When you click on links to various merchants on this site and make a purchase, this can result in this site earning a commission. Affiliate programs and affiliations include, but are not limited to, the eBay Partner Network.
The EMU black supports PWM fan control, and the only thing keeping us from using it is that our fan speeds are controlled via relay. I have a long term plan to source/design a solid state relay replacement which would allow us to accomplish two things:
1. Eliminate the failure-prone low fan resistor
2. Allow us to get rid of the fan speed states entirely, the fan would just ramp linearly up as needed.
3. Free up an output for whatever else we like
The current wiring doesn't allow this to work, the relay is triggered by the black and just switches the 12V, a large mosfet, a diode, and a pull-down resistor would probably make this work easily. I should look into doing this, could be a fun project
I wonder if it could PWM my water pump (with the right SSR or other device to handle the current - up to 15A) so I could remove the thermostat. I was planning to use the Davies Craig pump controller, but decided to just get a Thermatic switch to add an extra speed below the low speed and outside ECU control. My water temp gauge would work better below 195°, too, with full flow all the time....
I wonder if it could PWM my water pump (with the right SSR or other device to handle the current - up to 15A) so I could remove the thermostat. I was planning to use the Davies Craig pump controller, but decided to just get a Thermatic switch to add an extra speed below the low speed and outside ECU control. My water temp gauge would work better below 195°, too, with full flow all the time....
I'm pretty sure you can, would be more of a question of how well the electric water pump reacts to being fed a PWM signal. But that's a really cool idea, I think some of the really new modern cars are doing stuff like that. Always so gratifying to bring modern day tech to a two decade old vehicle, things the designers at the time couldn't have even hoped to have in a street car!
I'm pretty sure you can, would be more of a question of how well the electric water pump reacts to being fed a PWM signal. But that's a really cool idea, I think some of the really new modern cars are doing stuff like that. Always so gratifying to bring modern day tech to a two decade old vehicle, things the designers at the time couldn't have even hoped to have in a street car!
Davies Craig also sells a pump controller that does the PWM thing, so I guess their pumps are ok with it....
Lots of work has gone into the tune since I last updated
the car starts in any weather now, cold start fuel is looking good. Still has a slightly extended crank but putting that on the backburner for now.
main fuel table is dialed in, came up with some great tricks for making the autotune feature work on the road
knock control is 90% dialed, still a bit too sensitive but one more pass at that and it will be perfect. At least know I'm 100% sure that any knock will be caught by the sensor.
transient fuel is looking pretty good, this took me a very long time to get right but i'm pretty happy with how it looks now. Very few lean spots now :D
TPS sensor has been calibrated correctly, less noise on the signal now
Bidirectional CAN bus communication has been implemented between the ECU and the WMI module. The WMI unit can now put the ECU into limp mode if water flow stops for any reason.
Flat foot shifting works perfectly, limiter at 5500 RPM prob could go higher
Still lots more i'd like to get working:
Overrun pops and bangs, My car made a lot of noise on the factory ECU and I'd like to try and get that back
Traction control, the EMU black has its own TC logic, right now its all zero'd out so I have no TC. Not a huge deal but would still like to get this going again.
Lots of work has gone into the tune since I last updated
the car starts in any weather now, cold start fuel is looking good. Still has a slightly extended crank but putting that on the backburner for now.
main fuel table is dialed in, came up with some great tricks for making the autotune feature work on the road
knock control is 90% dialed, still a bit too sensitive but one more pass at that and it will be perfect. At least know I'm 100% sure that any knock will be caught by the sensor.
transient fuel is looking pretty good, this took me a very long time to get right but i'm pretty happy with how it looks now. Very few lean spots now :D
TPS sensor has been calibrated correctly, less noise on the signal now
Bidirectional CAN bus communication has been implemented between the ECU and the WMI module. The WMI unit can now put the ECU into limp mode if water flow stops for any reason.
Flat foot shifting works perfectly, limiter at 5500 RPM prob could go higher
Still lots more i'd like to get working:
Overrun pops and bangs, My car made a lot of noise on the factory ECU and I'd like to try and get that back
Traction control, the EMU black has its own TC logic, right now its all zero'd out so I have no TC. Not a huge deal but would still like to get this going again.
The AC button is mapped to CAN switch 19, so you can do funny things with it, i think I'm going to get a CAN bus keypad for functions like this. My car doesn't have AC so I don't need this button to work right, but it gets automatically enabled depending on outside air temp as I'm sure we're all aware.
