DATALOGGING PART III: BOOST
The real reason we're all here...
I think this is probably the big one everyone has been waiting for. Boost. You love the feel. You love the numbers. You want moreeee! Now it's time to learn. Turbos are complicated, and I'm not an expert. In fact, there's so much, today is really going to just about reading the logs and understanding WHAT is happening. But don't worry, as always -- more to come!
Boost is interesting because on one hand, looking at it is trivial. On the other hand, understanding the system as a whole is really complex. Since this series is titled "Datalogging" I want to stay true to that and start with what you NEED to know for datalogging. I've got lots of cool "tips and tricks" coming, but, one thing at a time!
Making Life Easier -- DataZap!
So up until now, I've been preaching "Excel! Excel! Excel! (and P.S. Google Sheets)". But, today I want to mention a tool that many of you know well: Datazap.me (that's a website if it wasn't clear). DataZap is a great way to really quickly analyze your data logs. While timing can be quickly analyzed by looking at raw values, the actual curve of boost is particularly important--in other words, we need a chart. DataZap allows you to upload a .csv (comma separated value) and turn it into an interactive chart. Bingo!
Now, the reason I started with Excel is because DataZap is so much MORE useful if you do some pre-processing, so let's walk through that.
Pre-processing the Datalog
While I think doing some pre-processing in Excel makes life SO much easier, if you want to jump straight to the good part you skip to the next section. I've provided a public link to an example data log with the pre-processing done and the core variables. It might help motivate you to come back and do the pre-processing!
[EXTRACTING THE PULL]: Even if you followed the "How to Log" article, chances are you started your log at least a bit early and/or stopped it after the pull ended. However, our chart is going to be rather hard to read if we have extra data points. Again, the idea is to simulate a dyno pull. We want to delete any data rows that aren't part of the pull, both before and after. For example, if you happened to log all of 3rd and all of 4th or something similar, feel free to keep more, or generate multiple files. But for the scope of this article I'll be assuming you just extracted you single "simulated dyno pull".
The top half of the data.
[BOOST]: If you're using the lovely (sarcasm) Eurodyne logger, your variables are not in the same units. References the same chart as the first article -- convert everything to PSI (multiple by 0.145038 or 0.0145038). Add two columns for target boost and actual boost, but now both in PSI. Even if you do this... you might notice something like:
For most data loggers, boost is measured in absolute pressure -- meaning it's boost pressure + atmospheric pressure. If this is the case, you're going to want to subtract atmospheric pressure (14.7psi). If you live at altitude (like me) atmospheric pressure is not 14.7psi. If this applies to you, you can use this nifty calculator (www.mide.com/pages/air-pressure-at-altitude) to find out what the pressure is at your elevation. Personally, I like to add two columns, one column which is 14.7psi corrected, and one this corrected for my exact altitude. This gives an idea of what the tuner probably expected (14.7 psi atmospheric), and what you're actually experiencing.
[BOOST "ERROR"]: The last calculated data point we want to add is "boost error", the difference between the actual boost and target boost. This will make analyzing the data easier, particulary when it's in chart form. The formula is simply "Boost Target - Boost Actual".
Most likely you've got a sheet with lots of columns right now, which is totally fine (no need to delete them, thanks to DataZap! However, the core variables we're going to want to look at today are:
I've done some coloring to make it a little easier to read because I like pretty things.
Deciphering the Log with DataZap
So to make the learning easy, open up this example log: datazap.me/u/mrconflicted/drivetribe-example-log?log=0&data=0-2-5-7
Select "engine speed", "boost actual", "boost target", and "boost actuator position" (select by clicking on the variables below the chart). Hopefully the first three are self-explanatory, but you may not be familiar with the last. The EA888gen3, like many modern turbo cars, uses an electronic waste gate actuator. When the waste gate is closed, we're forcing all the air through the turbo. In this case, we're asking the turbo to work as hard as it can. Inversely, if the waste gate is fully open, we're trying to essentially by-pass the turbo.
Example diagram. This diagram is for some other car (I believe), but it's the best diagram I could find in my perilous 3 minutes of Googling.
