Technical information on the Toyota D-4S system

[lurker]

yep, this diagram shows the mechanical operation of pump with relation to the spill valve. Thats the easy part to understand as the pump will either be in a state of suction,inactive or pressure sroke.
Switching the spill valve itself to either on/off coupled with the cam lobe timing will determine the above.

Just wondering if anyone has any evidence or theorys on how the EDU determines when to energise the spill to either on/off to determine pump or suction of fuel. i.e. it can be switched per 120' cam rotation, but pressure stroke would need to be done when all injectors are closed. so does the 120' point relate to the end of an injection timing event.

i guess without this knowledge full standalone control of d4s will remain unlikely.

Originally Posted by arghx7

All I've got for you is a service manual diagram...

[lurker]

PS every components used in the D4S system by denso had to have a patent done, this is a very good source for techical information on how the components and system work

example...

[arghx7]

Originally Posted by lurker

PS every components used in the D4S system by denso had to have a patent done, this is a very good source for techical information on how the components and system work

example...

Yes, patents can be a good source. You have to do some digging.

The patent you attached mostly covers shut down determination logic for fuel pressure control. When the ECU judges that the engine is about to shut down, it lowers fuel pressure. This reduces the peak fuel pressure during the following heat soak period, which then reduces the amount of fuel leaking from the injector.

There are some basic diagrams covering fuel pressure target. It's based on requested engine torque--no surprises there.



Here you have engine speed on the X axis, torque request on the Y axis, and then target fuel pressure zones. 1 MPa = 10bar . This is certainly not the late D-4S fuel system with PFI + GDI simultaneous injection. It may not even be for a Toyota application. There are references to stratified combustion judgment and dual injection pulse judgment. It's possible this could be used on the 4GR-FSE (IS250 engine).

There are no diagrams showing how the spill valve timing syncs with the crank and cam position signals, and how that relates to injection timing calculations.

[lurker]

yes, that document was cited as an example. The ones dated 2008 by denso (around 12 of them) relate to the d4s as used in the IS-F (same components as FR). They go deeper into the scientific and mathemtic workings of them system, but dont delve to deep into pump timing and demand, well the ones i have studied thus far

[breezy]

Lurker, why does the pressure stroke need to occur when all injectors are closed?

The pump timing doesn't seem to have (or need) a direct relationship to injection timing.

My guess is: based on the Tq_requested vs Ne table, there is a target rail pressure that dictates the total on-time of the spill control valve (note: the timing at which the valve is energized remains constant, while the timing at which the valve is de-energized varies to control the fuel delivery volume). I would think this is a closed loop system, where a pressure sensor on the rail sends current rail pressure info to ECU, which compares this to the target fuel pressure based on Tq_requested and Ne. If current rail pressure is lower than target fuel pressure, extend the spill control valve on-time. Then read current rail pressure again and this becomes a repeating feedback, response, feedback, response, etc. loop.

[lurker]

Originally Posted by breezy

Lurker, why does the pressure stroke need to occur when all injectors are closed?

The pump timing doesn't seem to have (or need) a direct relationship to injection timing.

My guess is: based on the Tq_requested vs Ne table, there is a target rail pressure that dictates the total on-time of the spill control valve (note: the timing at which the valve is energized remains constant, while the timing at which the valve is de-energized varies to control the fuel delivery volume). I would think this is a closed loop system, where a pressure sensor on the rail sends current rail pressure info to ECU, which compares this to the target fuel pressure based on Tq_requested and Ne. If current rail pressure is lower than target fuel pressure, extend the spill control valve on-time. Then read current rail pressure again and this becomes a repeating feedback, response, feedback, response, etc. loop.

Your describing the very basic fundementals of the fuel system..
i.e. ECU demands higher fuel pressure, EDU controls pump to supply higher fuel pressure.


What I am talking about is the actual Control of the Pump. This pump isnt simply switched on or off when fuel pressure is needed at the rail.

Control of this pump is via a spill valve. The spill Valve can only be activated during a certain window of time during each revolution of the camshaft. The spill valve on/off condition determines if

a. Fuel is Drawn into the pump
b. Fuel is fed to the Rail.




In the Diagram above ive tried to make this a bit clearer.

The plunger in the pump moves up and down in a cordance with the lobes on the cam shaft.

