it’s a Bosch LSU 4.9 wideband O2 specifically.

 

Ok so time for the reveal! The wideband O2 sensor shown in a previous post was a teaser for this:



This is an EMU Black standalone plug-n-play ECU from ECUMaster made directly for the R53 mini. I'm sure anyone reading this is aware of this product. I've been eyeing this for a long time now, pretty much since I got the car. I was considering the Bytetronik flash tune kit but at the end of the day the prospect of tuning this car without a dedicated closed-loop wideband seemed like hell on Earth. I was considering the MiniLink as well, I spent time with both software packages and the Link is definitely more mature, feature rich, and offers better control, but the EMU Black offers a more simplified environment for me to learn in at a lower cost. Both options still offer significant upgrades over the stock DME.

This will be the first standalone ECU I've personally worked with, I tuned my '19 Mustang with it's Gen 3 Coyote, and in comparison to that this is really easy. The complexity is low, the control scheme is fully documented and easy to understand.

So far I have the car idling and the wideband installed and working. However last night I realized the firmware on the EMU was severely out of date, so I updated that which gives me some new features. I was also severely sleep deprived and excited to get my new standalone ECU installed, so I likely made some mistakes. Today I'll be approaching this a bit more carefully to get the idle dialed in and the fueling dead nuts on target


Bosch LSU 4.9 harness made with PTFE insulated high heat wire.

ECU installed, will need to find a way to run this USB cable into the cabin.

Narrowband removed, wideband installed.
The downstream O2 is no longer needed, the ECM doesn't even have it connected up as far as I'm aware. Eventually I'll take it out and put a plug in its place, but for now it will stay as a very expensive plug

 

I was thinking the O2 sensor was for a tune.

Ok, I have followed a lot of R53S tuning threads and your ECU is a new one for me. That doesn’t mean a thing, really, as I haven’t looked into this sort of tuning or what is out there. Maybe others have seen that…

That said, I must say I’m looking forward to what you come up with.

 

yeah it’s definitely an oddball option, there’s not much evidence of its use other than a few threads on some other forums. It’s not even mentioned here at all. The Link has much more prevalence.

This ECU is an EMU Black pre wired for the R53, the EMU Black itself has much more use in the JDM world it seems. I don’t know why it’s not more popular here though, it has a ton of support behind it, very refined software and hardware, and free HPA videos about its operation.

My guess is that for a long time Link was the only option so that’s what the aftermarket picked up and stayed with. Lots of tuners spent lots of time developing base maps and don’t feel like re-doing it all over again.

 

Ok so I have an update!

The day the ECU was delivered last week i eagerly slapped it into the car. It was supposedly preloaded with a basemap that could use the factory narrowband O2 sensor. However I didn't even bother trying, I installed the wideband immediately. I didn't see any reason to stay on the narrowband even if just for a quick test. All of the sensors and control lines came correctly pre-defined in the base map so I didn't need to spend any time on that. The EMU black uses a built in automatic throttle learn routine that learns all of the characteristics of the drive by wire system on its own. So I ran that and basically like magic the throttle was working perfectly.

I switched the injectors to 380cc from the stock 330cc. I left the dead time untouched as I didn't have data for these injectors and I had no issues running them on the factory ECU. So far it hasn't been an issue. When I upgrade to larger injectors I'll be sure to buy some that actually come with a full set of data to make my life easier here.

The car cranks and starts no problem, I then spent last week learning about and dialing in the idle control system. I first dialed in the VE at the idle location, then I tuned the idle open loop so that it was approximately on target at all times (+/- 80 RPM), then I enabled closed loop throttle control, dialed in the PID loop, and finally enabled closed loop timing control and dialed in that PID loop. I now have a stable idle withing 25 RPM of target. The car likes to idle at 900-1000 RPM. Idle is a bit tricky to get right as it varies a lot with coolant temp, and this car heats up really fast, so I only get a minute or so at the low coolant temp cells of the idle table before it is warm again. I suspect I'll have to tweak this again in the winter, but its good for now.

