/speed range to achieve this.

Sorry, been slow to respond to things here.

Pinger-definitely no knock sensors on my car. I don't understand why as knock sensors were very well established by 1999 (as you say). All I can think was trying to save money on manufacture and sending the crappy ones to Australia.

I watched that video on lean of peak in aircraft. Found it very interesting. Obviously cruising in an aircraft is a different situation than car on the a road with varying roads, hills etc. Going with what Pinger mentions about the Chevrolet forum guy is pursuing, I know on the TAP that I've got you can set a 3d timing map when you're hooked up to the MAP sensor also, so that would allow setting advance at all rev and load ranges. I like this idea, but honestly don't have the energy at the moment to put the energy into more basic things.

So just to catch up where I am with the TAP. I've been running it for the past week on 7deg advance. Checking LPG running during the week it was showing lean on LPG vs petrol across most of the injection time range. I'm trying to get my head around whether this change in mixture is due to the TAP, as I figured if the TAP was affecting mixture it would change Pinj time, not just Gas time leaning. I'll have another look today. Its gone back to being fairly jumpy on acceleration and deceleration between about 1100rpm to 2100rpm, which I've found it doesn't do when mixture is correct. I'll have to plug computer back in to check O2 sensor is working fine, as this could be a reason (if its decided to die for some reason).

I did a drive up the mountain to a place about an hour away on the freeway, which I pushed the car at some stages. Consupmtion on the last tank (which included the freeway run) was 17.7l/100km which is much higher than the previous tank (14.5/100km).

I've been a bit too busy to keep up to speed with what's going on on-forum for a couple of days.

One aspect implied by a lean mixture is a slower burn, so it should be expected that if an engine purposefully runs a lean mixture it's ignition timing should be advanced.

Well known that a lean mixture can cause backfires igniting the intake plenum on a petrol/LPG carb or LPG mixer system... Some of such backfires are caused by the slower burn.

Checked running last night. O2 sensor fine, and adjustments I made to mixture when it was showing lean on lpg were spot on (thanks to your tutelage lpgc) so it's bang on with petrol inject time now. I'm thinking the TAP must have affected things as it was perfect before using the TAP.

I'll back the timing off in the next few days and see what that does to driveability and mixtures to allow me to compare.

With what you say above about lean mixtures and advanced timing, that would be advancing the timing in response to a lean mixture I'm gathering. What effect would you expect advancing timing to have on mixtures (as in other way around)? That's the part I'm trying to figure out. More complete combustion, if that is actually occurring, may indicate richer on the lambda sensor I'm guessing and possibly effect things

Edit: originally said lab mixture, was meant to say lean mixture. Stupid phone!

If you're doing lab testing you might be ahead of me, I only know from experience, thought experiments and theory (applied theory with the help of experience and some applied knowledge).

Would expect similar readings from too far advanced ignition timing, at least up to the point of any detonation. I don't know what results to expect if detonation were to occur but even then would still expect similar readings. We're only talking about lambda readings and lambda readings only relate to correct mixture for burning... which doesn't necessarily imply burn speed, burn pressure (with very slow burn speed maybe some of the burning is done in the exhaust under much lower pressure than cylinders) and I don't know if detonation would give the same (at least as far as oxygen sensor reading / lambda reading) result as burning/explosion. Might expect more NOX in detonation but I don't know.

Or to put things the other way around... I did say that leaner mixtures burn slower than stoch/rich mixtures, so the other way around is rich mixtures burn faster than stoch/lean mixtures, so with a very rich mixture maybe timing could be retarded a bit... But there's still yet another aspect in all of this - a rich mixture is less likely to suffer detonation, so on the flip side this means timing can be advanced. Burn speed is one aspect, the faster the burn the less advance is needed and a rich mixture gives a faster burn. Resistance to detonation is another thing, a rich mixture gives a faster burn so may need less advance for good results but is also more resistant to detonation and with more resistance to detonation may allow more advance which may give even better results. But it would be wrong to read this and think the richest mixture is best because it gives fastest burn and best resistance to detonation and allows most advance, because other aspects are spoiled when mixture is too rich or ignition is too advanced.

There are other ways to check 'more complete combustion' - the post cat lambda sensor readings give some insights, as do HC and CO readings on an exhaust gas analyser and to some extent NOX readings.

In a real engine it's really way more complicated than this, e.g. any cylinder pressure when the inlet valve opens could be expected to prevent fresh inlet charge entering a cylinder, and a slower burn could mean more pressure in the cylinder when the intake valve opens. This could imply some built in EGR which would lower volumetric efficiency but may further help with emissions but would raise inlet temperature and may increase NOX, higher intake temp may lower thermal efficiency. I doubt I have hardly touched on some of the real world complexities here and some of the would/could/maybe's may only be relevant for engines with certain bores/strokes, certain bore/stroke ratios, certain combustion chamber/head/piston crown designs, running at certain rpms at certain loads with certain mixtures at certain intake air temperatures on certain fuels at certain atmospheric pressures. There will be some rules of thumb but a lot of maybe's that rely on combinations of other factors.

