rusefi.com

See my edits above, your setup is already more than sufficient to run full sequential fuel and ignition.

I'm having a hard time articulating this, but I think the part you have wrong is thinking that more precision is better, when that's not necessarily the case. I'm not familiar enough with the math to prove this, but the variations in engine speed between pulses are negligible when considering the time scales that the ECU is working with. Every piece of rotating mass in the system, your crankshaft, flywheel, clutch, transmission, driveshaft, differential, axles, tires, even your accessories and belts, all add mass and damping to the system and reduce the jitter.

Yes, more pulses add more precision, but there are diminishing returns, and this won't remove the fact that the ECU will still have to look ahead and predict events long before the respective pulse comes around, because once it does it's already too late.

Will your timing vary by a couple degrees between rotations? Maybe. Will this affect your peak power? Eh. Will this damage your engine, or be noticeable to you, the driver? No.


Here's a great example, courtesy of Nissan:

The Nissan L series used points ignition, which was later updated to an electronic system. On the 4-cylinder engines, this system used a 4-pointed wheel that triggered spark. As you can probably figure out, 4 points turning at distributor speed = 180 degrees between trigger events. Advance was controlled via vacuum actuators, counterweights, and springs. Hardly an accurate system, but it works just fine, even for racing.

Can you send me your Honda TS project so I can try your settings in my car? I want to element ate the possibility that I am doing something wrong here.

filip5 wrote: ↑

Wed Apr 03, 2019 11:19 pm

Can you send me your Honda TS project so I can try your settings in my car? I want to element ate the possibility that I am doing something wrong here.

Tune is really out of date with current version, a lot of config-breaking changes since then. I'll take a look at the tune, and I'm sure a few others will as well, and will try and make suggestions.

Okay,

But do you have anything that works right now with your current version? Any engines running?

I prefer exact project not suggestions unless they are top secret?

filip5 wrote: ↑

Wed Apr 03, 2019 9:31 pm

How can one with 4 slots or 180 degrees resolution shoot right in a combustion engine when a piston jerks upward on compression and downwards in combustion? That is an impossible task in cranking and even could be a problem when the engine is running and there is a flywheel that helps, there still could be issues on sudden loads or gear changes. If advance is let say at happening at 50 or 60 degrees before TDC because of a sudden jerk or a load the engine would stop, wouldn't it? Are you guys hoping that the next spark would be good to shoot right and the engine will continue to run?

Plenty of factory ECUs from the 80s ran with a single tooth per cylinder (4 teeth for a 4 cylinder, as in your case). I had a 1985 Volvo 240 that had a 4 tooth + hall sensor inside the distributor, used for both ignition and fueling. Plenty of these cars are running the factory setup, even in racing applications. I had mine for two years and ran it on both the stock ECU and MegaSquirt 2 without issue.

Think about it this way: if the engine speed had enough phase noise to cause problems with a low-tooth-count trigger, then it would also have enough change in its speed that it would be EXTREMELY rough running. While the pistons and connecting rods are reversing course frequently, the rotational speed of the crankshaft is very very stable outside of cranking conditions. This is the very purpose of a heavy flywheel.

My newer volvos both had a 60 tooth wheel on the crankshaft (6 degrees per tooth), and I only saw around 20% speed variation during cranking. Once the engine was running (even at idle), there was almost zero. Any variation when off idle was immeasurably small.

filip5 wrote: ↑

Wed Apr 03, 2019 6:09 pm

I use encoder pulses to track packages till their final destination.

A spinning engine is a slightly different ball game.

There's a very good reason that more than around 60 teeth isn't a great idea:
Suppose you want to rev your engine to 7000 rpm. That's just under 60 cam revolutions per second. 360 teeth per cam revolution is 21,600 teeth per second, or a period of 46 microseconds. I sincerely doubt that early 1980s Nissan was actually decoding every single tooth at any speed past cranking. They may have used it for very low engine speeds only, but without custom hardware it just isn't physically possible they were actually using the 360 tooth wheel at any real engine speed. The only way it's possible that they were using every single tooth at all engine speeds is that they were using a custom ASIC (custom silicon in the ECU = $$$$$$).

My real answer is that I'm genuinely not sure why Nissan used a 360 tooth wheel. Exactly zero cars sold today use anywhere near that many teeth. 24, 36, and 60 have been common since the 1990s.

filip5 wrote: ↑

Wed Apr 03, 2019 9:31 pm

I was using separate coils and no distributor because I thought rusefi could handle that.

Indeed it can. Many engines have run with wasted spark and coil-on-plug. Have you verified timing with a timing light? Unplug injectors, set cranking timing to 10* BTDC or something reasonable, crank the engine, and see where #1 is firing. Only once you've adjusted the "trigger offset" to have #1 actually firing at 10* BTDC should you try to start the engine.

filip5 wrote: ↑

Wed Apr 03, 2019 5:07 pm

What you see in the video is my real nissan engine which is running at 480 degrees idle speed.

