The 2017 Honda Civic Si hit the roads about a month ago, though we've yet to see one "in the wild" (at least on the streets of the Atlanta area). So when we rolled up to Night Owl Performance, the guys there thought we had shown up in some sort of pre-production model because none of them had seen any Sis on the street, either. But they really seemed to like the looks of our Rallye Red Si coupe, and this particular shop spends more time working on pretty much everything besides Hondas, so that seems like a step in the right direction for Honda. I told them that with this new era of 1.5L and 2.0L turbocharged Honda engines, they should expect to see an uptick in interest from those seeking to tune their new Hondas.

We're not sure if the reasoning behind the relative dearth of privately owned 2017 Sis is due to Honda's limited distribution of them, or if it's due to a relative lack of interest in the car. It could be a little bit of both, but for those who are doing their due diligence before taking the plunge, we're here to present some interesting data.

Modern Electronics
So in the interest of killing two birds with one stone, we also tested the latest edition of Hondata's FlashPro product (more on that in an upcoming, separate article) and we drove the car to the dyno with the FlashPro calibration already loaded. The INITIAL plan was that we'd get our numbers from the FlashPro calibration, reflash the ECU with the stock program, and then get the baseline figures. Interestingly, Hondata's latest FlashPro includes a cool feature that incorporates the Si's "Sport" button, and this feature allows you to switch between a "stock" calibration (in the normal mode), and the chosen FlashPro calibration by pressing the "Sport" button and entering Sport mode. So we figured we'd get the numbers for the boosted mode, then hit the "Sport" button to take it back to "stock" mode, and then, finally, we'd reflash it back to stock and get the real numbers for the "stock" calibration.

If only it were that simple. Our very first pass went strangely. We saw a tremendous torque figure early in the RPMs, but midway through the run, the Si's dashboard lit up like the Vegas strip with multiple warning messages and then exited the "Sport" mode on its own. Unfortunately the dyno wasn't configured properly so it didn't record the boost readings for the first couple of runs. On the next pass, we saw significantly less torque at the peak, but about the same horsepower. At this point, we noticed that we could no longer put the car into Sport mode, no matter how many times we pressed the button, or restarted the car. It's almost like the ECU went into a failsafe mode. Boost levels never went above 18psi, though on the street we had been consistently seeing numbers in the 23-24psi range. We tried disconnecting the ABS sensors on the rear wheels and made a few more passes, but it did nothing to correct the situation.

So, we decided to try reflashing it back to the stock program. We did that, and the peak horsepower increased by about 7-10hp, but the peak torque was about the same. Boost had only gone up to 18.5psi, so something still wasn't quite correct. We also tried pulling the ABS/VSA fuse, but this didn't change the situation with all of the warning lights. This is the dyno plot that's shown here.

Click to enlarge

Something VERY Important
At this point Doug was free and he called me back and reminded me that I should be sure that the Knock Control Index reading was somewhere below 60% when I dynoed the car. This data can be found in the Flashpro's data logging tables. Doug explained that whenever the ECU is reset, this index resets to 100% and it limits the peak torque numbers as well as the peak horsepower figures, until the ECU can "learn" the true knock index. He explained that the best way for the car to learn was to put it on the dyno which had an adjustable load, and then load the dyno up to a point where it holds a steady cruise at about 1psi in a given gear. With this in place, the ECU typically "learns" the knock limits in 3-5 minutes. We were already off the dyno at this point, and Night Owl's Dyno doesn't have a load module, so Doug told me to drive it around on the street and to watch the figure. I found that he was correct in that the quickest way to teach the ECU seemed to be to try to hold a load of about 1psi if possible. I had my phone connected to the FlashPro via bluetooth so I was able to monitor this parameter in realtime via the Hondata Mobile app. On the street, this process took about 15 minutes of driving, but given more ideal roads with less traffic and fewer lights, this could probably be achieved in about half that time. I got it down to 54%, then James and I went for a ride. He seemed to be fairly impressed with the car, particularly in "Sport" mode with the +9psi boost.

The problem with our "Baseline" dyno - HUGE DISCLAIMER HERE
So after learning all of this (actually, I believe Doug has explained this to me in the past, and I'd forgotten the details) I realized that it was unlikely that the Si's ECU had sufficiently "learned" the Knock Index before we tested it with the stock program. In other words, the dyno plot you see here very likely shows numbers that are below what a stock Si is truly capable of putting down. Peak hp may not change too much, but we're thinking the peak torque numbers of a properly tested stock Si will be better than what we've recorded. The good news is that we feel pretty confident that we got good numbers for the FlashPro later in the afternoon, and I also devised a pretty good method for helping the ECU adjust the knock index to its proper levels while the car was on the dynojet, but I'll share that info in the Hondata FlashPro results article.

Conclusions
We hesitate to make ultimate conclusions since we feel lke we don't have the most accurate data, but we do have enough data to conclude that this engine seems to be fairly underrated. Rated at 205hp at the crank, we saw just under 205hp to the wheels. And Honda claims 192lb-ft of torque, while we saw 193lb-ft. This is already a good thing, and these numbers are quite possibly on the low side of what the engine will do in stock form, due to our testing issues outlined above. We may try to get another Si to dyno test to make 100% certain that we have a completely kosher baseline plot. Since last week's dyno session, we learned of a special maintenance mode that may address the dashboard errors we encountered, so hopefully our future dyno sessions will go much more smoothly.

Compared to Civic Sport 6MT Hatchback
Below is a plot showing the Si vs the Civic Sport 6MT Hatchback that we dyno tested earlier this year, and you can see that the two plots line up pretty closely, right up until 5500rpms, where the Civic Hatchback's turbo seems to start wheezing a bit. The two engines are exactly the same, but remember that the Si uses a different trim for the turbocharger, and you can see the advantage of this trim.

Click to enlarge

Compared to 9th gen Civic Si*
First of all, note that these dyno sessions took place nearly 4 years apart, and on completely different dynojets, so there is likely some variance attributed to this latter fact alone. So just for fun, we overlaid the best pass we had for a stock 9th gen Si against this baseline for the 10th gen Si. As you can see, the little turbo 1.5 makes a lot more power and torque all the way up to its 6600rpm rev limiter. The K24 goes on to 7300 rpm but that's hardly enough to make up the average deficit. Despite all of this extra power, on the road performance between the two cars is similar, however, owing to the boost sag encountered between shifts on the 10th gen Si (fortunately, Hondata has a feature to address that).

Click to enlarge