Engine makers had making most power from a given sized engine when fed with atmospheric air pressure well sussed by the late 1960's but then they were forced to start limiting emissions and increasing fuel economy and it's taken them a long time to get up to 1960's power levels while keeping within the limitations of emissions and economy. If in the 1960's power didn't matter but it was all about fuel economy I reckon they'd have got econonm sussed that early but we'd all have been driving very slow cars that kicked out loads of emissions. Or if they'd focussed on emissions back then they'd have been a lot cleaner but we wouldn't have had power or economy. As LPG installers we don't design the engines we can only focus on fuelling, when we start altering other aspects we are altering more than we might think. LPG and especially CNG have higher octane than petrol, had the engine designers designed the engine with that in mind they could have made it with more compression. In the old days engine makes went on rules of thumb then tested changes to their design, now most of that is done by supercomputer but the proof is still in the pudding and changes to a supercomputer design can give better results than the supercomputer design. .. We're probably not going to beat £multi-million engine development with our thoughts here but we can ensure mixture is correct and maybe use a few rules of thumb to gain 1% differences in power and economy with TAPs etc if we can't re-design engines with more compression (which might involve re-designing other aspects too, which might have another knock on effect which might mean we'd have to re-design other aspects, etc etc).

Thanks. I realised I put lab mixture in my post above yours, was meant to say lean mixture. Typing on phone.

On mixture and detonation, from reading I've done I was under the impression that rich avoids detonation due to the heat absorbed through vaporisation of the fuel, which would then only occur with liquid fuel (or droplets of liquid with injected). Are you just transferring knowledge of petrol operation over to lpg, or are you certain that this is the car with lpg also? I'm running through my mind rates of reaction when things are out of stoichiometric, but things in a test tube are very different to the rate of a cylinder firing under pressure.

But, contrasting to what I said above is what you say about burn speed. I gather this is converse to petrol, in that rich on petrol burns slower where rich on lpg burns faster like you mention. But isn't detonation just burning too fast (or extra extra fast), so there would be a sort of gray area??

The thing that Pinger mentioned regarding exhaust temp and combustion was interesting I thought. Unfortunately I don't have the equipment or the energy currently for testing any of those things to see the difference ignition timing has on my car. I'll have to just play around with things are see what difference occurs. So far I've had no pinging that I can hear (haven't heard it much in a long time, but was quite familiar with it on my older pre electronic cars from years ago)

Many things can affect detonation, there's probably a bit of detonation on a localised micro scale in most engines but it's only a knock if it becomes wider spread in the cylinder, squish areas in cylinders (particularly on older design engines) are likely places for such detonation but because they're areas without much internal volume they make it more difficult for detonation to spread. The cooling affect of evaporation will help prevent some causes of detonation but probably won't help in other causes. One thing that probably won't happen in a test tube is adiabatic heating.

Detonation isn't quite like burning, burning is triggered by flow of heat whereas detonation is triggered by a shock wave... kind of like the difference between an explosive like gunpowder and a high explosive like nitro-glycerine. There's more energy in a teaspoon of petrol than a teaspoon of nitro-glycerine.

I doubt any of us have all the gear to test all these aspects completely thoroughly, car makes don't have all the gear even with their supercomputers.

disco stu wrote: Sun Apr 25, 2021 1:38 am

On mixture and detonation, from reading I've done I was under the impression that rich avoids detonation due to the heat absorbed through vaporisation of the fuel, which would then only occur with liquid fuel (or droplets of liquid with injected). Are you just transferring knowledge of petrol operation over to lpg, or are you certain that this is the car with lpg also? I'm running through my mind rates of reaction when things are out of stoichiometric, but things in a test tube are very different to the rate of a cylinder firing under pressure.

Petrol (and to a greater degree methanol) offer the possibility of 'fuel cooling' that you allude to. Generally though - up to a point - richer burns faster.....

disco stu wrote: Sun Apr 25, 2021 1:38 am But, contrasting to what I said above is what you say about burn speed. I gather this is converse to petrol, in that rich on petrol burns slower where rich on lpg burns faster like you mention. But isn't detonation just burning too fast (or extra extra fast), so there would be a sort of gray area??

......which helps resist detonation by consuming the air/fuel mixture by an advancing flame front rather than an advancing compression wave.
Detonation occurs when the temperature in the end gas (unburned mixture ahead spatially of the flame front) is heated by compression from the expanding burning mixture above its self ignition temperature. To avoid detonation, the flame front has to get to the end gases before they have sufficiently heated from the advancing compression wave. It's a race!

[/quote]

disco stu wrote: Sat Apr 24, 2021 11:53 pm


With what you say above about lean mixtures and advanced timing, that would be advancing the timing in response to a lean mixture I'm gathering. What effect would you expect advancing timing to have on mixtures (as in other way around)? That's the part I'm trying to figure out. More complete combustion, if that is actually occurring, may indicate richer on the lambda sensor I'm guessing and possibly effect things

The mixture won't be directly changed as a response to altered ignition timings as it is calculated from air flow (and temperature) considerations but as LPGC mentions, gas flow may be affected by changes in ignition timing which could alter mixture.

Worth remembering that as a Lambda sensor is an oxygen sensor it will detect (and show as lean) incomplete combustion even when a mixture is rich but as LPGC points out this event will be accompanied by high levels of UBHC. Unlikely in your case as you are advancing timing but a phenomenon to be aware of anyway.