Backported gen2 BRZ throttle map for gen1

[callisto]

After my failed EU PI adventure, I came across a Reddit post where someone published the Subaru BRZ gen2 "Requested Torque B" map, the one that converts pedal% to a torque figure, which is then interpreted by the other maps. I've been trying to improve the smoothness of my 2019 for some time now and after my 2013's experiences with OFT (cc Toyota GR) throttle, and being eternally curious, I decided to try it.

Flashing it directly to gen1 does work fine, except that you'll hit WOT at no more than 50% pedal (as opposed to 85% pedal on stock Subaru) at any given rpm/load point. Not a very smooth experience when it crosses from 23% throttle at 1600rpm to 100% throttle at 2000rpm, but exactly the same as flooring it at 1600rpm. (I did study the throttle code extensively and it will not request more than Max Torque, no matter what Requested Torque says, so this is a safe test insofar as randomly flooring the gas at various rpm/load points is safe.) But, before it reached that WOT point, it was a lot smoother to drive it when operating exclusively below the dip.

Twenty hours of spreadsheet work later, I finished rescaling the gen2 map to gen1 and it's been running on my vehicle for a few weeks. It's absolutely dreamy — essentially, the new map smooths out the 'jaggedness' of the gen1 map and replaces it with a smoothed slope to 'peak' torque, while keeping WOT above 65-70% throttle (which is a bit earlier than the original 85%, so this probably isn't as suited for track use after this, where I would imagine it's more likely that 85% is shifted to 100% to increase pedal range and fidelity, but that's not a topic I'm familiar with.)

The gen2 map turns out 0-100hp from 28% and 100-200hp from 28-100%. I tried only rescaling the 100-250hp range but could feel a noticeable G-forces breakpoint whenever it crossed over, because the rate of acceleration would change unrelated to gas pedal input and that's precisely what I'm trying to get rid of. So I ended up having to rescale the entire 0-250 range down to our 0-200, with a lot of nuance in the 0-2400rpm / 28-100% throttle ranges since gen2's engine is just More Capable and a simple map-wide rescale wasn't cutting it. And then I redid that nuance in a rescaled map at 3% intervals. And then I cross-checked everything against its 'throttle% result' using the three throttle maps in mafscaling (two stock and one new Req Torq B). And then I reevaluated something or tweaked something or changed the nuance or completely revised how I was scaling. Repeat a thousand or so times.

Anyways, here's the resulting gen2-backported Requested Torque B table; note that both the X-axis and Y-axis have changed, so double-check them after pasting! Various images attached and copy-paste below.

First two images: The gen2 map from RomRaider, then comparing the gen2 map (rescaled to gen1 x/y axes) against gen1 facelift stock. Lots of +/- going on here that makes more sense when considering the following.

Next two images: The gen2 and gen1 maps presented linearly at 3% torque, 200 rpm intervals. This is perhaps the starkest difference between gen1 and gen2 throttle. For me, city driver who never tracks, the improved fidelity at 0-28% pedal is invaluable, and so I'm glad I put in all this work — but for gen2 drivers, it should be possible (and I imagine the pro shops have already done so?) to take the gen1 torque contour and apply it to gen2.

Caveats: This is only useful for non-boosted engines. I have a reddit screenshot of a Subaru BRZ gen2 table. I do not have the raw table values from gen2, because I don't have an unencrypted gen2 ECU image. Rounding errors of up to (est.) 0.4% may apply to any result value >=80 or so. If someone has an unencrypted gen2 ECU image, I would dearly love to get the precisely-accurate values out of it; please DM me.

