No start fresh build R33 GTST RB20DE Neo+T AEM Series 2 ECU 30-6620

[Neostead2000]

Heya people from down under,

As the title states, my project is at a point where I finally could crank the motor over to attempt a first start. I'm running an AEM Series 2 standalone ECU with a 33 connector which I repinned myself, car originally was spec 2 but the motor I put in was an RB20DE Neo from an R34. I've configured it to run via Speed Density using an AEM 3.5 bar MAP and IAT sensor. It's using the stock Neo coils and the stock Neo Hitachi plastic top CAS (360 slots for crank and 6 slots for cam ref). I have made extra certain that my re-wire and custom wiring suited the Neo but also copied a spec 2 harness which was given to me to compare against. I have double triple checked all the sensors and wiring which go to the ECU etc and they all follow the required inputs for the AEM ECU as per their pinout.

 

For the last 4 days I have been trying to start this motor with no success. I have amazing oil pressure, the walbro 255 primes and turns on while cranking, I get 3 bar of pressure at the regulator. I also checked #1 coil for spark, it does spark albeit slightly weak looking and orange in color (as opposed to bright white/blue). We also noticed that the injectors (R35 GTR 570cc) don't seem to be opening. I tried using a spare injector outside using brake cleaner through it and having it plugged into the loom, the motor cranks but the injector doesn't open. We've probably cranked it over 100 times by now (bad I know!) and the cylinders aren't even remotely flooded, mostly dry. That said, I suspect it's an issue with spark and fuel altogether.

 

The AEM EMS setup wizard has an option for "Stock RB 6 cylinder CAS wheel" as well as an option for "Stock R33/R34 Series 2 coils" both of which I have enabled and it apparently automatically sets the required dwell times and syncs. I do get a "CAS SYNC ON" status while cranking and I do get engine RPM signal (up to 400 or thereabout) so I know the eCU is receiving those signals from the CAS. BUt the timing is completely whack. With a timing light, with no adjustment of the electrical timing, and CAS set at the middle point, using a good timing light, the crank marks appear nearlt 60 degrees retarded. And even if I max out the crank advance setting (+73deg) and lock the base timing at 10 BTDC, the timing light still only gets to about 15 deg advanced of the 0 point on the crank. Basically it's erratic and not stable. I was very careful when timing the motor and turned it over by hand a few times to double check everything lined up.

Also the motor cranks extremely well, fast and strong. No banging or screeching or huffing. She's thirsting to fire but cannot for some reason. I've got experience tuning my Mitsubishi with another tuning platform but this AEM is kind of overwhelming and I fear I need some help. If anybody has an idea or pointers at what I could look at, please I would be super appreciative! I could also provide screenshots of the advanced pickups tab or anything else.

[GTSBoy]

I know nothing of the AEM ECU. But....is it a modern ECU? As in, does it have a built in oscilloscope so you can inspect the signal waveform from the CAS?

Do you have access to a handheld oscilloscope? They're <$200 from Aliexpress. You can use these to watch ECU input signals and output signals. So you can watch the pulse widths on the injectors. If you're not getting at least a recognisable pulse at the iinjector, then.... that would explain the no fuel. A noid light is the older way to do this.

And....weak spark sounds like it is a part of the problem. The question obviously has to be, "why is it weak?". Do you have power delivery problems at the coils? Do you have a bad earth? Is the ECU crook in some way? Does it just come down to not having a good signal from the CAS (I'd hate to think so.)?

Have you verified that the timing marks on the front are actually in the correct relationship to TDC? No point worrying about being 60° retarded if its just the pulley that has slipped. Then you have other things to worry about!

 

[WMDC35]

Assuming it's not the very late s2 wiring diagram/harness you've used, you've reversed the wiring for the cas from s2 to suit neo? 

Edited October 4, 2023 by WMDC35

[Neostead2000]

WMDC, the harness that was given to me for comparison was off a 1997 spec 2 gts4t poor man's GTR. I've made the pins correspond to a spec 2 wiring schematic while retaining all of the Neo features including the swapped Neo CAS wires. The 120 deg (cam) and 1 deg (crank) signal wires go to pins 41+51 and 42+52 respectively.

I've mentioned this before in another thread but I also had to adapt an RB25DET Neo Gloria engine harness to fit all this, and that one came without the ECCS relay next to the ECU like it is on a spec 2 harness, so I had to wire in a relay like that following the spec 2 diagram from that nice R33 schematic that can be found here.