During each revolution of the cam there is a certain window marked as X to Y in which a change of condition for the spill valve (to off) allows fuel to be sent to the rail. The spill valve off condition is met for a total of Z degrees

Now the only inputs the EDU recieves from the ECU are Injector control and Fuel pressure demand.

So in order to control the pump spill valve without any cam position information is via injector timing/opening/closing (the only information the EDU gets from the ECU) which would have a relation to cam position.

So in order to get the pump to function correctly, we must get the DI injector control as accurate as factory.

unless someone can correct me and say the have scoped the ECU or EDU and the Denso HPF pump has this switching inbuilt...

[breezy]

Lurker, your question is on how the EDU knows when (plunger position) to energize and de-energize the Solenoid Spill Valve?

I scanned the wiring diagram and noticed that you're right, the EDU seems to only receive pressure demand and injector control signals from the ECU.

I think there is a piece of the puzzle we are missing or not seeing currently. For the EDU to have to derive pump cam position from injector control doesn't seem like an efficient route to control the valve when the engine already has a cam or crank position sensor (which you could easily determine pump cam position from).

So I don't have an answer to your question, but to add to the discussion, I am familiar with a diesel high pressure pump by Denso that uses this same spill metering to control fuel delivery. The pump is much larger than this one used on the FA20, and features a cam position sensor to measure the position of the pump camshaft. I would think Denso would take a similar approach for this pump as well, though the sensor may be elsewhere or the signal derived from another sensor (potentially cam or crank position sensors).

[mad_sb]

Not sure if you guys have already reversed this info or not but Thought I would post it up as I am currently trying to understand the effects of DI Injection timing on torque output.. I have some dyno time next week and want to play around with the GDI Firing Angle.. However I'm not really sure what the degrees in the table represent, as it seems to contradict what was said earlier about the timing ranging from BTDC at high rpm to ATDC at low rpm




BTW, all of the GDI injection timing and ratio tables are set the same (hot, cold, warm, cold idle etc etc). After the reading I have done on DI, i thoght for sure the tables would vary so i was a bit surprised to find them all set the same.

[arghx7]

Originally Posted by mad_sb

Not sure if you guys have already reversed this info or not

I haven't seen this yet. Do you have any fuel pressure maps or anything else that may be relevant?

but Thought I would post it up as I am currently trying to understand the effects of DI Injection timing on torque output.. I have some dyno time next week and want to play around with the GDI Firing Angle..

If the timing is advanced too far, the fuel will hit the piston, with the worst case being an increased chance of preignition. If the timing is retarded too much, there's not enough time for mixture formation. Feel free to post the results of your testing for everyone's benefit. It might be a little difficult to get conclusive results. Remember, this engine was originally mapped on an engine dyno that could hit individual rpm and load points under controlled conditions. You can't realistically put a vehicle on a chassis dyno and hit 3600rpm at 1.10 grams/sec load, then try different 20 injection timing + port injection split combinations + valve timing combinations. That's what engine dynos are for.

However I'm not really sure what the degrees in the table represent,

It depends on how exactly the injection timing calculation is performed. Basically, those degrees probably don't represent actual physical crank angle degrees. There is probably an offset, say 40 or 60 degrees. The reason for this is the fact that the calculation is based on synchronization with the cam and crank position sensor from the last cycle. The sync point is, generally speaking, not at true TDC.

So, throwing some numbers out there: at low rpm low load (upper left portion of the map), we see a value of 320 degrees. Let's say there's a 40 degree offset. So in that case, 320 - 40 = 280 degrees BTDC start of injection. Or if the offset is 60 degrees, the SOI would be 260 degrees BTDC. Again, I made that 40 and 60 degree numbers up, but if you put an oscilloscope on the cam and crank sensors you could calculate the offset given enough other information.

At the lower right corner of the map, you have 370 degrees. That means SOI is 50 degrees earlier than at low load, low rpm (320 value). If the offset is 40 degrees, that would mean SOI at high rpm & high load is 370 - 40 = 330 degrees BTDC.

as it seems to contradict what was said earlier about the timing ranging from BTDC at high rpm to ATDC at low rpm

The trend of advancing SOI at higher rpm appears to remain the same, assuming we are calculating this right. It all comes down to the formula used... maybe you need to add or subtract 180 degrees or something to make it work. Whoever wrote the software and originally tuned these maps know. We have to make educated guesses.