Could be useful for like a track day type deal in a pinch!
The AC button is mapped to CAN switch 19, so you can do funny things with it, i think I'm going to get a CAN bus keypad for functions like this. My car doesn't have AC so I don't need this button to work right, but it gets automatically enabled depending on outside air temp as I'm sure we're all aware.
Could be useful for like a track day type deal in a pinch!
LOL!
A/C is mandatory in Azirona, but I have seen a number of different inputs in the docs where one could port this.
Was just discussing burble tuning with ChatGPT the other day, to get tuning tips on how to change the character of the burble away from pops and bangs and more toward proper British chuntering, mumbled/muttered grumbling that lingers a bit more.... Seems fitting for a MINI to chunter more than to pop and bang....
LOL!
A/C is mandatory in Azirona, but I have seen a number of different inputs in the docs where one could port this.
Was just discussing burble tuning with ChatGPT the other day, to get tuning tips on how to change the character of the burble away from pops and bangs and more toward proper British chuntering, mumbled/muttered grumbling that lingers a bit more.... Seems fitting for a MINI to chunter more than to pop and bang....
I got the big bangs you heard in the video with -25 deg overrun timing and +10% overrun fuel. But increasing the timing to like -19deg gives you more like a burble
I got the big bangs you heard in the video with -25 deg overrun timing and +10% overrun fuel. But increasing the timing to like -19deg gives you more like a burble
easy enough
Good info. Will look to extend the duration of the chuntering, grumbly burble as well....
Looking at the V3 proggy and getting familiar with the stuff that can be poked at. Imported the base R53 tune just to have some kind of idea of the baseline goodies. Looking at the PWM output settings and wondering what frequency the Davies Craig water pump would prefer....
This thing is bloody impressive....
Edit:
Just found the PWM EWP controller settings - looks similar to the D/C controller's scheme, but has two warm-up stages. Excellent!
Edit 2: CAN2BUS resembles MIDI....
Last edited by joe_bfstplk; Jan 19, 2026 at 07:57 PM.
Can’t believe I forgot to post this. I finally changed the fuel filter, 148k miles and likely original. Totally full of sludge replaced with all genuine mini parts because I don’t like to be questioning my fuel system. I think I will replace the pump too, it makes quite a bit of noise.
I don’t need to state which one is the new part I think it’s obvious
Last edited by junits15; Jan 21, 2026 at 08:03 PM.
I had gotten a used replacement pre-facelift one to keep the stock look, as I like the self-leveling Xenons for the low beams. Sadly, though it is in good shape otherwise, the little plastic piece that snaps onto the ball on the end of the height adjustment screw broke after just a month or two.
Ordered the new lights this morning. Soon, I will fit in perfectly with this crowd:
Last edited by joe_bfstplk; Jan 26, 2026 at 08:17 AM.
I had gotten a used replacement pre-facelift one to keep the stock look, as I like the self-leveling Xenons for the low beams. Sadly, though it is in good shape otherwise, the little plastic piece that snaps onto the ball on the end of the height adjustment screw broke after just a month or two.
Ordered the new lights this morning. Soon, I will fit in perfectly with this crowd:
they’ve been exceptional so far. Super bright and a crisp line across the road. With the brights and my driving lights on it’s enough to light up even the darkest roads
i will say thought that I don’t currently know where to source replacement bulbs if I need one, I took a look inside and it appears to be a standard LED module but I don’t know exactly what style it is. It’s too cold for me to wrench where I’m at right now
This has been quite a learning curve for me! Sometime I wonder if maybe I should have stayed with the factory ECU haha. I've learned a lesson that many armature tuners before me have also learned:
Tuning the car at wide open throttle is the easy part, the engine behaves quite linearly during that condition. The VE is near 100, the MAP increases slowly, the throttle is not moving, the air temp is consistent. Fuel delivery works as expected. Tuning the car at part throttle and low speed is more than 75% of the work. Its very easy to make a car that will pull hard its extremely difficult to make the car nice to drive otherwise.
The biggest issue I'm combating right now is cold engine fueling and enrichments. When the engine is warm, the measured lambda generally tracks commanded lambda. When the motor is cold this is not the case, the fuel swings wildly all over the place, making the car extremely difficult to drive until its warm. What I've uncovered so far is this
- Error exists in the VE table, error varies sharply in different areas of the table
- STFT takes up this error (usually +/- 5%)
- When low speed driving, MAP and rpm vary causing the fuel calculation to swing quickly around the VE table moving from areas of low error to areas of high error very rapidly.