 

GTSBoy, it may have an oscilloscope built in but at this point I am still getting to grips with it and figuring it out. It does show me cam/crank pulses in miliseconds, sync, tooth errors and all that. I've done a big 3 upgrade on the wiring and I'm certain this car has great power and ground, have also tested for electrical leaks and it's as clean as a whistle. All my joints were soldered, heat shrunk etc, no expenses spared.
 

I have verified that the motor is at TDC when crank pulley reads 0 (first mark). I also turned the cas by hand and it triggers the fuel pump and spark. In fact by hand trying to spin it, i am triggering those events much quicker than by cranking (logical I know).

 

I got to a dead end today (AGAIN!) and ended up calling one of the more renown tuners from my country who has experience with AEM and some nissan motors. He was very helpful and dedicated an entire hour remotely connecting to my laptop and messing around with the settings. The closest we could get the timing synced  was 5 deg retarded. 1 degree more than that, and the timing light started showing 60 degrees ADVANCED which made no hecking sense! He said if he kept trying he might be able to get it to run but highly recommended I make my own trigger wheel replica of AEM's and try using it. So I just drew one up on Solidworks and tomorrow I will look for the right material because the stock neo disc is I shit you not, 0.13mm thick! I think anything more than 0.2mm and it may hit the sensor head.

 

Also I will try running direct 12v to the ECCS relay and see if I get stronger spark and fuel when cranking just to rule out the relay or wiring. I will keep reporting back but in the meantime if anybody has more ideas I'm all ears.

[GTSBoy]

5 hours ago, Neostead2000 said:

Also I will try running direct 12v to the ECCS relay

There is a whole separate relay for the ignition coil supply.

[Neostead2000]

On 10/5/2023 at 2:48 AM, GTSBoy said:

There is a whole separate relay for the ignition coil supply.

You're right, I found it, the 1st one from top down to attached to the fusebox under the steering wheel. It functions and gets warm when I keep cranking the car.

I drew up the CAS disc no big deal, but I phoned up 20 companies and nobody has material that thin, not even 0.2mm aluminum or steel. So that's out of the question. And I called up some parts suppliers, nobody has the AEM wheel in stock, needs to be ordered from the USA...

I did get some progress made today but it was more me understanding how things worked and not so much making the car start. I took all 6 coils out on top of the manifold (without removing the neo Y pipe, im a master on this engine now) and put the original iridium spark plug in each coil.  I took the cas in my hand and started spinning it to double check if  I have synchronous ignition, and I do. I took a slow-mo clip of all 6 and here is how the ignition goes as I keep turning the cas:

Cyl 3 & 4 fire at the same time once
Cyl 2 & 5 fire at the same time once
Cyl 2 & 5 fire together once more
Cyl 1 & 6 fire at the same time once

Then the cycle repeats. I have no idea why 2 & 5 fire twice consecutively but this is my first inline 6 engine so maybe that's just how a straight six fires? I can upload the slowmo vid if anyone is curious. I could also hear all the injectors ticking away but I had pulled the fuel pump 10A fuse so I don't flood the motor.

Then we tried taking the 3 CAS bolts out and manually rotating it outside of its factory range, to see if we can get the timing closer to 10 BTDC (stock for 20neo). If we rotated the CAS clockwise (ignition RETARD) maybe a good 20-30 degrees outside of its normal range, and if I input "-10" on crank advance degrees within AEMTuner, we got close to 5 degrees retarded of TDC. But this still isn't good since the CAS could not be bolted on properly. Tomorrow we will experiment a bit more with this. Also we were using an old spark plug wire fed into the #1 coil and then going to #1 spark plug, with the timing light sensor hooked onto it as I know this is a more accurate way of measuring instead of at the blue loop at the back of the ignition loom.

 

[GTSBoy]

Firing order is 1-5-3-6-2-4

Something is wrong. 2 & 5 pulses should be separated by a 3&6 and a 1&4. But you will note that 1 & 4 are at either end of the sequence, so....they should happen consecutively. I reckon you have 2&5 and 1&4 outputs crossed up.

 

 

[Neostead2000]

8 hours ago, GTSBoy said:

Firing order is 1-5-3-6-2-4

Something is wrong. 2 & 5 pulses should be separated by a 3&6 and a 1&4. But you will note that 1 & 4 are at either end of the sequence, so....they should happen consecutively. I reckon you have 2&5 and 1&4 outputs crossed up.

 

 

I have traced each coil signal wire to its corresponding pin at the ECU, and they all match properly. I've double checked that my Crank 1* signal goes to pins pins 42+52 and my Cam 120 deg signal goes to pins 41+51. I cannot find an option within AEMTuner v3.2 to switch from batch fire to sequential like it should be. I'm literally stumped.