BTW, all of the GDI injection timing and ratio tables are set the same (hot, cold, warm, cold idle etc etc). After the reading I have done on DI, i thoght for sure the tables would vary so i was a bit surprised to find them all set the same.

Those maps could be carried over from different GDI software for a different engine program and basically be cancelled out, and then a different type of compensation would be applied specific to this engine. It's more of an indication that we don't completely understand the software. It's not uncommon for stock ECU's to have lots of random & seemingly useless maps & constants in there; that's because it uses common software with other engine programs, and certain functionality is just turned off.

[Shankenstein]

Alot of GDI work is still being developed. If they found a setup that works, and the performance in cold temps is OK... they probably put that investigation on the back burner, leaving the flexibility built into the code's framework.

How certain are we that the "Total Injection Ratio GDI port" represents the percent of total fuel mass that is sent to the GDI system?

That table could easily be the "Late Injection Ratio" that tells the injector how to split early and late injection. To clarify, many systems use an early injection that will be pre-mixed by piston movement. When the spark ignites it, cylinder temps rise and a late injection pulse can auto-ignite.

Many systems just do 50/50 mass split, but there are times where it's better to run GDI in early injection only mode (when cylinder temps would be too low to auto-ignite).

My question - are there other tables that have been identified, related to early/late injection for the GDI system?

If so, ARGHX's theory about an offset may be correct... or it could just be mislabeled variables.

[Ryephile]

Originally Posted by mad_sb

Not sure if you guys have already reversed this info or not but Thought I would post it up as I am currently trying to understand the effects of DI Injection timing on torque output.. I have some dyno time next week and want to play around with the GDI Firing Angle.. However I'm not really sure what the degrees in the table represent, as it seems to contradict what was said earlier about the timing ranging from BTDC at high rpm to ATDC at low rpm

...

BTW, all of the GDI injection timing and ratio tables are set the same (hot, cold, warm, cold idle etc etc). After the reading I have done on DI, i thoght for sure the tables would vary so i was a bit surprised to find them all set the same.

I agree with what @arghx7 said; it's likey a timing based on a formula or relativism that isn't explicitly called out in the datalogging software. I don't see a contradiction, just 1 piece of the puzzle.

Regarding all the tables having the same entries; it's likely the datalogging software doesn't show if there are multipliers or modifiers based on sensor data to manipulate the outputs.

[arghx7]

How certain are we that the "Total Injection Ratio GDI port" represents the percent of total fuel mass that is sent to the GDI system?

It does correlate with the chart from the service manual in post #22 of this thread.

That table could easily be the "Late Injection Ratio" that tells the injector how to split early and late injection. To clarify, many systems use an early injection that will be pre-mixed by piston movement. When the spark ignites it, cylinder temps rise and a late injection pulse can auto-ignite.

There's not enough time for multiple direct injections at high rpm though. Even BMW goes to single injection at high rpm. Here's some info they've published about the N54 (which has very fast acting piezoelectric injectors). See attached image.

If so, ARGHX's theory about an offset may be correct... or it could just be mislabeled variables.

I have seen an offset used in other manufacturer's software, so there is precedent at least.

[mad_sb]

I do recall reading in one of the tech docs this is a single injection event system not multiple. I think I am going to start with the GDI Ratio... One of the posted papers from toyota on the combined system have claimed best torque at 30% DI 70% PFI for low rpm leading up to 60%DI 40% PFI at 4,000 rpm (right at the bottom of the torque dip and in the middle of DI only on the stock map).

[Shankenstein]

I had no idea triple injection was a thing. According to Delphi, the first injection happens during the upstroke. The second targets the piston bowl. The third is approximately at TDC when temp and pressure are high.

Apparently, it stratifies the mixture (makes some areas unused) and allows more aggressive combustion timings for a given noise output. With 60% of the mix in the second injection, they got ~50% thermal efficiency... which is pretty awesome (even with the PM issues, atleast it's not NOx).

Thanks for posting this up. LINK to Delphi paper. They did alot of single-cylinder testing and KIVA simulations. When I worked at University of Houston, I tried to get our KIVA program off the ground... but research funding was moving away from combustion modelling. Most of our stuff was live-engine with production sensors (and a few more).