[Table3D]
0.0	0.47999998927116394	2.0	4.5	7.0	10.0	14.0	17.5	21.0	28.0	35.0	40.0	55.0	75.0	85.0	100.0
800.0	0.0	1.0	11.9609375	14.8984375	24.734375	42.8671875	54.265625	58.3984375	77.4375	95.734375	112.140625	114.34375	120.9453125	129.7421875	134.140625	139.0
1000.0	0.0	0.046875	2.9921875	7.4453125	19.7890625	29.7578125	40.59375	54.8828125	75.6953125	95.734375	114.40625	117.90625	128.4140625	142.4140625	149.4140625	158.5
1200.0	0.0	0.046875	0.46875	5.9609375	14.84375	28.5546875	40.25	56.421875	76.46875	95.734375	133.4921875	135.875	143.0078125	152.53125	157.2890625	164.4296875
1600.0	0.0	0.046875	0.5	4.96875	13.8515625	29.5390625	37.1015625	51.8984375	74.2421875	95.734375	135.2734375	138.453125	147.9921875	160.703125	167.0625	175.4453125
2000.0	0.0	0.046875	0.1015625	3.96875	11.875	27.96875	33.8359375	47.1171875	71.828125	95.734375	139.0625	152.8984375	161.546875	173.078125	178.84375	186.46875
2400.0	0.0	0.0390625	0.1015625	3.5546875	10.21875	28.5546875	35.234375	41.09375	68.8203125	99.5625	139.171875	167.1484375	180.8515625	190.375	194.109375	211.0
2800.0	0.0	0.0234375	0.1015625	0.4296875	6.9609375	29.265625	35.78125	42.2421875	69.40625	101.0	137.21875	162.453125	182.6796875	202.59375	208.90625	213.5859375
3200.0	0.0	0.0234375	0.1015625	0.4296875	4.5546875	27.421875	34.0390625	40.6171875	68.6328125	103.8671875	138.09375	161.515625	183.046875	207.9765625	217.609375	222.0625
3600.0	0.0	0.0234375	0.1015625	0.4296875	4.1328125	25.4921875	31.609375	37.6875	68.1484375	100.515625	136.1953125	160.5703125	181.5859375	207.9296875	217.609375	222.9140625
4000.0	0.0	0.0234375	0.1015625	0.4296875	3.6015625	24.1953125	30.8203125	37.3828125	69.40625	102.4296875	136.375	160.5703125	179.9375	200.40625	213.2578125	217.828125
4200.0	0.0	0.0234375	0.1015625	0.4296875	3.1953125	23.28125	29.671875	36.015625	68.2421875	99.5625	135.3671875	160.5703125	180.1171875	197.6640625	208.03125	211.890625
4400.0	0.0	0.0234375	0.1015625	0.4296875	2.9765625	21.6640625	28.0390625	34.359375	66.890625	98.6015625	135.28125	160.5703125	179.9375	197.28125	214.125	218.671875
4800.0	0.0	0.0234375	0.1015625	0.4296875	4.6015625	17.6875	25.8359375	33.9140625	65.2421875	95.734375	134.40625	161.515625	177.5546875	199.765625	216.7421875	221.21875
5200.0	0.0	0.0234375	0.1015625	0.4296875	2.3984375	12.15625	20.8359375	29.4453125	59.3359375	88.359375	131.515625	161.515625	175.90625	198.8046875	215.0	219.5234375
5600.0	0.0	0.0234375	0.1015625	0.4296875	2.375	5.90625	15.328125	24.6484375	53.7265625	82.0390625	127.75	159.6328125	178.8359375	205.2109375	214.125	216.9765625
6000.0	0.0	0.0234375	0.1015625	0.4296875	2.3671875	3.375	12.671875	21.8828125	49.5625	76.5859375	120.1328125	150.2421875	177.5546875	205.96875	214.125	219.5234375
6400.0	0.0	0.0234375	0.1015625	0.4296875	2.359375	6.0390625	10.3671875	21.9453125	47.234375	71.796875	112.578125	140.8515625	179.75	209.0546875	211.515625	216.1328125
6800.0	0.0	0.0234375	0.1015625	0.4296875	2.3671875	6.078125	10.7421875	20.6328125	44.4296875	67.5859375	105.9375	132.40625	177.0	206.390625	208.90625	213.5859375
7000.0	0.0	0.0234375	0.1015625	0.4296875	1.09375	2.2890625	6.3125	16.90625	41.625	65.671875	103.0546875	128.6484375	170.953125	201.9140625	204.5546875	209.3515625
7200.0	0.0	0.0234375	0.1015625	0.4296875	0.6796875	1.0546875	2.3515625	12.1484375	38.3359375	63.8515625	100.1875	124.890625	162.7109375	194.0078125	197.5859375	212.0
7400.0	0.0	0.0234375	0.1015625	0.4296875	0.5390625	0.640625	1.3828125	9.1328125	35.9140625	62.1328125	97.4609375	122.0703125	153.1796875	183.3046875	185.40625	189.859375
7401.0	0.0	0.0078125	0.1015625	0.4296875	0.5390625	0.6328125	1.3828125	9.1171875	35.9140625	62.1328125	97.4609375	122.0703125	153.1796875	183.3046875	185.40625	189.859375
7402.0	0.0	0.0078125	0.1015625	0.4296875	0.4921875	0.6328125	1.3828125	9.1015625	35.9140625	62.03125	97.453125	122.0703125	153.1796875	183.3046875	185.40625	189.859375