[Neostead2000]

Edit: After spending the entire Saturday reading about cam synchronizing and discussing it with a great fella from NZ (props to Zeth) as well as reading through the inbuilt AEMtuner info boxes, I have a slightly better understanding of what may be happening.

 

The AEM EMS is hard coded to read a 24 tooth wheel for 1 full engine cycle (720*), or 12 teeth per 360* revolution, period. AEM have inserted a "stock 6 cyl Nissan CAM disc" option for some reason which loads up with 6 teeth for spark/fuel/wheel. When you view information about the "Spark Teeth" option, AEM say: "Is usually set to the same value as option Wheel Teeth for sequential ignition and set to half of Wheel Teeth for waste spark ignition. "

So, if you wanted to run waste spark/batch fire, you would set the spark teeth to let's say 12, and wheel teeth to 24. Otherwise if both options have an equal amount of teeth input, it should work as sequential. ODDLY, the stock setting of 6-6-6 (teeth) right now on my car makes it do wasted spark, but that is because the magic behind all the calculations and the way AEM have built their transcoders or whatever they're called. In other words, these coils will always do batch fire if you run the stock cas, simply no way around it. And that is if you manage to even start the car at all! This explains why my coils are firing the way they are.

There is still much to learn, but I have already lost an entire week trying to make it run. It almost looks like there is no way to make the AEM read the stock CAS properly, so I will have to resort to making my own trigger disc by getting the thinnest piece of metal I can, then manually lapping its face for hours until I get the desired thickness. I've been told by an AEM user that he's done this to at least 5 different cars with custom trigger discs and they're still running flawless.

I'm writing all this so that people (like me) in the future could have an up-to-date reference about this issue without having to read 20 different posts and talk to 10 different people and watch countless youtube videos.
 

Will keep updating!

[GTSBoy]

In all seriousness, at this point, just throw the Nissan CAS in the shitcan, where it belongs, and replace with a proper trigger setup. There's so many available off the shelf now, and there's really no reason to make yourself suffer with the stock CAS when you have an aftermarket ECU that doesn't need it. Timing will be much more stable with a proper trigger setup.

[Neostead2000]

HUGE UPDATE!!! SHE STARTED!

In the most random fashion, the universe sent a savior right to my doorstep. A guy showed up who apparently doesn't even do cars for a living but he has experience with AEM 1 and 2. He took a quick look at my setup and found 2 problems right away. Firstly he said I needed to change my windows settings to allow the decimal  point to be "." instead of ",", then to allow it to run up to 2 decimals. Apparently this makes AEMTuner let you do precise numbers and various other things.

Secondly, he said I couldn't set the timing right because I was using the wrong resolution for the teeth sync. Meaning, since the EMS expects 6 teeth, the synchronization will be between 0 and 6 teeth maximum. And what I was doing until then was inputting big values such as 20, 50, 100, 180, 256 even (max value) and always wondering why the timing was either +60 deg advanced/retarded or marks not even visible...

So I set my "Ignition sync" option to 1 and started measuring the timing while cranking. We slowly upped that value to 1.395 teeth which essentially tells the ECU that the CAM reference signal is 1.395 (out of the 6) teeth away from true sync. At that point, the timing was very close to 0, I think we left it at 5* retarded of TDC. The guy was 100% certain that it should fire up if we got the timing at least close to 10 BTDC. I honestly was not expecting it to start up so I didn't even have my phone in my hand, but as my friend went to crank it one last time, the lil' RB came to life like it was nobody's business!!!! It started so nicely and quick, and it held I think 1600 RPM super smoothly.

We quickly saw that there were a few minor issues such as an oil leak at the brass T fitting for the oil feed, #4 injector not firing (tested it in different ways, we suspect its blocked), and it seems i have overtightened all the belts so they were squealing. But she fired up and ran, that's what matters. From now on I can take it from there and set idle tune, AFRs and take care of the belts and leak.

Additionally, Zeth from NZ worked out that the injector/coil phasing was in the right order, but the tooth count was incorrect (set incorrectly by AEM, as with many other things, which I highly suspect was so that you are forced to buy their disc anyway). He told me to set my phasings from 0 to 6, again, using the logic that there are only 6 physical slots from which the ECU reads from.

So, for anyone reading who wants to start his RB on an AEM Series 2 ecu with STOCK 360+6 Nissan CAS wheel, here are the settings which hopefully will help you out and save you time and frustration!