I took the car for a light gentle test drive and noted the following:
- The pops and bangs are gone, there's no overrun on this version of the ECU software so enabling pops and bangs will have to be manually done or wait until I upgrade to the latest version (more on that later).
- The basemap does not do closed loop fueling when you are at wide open throttle.
- the base VE table is about 10% RMS off target at all times.
- the base timing table is low, never knocks (good) but will need to be dialed in for power

I suspect the reason the basemap was set to not do closed loop fueling 100% of the time is simply a side effect of it being written for a narrowband O2 sensor. I modified the settings so that it will be applying a fueling correction all the time even at wide open throttle. This is modelled after the stock strategy using in my mustang, and many other modern cars. I'm not sure how the EMU black will behave, I suspect I'll need to dial in the closed loop fuel PID to make this work better. This should help bring down that 10% error int he VE table and more importantly, keep the motor safe at all times.

The VE table is likely off because my car is a JCW and this basemap is for a standard R53, so my ported head is likely enough to change the VE of the motor. I found that in the low RPM areas I am flowing more air than the VE map expects. This is what I'd expect from a ported head.

This ECU and all EMU Black ECU's right now come preloaded with Version 2 of the operating system. There is a Version 3 beta available that enables a lot more cool features, and my long term goal is to switch to that OS. Unfortunately nearly all of the currently available training material is written for Version 2. So I'll be leaving the ECU on V2 until it is completely dialed in, and then I'll upgrade to V3. There are enough differences to where I, a novice tuner, is struggling to make sense of it all.

Overall this was definitely the right choice, I briefly looked at some of the tables available in the Bytetronik software and its just impossible for me to make sense of. In comparison the ECUMaster software is clearly laid out, and every table comes with complete documentation and often suggested values. Its a world of difference.

 

A “novice”? You clearly underestimate yourself. Sounds like good progress no matter what level you consider yourself.

 

Well I sure feel like a novice on this platform lol. Using a standalone makes things much easier, since its the same interface, it means that solutions that someone else has implemented on say a Toyota can be leveraged on my Mini. However there are a few hangups, the biggest one is finding the right timing for the car. I'm actively on the hunt for a Stock JCW tune file in some form of human readable format so I can know what the stock fuel and timing are set to.

I can manually tune those things, but without any starting point it will be a very long and difficult process.

 

Small update:
I loomed the wideband wires with Tessa tape. Which is a synthetic fabric anti-abrasion tape, I could smell it melting under the hood near the header. So I swapped it with this fiberglass sleeve. Problem solved.




the wires were never in danger, the teflon insulation is very strong.

 

I've come up with a method of tuning which I've been referring to internally as "commuter tuning". On my ride into work I collect data, at lunch I look at the data and make changes, then I drive the new revision home and make changes when I get home. The next day I test that revision, and rinse and repeat. Theoretically this allows me to make 10 revisions a week. In reality I take my mustang on e85 when the weather is too nice to ignore.

For anyone wondering this ECU does come with a base-map, it is designed for factory R53 with the narrowband sensor still installed. If you do what I did and immediately switch to wideband, you'll find that all of the wideband functions are completely unmapped.

This ECU uses a speed density model and a VE fueling table. Pretty standard stuff, the VE table models the airflow capacity of the motor relative to ideal at a given manifold pressure and RPM and then the fuel control system uses that to calculate the appropriate fuel dose. The VE table is never perfectly accurate, and depending on temperature and elevation the VE table may be inaccurate from one day to the next. The ECU uses the wideband sensor to measure the mixture all the time, and if the tuner desires can apply a rapid short term fuel trim correction to the fueling to account for errors in the VE table (within reason ofc).

This is the stock VE map, I believe it is a stock R53 map:

Base fuel table

Here is my fuel table after modifications to account for the JCW head.

fuel table tweaked for JCW head


Its not a huge difference but it is a difference, enough to where I could feasibly explain it to be directly caused by the head. This fuel map is reasonably accurate with STFT disabled, and with STFT enabled yields +/-5% trims at all times. I'd say this is good enough. My engineering brain told me I should smooth this map out, but in practice I found that made things worse, so I left it as you see it, jagged and ugly but functional.

