Honda Power Unit Hardware & Software

[subcritical71]

They must have 100 times more resolution on injector pulse width.

In general the McLaren ECU samples at 10k samples per second. Four channels can do 100k, and some ‘slower’ ones only 1k. I don’t know how much of that is logged or even transmitted in the telemetry as I’m sure if you max everything out the bandwidth for transmission isn’t there. But still much better than you and I can get on our $500 tuners hooked up to archaic automotive ECUs!

[63l8qrrfy6]

GDI piezo injectors can do pulse widths below 0.5 ms and run very high rail pressures compared to PFI.
Even though they inject against a high cylider pressure, the pressure difference across the injector is still huge compared to a port injector so for the same orifice configuration they will flow much more.

[gruntguru]

Does the moment of injection matter in this respect? My (hobby) (simple 4 cyl inline wasted spark) road car does have fully sequential injection at this time, but no cam sinc. So at this moment I've set it to (port) inject it at 90 deg BTDC. I just don't know what stroke it is. 1 could be in compression and 4 in outlet or vice versa. I can change this, as long as I stay later then the first trigger, which is at 138 deg, but earlier then 60 deg BTDC. Al I know the inlet valve is closed on injection when the pulsewidth is low. On wide open throttle, high RPM, the duty cycle of the injectors is 70%, so then it pisses all over, but that doesn't really matter then anymore. Or does it?

If you don't have cam sync, don't fire the injectors sequentially. Although every cylinder gets its fuel at the same time, next time you start the engine, the injection timing could be 360* different.

[63l8qrrfy6]

I would assume he is running it semi-sequentially firing injectors in pairs two times per cycle?

[Bandit1216]

No, fully atm, cam sensor is in the planning

[63l8qrrfy6]

But you realize that without cam sync you could be firing the injectors at a closed valve every single time?

With semi you are guaranteed to fire 50% of the fuel at the correct time even though the other 50% of the time it will be at a closed valve.

[Bandit1216]

Yes, I do realise that. Some advise to inject on closed inlet valve to give time to evaporate. I’ve always doubt that because of “time”. What’s 20 ms when you talk about evaporation? But with batch you inject every single stroke so you know for sure all is at least eqaul. With semi you also don’t know in which of the upwards stoke you are. Main reason I use full seq is to get max bandwidth at idle. I’m at 2,7 ms at idle now with the bigger injectors. I’m not sure though whether one decimal is the max software shows, or it actually is the max resolution. Perhaps it’s 1/255 resolution, because most is based on one bit. So I’m afraid batch will put me back to 0,5 pulsewidth an I basicly have to choose between lambda 1 and 0,8 because of lack of resolution. I can however write multiple configs and just try.

[PlatinumZealot]

Yes, I do realise that. Some advise to inject on closed inlet valve to give time to evaporate. I’ve always doubt that because of “time”. What’s 20 ms when you talk about evaporation? But with batch you inject every single stroke so you know for sure all is at least eqaul. With semi you also don’t know in which of the upwards stoke you are. Main reason I use full seq is to get max bandwidth at idle. I’m at 2,7 ms at idle now with the bigger injectors. I’m not sure though whether one decimal is the max software shows, or it actually is the max resolution. Perhaps it’s 1/255 resolution, because most is based on one bit. So I’m afraid batch will put me back to 0,5 pulsewidth an I basicly have to choose between lambda 1 and 0,8 because of lack of resolution. I can however write multiple configs and just try.

For sequential fuel the Crank angle sensor alone is not enough. The ECU needs to know where the respective TDC at the each cylinder is. So set that up first the cam position sensor and check it.

[Bandit1216]

Yes, I know. But without cam sinc one does know the inlet is closed. Because at 90 deg BTDC you know it’s closed because it will either be compression OR outlet stroke and at 1200 idle 5 ms is only 36 deg. So with cam sinc only advantage is at higher rpm and load, and that’s where one doesn’t need it. That’s what they told me at least. Come to think of it, the 100% duty cycle event I had this week acted just like rev limiter, so I think that’s a safety in software. Perhaps 25,5 (one bit) ms is the max pw. It’s a vems. The computer has all features one can think of and more for half the prise of eg halltec or dta, but support is bad. No one to answer these kind of questions.

[godlameroso]

Oil pressure issues are hell to fix. You either need a stronger pump with more parasitic drain, or you play with the pressure relief valves on the oil circuit to allow the part you modified to have enough oil pressure to function properly, without reducing oil pressure on other parts of the engine. Ugh. To make matters worse you have water like oil viscosity which makes maintaining precise oil pressure more difficult.

