Veloster Turbo Forum banner
1 - 5 of 5 Posts

·
Registered
2016 Veloster Turbo DCT Ultimate
Joined
·
397 Posts
Discussion Starter · #1 · (Edited)
Limits
As many of you know, there are 2 air pressure sensors* on the car: One on the Intake Manifold (MAP) & one on the stock intercooler (Boost/Upstream Throttle). They used to be 2 different part numbers I believe are now the same part and interchangeable. Both sensors are 3bar.
1bar = 14.5psi. 3bar = 43.5psi.
Technically, 3 air pressure sensors. There's the barometric sensor on the grill. It's used to report atmospheric pressure to the ECU. All 3 are tested against each other to make sure they are within spec.

The pressure sensors report in absolute pressure values. You have to subtract 1bar for the atmospheric pressure to get what we care about, Boost.
3bar - 1bar = 2bar. 2bar = 29psi in boost

You'll notice if you raise the boost on your car, if you're logging the intake manifold air pressure sensor (referred as MAP sensor), it will max out at about 21.x psi.
This is because of 3 things:
  1. ECU stores the value in 8bit. Meaning 255 is the highest this value can be. Hard limit. Blame an engineer at Bosch.
  2. ECU sets the max observable limit for both sensors at 2500hPa. Soft limit. This value can be changed.
  3. ECU has offset "0" to account for vacuum by 1bar.
    1000hPa = 1bar. 2.5bar = 21.75psi boost & 14.5psi vacuum
If you're logging the "Boost" sensor, you'll notice the car will continue to read to 43.5psi* (unless you have a built block or a death wish, you're unlikely to see this). The boost sensor isn't offset for vacuum, so 0 is atmospheric. With this sensor, you can monitor boost up to 29psi. Although both sensors are 3bar sensors, they are both programed to read as 2.5bar.
max observable limit ≥ sensor limit
43.5psi might be incorrect. If the max observable limit applies here too, 36.25psi would be the limit. Alternatively, the programming logic might still report the value up to 3000hPa (16bit value for this sensor). Then the value meets a gate with the max observable limit, reporting the lowest value of the two. In this way, nothing higher than 2500 gets to the numerous functions that depend on this value.

For the purposes of tuning, as you start to approach the max observable limit the ECU begins to freak out. When actual boost pressure ≥ max observable limit, the ECU has no value to process for any of the functions, maps, safeties any further. Most importantly the Boost PID can't regulate properly at that limit. Leading to all sorts of cuts. To solve this you can raise the max observable limit, while trying to keep your requests below that limit. This solution is plenty enough to satisfy the needs of 98% of tunes, as most will never run more than 29psi on a stock engine.
PID is an general engineering concept. If you want to understand it more I found this video helpful, just imagine the object is target boost. Here it is applied to control boost. Which turns out, is a very delicate balancing act from an engineering point of view!

Bosch uses a good ole (y=m(x)+b) function to determine converting voltage from the sensors to hPa for the ECU.
Rectangle Font Circle Parallel Magenta


In the case of the Veloster:
m = 628.6250hpa/v
5v * 628.6250 + b = 3143.125hPa + b
So, for 5volts the sensor should be able to report ~45.59psi? Not quite. Although there's the offset b to consider, it is a small adjustment. What really comes into play are the mechanics of such sensors. As you approach 0v or 5v, linearity gets worse and the sensor starts to under/over shoot values. For this reason, a dead zone is left at either end, making 0.5v-4.5v the "usable range" (~2501.88hpa).

You can continue to extend until max observable limit = sensor limit . Congratulations, you've maxed out the sensor and you've managed to keep all of the pistons inside of the block at ~31psi!
 

·
Registered
2016 Veloster Turbo DCT Ultimate
Joined
·
397 Posts
Discussion Starter · #2 · (Edited)
Theories

Once you max out the sensor. From my research on other platforms with this problem (but with bigger tuner markets), aside from moving to a standalone ECU, there are only 2 ways forward:
  1. Under-scale the current 3bar sensor
  2. Upgrade to 4bar sensor & re-scale

Under-scaling:
Under-scaling is where we change the math so that actual 40psi = ecu 31psi. This is essentially how alot of piggyback tunes handle increasing boost. Except, instead of changing the math, piggybacks just reports smaller voltages from the sensor. On the ECU side, we're decreasing the hpa/v. By decreasing m = hpa/v we can make actual 40psi = ecu 31psi, in turn making actual 14.5psi < stock voltage for 14.5psi . There are several problems with this approach:
  • If 31psi is actually 40psi, you have to go and change EVERY map related to this value. Too Time intensive.
  • All 3 pressure sensors are tested against each other. You have to get the numbers right for the car to even start
  • None of your values/data have any meaning
    • there are many models to calculate things like: EGT, load, air charge, torque etc...
    • changing this variable affects the output of all of them. Too many unintended outcomes
    • you can't compare any single datalog against any other car now that the numbers mean nothing
  • You lose of resolution from the sensor, making tuning more difficult
It is likely not worth it to go down this path because you've already spent all this money on a block that can handle 31psi, why fly blind to 40psi?

Upgrade the sensor:
If you were to upgrade the sensor, there would be no need to fudge the numbers. The sensor itself is constructed to report 4bar as 5v. The sensor would report actual 14.5psi ≈ stock voltage for 14.5psi This is more desirable for several reasons:
  • You only need to tweak the numbers to calibrate, not misreport.
  • You aren't required to touch every subsequent map related to the air pressure since actual 20psi = ecu 20psi.
  • With more available sensor resolution, you can re-scale various maps to address areas in that range of load.