[callisto]

Notes:

This map should be compatible with any non-boosted 2012+ 86; the table dimensions are the same in all of gen1 (and apparently in gen2?). Pre-facelift / EU vehicles, if you use this unmodified you will reach wide-open throttle a few percent earlier in the pedal travel (at, say, 60% rather than 65%) at 2300+/-100 and 7200+/-200 rpm. You could rescale the map with a mult(iply) by 0.95 in RomRaider (leave the 100% column intact) to work around that, or redo the entire remapping process, but I suspect it's unnoticeable in non-track use.

If you have a boosted gen1 86 and you apply this throttle map, it probably won't break the car, but it will definitely cap your gas pedal torque at ~220. That's not necessarily a bad thing but most people boost this engine for higher torque, I assume?

After flashing, remember to do the throttle travel relearn process — off, pause ~5s, on-not-running, pause ~10s; depress gas fully, pause ~1s, release gas fully, pause ~1s; repeat ten times; off, pause ~1s, done — so that you're comparing apples-to-apples with the throttle before/after the flash.

To calculate target throttle%, open Kodename's mafscaling tool to tab Throttle Maps and paste in all three tables — Requested Torque Base (right-click, Paste Vertical), Requested Torque B (stock, this one, etc), and Target Throttle Plate Position Maximum; click GO. The result lets you see what throttle% you'll get for, using the Req Torq B table of your choice, given a pedal% and rpm. The ECU hard-coded max throttle% is 102.3816 and the stock map reaches that by 85% pedal at all rpms. (Similarly, the 100% column in gen1 stock, likely gen2 stock, and this rescaled map all target 105% of the Plate Pos Max value for any given RPM; some bilinear scaling was used to pin this down accurately.)

This torque map expressly prioritizes smooth changes in torque (and thus G-forces) over smooth changes in throttle; while the rescale tends to be slightly sharper at certain transitions around the dip, both stock gen1 and gen2 treat the throttle as 'something that can be varied by tens of % without any change in pedal input'. You can still see a bit of evidence of the gen1 'tuning fork' around the dip as the map approaches max throttle. (Ignore the bolded cells, that's mistakenly leftover formatting.)

If you let your car auto-learn fuel trims and knock correction, great. Otherwise, there could be unforeseen drawbacks of this map if you have fuel or knock learning 'dialed in' through changes to your timing/etc maps and have disabled STFT / LTFT / FBKC / FLKC (this is not common). While this map change does not alter the engine timing advance, it does alter the times in your driving load/rpm cycles at which the engine will change the throttle for the same foot-pedal inputs you're used to; please monitor your knock logs for a while afterwards.

I didn't rescale the gen2 Toyota GR86 throttle map because it's described as 'aggressive' and strongly resembles what I remember seeing from OFT in gen1 pre-facelift: namely, an even sharper onset of throttle at the slightest hint of gas pedal. In the above linear map images, I suspect it would look like the map jumps from white to yellow and straight on to red from there.

I didn't post the full spreadsheet because it's an absolute clusterfuck and having it would not help reproduce my results. I recognize this isn't optimal but, as Wonka famously said: "Invention is 93% perspiration, 6% electricity, 4% evaporation, and 2% butterscotch ripple"; a notable chunk of what makes this map better than a straight rescale (I tried that first, tens of hours ago) is me, manually annotating and editing or interpolating specific cells based on feelings. Here's a screenshot of the spreadsheet, though:

[Ruben_86_]

I have a Gen1 with OFT STAGE 2+, I never liked how progressive the accelerator pedal is, I tried to load the values of the original map and it seems to me to work better. Will this work correctly or do I have to modify something else?