Do you have a wideband O2 sensor connected? It sounds like your engine is running pretty rich, and with pretty far retarded timing. A stock cam KA24 will probably idle happiest around 15-20 degrees timing. For now, it's safe to just set the whole ignition timing map to something like 15 degrees to get the engine idling.

I posted this argument earlier but somewhere got lost and I can not find it.

Here is the post again. The earlier Nissan runs fine with 4 slots and vacuum gadgets to control advance because the actual TDC is physically tied to the slot that is a distributor and does not move even when the engine hesitates or jerks. No meter what happens to the engine the slot is there together with its TDC.

It is different story with the EFI ECUs such as Rusefi in this case. For Nissan the slot is positioned at 630 degrees off from TDC for the previous rotation or past 90 degrees off TDC the current rotation whatever one wants to consider it. That is how distributor is mounted. The reason why they do it is for the ECU to have the slot actually pass the actual TDC and then follow with the process such as sparks etc. So the ECU assumes that TDC comes after 90 degrees it has passed the actual TDC. But here is the issue how does ECU calculate where TDC is ? It divides the time that one slot passes to the next by 180 degrees. If engine is running normal that one could see would be okay but there is still a possibility of problems if engine is put under load immediately.

Now imagine out of 180 degrees from one slot to the other the first 90 degrees are past at 1 ms let say during cranking but the next 90 degrees are passed for 3ms because of the compression resistance. In total you have 4ms that divides by 180 gives an average of 0.0222 ms/degree. But the first 90 degrees were past at 1ms or 0.0111ms/per degree and the second 90 degrees at 3ms or 0.033ms/degree. The actual position of the TDC is based on the second part of 90 degrees let assume or after 3ms but ECU is telling you no, the TDC is at 90 degrees x 0.0222= 2ms. But it was actually at 3ms not 2ms off actual TDC . So 1 ms gives you about 30 degrees off for that part of rotation. ECU is telling you wrong TDC is 30 degrees off. That is why much better resolution is needed. 60 slots might work much better than 4 but still that is a plus mins 3 degrees of or 6 degrees off and if it is 2/60 that is at least 12 degrees off assuming time is absolute equal and not if jerks in the engine happens which even 2/60 would cause issues because variance of actual TDC from assumed TDC could be much larger than 12 degrees off.

Did this thread really start on April first?

I suggest you first use OEM spark. Get fuel figured out. After fuel is proper, then migrate spark, one cyl at a time.

From a hardware stand point 360 teeth per rotation is bad. It will be very difficult to relibily get the signal to the ECU. Especially at higher RPM's. It's not the first or last time an auto MFG has made a mistake.

Nissan stopped using this wheel because it was a bad idea.

Your offer for us to come to your shop does not help. We have trouble getting to our own garages.

I understand.

I guess your engine is not taking you there.

A 4 hole wheel has 4 holes, one corresponding to each cylinder. Since the disk rotates at half crank speed, 180 degrees of crankshaft rotation occur between two adjacent holes.

At cranking speed, suppose the speed of the crankshaft is actually varying by ±10% as each cylinder hits compression. Worst case occurs if the trigger tooth is at bottom dead center, so the engine will run 10% slow for the first half of each interval, and 10% fast for the second half. The worst error is going to be exactly in between teeth at TDC, when we've been 10% slow since the last tooth. 90 degrees of rotation have occurred since the last tooth, so we could be up to 9 degrees off (worst case) in between two teeth.

When we're running and the variation is smaller, perhaps ±0.5%, we can only be 0.5% of 90 degrees off, which is 0.5 degree. Half a degree! In reality the engine is much smoother than ±0.5%, as that would be a pretty extreme vibration, so it will be far less.

In conclusion, there is zero concern about the ECU having difficulty predicting when the spark should occur with a small number of trigger teeth. OEMs have been doing it for 40 years. It's a proven solution. Somehow I drove my car a few thousand miles with an ECU running with a 4 tooth, distributor-mounted wheel, just like yours. I never had issues starting the car, or with unstable ignition timing once running.

I can take a look at your tune too when I get home and see if I find anything obviously wrong.

mck

Don't look at the tune, I don't think that will help anything but guess. Look at the the actual engine sniffer I have posted to see how your assumptions pan out.

I am done for the day.

@, you seem to be 100% confident in what you are doing and seem to refusing advice.

your cranking fuel / CLT curve is flat and 4.1 *2 = 8.2 value is probably on the lower side, especially if you do not warm up your engine with stock ECU
your cranking duration coefficient 2.5 is also pretty flat but it gets total fuel to reasonable values I guess.

What is your wideband sensor showing while you idle at 480 RPM?
What is your rationale for 480 RPM cranking?