[stevesingo]

Yes, I do realise that. Some advise to inject on closed inlet valve to give time to evaporate. I’ve always doubt that because of “time”. What’s 20 ms when you talk about evaporation? But with batch you inject every single stroke so you know for sure all is at least eqaul. With semi you also don’t know in which of the upwards stoke you are. Main reason I use full seq is to get max bandwidth at idle. I’m at 2,7 ms at idle now with the bigger injectors. I’m not sure though whether one decimal is the max software shows, or it actually is the max resolution. Perhaps it’s 1/255 resolution, because most is based on one bit. So I’m afraid batch will put me back to 0,5 pulsewidth an I basicly have to choose between lambda 1 and 0,8 because of lack of resolution. I can however write multiple configs and just try.

On my 4cyl hobby with sequential (cam sensor fitted) I just tuned injection timing for lowest Inj PW for the given target lambda. I figured this gave me the lowest unburnt fuel. It ended up with end of injection in the region of 205-150 deg BTDC from 3k to 8k rpm with Inj PW of 5-9.5ms.

3k rpm Inj PW 5.7ms - Inj 89-155deg ATDC intake stroke

4k rpm Inj PW 6.95ms - 18deg - 185deg ATDC intake stroke

5k rpm Inj PW 8.65ms - 54deg BTDC - 205deg ATDC intake stroke

6k rpm Inj PW 9.05ms - 120deg BTDC - 205deg ATDC intake stroke

7k rpm Inj PW 9.15ms - 180deg BTDC - 205deg ATDC intake stroke

8k rpm Inj PW 8.95ms - 224deg BTDC - 205deg ATDC intake stroke

At lower rpm the injection is occurring entirely in the region of lowest cylinder depression.

In all cases injection is complete before intake valve closes the SOI is advancing with rpm.

[Bandit1216]

Yes, I do realise that. Some advise to inject on closed inlet valve to give time to evaporate. I’ve always doubt that because of “time”. What’s 20 ms when you talk about evaporation? But with batch you inject every single stroke so you know for sure all is at least eqaul. With semi you also don’t know in which of the upwards stoke you are. Main reason I use full seq is to get max bandwidth at idle. I’m at 2,7 ms at idle now with the bigger injectors. I’m not sure though whether one decimal is the max software shows, or it actually is the max resolution. Perhaps it’s 1/255 resolution, because most is based on one bit. So I’m afraid batch will put me back to 0,5 pulsewidth an I basicly have to choose between lambda 1 and 0,8 because of lack of resolution. I can however write multiple configs and just try.

On my 4cyl hobby with sequential (cam sensor fitted) I just tuned injection timing for lowest Inj PW for the given target lambda. I figured this gave me the lowest unburnt fuel. It ended up with end of injection in the region of 205-150 deg BTDC from 3k to 8k rpm with Inj PW of 5-9.5ms.

3k rpm Inj PW 5.7ms - Inj 89-155deg ATDC intake stroke

4k rpm Inj PW 6.95ms - 18deg - 185deg ATDC intake stroke

5k rpm Inj PW 8.65ms - 54deg BTDC - 205deg ATDC intake stroke

6k rpm Inj PW 9.05ms - 120deg BTDC - 205deg ATDC intake stroke

7k rpm Inj PW 9.15ms - 180deg BTDC - 205deg ATDC intake stroke

8k rpm Inj PW 8.95ms - 224deg BTDC - 205deg ATDC intake stroke

At lower rpm the injection is occurring entirely in the region of lowest cylinder depression.

In all cases injection is complete before intake valve closes the SOI is advancing with rpm.

Oké, I would have to get into that. I have the option to tune this with injector angle curve, but not looked into this jet. This would mean I need to fix camsinc first. I had pw from 4,2 to 18,7 with the smaller injectors. But of course 18,7 is too long unless your redlight is at 4000. Now 2,5 at idle and should be something like 12 at 7000. A lot of diff from idle to full power for an natural aspirated engine. Can I move this really of topic part to engineering myself?

[JordanMugen]

No, fully atm, cam sensor is in the planning

I take it this is an older engine that originally came with a carburettor or single point fuel injection?

[63l8qrrfy6]

How can a direct injection system lean mixture that have failed to combust (misfire) will be/can be re-ignited on the next firing stroke? In a direct injection system any fuel injected that have failed or not to ignite/combust will be expelled with the exhaust gases (exhaust stroke) in between firing strokes.

Not all of it.
Because of lack of combustion, the cylinder pressure is low when the exhaust valve opens so the mixture will not be evacuated efficiently.

[godlameroso]

Anyone want to take a guess what Honda is doing to improve MGU-H recovery, and by extension, increase MGU-K deployment? They had a marked improvement in deployment and energy management in the last 5 races. Next year the ability they have refined in the last 1/3rd of the season along with an improvement in thermal efficiency(higher compression ratio) should amount to a nice bump in performance. Apparently this "strategy" only has a minimal increase in fuel consumption, and the thermal efficiency gains should in theory nullify this penalty making the strategy more flexible.

Pretty excited to see this new power unit on track. I'm sure everyone else is making good gains as well, and what Honda is doing is no secret at this point.

It's a shame that Honda is leaving just as it's hitting its stride, as I feel they have a good grasp on what's needed for this current engine formula.