Shainiac

iTrader: (7)

Hi everyone,

I’m a bit confused with how the Required Fuel Value works with different sized primary/secondary injectors.
I have a MS3X V3.0 in my car and running well. I have an okay VE map that runs well enough to daily drive 50 miles a day and can hold a decent AFR under boost but idles poorly and has poor drivability at low PW.
Currently, I have 4 x 2200cc injectors firing in sequential.
I originally used a required fuel value of 2ms (Aaron recommended 8ms for 4 x 550cc, I have 4x the fuel). I’m also using Aaron’s recommended engine displacement value of 2600cc.

When I first started the car, the VE values were HALF of what I’d expect to get reasonable AFRs, so I doubled the required fuel to 4ms and my old MS2 VE map was pretty dang close. That's the value I'm using now.
I thought I remember seeing that in current firmware, rotaries now use the actual displacement of 1300cc. Is that correct?

Also, I just purchased 2 x ID1000 injectors to try to get a leaner and smoother idle and cruise. Since I’m keeping the 2 x 2200cc injectors in the secondary position, multiply my current required fuel x2.2 (2200cc/1000cc)?

To sum it up:
Is Required Fuel just tied to the primary injector size? What about the secondary size?
Is the correct Engine Displacement with current firmware 1300cc or 2600cc?

Attached is my MSQ for anyone interested.

If anyone has had luck idling and driving a car on 2200cc injectors, let me know. I can’t get it to idle much leaner than 12.5 stably and my high vacuum area is very erratic and required a ton of additional fueling to be happy. I’d like to keep the 2200s if possible since I’m already at 46% Duty Cycle on pump gas and haven’t really even turned the boost up yet. ID1000s won’t leave me any room to go E85.

mhr650

With the newer firmware you use 1300 not 2600 like you used to have to.

Shainiac

iTrader: (7)

Thank you. That explains why the suggested required fuel value was double what it needed.
I also found an earlier response from Aaron stating that the ReqFuel only references the Primary injectors.


Looks like I’ll change my displacement to 1300 and my required fuel to 2ms and see if it drives the same.
That answers both of my questions.

Now, does anyone have any advice to get 2200cc injectors to idle lean? I’m using Bosch NGI-2 injectors which are the base injector for the ID2000. The dead times and voltage offsets are well documented and that is what I’m using (0.55ms @ 13.2). I also “verified” the dead times are correct by putting the car in batch fire and switching between 2/4 squirts per cycle to see if there would be any change in AFR from having additional total dead time with more injector openings. It didn’t budge.
It has a pretty large street port, so I’m not sure if it’s all the exhaust gas dilution from the overlap making it run like crap, but, it won’t idle leaner than low 12s when fully warmed up. High vacuum cruise (30-40kpa) is very lean and stumbly. 2 x ID1000s /2 x 2200cc are going to be borderline for my goals, but I’d rather have the car drive nice and make a bit less power.

wankelbug

does this only apply to ms3 or ms2 as well with the latest firmware?

Shainiac

iTrader: (7)

So, I switched my displacement to 1300cc and my req fuel to 2ms and it ran WAY too lean.
I changed the req fuel back to 4 (still 2x what it should be?) and it ran like it used to, except my drivability is somewhat improved.
I still have trouble getting a lean idle and my high vacuum areas of the cruise map are still somewhat erratic, BUT It fixed an issue I had been having since the beginning.
I would get horrible lean spikes on WOT when spooling the turbo. I tried increasing TPSdot AE, MAPdot AE, upping the VE in the lower KPA boost cells, NOTHING worked consistently. Changing the displacement from 2600 to 1300 and my old TPSdot AE settings were pretty dang close.
I turned off MAPdot AE and leaned my 100-200KPA cells back out and it drives much better.

Aaron, would it be possible to update your writeup for the MS3X/MS3-Pro to use the current displacement. I believe 2600kpa was the correct displacement for the firmware when you wrote that, but I’m not sure when it changed.