 

For the phasings, RB fire order is 1-5-3-6-2-4, but the phasings must be like so in order to retain that order:

Inj/Coil #1  = 0.000

Inj/Coil #2  = 4.000

Inj/Coil #3  = 2.000

Inj/Coil #4  = 5.000

Inj/Coil #5  = 1.000

Inj/Coil #6  = 3.000

 

Then you start cranking your motor with the timing light (blue loop on the back of the harness is perfectly fine) and start increasing the "Ignition Sync" value slowly. Maybe start with 0.5 and go up to 1-1.2 which should make the balancer marks within view. Then adjust it by the decimals so 1.255, 1.258 etc until you get it to your desired timing. For a 20 Neo it's 10 BTDC, for a 25 I think it's 15. You may also have to adjust your CAS clockwise (retard) or anti clockwise (advance) to get it super close.

That said, I'm aware this is NOT the optimal setup since it's still doing batch fire which can get sketchy and I assume I may have issues at very high RPMs, but for the time being, the important thing is the car can start and most probably drive fine with more tweaks to the maps. I will end up making my own CAS wheel once I get the material.

I hope this will help somebody 👌

[Duncan]

Great to hear you got this running....but what I really took from this don't try and use AEM with Nissan CAS....

[GTSBoy]

Yeah. I still don't know the AEM ECu from a bar of soap, but.....I'm decreasingly likely to ever want to.

But definitely congrats on getting it going.

[joshuaho96]

19 hours ago, GTSBoy said:

Yeah. I still don't know the AEM ECu from a bar of soap, but.....I'm decreasingly likely to ever want to.

But definitely congrats on getting it going.

AEM hasn't even supported the series 2 in a very, very long time. To my knowledge it doesn't even support speed density. AEM has completely exited the ECU market as well. The last ECU they shipped was the Infinity and frankly speaking it was a buggy mess, the linked thread shows some things but my personal experience with it was not great either: https://www.miataturbo.net/aem-59/aem-infinity-piece-junk-lets-find-out-together-91346/ 

[Neostead2000]

It very much supports Speed Density and that's how I have it configured and will be tuning the car. You need an accurate and working IAT sensor in order to get the right density calculations, though. I opted for a BMW E46 IAT M10x1.5 threaded sensor since that one is identical to the one used on an R33 GT-R, direct replacement.

[joshuaho96]

On 10/9/2023 at 11:38 AM, Neostead2000 said:

It very much supports Speed Density and that's how I have it configured and will be tuning the car. You need an accurate and working IAT sensor in order to get the right density calculations, though. I opted for a BMW E46 IAT M10x1.5 threaded sensor since that one is identical to the one used on an R33 GT-R, direct replacement.

I'd have to look at the series 2 in specific but when I worked with the EMS-4 in AEMTuner it did not work in speed density. Yes it uses manifold pressure as a load signal and there is an IAT/CLT sensor that is used to adjust fueling but all the tables ultimately operate in raw injector pulse width, it has no conception of volumetric efficiency. 

[GTSBoy]

11 minutes ago, joshuaho96 said:

I'd have to look at the series 2 in specific but when I worked with the EMS-4 in AEMTuner it did not work in speed density. Yes it uses manifold pressure as a load signal and there is an IAT/CLT sensor that is used to adjust fueling but all the tables ultimately operate in raw injector pulse width, it has no conception of volumetric efficiency. 

What you talking about Willis? Volumetric efficiency is not a requirement to implement speed-density.

So, the outputs of the injection table are in raw ms. So what? So have almost all older "speed-density" ECUs going back into the 1980s. We all used to tune the by adjusting ms values in the fuelling table. That has absolutely nothing to do with whether the ECU is using speed-density to work out where on the map it is.

Speed is engine speed. No reason to keep talking about that.

Density is density - if you want it to be in units that relate directly to density. But if you don't want to, you can just use MAP in pressure units with a temperature compensation. You can apply the temperature compensation before you go reference the map axes, or you could do it afterwards, depending on the way your brain works and whether you care if the load axis being not quite right is going to upset people (which it would be not quite right if you applied the temp correction to the pressure before you looked up). The pressure, once temperature compensated, is absolutely equivalent to an outright density number, and serves the purpose of looking up the fuelling map as well as any other load index. There are no calculations required by the ECU or by the tuner to work out what that manifold pressure at that rpm needs in terms of fuelling. That's what the dyno and the wideband are for. The fuelling model is a function of reality.

Back in the early 90s we were tuning Microtech LTXs and the like, with psi on the load axis and a temperature correction applied to the resulting ms value that came out of the table. Was it speed-density? Well, not if you require your definition of speed-density to have actual density calculated and some sort of VE model exposed in a table. But was it using the density of the air in the manifold as the (effective) load index? You bet it was. The VE table is encapsulated in the raw ms values. Is a given cell at 90% VE? 95%? 112%? Dunno. Doesn't matter. It's at 10.7 ms and that's the right fuelling at that load point regardless of what the VE actually is.