[callisto]

Originally Posted by Ruben_86_

I have a Gen1 with OFT STAGE 2+, I never liked how progressive the accelerator pedal is, I tried to load the values of the original map and it seems to me to work better. Will this work correctly or do I have to modify something else?

Assuming you have previously loaded, saved, and flashed tunes with RomRaider; then, yes, pasting in this table to Requested Torque B should work correctly, same as your prior revert-to-OEM edit; for a quick double-check that you pasted correctly, look at five places: rpm top row 800, bottom row 7402; pedal left col 0.0, right col 100.0; and, value 4200rpm @ 55% throttle is 180.1 torque. If you're missing one of those values then your paste failed, revert and try again. Do the throttle re-learn after the flash (it's above in the notes).

You won't be editing any other torque table but Requested Torque B (Accelerator Pedal); your tune already has functioning Base and Max tables and they don't need to be modified for this to work.

Note that I'm not sure what OFT Stage 2+ and intake stuff does to your max torque; it's possible that your OFT Requested Torque B table after rescaling to the new x-/y-axes, has in column 100% target torque values that are higher than (1.05 times) what's in either your OEM table or my table here. That shouldn't much matter — if it did, you would have noticed when you reverted to stock Req Torq B! — but if it does and you find you're not reaching WOT, either on my backport or on stock — or if you feel perfectionist / wish to verify as I often do — you can calculate your tune's precise 100% column values for Req Torq B as below. If your values are significantly different by more than a few percent, this map probably isn't a good fit for your tune; see the notes for boosted engines above, and consider working with a professional tuner (I'm not, sorry!) to reinterpret this all / design a better fit for you.

1. load mafscaling, switch to table rescale, paste your Requested Torque Base table
2. in the lower pane, edit the rpm column headings to match the rpm list in the new Req Torq B map (not the old one!); should be 800..7402, and 7400-7401-7402 will have the same value.
3. check that the bottom rescaled table has a sensible value for 4200 rpm, should be halfway between whatever you see for 4000 and 4400 in the top
4. right click, select all, right click, copy
5. go the to throttle maps tab, right click the first cell of Reqd Torque - RPM, select Paste Vertical
6. select the entire torque values column only (not the rpms), right click, Copy
7. open a notepad, type [Selection3D], hit return, Paste the clipboard
8. select all, Copy to clipboard, switch to Romraider
9. select all values only in the 100% column, in the table's Edit menu, choose Paste
10. leaving the now-altered values selected, in the top toolbar just left of the Set Mul buttons, enter 1.05 and click Mul one time only; confirm that the value for 4200 rpm is now 1.05 times larger than the value checked in #3 above

Standard ECU editing advice (that is not unique to my table change) for anyone reading this thread — If you have not previously edited and flashed your tune in RomRaider, then you need to work out the usual 'get the correct definitions file from your tuner' process and make sure that when you open up the rom for editing, you see reasonable values in all the torque tables before you go around pasting anything. After flashing, remember to do the throttle travel relearn process (and if you desire/need it, the much longer idle relearn process). If you're flashing your daily driver, be aware that it takes around five days to get a used ECU in from eBay in order to replace yours if you brick it.

[zc06_kisstherain]

a lots of good work here. i am no expert in tuning though

[DarkPira7e]

Thank you for continuing to innovate!!! Very cool

[Compelica]

Have tested the back port, including doing the rescale for the max torque and coming from the OFT Stage 2+ throttle mapping; the differences are rather perceptible.

Subjectively by feel - OFT's mapping feels like an exponential curve where the first 50% has not much difference in throttle opening, but ramps up after. In my opinion this allows the driver to modulate torque demand easily by allowing additional travel from 0% to 50%, especially for novice drivers who are still too deliberate with the throttle. This also comes in handy in certain driving situations eg. taking a sharp turn from standstill on low grip surfaces, or modulating throttle when doing donuts for gymkhana.