Thanks,
Alex

arghx

iTrader: (3)

Displacement and req'd fuel are ultimately arbitrary numbers. What I mean by that is, pick the number that works best. Violate rules of thumb if you need to. The megasquirt code is still a very simplistic model of the engine. Just make it work.

First things first: remember that these old pre-Renesis rotaries are the equivalent of musclecars with aftermarket cams in terms of the port timing. They have a lot of overlap, and that recirculation of exhaust gas affects the idle quality. Porting makes it worse. On a stock engine, the air pump injects fresh air into the exhaust ports so that it is sucked back into the intake stroke during overlap. This stabilizes the combustion by diluting that exhaust gas. You're not going to run at a normal piston engine idle speed without an air pump, unless your idle speed is very high.

Speaking of idle speed, what idle speed are you running? The lower the idle, the less stable it will run, in general, and the richer it needs to be.

Did you play around with spark timing? Did you play around with injection timing?

Shainiac

iTrader: (7)

Thanks for the reply. I have had a couple street port engines and this one is by far the largest. That said, it is still a street port and I did not open the exhaust any later.

What I notice is that I can get it to idle as low as 600rpm, but at almost 10:1. My assumption was that this is as small a PW as I could get the injectors to open at. If I trimmed any more VE, it would swing massively lean/rich. I have the idle now set to 1000rpm when hot and AFRs in the high 11s-low 12s. I might be able to trim that a bit, but occasionally it'll hunt when I pull to a stop and the engine is heat soaked.

I believe I'm idling the engine around 18*. If I go less, it idles lower and at less kpa, but I can get it leaner. Leaner meaning higher AFR, but it's usually the same or higher duty cycle also lol.

As far as injector timing? I have no clue what to use. I read somewhere that Haltech recommends 270* For rotaries. I tried for a while, then 300, then 330. I saw little difference at all. If anyone has any insight on that, I'd love to try something different.

Aaron Cake

iTrader: (1)

With 2200CC injectors, you're looking at about a 0.2 - 0.4 mS pulsewidth to inject the correct amount of fuel. The injector is probably not physically capable of this.

Shainiac

iTrader: (7)

My data logs are showing approximately 1.5ms with a idle AFR of 11.7 at 1000 rpm. This is pretty close to what I was seeing with my 1100cc primaries on MS2 in batch fire also.

arghx

iTrader: (3)

Remember that the stock idle timing is -5L and -20T (when it is in a fixed mode). Granted that is for stock ports, but a lot of people forget how the stock timing works. So you are advanced 23 degrees from stock. The idea is that I get more torque with more spark timing, which makes sense. However if you look at stock tunes on cars in general, they are always retarded from MBT spark. Part of that is for emissions, but the other part of that is for idle quality, as strange as that may seem.

Basically it's retarded spark, requiring more air for baseline idle. Then spark can be quickly advanced if there's additional load on the engine or some sudden instability. It takes longer for airflow to catch up (changing idle air control valve or electronic throttle, due to time delay of manifold filling). Now there's more than one way to skin a cat, but if we use these stock tunes as inspiration, here's something to try:

You're at 1000rpm right now. Walk your leading spark timing down in 5 degree increments. Try to get anywhere from 0 to 10 degrees leading. Split doesn't matter that much but you can do 20 split like stock if you want to. If your idle drops below 800, add more air through the idle speed control system (not sure if you deleted your ISC/BAC valve) or by adjusting the bypass air screw if you have one.

Even if your injection pulsewidth doesn't go down and the fuel mass injected is about the same, you may be getting leaner as measured by a wideband because the engine is burning better (more CO2 emission, less CO and HC emission, indicating higher quality combustion and fewer misfiring cycles).

arghx

iTrader: (3)

In your sequential settings, are you set to end of squirt? If not, change it to end of squirt. This isn't a huge **** to turn but sometimes you can get a better lean limit at idle and low load with different injection timing.