Alright, let's try to understand what's happening in this log! First off, when we hit the gas, boost has to "build" or ramp up. During this period, we close the wastegate, directing all exhaust to the turbo to spool it up as fast as possible. It's important to note that some loggers will display closed at 0(%), while others will display it as 100(%). This first part of the graph during ramp up lets us know which way our logger is displaying it. Seems complicated, but hopefully is simplified in the annotated graph below.
Some initial notes.
Another thing to point out is that the ECU is "closed loop". As highlighted in the graph in red, the ECU is constantly adjusting the actuator to try to maintain targets. The actuator position gives us a really nice idea high level view of the system. Check out the example below.
A different data log of what we DON'T want to see. (No, this isn't DataZap)
In the plot above, the tuner has requested a steady 22psi all the way to ~6000RPM. However, up here at altitude, it turns out the IS20 can't make 22psi at high RPM's very easily. The actual boost starts falling below the target, and the ECU adapts. The ECU begins closing the wastegate, pushing the turbo harder and harder in an attempt to hit boost targets. For the safety of this man's car, I told him not to push further, but would he have kept going we likely would have ended up with a fully closed wastegate at 7000 RPM -- prime conditions for turbine shaft failure.
In this example the waste gate position is dropping into the "danger zone" (hope you used your Archer voice). This is an issue of a boost curve not matching the car's supporting mods and external conditions. The zones, green, yellow, red areas, are just a general "rule-of-thumb". Waste gate more than 30%, no sweat. 15-30%, you're starting to push it, but it could still be safe if done right. 0% at high RPM's? We're likely going to overspin the turbo and cause failure. Note that is much more relevant for smaller turbos as we'll see in future articles.
Now, of course there's outliers. Some people will be able to push their cars into the "danger zone" without issue, and other turbos will fail in the "safe zone". Don't let that deter you thou, it's still a great estimator. Again, we'll see why when dive into compressor maps.
[ANALYZING BOOST DIRECTLY]: But Karl, I don't have this magical wastegate actuator position variable! What do I do? Remember how I told you to calculate the "boost error"? That's a great way to also verify the health of the setup. The first log we looked at on DataZap was a positive example, let's look a negative one. While I've kept the wastegate variable on there, it's so you can see the strong correlation between wastegate tending to 0 and boost error increasing (hits 2psi!) in the upper RPM's. For me, underboost of more than 1 is questionable, underboost of 2 psi is definitely less than ideal. Check it out:
Example of high boost error (2 psi) and a ful
The last thing I want to touch on is what we moved to charts, which is the true "curve". If you noticed, we haven't talked much about the "Engine Torque" variable yet. It's Engine Torque and boost are very strongly correlated, so placing it in the boost article seems most logical. To show the correlation, check out the earlier excel snippet, or this graph: datazap.me/u/mrconflicted/drivetribe-example-log?log=0&data=2-8&solo=2
The strong correlation between torque and boost.
While engine torque is a simulated variable (meaning its an approximation done by the ECU) it's basically a real data point for the butt dyno. While this isn't perfectly smooth, it's overall not bad. As expected, the IS20 is pumping out loads of torque from 2000-4000RPM in its peak efficiency zone, and droopping steadily as it approaches redline. Changing the boost curve changes this torque curve. Gives us more mid-range, or more top end. It's a great way to look at holistically how the tune is setup.
This article is about twice as long as I intended, but I want to kind of leave you with three main things to look for. Firstly, if you have access to the wastegate actuator variable, make sure it's not dropping down to low (or is pegged at closed at upper RPM's). Second, look for underboost (or overboost). More than 2 psi is definitely cause for concern. Lastly, take a look at the overall curve. Does it look choppy? Is it nice and smooth? These are all things that make up a well-sorted tune.
I want to note for those of you who may already know all this -- there's a lot of "astricks" that could or should be placed throughout this article, but we gotta take things a step a time. Hopefully we get through them all, so if you keep reading, I'll keep writing! Stay tuned for a video version!