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.
An updated tune is attached. Pushing the timing a bit. Knock detection dialed in a bit more. Spent some time on startup tuning.
I'm finding that after a priming pulse to wet the walls, the car starts best with no cranking enrichment and very little after start enrichment. It also needs zero acceleration enrichment to hit the AFR target within 0.1 seconds after a very fast throttle change. All due a properly tuned VE table and the very fast MAF that I'm using. Overall, tuning with the MAF has been really easy.
Getting overrun/decel/tps/idle/MAP thresholds dialed in took more time than it should have thanks to some improperly labeled and documented fields. Once I ignored the descriptions and documentation, and instead worked off of empirical results, I got it all sorted out.
Another thing worth mentioning is that TunerStudio has a coolant temperature sensor curve for our sensors. It's listed under the BMW E30 sensor in the drop down list, but it's the exact same specs as in the 928 factory service manual. However, the factory specs have a range of resistances for each temperature, and my sensor tested out close to the middle of the range, but not quite. It's pretty easy to adjust the calibration with everything installed to get a very accurate coolant temperature reading in the ECU.
If I had a stock 928 that had a failed LH or EZK ECU, or even if I just wanted to make some minor improvements, I think I would go with a MicroSquirt. Cheap and supports everything on the 928. The only two things you would need to add are a voltage converter for the stock MAF and an O2 sensor controller. You could even make it plug and play without having to replace any stock wiring. I'm sure there are other modern aftermarket ECU's that would serve just as well at a similar price point.
At this point, the car is done, other than my incessant need to tinker. I'll keep pushing the timing and finding things to adjust here and there, but they are things most other people wouldn't notice or care about. The car drives better than it ever did with the OEM ECUs (whether OEM tune/configuration or SharkTuned with the turbo). I couldn't be more pleased.
Install Tuner Studio, then save the MSQ file somewhere and double click it. It should ask you if you want to open it in a temporary project. Select to use the temporary project and you can look at all of the settings in my tune. What the MSQ won't have are things like the coolant sensor calibration, or the custom log channels I created, or my custom dashboards. You would need the full project to get all of that. I'm happy to zip that up and post it if someone wants it.
I should add the caveat that I haven't been able to tune startup related things for sub 70 F temperatures. It just hasn't been that cool here since I installed the ECU. As we move into winter, I'll finish dialing in those last few rows in the startup tables.
Attached is an updated tune. The main changes are to the priming pulse and cranking pulse maps. Each morning I start the car cold, note the startup performance vis-a-vis the previous starts, and adjust the cranking and priming pulses accordingly. After the first start, subsequent restart data has decreasing validity. Once the car has been running for a few seconds, the conditions of the intake and cylinders are different than if it hasn't been started for a couple of hours, even if the coolant temperature is the same (which is what the startup maps are referenced from).
I repeat this once or twice more through the day.
My ideal objective is to have the car fire by the first compression stroke. My "tolerable" standard is for it to start by the second or third compression stroke. And it should do that whether it's been sitting for a day, or a few seconds, regardless of temperature. I'm closing in on my "ideal" in most scenarios, and have achieved "tolerable" for the rest of the map at the temperature ranges available to me (70+ F).
I started my cranking pulse and priming pulse maps by using the factory maps as extracted with the SharkTuner. They turned out to be way too rich for my car. Something to keep in mind as a data point for anyone doing startup tuning in the future.
Now for the "fun" stuff...I abandoned the Android tablet approach to instrumentation. The "ShadowDash" application hasn't been updated in years, and the developer said there are no plans for an update. I kept having to either work around, or just give up on, various instrumentation features due to the Android app not supporting things implemented by the current desktop tuning application. I also found the update speed to be barely tolerable on Android.
So, I went with a Mele Quieter3 fanless mini-pc tucked away on the parcel shelf (hidden by a leather closeout panel), a timer relay to manage power, a voltage regulator, and a touch screen monitor mounted where the tablet used to be.