Here's some background on injection timing I have mentioned in the past:



The diagram above shows intake port opening area on a 4 port rotary engine (stock ports) vs crank angle degrees. From left, the four strokes are: expansion, exhaust, intake, and compression. The crank angle degrees are in this case numbered degrees BTDC firing, the same units used for spark timing maps. This is analogous to a cam lift vs crank angle degree diagram on a piston engine (shown here with a lot of overlap):



Keep in mind that piston engine has 180 degree strokes while a rotary has 270 degrees. That diagram is counting forwards from TDC firing, rather than using a BTDC numbering like spark usually does. Some ECUs do it that way for injection timing.

If the injection timing is based on start of injection, you locate the crank angle degrees (assuming it's BTDC firing like spark timing) listed. Then you move to the right in the chart based on how long the injection pulsewidth is (pulsewidth must be converted to eshaft degrees based on time and engine speed). If it's end of injection, you move to the left.

There are three basic types of injection timing strategies to consider: low load, open port injection (open valve on piston engines), low load closed port injection (closed valve on pistons), and high load/high fuel flow injection (high duty cycle).

With open port/open valve injection, air is rushing into the intake port as the injector sprays. The idea is for the air motion to help mixing. With closed port/closed valve injection, the fuel has time to sit in the intake port and vaporize before the port/valve opens. The best strategy depends on the engine. In general, if you don’t care about emissions, moving the injection timing around at low loads is something you do for smoother idle and more stable combustion.

At high injector duty cycles, the injector is spraying so long that it doesn’t matter much what the injection timing is. If the injector hardly turns off (90% + duty cycle, meaning almost the full crank angle range) it doesn’t make much of a difference.

If I have an End of Injection calculation, like a Haltech or an Rx-8 stock ECU, you will see the number 270 BTDC thrown around a lot. This is the end of the intake stroke, meaning that the ECU counts backwards from there according to the injection pulsewidth (on-time) and maybe an injection advance map, typically advancing end of injection a bit at higher rpm. Here is a diagram for the Rx-8:



This means that the Rx-8 starts at BDC compression (right before entering the compression stroke) to determine the injection end point. If the engine speed is higher, it will end earlier (or is it later? what does "lower" mean? bad translation). To figure out the start of injection time, it counts backwards from the end point based on how long the pulsewidth is. The result is mostly open-port injection. As mentioned before, at high injector duty cycles the timing doesn’t matter a whole lot because the injector hardly turns off.

FYI for anyone curious: direct injection engines (currently only in production for pistons) are very different from this.

Shainiac

iTrader: (7)

Thanks, that's a really good explanation of how the injector timing works. I had originally use 270 based on what I had read from Haltech guys. I saw some different values on other people's MSQs and had been experimenting but haven't noticed a huge difference. Do you think it'd be worth adding a few degrees to the 270* value since my engine's ported?

I did, however, make a huge improvement in my AFR stability at low load/high vacuum. I noticed that my TPS signal looked "fuzzy" in datalogs sometimes if I zoomed all the way out. at low TPS positions, there seems to be noise in the signal. It will fluctuated +/- 2% or so. What I didn't consider was the TPSdot. It was causing Accel Enrichment to engage rapidly and repeatedly, making the AFR's swing rich/lean, rich/lean rhythmically. I upped the TPSdot lower limit in Accel Enrichment and my AFR's are rock solid. I still get misfires at very high vacuum, but the AFRs don't change now.

That said, my fuel mileage still SUCKS. I did 100 miles on the interstate last night averaging 70mph and returned 14.9 MPG at 14.5:1 AFRs. It was somewhat hilly ground, but nothing treacherous.

Granted, it was night time and I had my headlights up. I doubt EPA added that to the average Mazda put on window stickers. I also have a JDM trans with the 0.806:1 5th gear, so its humming along at about 3200 rpm. My EGTs were also around 1550 at cruise, which seems excessive.

Funny enough, 50 miles of my drive last night was without the TPSdot corrected and it returned slightly better fuel economy lol.

arghx

iTrader: (3)

Take a step back and check your sequential settings. Does it say "end of squirt" or not? If it says end of squirt, you're spraying into an open intake port. If it's start of squirt (maybe even middle of squirt), you're spraying right after the intake port closes and the fuel is sitting in the port until the next cycle.

