Speeduino is now on Github Sponsors (Rather than Patreon): https://github.com/sponsors/noisymime
Help with building your Speeduino, installing it, getting it to run etc.
By BasilF
#57465
I’m new to fuel injection and am having some problems. I’m using an NO2C for fuel and ignition on an 810cc Briggs and Stratton V Twin. It is fitted with two throttle bodies from a motorcycle and a 60-2 wheel on the flywheel with a VR sensor and a VR conditioner and that side of it seems to be working fine. However I’ve managed to start it and can only tune small throttle settings as it has no load on it at present. The problem is that after idling steadily for a while the AFR will suddenly go very high and the injector duty cycle drops even though the throttle position and revs are steady.
I’d like to know what is causing this.
Also;
I imagined the AFR target field would be the value read off the table for the revs and TPS position but it sometimes shows values (like 16.6) which don’t appear anywhere in the AFR target table.
Is there a field that shows injector firing better than duty cycle, something like milliseconds open per revolution?
Both exhausts feed into a common silencer where the O2 sensor is. If one cylinder stops firing (due to a fouled plug for instance) will the O2 sensor show a weak mixture due to the free oxygen from the non-firing cylinder even though the mixture is correct?

Thanks for the help.
(1.15 MiB) Downloaded 26 times
(81.07 KiB) Downloaded 25 times
By LPG2CV
#57469
Hi :)

1/ Could you point timewise to a particular instance of it occurring in the log.

2/ AFR, an example please

3/ yes. look at PW (pulse width)

4/ yes, a fowled plug will most likely cause a lean misfire

you may want to turn off DFCO until tuned
WUE. Only the last cell should be %100
User avatar
By jonbill
#57470
Hi,
first part - those AFR target values that aren't in your table are actually the AFR values at that time - Speeduino sends them during the o2 sensor warm up. This is configured in tuning->afr/02 - ego delay after start is set to 15 seconds, so for the first 15 seconds, AFR target is reported as the actual AFR.
in terms of why it goes lean after a time - can confirm when in the log you're looking? if we look at 579.303, we can see that the AFR starts to climb - this is because it's fallen from a VE = 18 area to a VE = 12 area, which is quite a big difference. the 12 is is from the cell at 0 TPS, 1200 RPM. If that cell was 18, you wouldn't get that jump.
It's interesting that there's almost no difference in the other logged values between 579.303 and the records after, i.e. tps 0.0 and rpm 1168 go to tps 0.0 and rpm 1175 causes the VE interpolation to go from 18 to 12 - maybe there's an unseen change in the TPS ADC value? (anyone?)
anyway, for practical purposes, if you change that cell at tps 0, rpm 1200 to 18 you wouldn't get that AFR jump.
By BasilF
#57472
Thanks for the pulse width (soo many abbreviations to learn) and the delay in target AFT. I can also see that AFR lags by several seconds after a change in other parameters so I'll consider this in future.

The events I'm puzzled by occur at 176.346 and 218.381seconds where the pulse width goes to zero without any change in VE table or TPS followed by the AFR climbing (well it would).
Also the event at 421.761 where the PW stays constant but AFR goes from 11.7 to 14.3.
Finally at 612.645 PW and VE goes apeshit. There are definitely no values above 100 in my VE table so where do they come from?
Thanks for the very helpful comments so far.
By BasilF
#57473
One further question. These very low power runs have the pulse width hovering around 2ms which I assume is 2ms per revolution. The engine is flat out at 4000rpm so that gives 66revs per second or 15ms per rev. With a maximum injector duty cycle of say 90% that gives 13.5 ms available for fuel injection. That ratio of 7 to 1 between idle fuel and WOT fuel doesn't seem very large. I'd imagine that the motor is producing between 1 and 2 hp to keep it running at 1000rpm under no load and I'm expecting about 35hp at full revs and throttle so wouldn't I need about 17 times more fuel between idle and full power. However if I fit larger injectors then I have very small PWs at idle. Mightn't this cause problems with injector opening time issues, e.g. 1ms opening time when 0.5ms PW is required?
User avatar
By jonbill
#57474
BasilF wrote:
Tue Jun 21, 2022 1:08 pm
Thanks for the pulse width (soo many abbreviations to learn) and the delay in target AFT. I can also see that AFR lags by several seconds after a change in other parameters so I'll consider this in future.