- This causes rapidly changing fuel error, showing in the "Lambda Error Mult." PID
- the STFT tries to constantly correct this error, but it is not fast enough and the rapidly changing fuel error causes the STFT to overcorrect and make fueling even worse
So this all stems back to VE table error. The VE table really needs to be very close to correct all the time. I've been using Autotune to tune the VE table, and it has worked really well. Up until this point I've been using autotune with STFT enabled, and ECUMaster has indicated that this is fine. I think they aren't lying, the algorithm takes into accound the amount of STFT applied. However, I think in the situation I am having at cold, it cannot resolve the data accurately enough. With the fuel wildly swinging all over the place caused by overcorrection in the STFT I get less than ideal changes out of the Autotune.
So what I've done is disable STFT, the fuel table I'm using is pretty close even in open loop. By disabling STFT the fuel error is more consistent and easier for the Autotune to resolve into a VE table change. The majority of the issues occur between 1k-3k RPM. I put the car in 6th gear at low speed and low RPM and gently leaning on the throttle, this causes the car to slowly rev up at very low MAP and slowly cover the problematic section of the VE map. I do this multiple times and lean into the throttle more and more each time to really well tune the low RPM areas of the map.
I'm going to do this for a few days and see how it helps the cold fueling. I think it will be a massive improvement.
The reason I say "maybe I should have stayed with the factory ECU" is that this area of the factory tune does not need to be modified. So I could have moved onto making power by now, but I also wouldn't have learned as much as I have so far. I think it was the right choice no matter what
they’ve been exceptional so far. Super bright and a crisp line across the road. With the brights and my driving lights on it’s enough to light up even the darkest roads
Nice. That was one thing I liked about the Xenons - the sharp cutoff so they don't blind oncoming drivers. Just gotta aim them well, I guess. Will know in a few days after they arrive....
EDIT: They arrive tomorrow...tomorrow, I love you, tomorrow, you're always a day away.... :D
Originally Posted by junits15
i will say thought that I don’t currently know where to source replacement bulbs if I need one, I took a look inside and it appears to be a standard LED module but I don’t know exactly what style it is. It’s too cold for me to wrench where I’m at right now
Hopefully won't ever need to.
Originally Posted by junits15
Tune update:
This has been quite a learning curve for me! Sometime I wonder if maybe I should have stayed with the factory ECU haha. I've learned a lesson that many armature tuners before me have also learned:
Tuning the car at wide open throttle is the easy part, the engine behaves quite linearly during that condition. The VE is near 100, the MAP increases slowly, the throttle is not moving, the air temp is consistent. Fuel delivery works as expected. Tuning the car at part throttle and low speed is more than 75% of the work. Its very easy to make a car that will pull hard its extremely difficult to make the car nice to drive otherwise.
The biggest issue I'm combating right now is cold engine fueling and enrichments. When the engine is warm, the measured lambda generally tracks commanded lambda. When the motor is cold this is not the case, the fuel swings wildly all over the place, making the car extremely difficult to drive until its warm. What I've uncovered so far is this
- Error exists in the VE table, error varies sharply in different areas of the table
- STFT takes up this error (usually +/- 5%)
- When low speed driving, MAP and rpm vary causing the fuel calculation to swing quickly around the VE table moving from areas of low error to areas of high error very rapidly.
- This causes rapidly changing fuel error, showing in the "Lambda Error Mult." PID
- the STFT tries to constantly correct this error, but it is not fast enough and the rapidly changing fuel error causes the STFT to overcorrect and make fueling even worse
So this all stems back to VE table error. The VE table really needs to be very close to correct all the time. I've been using Autotune to tune the VE table, and it has worked really well. Up until this point I've been using autotune with STFT enabled, and ECUMaster has indicated that this is fine. I think they aren't lying, the algorithm takes into accound the amount of STFT applied. However, I think in the situation I am having at cold, it cannot resolve the data accurately enough. With the fuel wildly swinging all over the place caused by overcorrection in the STFT I get less than ideal changes out of the Autotune.
So what I've done is disable STFT, the fuel table I'm using is pretty close even in open loop. By disabling STFT the fuel error is more consistent and easier for the Autotune to resolve into a VE table change. The majority of the issues occur between 1k-3k RPM. I put the car in 6th gear at low speed and low RPM and gently leaning on the throttle, this causes the car to slowly rev up at very low MAP and slowly cover the problematic section of the VE map. I do this multiple times and lean into the throttle more and more each time to really well tune the low RPM areas of the map.