Set it to end of squirt. Start at 270. Lean the VE table to lean limit at idle - right before it gets unstable. So if it's unstable at 12.2:1 , set it to 12.1:1 ish (this is not exact, but you get the idea). Adjust end of injection timing. Try 30 degree increments in either direction, assuming it's not very sensitive. With a fixed spark timing, see if you can lean it further. Go about 90 degrees in each direction, so:

270
300
240
330
210
360
180

Save datalog.

Begin new datalog. Pull 5 degrees of spark timing and do the above injection timing sweep again. Save datalog. Begin new datalog. Pull another 5 degrees, sweep injection timing again. If that doesn't do anything, you're probably barking up the wrong tree, ie you've tested and determined that the engine is insensitive to injection timing at idle and that's the result of your study. I realize the engine will heat soak a bit as you sit there idling, but there's not so much you can do about it.

If that sounds repetitive/time consuming, remember that if you take it to a pro, unless you give them a blank check/pay hourly they will never give you that level of detail to your tune, because they would lose money. So be happy you can be so thorough.

Shainiac

iTrader: (7)

Sorry, I forgot to include that I am using end of squirt and have been since the beginning.

I'll try to test different timings with locked ignition timing. But on a day when it isn't 95F and beating sun lol

arghx

iTrader: (3)

The basic EPA city and highway cycles date back to the 1960s and do not have any electrical load on it. The newer tests include a cycle which has A/C and electrical load, as well as more aggressive driving, but it's weighted into the total results. The EPA applied a correction factor to the window sticker values of older cars a few years ago. Rx-7's all went down, which should be no surprise.

The current corrected EPA values are 15 city, 22 highway, 17 combined for an 87 T2. So you are achieving EPA city mileage. That's not that surprising, considering you've got an ECU which probably over enriches on tip in compared to stock, and you've got no airpump and more overlap, and you're flowing more air when you get into the throttle due to mods.

Shainiac

iTrader: (7)

Right, and I'm sure their mileage is based on a 55mph highway speed limit, not over 70 and with a sorter gear ratio. Like I said, I'm sure the headlights are terrible for aero and the fact that I have much wider tires and wheels doesn't help. I was really hoping that switching to sequential injection would see some improvement though. This is basically the mileage I was getting with worse injectors in batch fire.

elturbonitroso

this is where having the right deadtime comes to play.

Shainiac

iTrader: (7)

I'm using the advertised deadtimes for the injectors. They are Bosch NGI-2 injectors, same specs as ID2000s. I'm using ID's voltage curve for dead times as well. 13.2V is 0.55ms dead time. I'm starting to think it's just the overlap of the ports that make idle unhappy.

elturbonitroso

are you using the voltage curve they have specifics for ms.

Shainiac

iTrader: (7)

Yes, I took Injector Dynamics voltage curve and extrapolated it out to what % of the latency at 13.2V was. I've attached a screen shot of the graph I made. ID2000s are the curved blue line. The green line is 13.2V which is the value TS uses as the base dead time.

Aaron Cake

iTrader: (1)

Ha! I wrote 0.2 - 0.4 which of course was meant to be 1.2 - 1.4.

Saying that once you get down to duty cycles that low, injectors become rather non-linear and with injectors that large, you don't have very much resolution to begin with.

I don't remember what your porting is but if is the typical street port you may not be able to idle it much leaner than 12s. Try around 5 degrees of timing to start with at idle.

arghx gave good advice on trying to get your injector timing closer. You wan to set up that fuel cloud when the port is closed, not fire into an open port.

Basically what I do is get a stable idle then start moving injector timing until the AFRs begin to richen up. Then pull out that fuel and keep moving the timing. At some point you will reach the opposite...it will begin leaning out again as the timing isn't optimal for atomization. Back off a bit and you've found your injector timing...for that RPM and load anyway.

Do you have the primary injector air bleeds hooked up and functional? They make an enormous difference.