The resulting functionality:
1) Turn key to on position and PC starts and loads TunerStudio in full screen mode. 2) Turn key off and monitor turns off after 15 seconds (configurable), but PC remains on. 3) Touch the screen or use the wireless keyboard and the screen comes on and stays on until there has been no input for 2 minutes (configurable). 4) Turn key on and monitor turns back on within 5 seconds (configurable). 5) Turn key off and PC gracefully shuts down after 4 hours (configurable).
This lets me run errands without having to wait for the PC to boot each time, but also shuts down the computer without losing data if I don't use the car for a long period of time.
Logs are automatically synchronized to my private cloud. That lets me review them from my desktop without any extra steps. I can also remotely access the car computer from my desktop if I want to make configuration changes, without having to take any extra steps. Just shut the car off and walk away.
Boot up time is pretty typical for this kind of thing. Rasberry Pi is roughly the same. The only way to get it faster is to either cheat and claim "booted" when I see the desktop, (which is where misconceptions about Rasberry Pi boot times come from), or use much more powerful hardware (which would use much more power and generate more heat).
Here are the bootup times: -POST complete: 5.5 seconds -Desktop loaded: ~22 seconds (from power on) -TS loaded with dash in full screen: ~43 seconds (from power on)
Obviously, that "lengthy" startup time is only if everything is off. If the PC is already on and just the monitor is off from stopping to run errands, it only takes a few seconds for my program to detect the ECU is connected and turn the monitor back on. That delay is configurable, though.
Here are the loads:
1) Everything off, timer relay has full time power: 0.01 amps at most (it's less than my multimeter can measure) 2) Timer relay activated, mini-pc powered off: 0.07 amps 3) Mini-pc on, monitor off: 0.35 amps 4) Mini-pc on, monitor on: 0.7 amps (but it does surge to 1 amp occasionally if the PC is working hard, like booting or processing a lot of data)
You could also use two diodes and wire it so that the timer relay gives itself power from the full time power source, so that once the timer runs out it turns its own power off as well.
Making this all work required quite a bit of customizations. Attached to this post is a ZIP file with most of that. A Powershell script I wrote, task scheduler task exported in XML, TunerStudio configuration screenshots, etc. Most people won't care about those things, but for someone that does, feel free to send me a private message if you have questions.
The screen is on a magnetic mount. I can remove it in a few seconds.
A few photos:
Voltage regulator (silver box on left) and timer relay (black box on right):
The mini-pc (top left), 8" touch screen monitor (top right), Bluetooth keyboard and trackpad (bottom middle):
Booting up with customized logo (no blue windows squares):
Dash loaded and ECU disconnected (hence the greyed out appearance to the screen):
I use an Asus 12 inch tablet running Windows 10 as seen in the photo above in my race car. I use it in the 928 with the add on keyboard. About the only programme on it is Tunerstudio and Megalog Viewer so it's rocketship fast! Battery life is excellent. I run it all day on trackdays with no problems.I looked at your cranking and priming pulses and they are a fair bit different to mine as expected. No car is going to be the same and yours being turbocharged etc is not going to be the same as mine with the intake set up I have. The start ups later in the day will vary too from heat soak.
I use an Asus 12 inch tablet running Windows 10 as seen in the photo above in my race car. I use it in the 928 with the add on keyboard. About the only programme on it is Tunerstudio and Megalog Viewer so it's rocketship fast! Battery life is excellent. I run it all day on trackdays with no problems.I looked at your cranking and priming pulses and they are a fair bit different to mine as expected. No car is going to be the same and yours being turbocharged etc is not going to be the same as mine with the intake set up I have. The start ups later in the day will vary too from heat soak.
Yes, startup tuning is a labor of love with no real upside At worst, I would get 0.8 second startup times. My goal of sub 0.2 seconds at all temperatures and all delays since last start is becoming a hobby at this point. But, when it's rainy outside and you need an excuse to tinker....
My startup differences might have more to do with my injector size than the turbo. The turbo isn't doing anything at startup. And the compressor should add no meaningful restriction beyond what the closed throttle plate is doing. I wonder if the aftermarket starter makes a difference too? Do you know what your cranking speed is before cranking firing begins? Mine is between 120 and 180.