I then spent an exorbitant amount of time on the STFT PID loop. I've always conceptually understood PID loops since I first brushed up against them in control systems class during my undergrad, but I never really spent time with one to learn how to intuitively tune one. This article explains PID loops pretty well if anyone is curious to learn more about them in detail: PID loops - wikipedia The real thing to know is that the PID loop is what lets the STFT react quickly to fueling errors without overshooting, oscillating, or generating unpredictable results.

The EMU black has the following options available for the STFT PID:




If you look at the PID gains window (2nd screenshot) you'll see that the D term is distinctly lacking from this page. I believe that the D component of this PID loop is actually the lambda delay table (screenshot 3). This setup is prone to integral windup, so I am still dialing in the integral limits. Right now I have the STFT pretty good, its very fast to react and reasonably tight to the the target at all times. However it is very oscillatory, overshoots and undershoots a lot, and during rapid fuel changes swings away form target wildly. I'll be dialing this in more to see how good I can get it, I think I will get it there in only a few more revisions.

Knock has been an issue as well, but not for the reasons you may think. I have the knock sensor set to flash the check engine light if knock is detected, this is a useful tuning tool but probably not suited to a daily driver which will sometimes experience mild pinging. I was very surprised to see knock registering, at part and wide open throttle in 2nd gear. Immediately I knew I had to rule out of this was actually knock or erroneous readings from the knock sensor. I added 18 oz of MMT based octane booster (VP Octanium unleaded) to the tank, allowed it to mix and drove again. This mixture is equivalent to 98 octane, and with this fuel in the tank I was still registering knock at the exact same intensity and frequency as before. This told me that this was not a tune issue but rather something in the motor make a sound that was registering as knock.

This is what the "knock" looks like on the log, sharp spikes inching above the engine noise threshold:



At this point I recalled the blown lower motor mount that I had happily ignored many months ago:


I installed a new mount with some SuperPro polyurethane bushing inserts for added stiffness. I'm no longer able to trigger any false knock at all, the graphs are quiet and much less spiky.


Also my vanity plate was accepted


Thats all for now!

 

All of this from a “novice”…

That was a great catch with the lower engine mount. Not sure many people would make the connection between that and the knock sensor output.

And - Nice plate

 

I am very much still a novice, some of the people I watch online are just wizards at this stuff. They make connections and understand these systems so well, I can only hope to know that much some day. I made the connection between the knock sensor and the motor mount because I come from more modern cars where knock is normal all the time. My Mustang and my old Focus were designed to knock even at wide open throttle, but excessive knock would rob you of power. We would spike the gas with octane booster or a small amount of e85 and see if the knock went away. If it did you knew it was real knock because the high octane fuel stopped it, if it didn't you knew it was a mechanical "false knock".

I was happy to get the plate, it really ties the whole car together!

 

Some Sunoco stations around here have 94 octane. It does run nice in the MINIs, although I don’t see a difference on the track performance. I wonder if would the difference from 93 to 94 would be seen by that program?

 

while under the car to do the lower motor mount I found this coolant leak. I pulled the front off and found the water pump failed.

i decided I’m going to delete the AC for now because the compressor bearing is bad, the condenser is shot, and the lines are shot. I don’t have the funds to fix this right so I’m going to put in the delete pulley and in the future overhaul the whole AC.

i decided since I had the front off it’s also a good time to take down the front subframe and overhaul all the bushings and bearings, and finally fix the power steering leak.

lots of parts on order, I’m going all powerflex where I can. These PS lines are so expensive!

 

It would be noticeable if the engine is actively pinging

I believe the stock setup does ping a small amount so theoretically it could help prevent that. I don’t know enough about how the stock ECU works to say definitively if it would do anything.

I know there are adaptations on the stock ECU, if there is a knock based adaptation having higher octane fuel would be helpful.

on more modern setups, cars can automatically add in timing if knock is not detected. So in that application the car would make more power.

this standalone ECU doesn’t do that, just reacts to knock when it happens.

 

Interesting.

 

Refreshed the front subframe! All powerflex race bushings, all new power steering lines, all new ball joints. This thing is nearly as fresh as it was when it was new! There was so much old oil and road grease on it that it barely had any rust at all.