The BRZ gen2, as expected from previous explanation is a lot more linear. You would be doing a lot more throttle opening in lesser pedal movements, which certainly feels more responsive as a result. Throttle blips are a lot more easier on this mapping as well, since you only need to tap a little for the desired throttle opening to raise RPM, whereas for OFT you would need to stab deeper/longer.

Will try the gen2 mapping for this week, though an additional concern on my end would be fuel consumption given that the throttle mapping likely would be more susceptible to moving from CL to OL.

[callisto]

Yeah, it's entirely possible that it triggers open loop more often; now that I've stabilized and am not flashing a new throttle map every couple weeks, I can collect long-term logs and see how much open loop is triggering below the slope-cutover point and take steps to deal with that if necessary. (Steps that aren't simply available with a throttle map.) Four fillups so far on this map and no obvious change in gas consumption, so at least it's not going to be severely worse.

I also found an analogy that may be familiar to others: Xbox Pro controller thumbstick curves. (These numbers are all approximate for analogy purposes, I have not done any complex calculations to try and collapse the pedal-torque vector fields into linear equation sets like below.)

The gen1 BRZ is akin to the controller's 'Aggressive' curves: very steep incline y=3.0x with OFT, at { 0 < x < 28% } pedal, soft incline y=0.3x at { x > 28% } pedal; you lose a great deal of fidelity in the high-torque space. One example of this in the above linear maps is at rpm 2600, pedal 10-13%; gen1 BRZ jumps from 13 to 61 torque, while my map jumps from 29 to 34 torque. As you note, this means you're potentially using more fuel in this range; in this region, the OEM map is a bit more fuel-sippy below 13%, but if you're driving in a curve and your pedal inputs increase slightly from 10 to 13% and the car tries to add +25% torque as it does here, then you have a much worse experience in the curve due to the sudden sharp increase that made the fuel-sippy region possible.

I imagine OFT is similar, at more like y=6.0x at { 0 < x < 10% }, very soft incline y=0.1x at { x > 10% }, with even more fidelity in the high-torque space in exchange for essentially not having a low-torque space. And toyota GR's curve is basically all the way to the Xbox 'Instant' setting: either you're using gas at full steam ahead, or you're not using gas at all, which is certainly great for efficiency but has some stark drawbacks in sloshing the vehicle's G-forces up or down each time you cross that pedal crux point.

The gen2 BRZ's curve doesn't really have an easy analogy in this Xbox description, because instead of having _/‾ two linear zones joined by a transition zone like gen1 BRZ, or _¦‾ two linear zones fully disconnected like toyota GR, it just has _╱ two connected linear zones without a transition zone at all; soft incline y=0.3x at { 0 < x < 28% }, steep incline y=3.0x at { x > 28% }. There's still a G-forces transition across the 28% pedal crux point, but rather than being +20% torque like in gen1 BRZ, it's only +3% torque.

So for me, with my prior-known preferences for the gen2's curve style rather than the others available, both in thumbsticks and as I suspected in pedals, stabilizing the G forces of the car in response to fractional pedal inputs makes the weight shifting more predictable for my particular proprioception. I absolutely assume that for stock vehicles on stock tires, the 'breakaway' nature of the car's desire to slide in gen1 was specifically due to the pedal inducing 'breakaway' through these sharp transitions at very low inputs. For my purposes, though, since I've spent most of my time on this platform minimizing sliding — I learned to drive in heavy night rain on interstate highways with loaded logging trucks in the next lane, so I am averse to my street car entering a slide as readily as it was sold to! — shifting to the gen2 curve style makes much more sense for my personal case. It's also made the car a lot less likely to spring undesirable G-forces on me in the middle of long curves in high gears for smooth and gentle pedal changes.

Now that I have linear throttle maps, my next analysis is to overlay the six gear ranges onto the linear map grid so that I can see about softening the fuel usage along the 'coasting' left edge of each gear — and, Compelica, thanks for trying to get through to me about this — I will also see if I'm causing OL 'hot spots' where the car ought to just stay in CL. Not sure if I can adjust those with throttle maps alone, but I've technically read through the CL/OL code a bunch of times for other quests (with varying levels of comprehension), so maybe on a future read-through I'll find some hidden map that lets me inhibit throttle-map blips without impacting pedal-actual requests.