I'm going to do this for a few days and see how it helps the cold fueling. I think it will be a massive improvement.
I have been employing ChatGPT to reverse-enginneer Bytetronik table labeling and find what I need to adjust for certain problems (p1689, anyone?) and it has proven valuable when double-checked. It's great at finding issues caused by tables being weirdly lumpy, and at targeting the cells hit in logged data for adjustments. Just gotta keep an eye on it, and keep to one subject per chat to keep the signal to noise ratio good in its input.
Originally Posted by junits15
The reason I say "maybe I should have stayed with the factory ECU" is that this area of the factory tune does not need to be modified. So I could have moved onto making power by now, but I also wouldn't have learned as much as I have so far. I think it was the right choice no matter what
It's absolutely the right choice. The factory stuff is a mystery wrapped in an enigma because they're afraid of you stealing decades-old "secret" techniques and technology. EMU Black wants to help you because it helps them sell their products....
Last edited by joe_bfstplk; Jan 29, 2026 at 12:03 PM.
I have been employing ChatGPT to reverse-enginneer Bytetronik table labeling and find what I need to adjust for certain problems (p1689, anyone?) and it has proven valuable when double-checked. It's great at finding issues caused by tables being weirdly lumpy, and at targeting the cells hit in logged data for adjustments. Just gotta keep an eye on it, and keep to one subject per chat to keep the signal to noise ratio good in its input.
It's absolutely the right choice. The factory stuff is a mystery wrapped in an enigma because they're afraid of you stealing decades-old "secret" techniques and technology. EMU Black wants to help you because it helps them sell their products....
You inspired me to try this, I've been using Gemini, and i think that its a great tool for this. It scours the net and makes info much easier to find. Yesterday I used it to help me understand how the accel enrichment should be scaled with CLT. It doesn't just give the answer obviously, but it shows me examples of what that table should generally look like, which gives me a starting point. I used it to understand how to scale enrichment with map, low map is actually more enrichment than high map. I never would have guessed it works like that otherwise.
I've also found that i helps me learn what to seach for on google to better solve an issue. I'm dealing with bad fuel at cold, and thats just such a generic thing. But conversing with the AI allow me to see what it could be, then go off and search for those topics on forums.
I never expected AI to be useful for tuning, but it really is.
You inspired me to try this, I've been using Gemini, and i think that its a great tool for this. It scours the net and makes info much easier to find. Yesterday I used it to help me understand how the accel enrichment should be scaled with CLT. It doesn't just give the answer obviously, but it shows me examples of what that table should generally look like, which gives me a starting point. I used it to understand how to scale enrichment with map, low map is actually more enrichment than high map. I never would have guessed it works like that otherwise.
Yeah, it's a good helper, as long as you keep your eyes peeled for hallucinations and lost references from earlier in the conversation. I have been using it as a substitute for the manual/documentation/tutorials that Bytetronik lacks. Pretty much dialed out the p1689/limp I was getting due to how much better the TVS900 moves air at lower RPMs. Stretched the top end of the fuel tables' airflow axis so that it's operating within the map instead of off the edge where the dragons be.
Originally Posted by junits15
I've also found that i helps me learn what to seach for on google to better solve an issue. I'm dealing with bad fuel at cold, and thats just such a generic thing. But conversing with the AI allow me to see what it could be, then go off and search for those topics on forums.
I never expected AI to be useful for tuning, but it really is.
Yup. It's a great tool for research and number-crunching. Also great for finding unintended ridges and/or cliffs from table edits (including its own suggested edits - gotta watch those and suggest smoothing sometimes), so the ECU doesn't freak out or do weird stuff....
My original fuel pump was quite noisy, I don’t know it’s that’s normal or not but I don’t like it and after nearly 150k miles I felt it prudent to replace it. Since this car is trending towards more and more power I opted to go for an upgrade.
enter: Deatschwerks 65v 265 lph drop in fuel pump.
new pump
basket and original pump removed form the car
new pump installed in basket
Level sender and wiring re-attached.
it fits really nice, it is ever so slightly shorter than the original pump so it wasn’t fully snug in the basket. I’m sure it would have been fine but I added a second o-ring on the top just to force it down and keep it snuggly in place. I also peeled off the shrink-wrapped label on the pump that you see in the first pic. I couldn’t 100% verify that it was fuel compatible and if the intention was for it to be removed or not. So out of an abundance of caution, I removed it. It means the pump is now just a nondescript silver tube and is not immediately identifiable as upgraded, but I don’t care about that.
much quieter too, fueling behavior is 100% identical on logs so that tells me the regulator has no trouble handing the added fuel flow.
i think it’s a great upgrade, with this and the 550cc injectors I’m e85 ready not that I’ll likely ever do that, the WMI kit will provide nearly the same gains, but if I ever go to a better SC or a built engine the fuel system is already ready to go.