 

[joshuaho96]

56 minutes ago, GTSBoy said:

What you talking about Willis? Volumetric efficiency is not a requirement to implement speed-density.

So, the outputs of the injection table are in raw ms. So what? So have almost all older "speed-density" ECUs going back into the 1980s. We all used to tune the by adjusting ms values in the fuelling table. That has absolutely nothing to do with whether the ECU is using speed-density to work out where on the map it is.

Speed is engine speed. No reason to keep talking about that.

Density is density - if you want it to be in units that relate directly to density. But if you don't want to, you can just use MAP in pressure units with a temperature compensation. You can apply the temperature compensation before you go reference the map axes, or you could do it afterwards, depending on the way your brain works and whether you care if the load axis being not quite right is going to upset people (which it would be not quite right if you applied the temp correction to the pressure before you looked up). The pressure, once temperature compensated, is absolutely equivalent to an outright density number, and serves the purpose of looking up the fuelling map as well as any other load index. There are no calculations required by the ECU or by the tuner to work out what that manifold pressure at that rpm needs in terms of fuelling. That's what the dyno and the wideband are for. The fuelling model is a function of reality.

Back in the early 90s we were tuning Microtech LTXs and the like, with psi on the load axis and a temperature correction applied to the resulting ms value that came out of the table. Was it speed-density? Well, not if you require your definition of speed-density to have actual density calculated and some sort of VE model exposed in a table. But was it using the density of the air in the manifold as the (effective) load index? You bet it was. The VE table is encapsulated in the raw ms values. Is a given cell at 90% VE? 95%? 112%? Dunno. Doesn't matter. It's at 10.7 ms and that's the right fuelling at that load point regardless of what the VE actually is.

 

I guess the correct distinction is VE vs injection time, regardless working in injection time is just kind of a pain. It's kind of like old style transient fueling vs x-tau.

[GTSBoy]

From a tuning perspective, there is no difference in pressing pgup or pgdn on a cell reading ms or % or some arbitrary index number (ie 0-255). Who cares? Number get bigger, fuel gets morer, 3D view of map gets taller, colours change appropriately.

[Neostead2000]

That's actually a good point you two discussed because when I tuned my Mitsubishi, the VE table was a true % efficiency table with psi load on the Y axis and RPM on the X axis. It was very straight forward; I knew that *in general* my motor was close to 100% efficient around 5000-6000 RPM so I knew around those load cells I would have to aim for pretty much 100% VE as a value, and when I was done tuning it, it took a very nice form and had logical steps from cell to cell.

Now when I jumped onto this AEM, Josh is right, it doesn't have a "true" VE table. It just has a fuel map that you can set to be based off TPS or psi load (from the MAP). I chose MAP-based of course, but I noticed the values did not correspond to VE percentage, but rather you could choose, as GTSBoy said, whether to display the values in injector ms or RAW. And the RAW value apparently just goes up to 255 because that's the max value programmed into AEMtuner.

 

This threw me off when constructing my initial base fuel map because now I had no bearing on the value scales. However, after further tweaking with it, I understood that it's just arbitrary numbers. What's important is to have a decent and logical AFR map, then have a good, calibrated O2 sensor to compare against and simply start adjusting the fuel map on a cell by cell basis, until my target AFRs match my real AFRs. In the end, the map should take a similar shape as if doing standard VE table, and the VE is still static based on the hardware so there was not much point in getting hung up on that. I'm still learning, this is only my second car that I have started to tune so I'm in no way an expert and clearly lacking some further understanding, but I will get there eventually.

 

Also slight update, I ended up buying a metal spatula from the general store. It's normal steel, 0.3mm thick precisely. I took it to a laser cutting studio and had them cut out the disc from it. It was near perfect, just needed a light wet sand with 120 grit to get rid of any potential burrs from where the laser starts it's beam. It looks proper and fit incredibly snug onto the shaft, slightly better than the OEM disc I might add. There is also plenty of space on each side of the disc inside the optical head.

And of course, AEM failed me once more. I loaded up their custom settings for their custom disc and guess what? NO START! I swear with this car, it's 1 step forward, 2 steps back. I had a starting,  running, perfectly idling car with the stock disc and custom settings. And now with an exact copy of the AEM disc, and with their proper settings which are meant to be PnP, it doesn't even  get spark or fuel. If I set the crank signal to read from both rising and falling edges (which normally you shouldn't do), then it makes spark but no fuel. And when I try to time it with a light, I only get a flash at the start of the crank and at the end.

Gonna give it 1 more day today and if I can't make it work, I'll throw in my stock CAS disc...