Steps to do this:
  1. You need the specs for your new sensor. min(i volt = i hPa) & max(i volt = i hPa)
  2. From the previous post's equation #1, there are 4 variables that need to be changed.
    Code:
    max hpa, max volt, min hpa, min volt
  3. This gives us our new m = hpa/v
  4. equation #2 gives the offset
In total, 6 values need to be changed, per sensor.

That all sounds great! Why hasn't it been done yet?
  • Demand
    • There really isn't a need for this. 98% of enthusiasts will never see more than 23psi. The block is more of a concern than making +29psi.
  • Software Limitation
    • From what I've seen from the Bosch FR's and in the bin, the values that store map and boost pressures are 16bit and can store values up to 5120hPa.
      • the ends of those paths can also store 5120hpa
    • Theoretically I don't see any reason this can't be done.
  • Hardware Limitation
    • There isn't a plug & play option on the market for this
    • Not too hard of a wiring job provided you can find a 4bar MAP or TMAP sensor that fits into the existing hole
      • Or take the path of no return and increase the size of the holes to fit.
It's a very interesting experiment I'd like to try out. It's just outside of the scope of my goals with the car. Maybe as a proof of concept.
 

·
Registered
2016 Veloster Turbo DCT Ultimate
Joined
·
397 Posts
Discussion Starter · #3 · (Edited)
Proof of Concept:

In the testing I've been doing, I noticed I was having issues with either surge or overboosting on shifts. Either is a problem, but logging with torque pro meant the resolution of my data was not great. Not knowing if you're over boosting or by how much is another big problem. Below isn't an example of the overboost issue I was having, but is an example of what the car was observing.

Note: The "turbo boost pressure s2psi" is an absolute pressure value. It includes atmospheric. I'm not actually making 40.83psi.
40.83psi - 14.5psi = 26.33psi

Slope Font Rectangle Parallel Pattern


So with the max observable limit at 2500hpa, the car probably would've shut the party down as I reached 36.25psi because there isn't enough information for the PID to make informed decisions on boost control. But since I've extended the max observable window to about 2800hpa here, the car has the information to decide if this amount of boost in controllable. This solved 80% of my problems in the middle of run

Below is an example of my overboost problem on shift. Just before shifting, the peak boost is about 39.38psi.
Slope Font Rectangle Parallel Plot


When the car goes to shift, the wastegate full opens to kill power, likely due to my BOV setup. My logging resolution isn't fast enough to catch what happens but I imagine there is a rather large spike in that fraction of a second. the gradient it approaches the max observable limit, or passes that limit, is likely alarming to the ECU and it kills the party.
The reason I was having this issue is because I run a Tial Q50 BOV and I bypassed the BOV solenoid and ran it straight to the intake manifold. I know better now but the ECU uses the BOV+solenoid to regulate boost as well. With the BOV to the IM, it only operates if you take your foot off the pedal, or the throttle closes. If this was a manual, I probably wouldn't have noticed this problem as much. But with the DCT I don't take my foot off to shift. The ECU can anticipate a shift and use the BOV+solenoid, along with other things to make sure pressures don't spike. Without that control, the ECU can only react to the pressure spike. Lesson learned: Keep the solenoid.
Below is an example of raising the max observable limit near the sensor limit (and I also fixed the hardware problem). The ECU and log should be able to read up to 43.5psi now.

Brown Font Slope Parallel Rectangle


And when the car goes to shift, 41.99psi. At the very least I was experience a 3.2psi spike on shifts. And the logs could still be missing more! I imagine before fixing my BOV setup it was seeing more. You'll also notice the waste duty cycle is ALOT smoother, because the ECU has more control with the BOV and more information.

Now it's hard to say I've proved anything yet, because this is the most boost the car has ever made. So perhaps reading up to 43.5psi is already possible without the changes I've made. But I do believe it helps keep the ECU happy up there.

I'm more inclined to try this out with a 3.5 or 4bar sensor this summer. I know my friend will finally have his fully built ecoboost mustang back out of the shop. If the 4bar sensor on his car has the same bosch connector, I'd like to test it out. As the sensors are +100$ I'd like to throw away money at a failed experiment.
 

·
Registered
2014 Veloster Turbo Automatic
Joined
·
204 Posts
This is lovely. Would a more granular sensor allow for smoother delivery of boost, even though you don't need the extra headroom?I'm curious if the upgraded MAP sensor has any desirable benefits to people not pushing huge power
 

·
Registered
2016 Veloster Turbo DCT Ultimate
Joined
·
397 Posts
Discussion Starter · #5 ·
I'm not so sure, I've seen arguments for both sides.
  • Having a bigger map sensor is fine.
  • Having a bigger map sensor will reduce resolution.
I guess it depends on the ECU. If it doesn't have a high resolution, reading 3bar in 0v-5v is fine. But trying to read 3bar in 1v-2v might be a struggle. If the ECU can resolve 1v-2v all the same, then... there isn't a reason to spend +100$ on a sensor and whatever on tuning lol.

I don't think it would change the experience for drivers or tuners if you aren't above the 3bar limit.
 
1 - 5 of 5 Posts
Top