The events I'm puzzled by occur at 176.346 and 218.381seconds where the pulse width goes to zero without any change in VE table or TPS followed by the AFR climbing (well it would).
Also the event at 421.761 where the PW stays constant but AFR goes from 11.7 to 14.3.
Finally at 612.645 PW and VE goes apeshit. There are definitely no values above 100 in my VE table so where do they come from?
Thanks for the very helpful comments so far.
at 176.346 PW is 2.384, but it was 0 a second or so earlier - because DFCO was on. at 218.381 DFCO is on again.
re 421.761: not sure, but seems to be correlated with the spark advance became very erratic at 421.0 - I can't see what would cause that.
similarly, at 612.645, I can't see what would cause that. seems very strange. There is a correlation with MAP values around there - MAP is stable at around 50 before 612 then starts oscillating up to 60 or so with the same cadence as the VE swings.
User avatar
By jonbill
#57475
BasilF wrote:
Tue Jun 21, 2022 1:23 pm
One further question. These very low power runs have the pulse width hovering around 2ms which I assume is 2ms per revolution. The engine is flat out at 4000rpm so that gives 66revs per second or 15ms per rev. With a maximum injector duty cycle of say 90% that gives 13.5 ms available for fuel injection. That ratio of 7 to 1 between idle fuel and WOT fuel doesn't seem very large. I'd imagine that the motor is producing between 1 and 2 hp to keep it running at 1000rpm under no load and I'm expecting about 35hp at full revs and throttle so wouldn't I need about 17 times more fuel between idle and full power. However if I fit larger injectors then I have very small PWs at idle. Mightn't this cause problems with injector opening time issues, e.g. 1ms opening time when 0.5ms PW is required?
Your VE table suggests that between 5 and 7 times more fuel is required at WOT @4300 rpm compared to closed throttle at idle. so that suggests it will want to use maybe 12ms of PW at WOT, which would fit within your budget. However, the log doesn't go there.
open time is added to desired PW, so as long as open-time is accurately set the delivered PW should be correct. However, any jitter in open/close times do become massively more impactful with short PWs.
User avatar
By PSIG
#57476
I wouldn't get too focused on specifics yet, as there is a lot of housekeeping to do in your tune first, that will make it much easier to "see" issues and potential causes. For example, I see a general trend toward leaning after apparent warmup, but also note you have no injector voltage compensations as your average voltage drops across the run. Coincidental? Perhaps, but is the type of setting that will help issues either go away, or become clearer.

Knowing almost nothing about your hardware setup, we can only mostly guess or look for trends. This relies on the data streams, and you have 41 of them in the log file. So, I would suggest reviewing each of those data streams for function, accuracy or at least reasonable numbers, if you (and we) are to rely on them. For example, Semi-Sequential and 4-squirts might perform better with consistent injection timing, but we can't see clear results in data well enough yet to diagnose those experiments. Yet.

Every setting in TS has a purpose and best value. I would start with reasonable values for each. Housekeeping of your data sources and settings is all just stuff to make your job easier. ;)
By BasilF
#57478
PSIG wrote:
Tue Jun 21, 2022 2:16 pm
For example, Semi-Sequential and 4-squirts might perform better with consistent injection timing, but we can't see clear results in data well enough yet to diagnose those experiments. Yet.
I really don't understand the injection settings. When applied to a two channel ECU on a two cylinder engine .