Before

After

the sick irony is that the oil leaks kept this 20 year old subframe rust free up here in the great northeastern United States. But now that this car is actually being loved and the oil leaks are being fixed, this subframe is destined to rust. Ah well, what’s that saying? “To be loved is to be changed”

 

Gonna swap out some of the tired coolant lines before I put the subframe back into the car.

I’m also in the process of deleting the AC, my brother printed me these nice block off plates for the evaporator.

 

The plugs are great.

 

yeah I gotta get my own 3d printer, it’s just so cool

 

And learn 3D drafting…

 

Ahhh whats one more hobby

 

Subframe is back in the car, what a bear of a job that was. Like all things on this car none of it was hard it just took forever. It was 90f today and high humidity.

Power steering is filled with fresh fluid, still working out all the air bubbles. Cooling system is full of fresh new blue coolant, bleeding this car is so much easier when it heats up fast in the warm weather.

AC delete is working flawlessly pulley is quiet no issues at all.


I had a ton of trouble getting the steering linkage reconnected. It seems like it was a bit bent so it was tighter than it should be. I had the subframe up and bolted and then I had to take it all the way back down and hit the rack and the u-joint with a flap disc to blast off the rust and give it a fighting change of going back together. I still had to beat it on with a hammer.

replaced the metal coolant pipes that go to the oil cooler, just because it seemed really easy with the subframe down. It was easy, took 10 mins tops and now I know that pipe is good for a long time.

the oil cooler hoses are in sorry shape I need to order new ones before they pop.


haven’t driven the car to feel the new bushings yet but I’ll give it a ride tomorrow. I’m beat.


Subframe ready

Gradually nudging it back in

Bolted up

Old front crank seal

New seal in

Original harmonic balancer in excellent shape

Harmonic balancer back on, new bolt

New lower coolant pipes. Old one was probably fine despite being rusty, replaced “while I was in there”. I’ll clean up and repaint the old part as a backup

AC delete block off plates

This was just hanging around the steering linkage, it doesn’t appear on realOEM. The original boot was there just not seated right. Weird.

AC delete pulley

Coolant purged, back on the ground.

WMI nozzle in.

 

I've kind of fallen down a rabbit hole on this parameter. For context this article explains what injector dead time (IDT) is: Injector dead time - Haltech

My issue is that the basemap included only has injector data for stock non-JCW R53 injectors. Even worse I'm using VW1.8T 380CC injectors with unknown dead time. IDT is a critical parameter, if it's wrong the injection pulse will happen at the wrong time and the VE model will be wrong. If the VE model is incorrectly linked to the injector's like this, you can't easily swap injectors without needing to re-tune VE every single time.

Do not do this knock sensor method its completely wrong, there is no simple correlation between when he pintle clicks and the dead time. Use this method: Injector dead time testing - X-intercept method.
I tried a method to measure injector dead time using a knock sensor and an oscilloscope to listen for the pintle clicking down into the seat. That seemed to work I can accurately measure the dead time and it behaves as expected. IE lower dead time at high voltage and higher dead time at lower voltage. This was measured free air (no pressure), and that isn't accurate, it needs to be measured under the same pressure the injector sees when operating.

I've decided to build my own injector tester which will let me measure flow rate and IDT under pressure. I'll post it here when its working t

 

Oh, this really is a big, deep rabbit hole…





Pictures, please…

 

After a LOT of work I was able to characterize the IDT of the Bosch 0280156063 Audi injectors I'm using. Came out to 0.5ms at 12V and 0.4ms at 14V. A quick video of one test case:
After updating the dead time I'm getting much better fueling, way less overshoot and generally much more accurate and fast response across the board. The base VE map needs way less editing to be correct now.

Good thing it worked because, I was getting sick of testing injectors. The method that i used before with the knock sensor is completely wrong, nobody should ever do that its like not even in the same ballpark. The right way is to do a flow test. Additionally the flow rate of the injectors measured at 374cc/min at 50% duty cycle, which is exactly what they're quoted at. I didn't test all four, the ECU doesn't have provisions for multiple dead times so there's really no point. I could flow test them all but honestly I don't see it being worth it, we're not making that much power here lol.

Also snagged these R56 rear trailing arms and bushings for $25 each from a scrap yard:

Got some better exterior rear trim too, the rubber was falling off mine.