Well this is it! The first stable tune release! Cold engine performance is still so-so and could be improved, VE table is close enough to drive without worrying but still needs some refining in the low RPM low MAP areas. The car starts every time, in all weather, tested down below 0F.
But it works! If you opt to use this please please check everything, i make absolutely no claims about anything in this map. Also I have the air conditioner button set up to switch between the two timing tables, this can be disabled if need be.
Well this is it! The first stable tune release! Cold engine performance is still so-so and could be improved, VE table is close enough to drive without worrying but still needs some refining in the low RPM low MAP areas. The car starts every time, in all weather, tested down below 0F.
Logging during regular driving is key here. Got my Specified Target Intake Air Pressure table sorted nicely using ChatGPT to pore over the log data and develop a "hit chart" showing which `RPM × ThrottleAngle` cells got hit most over several days' logs, then used the log data to suggest expected ManiPressureHPa values for each cell. There is some overlap in the sub-2000RPM range with first gear starting/creeping, though, so I had to be aware that the table will be better if optimized for first gear in those RPM ranges.... Likely this has some sort of analogue in your VE setup.... I can't wait to pore over the file and see what you got working and how!
Originally Posted by junits15
But it works! If you opt to use this please please check everything, i make absolutely no claims about anything in this map. Also I have the air conditioner button set up to switch between the two timing tables, this can be disabled if need be.
Logging during regular driving is key here. Got my Specified Target Intake Air Pressure table sorted nicely using ChatGPT to pore over the log data and develop a "hit chart" showing which `RPM × ThrottleAngle` cells got hit most over several days' logs, then used the log data to suggest expected ManiPressureHPa values for each cell. There is some overlap in the sub-2000RPM range with first gear starting/creeping, though, so I had to be aware that the table will be better if optimized for first gear in those RPM ranges.... Likely this has some sort of analogue in your VE setup.... I can't wait to pore over the file and see what you got working and how!
I've been relying heavily on the "autotune" feature built into the EMU black PC client. Its decent but honest not perfect. I think you are right, i will most likely see better results if i process the data on my own. The Autotune feature is very simplistic. It monitors RPM, MAP, transient status, and lambda error. It then bucketizes the data and averages the lambda error/cell and gives you a recommended change. A transient is things like fuel cut, acceleration enrichment, the system ignores the data after a transient for a configureable time.
The issue is that if its not in or after a transient the data is considered good, this means that time spent with key on engine off is accounted for, very logn transients cannot be ignored either. This means the interplay between short term trim, and lambda error causes bad data. The autotune can get you close, but its really designed for use on a dyno.
I've been relying heavily on the "autotune" feature built into the EMU black PC client. Its decent but honest not perfect. I think you are right, i will most likely see better results if i process the data on my own. The Autotune feature is very simplistic. It monitors RPM, MAP, transient status, and lambda error. It then bucketizes the data and averages the lambda error/cell and gives you a recommended change. A transient is things like fuel cut, acceleration enrichment, the system ignores the data after a transient for a configureable time.
The issue is that if its not in or after a transient the data is considered good, this means that time spent with key on engine off is accounted for, very logn transients cannot be ignored either. This means the interplay between short term trim, and lambda error causes bad data. The autotune can get you close, but its really designed for use on a dyno.
I'll see what I can do...
The Autotune is doing the hits, but likely not smoothing them into the table with a "halo" of lesser adjustments around them. Discontinuity tends to flip things out in interpolation from table data, so you have to look for "ridges" and "holes" in the table data. Some are intended - like the ridge of richer fueling below the torque peak that helps prevent detonation. Others can freak the algo out and get torque monitors twitching and tossing p1689s at you like they're on a fire sale.... I tend to use linear interpolation between changed cell and the cell beyond the neighboring cell if the suggested change is more than a couple percent....
I am, essentially, doing the same bucketing with my log data, but going a step beyond and making the table surface less craggy....