"1. Paired - 2 injectors per output. Outputs active is equal to half the number of cylinders. Outputs are timed over 1 crank revolution. 2. Semi-sequential: Same as paired except that injector channels are mirrored (1&4, 2&3) meaning the number of outputs used are equal to the number of cylinders. Only valid for 4 cylinders or less. 3. Banked: 2 outputs only used. 4. Sequential: 1 injector per output and outputs used equals the number of cylinders. Injection is timed over full cycle."

From the above it would seem that Sequential describes my setup with one injector per channel, two channels and two cylinders, but I understand that Sequential requires the use of a cam timing wheel which I don't have.

I'll apply a battery voltage correction although the suggested corrections are small. However your comment has shown me that the charging system doesn't seem to be working as it should be up at 13.8 volts after a few minutes of running.
User avatar
By PSIG
#57510
You may have missed my point, that you have many settings that are not set well or questionable, and they are affecting your operation. It is also making tuning more difficult, by clouding engine responses. The general path is to work with wide-scope changes to alter overall trends, finally working down to detail areas and items as they become clear. Most settings are general and affect the overall tune.
BasilF wrote:
Tue Jun 21, 2022 3:38 pm
I'll apply a battery voltage correction although the suggested corrections are small. However your comment has shown me that the charging system doesn't seem to be working as it should be up at 13.8 volts after a few minutes of running.
I'm glad you have found a potential weakness to investigate. However, I would not categorize dead-time (latency) voltage corrections small. I pulled-up the first random injector example from my files of a popular and very quick 600cc/min type. Between your "normal" voltage of 13.8V and your log showing 11.9V during the run; this random injector would be supplying only 78% of the fuel you thought it was. :shock: That is not a small error, and this random injector example is better than most. Could 27.5% less fuel cause a leaning effect? ;) Multiply that "small" error with several other small errors, and perhaps you can see how the run could be negatively impacted. Do some review of all your settings. Repeat the review regularly.

I understand your concern with the different modes of injection, and what they mean. However, that is not your primary issue right now, and you couldn't effectively test and compare different modes at this point in tuning. I would continue on your current path, making one change at-a-time. Again my point is to get the tune and settings as stable and effective as possible. Then you can see more clearly how different modes and test-changes affect the operation.
BasilF wrote:
Mon Jun 20, 2022 9:50 pm
Is there a field that shows injector firing better than duty cycle, something like milliseconds open per revolution?
Use the PW field for milliseconds per pulse. Load the Calculations.dash dashboard (in the firmware download package) to see real-time values that result in the current PW.
BasilF wrote:
Mon Jun 20, 2022 9:50 pm
Both exhausts feed into a common silencer where the O2 sensor is. If one cylinder stops firing (due to a fouled plug for instance) will the O2 sensor show a weak mixture due to the free oxygen from the non-firing cylinder even though the mixture is correct?
Yes, you are correct. You have some misfires that are showing as lean following each event as examples of this. You can use a combination of MAP, rpm/s and other data indicators to help find partial and complete misfires in logs. Learn to use the filters in MLV. ;)

Note here that this is one reason "live" auto-tune (VEAL) is not suggested early in the process, as even partial misfires can show lean, causing an auto-enrichening of the tune at that point. As the misfire might be caused by a rich condition, not only is the tune now incorrect, but can make the situation worse. I suggest using MLV with VEA mode to review each log before allowing table alterations.

You are on the right track, but tuning takes work. Like any skill, there is a lot to learn and practice. General initial trends working down to details, as the engine gives you feedback. The more work you do, the better it performs, and the better and quicker you can do it. Keep going until it is "good enough" towards perfect in your judgment, or it meets your project goal requirements. 8-)

Ensure you have hardware test turned off.

Simulator

Just following up a little. Is there a "Spee[…]

So I saw this before but I couldn't find any forum[…]

The factory-equivalent table is only to get you st[…]

Still can't find what you're looking for?