Would you mind sharing your tune? I would love to take a look.
I would have preferred a tablet like yours for a lot of reasons. But, the smallest new one you can buy right now is 10", and that's just too big to suit my personal aesthetic.
Thank you for sharing. Your priming pulses do look a lot different from mine. They start smaller and get much bigger. And with much smaller injectors.
What's really interesting is your cranking pulses. You're actually pulling fuel out when it's warmer than 50 C. That's really unusual from what I've seen. To only use 40% of the normal fuel for cranking when the engine is fully warm seems like not very much fuel at all.
We have very different intake manifolds, which might account for some of the difference? I don't know, but it's really interesting to compare the two tunes.
I also see that you're using 14 degrees for the cranking advance. How did you arrive at that number? I started at 12, and it works fine, but I've been playing around a bit with lower advance and haven't noticed much difference. More advance would give a snappier start, but risks kick back, so I've been hesitant to give it more. Maybe I'll try going a bit higher and see what happens.
Am I reading your configuration correctly that you're somewhere in the neighborhood of 17 degrees advance when idling?
My car is particularly easy to start and seems to fire on the first revolution. I could spend a bit of time on the after start enrichment but I'm inclined to leave things as they are. The cranking advance is 14 simply because that was what my other car with a Microsquirt needed to start with a supercharger and E85 fuel. I started with 14 and never thought any more about it. The injectors are sized fairly well, I just used the same size as other 928 owners that have done Megasquirt conversions. Yesterday, when I checked the data log the duty cycle at its highest, was about 52%. The idling advance is a bit high at the moment and I shall be playing around with that in the next couple of days, dropping the advance and adding a bit of fuel to keep the idle speed the same but run it richer. It was about 14.7 at idle and when its idling in the high 13's and low 14's it has a smoother idle. The only downside of batch fire is the need for a richer idle.
The throttle body I used is off a V8 Commodore, GM product, and it's only 55mm which is a fair bit smaller than the Porsche item. I asked a local dyno operator what is the best and he suggested this one. Limited to 450hp but perfect for a 300hp car. The spider intake is polished and match ported and all the crankcase gasses are diverted which I think makes a big difference to this car.
NOW you tell me the cranking advance was arbitrary I went ahead and slowly increased mine to 14 as well, and it did make a small difference on cranking. So, thank you
I wasn't implying your injectors were sized improperly, and I hope it didn't come across that way. I was only noting the difference between yours and mine. The deadtimes that we have configured in our tunes are very different, and our injector sizes are different, so our priming pulse lengths will end up being very different as well.
With my 60 lb injectors, I'm hitting a max duty cycle of 76% on spikes. 60% to 70% is my "normal" max range, though.
Why did you use the GM throttle body instead of the Porsche throttle body? I'm just guessing...drive by wire or TPS mounting requirement? I see that you don't have an idle air valve configured in your tune, so I'm guessing that it's manually set with a screw?
You've put a lot of work and thought into your setup and I bet others will benefit from it in the future!
The GM throttlebody was chosen because it was cheap and readily available but also because of the throttle radial quadrant/TPS and it has one of the better idle air valves. The stock throttle body on the early 928s is basically just a lever and is unsuitable for accurate TPS operation. The Megasquirt/Microsquirt uses the GM sensor values as a default to so I run all the sensors from the GM line except the VR which is Ford. The IAC does not show in my tune because it's deactivated. I found it was a pain and unnecessary with both my cars. The Fiat runs a GM throttlebody as well. The IAC plug has been pulled and the valve is opened just tiny amount and just happens to be about right for the idle speed. I fine-tuned the idle speed with the VE table and ignition timing. Having the lowest speed column on the timing map a little higher than the normal idle speed bumps it back up if required. Yesterday when I took my wife shopping I spent some time adjusting the idle and now it's rock solid steady at 14.3.
08-25-2022, 09:50 AM
At worst, I would get 0.8 second startup times. My goal of sub 0.2 seconds at all temperatures and all delays since last start is becoming a hobby at this point. But, when it's rainy outside and you need an excuse to tinker....