Shainiac

iTrader: (7)

Hi Aaron,

I ported my engine with a Pineapple Racing "Large" streetport template.
The exhaust ports I opened approx 3mm earlier and cleaned up a bit. Still stock closing timing.

I do not have air bleeds or diffusers. I am using Full Function Engineering's billet o-ring'd inserts for Injector Dynamic style injectors. The EV14s have much better atomization than old EV1 pintle-style injectors, so I'm not sure how critical that is.

Sooooo, I had a look at my timing map and I'm running 22* at idle. I recall setting the timing at first startup to get the max vacuum and that's where I left it. I'm using the FFE 36-1 trigger kit and timing has been verified with locked timing at -5*

I tried tonight to play with timing and get the idle to be a bit leaner. I was able to get a steady 12.5:1 idle at around 850, but I noticed it would want to stall when trying to recover from blips of the throttle. This is because of a couple reasons. I had to increase the AccelEnrichment minimum to 100 %/s because I have noise in the TPS signal and AE was activating at steady low throttle on the highway. The signal was oscillating +/- a few percent, but at a high enough frequency to activate AE. Increasing the minimum value to 100 %/s masked that issue, but now at throttle tip in at low rpm, it goes lean and will stall. Not exactly sure how to eliminate the noise, since I have the TPSref 5v wiring going to 5 different sensors lol.

Long story short, I set timing to 17* to put a little more load on the engine, and I guess I'll live with 12.5:1 AFR at idle.
If anyone has ideas on how to fix noise in the TPS, I'm open to ideas. I believe I may have noise in the CLT also as it seems to move around quicker than I'd expect. MAT seems stable, oddly.

A log is attached.

Aaron Cake

iTrader: (1)

Sounds about the same as the last street port FC I tuned with an MS3X then. RB big street port templates, RB exhaust ports (opens earlier, closes stock). Primaries ported out to roughly RX-5 Cosmo timings. Ended up that the best idle was in the high 12s w/15 or so degrees of timing.

They are critical. Even the RX-8 with the new style injectors improved on the bleeds by moving them from the diffuser to a small tube sticking out of the port runner actually blowing onto the injector nozzle.

As a point of reference, I know someone who did something similar to make adapters to put top feed injectors on an FD. Street port to the same timings. The adapters eliminated the bleeds. The car will not idle leaner than about 11:0 no matter what is done. Injector firing angle, timing, nothing. One can actually see the runner load up with fuel on the wideband. Without injector time increasing, AFRs will begin to dip until the BAC opens more to compensate for the drop in engine speed, which will lean it back out to the 11s, and then the whole process will repeat.

Your TPS has a little jitter, but about what I would expect, it not just slightly noisy.

Your CLT is all over the map. As in I see areas where it reads 187 then the next capture it reads 180. A 7 degree change between snapshots. You need to check the wiring to the sensor and the sensor itself. Seems like it may be intermittent.

Did you ground all sensors to the dedicated sensor ground?

Also, from the sections of the log I looked at, I didn't see any TPS based AE events. All MAP based events. Your AE seems to be configured to trigger on manifold pressure. Not a setting I would recommend unless we are talking an S4 TPS or you don't have a TPS.

Shainiac

iTrader: (7)

Aaron,

The ports are one of the more aggressive templates around, without bordering on eating side seals. I believe 1000rpm idle is around 60kpa.

At this point, I'm not sure I can salvage the air bleeds using the FFE inserts. I had to modify my injectors by clipping off a plastic o-ring retainer, so they no longer work with Denso "donuts".

I think I'll just settle for 12.5:1.

I'm not sure why the CLT is so jumpy. I do have every sensor tied into Sensor Ground and the entire MS grounded to the battery. It's a brand new AC Delco Corvette sensor which has the same cal and the 3/8" NPT GM sensors, but the same thread as the OEM Mazda sensor. A new one is inexpensive, so I may try that first. The connector was a brand new crimp-on connector from DIYAutotune.

As for AE, I had the MAP>TPS slider set to 99.8% TPS instead of 100%. But I had all my MAP-dot adders to 0, so it still should have been all TPS enrichment. See MSQ attached.