RomRaider

I made this post responding to a different topic that I didn't realize was in the ECU Analysis forum, and while it did answer some questions people had in that thread, it probably isn't the right place. So I'm replicating that post here...

-----------------------

Injector Dynamics took away the graphs from their site that Jason used to estimate low pulse width info for the ID1000s, BUT they added something even better... Plug and play data for various ECU tuning software in Excel Spreadsheet form! Crazy awesome.

Of course nothing we use on a regular basis is listed, but the GM_EFI data has 60+ data points to describe the low pulse width characteristics of each injector. LOTS of data that can be used to model any other software. I modeled the data for ID1000 / ID1300 / ID2000 and what the values would translate to for 32bit and Carberry ROMs, using some of Jason's data from the ID1000 to fabricate / guess at / make engineering assumptions of data at voltages other than 14v.

These values are just theoretical at this point, I haven't tried them in a ROM. Note that the ID plug and play values use a small pulse threshold of 3.0 for ID1000 and 1300 and 3.25ms for ID2000. These are VERY considerably larger than I've seen for other setups, which seem to turn off corrections around 1.2-1.5 ms.

Incoming picture-bomb of the Excel sheets after I worked on them to translate them to our ROMs
First we have the GM_EFI_Live data ID provided for the ID1000, ID1300 and ID2000:

Looks like ID1300s use a little different strategy than the 1000/2000 with overdriving then settling from the negative direction.
Even though the % delta at most pulse widths is lower for the ID1300 and 2000 than the 1000, I put my idle data in the screen cap to demonstrate that the flow scaling of the larger injectors means they're still seeing significantly larger errors because they're operating at lower pulse widths than 1000s are at idle and cruise conditions.

Next low pulse data on ID1000s:


ID1300s:


And Finally ID2000s:


The primary issue with getting the high pulse width data well fit is that the ROM takes about 0.78% steps in the data. Both the CarBerry and 32-bit ROMs do this. I chose the breakpoints available in the ROMs and moved latencies around to get the best fit I could. I didn't do any math on overall fueling errors, so it probably could be better, but this eyeballs out pretty well with most areas seeing well under 0.5% overall error from the ID data. CarBerry gives enough values to get a VERY good fit. 32-bit has much fewer cells available, so the fit isn't as good, but I think I did pretty well given it has to be modeled with 8 data points.

The next question is how do we get other MFRs to provide this data... or where do we send other MFR injectors to get this kind of detailed data for a reasonable cost? I think it'd be adequate to have about half the data points ID provides (about every .1ms step size). As long as the equipment is repeatable enough.

Very interesting stuff!

I have the 2200s so this should come in handy.

Hey Concillian, do you mind doing a table for the DW1000 injectors? They also use bosch ev14 cores and are matched, but are slightly different to IDs.

Attached is the pdf document off DW's website.

Copy and paste for Carberry ID1000:

quick pass gives me this for 14v (8v - 12v and 16v are assumptions.):


The "phase" could be off depending on when the data really happens relative to the latency. That is these could be easily + / - 0.05ms of where I put them. Good news being that you don't really get below 0.7ish very often on this size injector, so it's a relatively small issue.

I imagine there is quite a bit of variation vendor to vendor given that they generally cut the tips off and put their own tips. This is sure to affect how the pintle actuates and settles, which I think is a reasonably large factor in how the injector behaves at low pulse widths.

Thanks!

Are you able to tell me or share with me how you came up with these numbers? Then I can hopefully, tidy up the table my self without bugging you again

Estimate latency at 13.5v (voltage used for test) by plotting out 8v --> 16v latency @ 3 bar, curve fit and interpolate.
Subtract latency from DW data. This becomes the new zero.

Then I plot that
Then I create 18 cells that are in the Carberry ROM. The lower % cells I put in just about all the breakpoints (which are every 0.78%), then the upper % cells can skip some based on the curve. Then change around latency for each cell so the overall error in fueling is minimized from an "eyeball average" of the DW data. Basically getting the red line (Carberry table data) overlaid on the DW data (blue) to match closely.

Haha...that is way too complicated for me...I might need to read more before I can understand.

Just wanted to say thanks for compiling this data. This should be a 'sticky' I think.

I did notice that the Carberry ROM doesn't let you enter in negative percent values for low IPW comp. Should we be able to get away with just using 0 instead and tuning around it?

Do you mind doing a table for the DW1300?

x2

Or maybe it is not needed? since the DW1300 is based on the stock Denso injectors?

Does anyone have a table for these injectors? Thanks!

http://www.fiveomotorsport.com/56010imp ... i-pigtails

I can put in negative values just fine in the ROMs I've tried (just tried again on 3.8.5 as I'm migrating my tune to that version.)

ID1300s are designed to need negative comps. They are clearly using an "overdrive" style of control vs. the ID1000 / 2000 "underdrive" style of control. The good thing about this is that the comps get close to zero very quickly. They have a large change in comps where the ID1000s and 2000s don't, but if you stay above these injector pulse widths, then you're pretty good to go. Even on pump gas idles, the open times are needing only -4% compensation, which means life should be pretty good even without any low pulse width compensation.

Find me data and I can put the values in for any injectors. Most MFRs don't provide this detail of data, so for the blackOPS or DW1300, I can't do the tables without someone sending injectors for detailed flow testing at fine steps. I believe DW recently started offering this as an optional service for an additional fee when doing flow testing.

The question is whether this actually works well for ID2000 style injectors that are a real problem here. It works fine for my ID1000s, but I never had any major issues with them to begin with. ID2000s are a different story.

And even though they're the same style as stock injectors, they may still need low pulse width compensation just because they're bigger. Let's say it takes 10cc of fuel to idle... 575cc injectors put out 10cc of fuel at almost 1.7% IDC, but 1300cc injectors put out 50cc of fuel at ~0.7% IDC These are totally different ms on the scale.

Hmmm, I just tried it with ECUFlash and you're right, I can put in negative values. I don't know what happened that made me say that. Thanks.