I will be making a wheel with 60 teeth. I guess the wheel will have 58 teeth and 2 teeth are as a slot , right?

filip5 wrote: ↑

Thu Apr 04, 2019 1:13 pm

I will be making a wheel with 60 teeth. I guess the wheel will have 58 teeth and 2 teeth are as a slot , right?

60 with 2 skipped is the popular kind but it's flexible. it't just that 60 is currently the limit for rusEfi for reasons of RAM usage

Removed

Removed

Removed

Removed

removed

removed

removed

removed

Custom wheel trigger should allow to run 60/2 on cam. One way or another, there is definitely a way to run 60/2 wheel on either crank or cam.

it is running only when inputs of Hall are connected to cam sensor inputs it is not working when connected to crank inputs. Being that cam sensor for 4 cycle thinks the crank is running at twice the speed how is this corrected?

I see it, "4 stroke without cam sensor" should be selected and still be sensor be connected to cam sensor inputs.

This is much better because it gives me twice more resolution than 60/2 connected to crank.

It does not print degrees in the engine sniffer screen, it prints only n/a

I know why you are confused abd its rusEfi fault. There are some plans in motion to make it less confusing on Frankenso and Tuner Studio.

The trigger looks good now with this 60/2.

filip5 wrote: ↑

Thu Apr 04, 2019 6:32 pm

The trigger looks good now with this 60/2.

This is pretty vague. What does it mean looks good? Using real hardware? Using what metric? Are you vague on purpose?

The engine sniffer trace looks fine. The engine sounds like it's running fine. The only things I see wrong in your tune are that you aren't running nearly enough ignition timing at idle (it needs 15-20 degrees), and the trigger offset should be 0 instead of 720.

Again, you need to validate your assumptions. The engine sniffer can only show the ECU's estimate of where the engine is. You need to check the REAL timing on the REAL engine with a REAL timing light. If it's truly jumping around a lot like you claim, then there is an actual problem. If it's not jumping around, then the ECU is working fine.

mck1117 wrote: ↑

Thu Apr 04, 2019 8:36 pm

The engine sniffer trace looks fine. The engine sounds like it's running fine. The only things I see wrong in your tune are that you aren't running nearly enough ignition timing at idle (it needs 15-20 degrees), and the trigger offset should be 0 instead of 720.

Again, you need to validate your assumptions. The engine sniffer can only show the ECU's estimate of where the engine is. You need to check the REAL timing on the REAL engine with a REAL timing light. If it's truly jumping around a lot like you claim, then there is an actual problem. If it's not jumping around, then the ECU is working fine.

I thought ignition timing at idle should be near 0 or negative due to low RPM?

15 to 20 degrees advance in idle is okay you mean?

I am using this 60/2 wheel right now. The missing teeth are placed about 90 degrees passing the TDC clockwise and engine turns clockwise. What should my offset trigger be?

I am still having trouble starting this engine. When it starts and gets turning stable it runs ok. Attached is a picture of the engine sniffer for 1 full camshaft rotation. The fuel is injected about 360 ahead of the ignition. The injectors are Bosch that throw 8mg/ms and the engine cyl size is 464cm3

The mixture during cranking is a little reach , any idea that anyone can suggest? Attached is the tune and the engine sniffer pict.

Any recommendation what injection table phase should be? I see that 355 to 359 degrees responds somewhat better than any other value.

russian wrote: ↑

Thu Apr 04, 2019 6:53 pm

filip5 wrote: ↑

Thu Apr 04, 2019 6:32 pm

The trigger looks good now with this 60/2.

This is pretty vague. What does it mean looks good? Using real hardware? Using what metric? Are you vague on purpose?

For some reason I see you doubting others if we use real hardware or some type of simulator instead and showing it to you here somewhat as real.

Why would I not use a real hardware and tell you that I am using real hardware? I just don't get it.

I think you don't fully grasp how blind we are to your particular engine. We do not know many things which you do know. Seeing your setup fills in many gaps for us. It is common that other people have posted both simulation situations as well as real engine situations. This might not be your particular scenario, but it has happened in the past. With a history of people doing it, it causes us to ask questions which might seem really basic. Again, you know this because you can see and touch it, while we do not know these basic things. If we make assumptions, it generally causes more frustrations and more delays than any of us care for. So we commonly ask really basic questions.

Did you know that rusEFI has self simulation capabilities? AKA you can very easily switch between simulation and the real world.

It is really handy to have a video, saying this is what I'm doing, and this is the problem, as well here is the log file of what the software captured during this video. Something like that is really helpful in removing our blindfold.

I suspect you are trying to keep you engine setup semi-secret. I also suspect that is some how related to a racing thing. I've seen many race folks with some odd requests to not share pictures and such. It seems those requests happen because they feel it creates some competitive advantage for their competitors. When this happens, it can be really difficult for those of use who are trying to help. I would like to suggest you post more about your application, and you post your log files with more detailed information. If you do that it will make it much easier for us to help you over the hurdles.