I get that, I had it happen to me. I thought that people who have fiddled with the steering column and placed the steering wheel 360 deg out of position get errors from the active handling system. Maybe this isn't true and it's when it's only out a few degrees.

 

There's no index signal coming from the SWPS (at least for '01-04 cars), and oharal's description aligns with my understanding.

Just as you are, I'm pondering how to ensure that what looks like "center" IS center.
Currently my code requires the steering wheel to be centered within +/- 45 degrees whenever the device is powered on (otherwise it might incorrectly interpret center). To help combat this I've got a placeholder for a center-detection algorithm that should reliably extend the current +/- 45 deg range to +/- 135 deg.

I determined that for how I'm planning to manipulate the SWPS signal, proper center calibration is also imperative to ensuring the converter device's output properly matches what the car should be seeing.
To that end, yesterday I implemented a center-calibration technique: on the converter's first power-up (and any time later) the user would center their car's steering wheel and press a button on the converter device. This would lock in an offset variable derived from any discrepancy between steering wheel center position and SWPS electrical center/zero position.

On another note, now I'm leaning toward the device utilizing a LCD screen full-time... originally I planned the LCD was only for my own development/prototype. These are the items I've come up with so far that can be displayed to the user:

• SW angle
• which angle kit is selected (including stock / no angle kit)
• calibration status & reset calibration
• raw data (for troubleshooting).

Lastly, today I received a shipping notification for my FDF angle kit! Super exciting.

At least for the '01-04 SWPS, the signal repeats every 360 deg:



The only issue is if it were installed without the centering pin, otherwise its center position should always be within a couple degrees of the steering wheel's center position.

 

97-00 had an extra index signal. More or less a multi-turn pot.

 

I don't think you need your converter box to know what center is because the car will be interpreting "straight ahead" as usual just by reading your compressed output signal from the box. I'm curious about your thought process about it though in case I am missing something.

 

He is converting the SWPS signals into a steering wheel angle and then converting the angle back into simulated SWPS signals. So, the box needs to know zero to be capable of doing that.

You are correct that it doesn't need to truly know zero, but it does need to be outputting a signal that is close to what the EBTCM expects for zero when the wheels are straight ahead. The EBTCM will throw a fit if the signals are too far off when going straight down the road.

I think you could assume that red=0V and blue = 2.5V is zero and work from there to do a linear compression of the signals. However, I suspect that a non-linear compression will be required and I think MetalMan2 is assuming that too and that he'll have to account for it.

 

@lionelhutz nailed it, thanks for saving me some typing haha. I fully expect that with an angle kit the SWPS signal manipulation will have a non-linear relationship, and my planned manipulation strategy won't work if the proper steering wheel center position is unknown/wrong.

For fun I made up a graph to show a guess at what stock and angle kit could look like, along with trendlines/equations that I'd generate to compare them:



It wasn't my plan to start working through this next part yet but it's relevant to our current discussion and would need to be done eventually...
I developed a process that manipulates the SWPS signal while used with an angle kit. The outcome of that is the graph below, which compares "stock" and "modified" (angle kit) SWPS signal outputs vs SW position.
Keep in mind these are just based off the above guesstimate graph and so aren't real. However once I measure actual stock & angle kit wheel angle vs SW position this process will be applied.

 

You explained it better though, with graphs.

 

Have you tried to contact the manufacturer of the kit to see if they have any engineering information on how the steering angle vs steering wheel position is changed when the kit is installed. I assume they must have done some measurements. They might not want to share it but it may be worth a shot

 

Short answer: no.
Long answer: I contacted FDF (via their online form) several weeks ago to see if they might be interested in partnering with me on this device, but no response yet. I'm sure they're swamped this time of year. That tells me they might not respond to any inquiries other than order status.
Also if they did have such data and were willing to share it, I wouldn't expect it to be of the precision/fidelity I need. I've got a solid game-plan in place for collecting this data; some of my thoughts are sprinkled across several posts such as post #33.


BTW guys, I really appreciate you following along with and contributing to my progress!

FDF Mini Mantis angle kit is scheduled to arrive tomorrow!

 

I love this thread, where it is going, and the ingenuity to reverse engineer oem signals and then spoof them in an attempt to make the AH function. I am absolutely subscribed and would do so even if I wasn't thinking about a quick steer kit.

The neutral zero concept. What if the signal has no non-linear aspect since oem resets or determines the range of zero at start? Maybe then all we need to do is convert the ratio of the quick steer signal to the oem range and the bcm will adapt since it is just processing the raw feed.

Also, let's say we get the signal to be properly adapted so that AH believes it is within range. If it using the information provided, as a random example, that it believes the wheels are pointed to 45* and the car is sliding, and it begins to apply the factory programmed correction, but the wheels are actually at 60* and we have applied far more inertial shift to the vehicle, would AH then be essentially useless or more dangerous? Or does AH have other inputs which help it determine the physics applied to the car?

 

If you're investing the effort, you might also contact more than one vendor to see if any of them would be interested in working with you to support the product. Ultimately, the fix you are developing can be adapted to any of them but it wouldn't hurt to have material, intellectual, or financial support during development.

GM being what it is, I wager once you the C5 figured out they probably have similar sensors for the C6 at least. Though I bet they use them differently, again "because GM."

 

This will correct it so the car knows the angle of the wheels. The problem right now is that the car things the wheels are turned less than they really are.

 

I'm not sure I am following what you are asking. Ultimately for adapting to quick steer, the original SWPS sawtooth voltage signal still needs to be a sawtooth voltage signal, but it needs to occur faster. And how much faster it cycles increases the further the steering angle gets from center.

@lionelhutz right again!.
For Tusc's post, the description actually sounds to me like how the car is in factory form with a quick steer kit.

Yeah I've got placeholders for the FDF Mini Mantis, SLR Speed, and Vettenuts angle kits. The numbers will differ since the tie rod mounting locations aren't the same, but the device will certainly apply to all of these.
My only holdup is that I can't collect data on angle kits I don't have installed. If I didn't have any vendor support (and since I don't have plans to buy multiple angle kits) it might be possible for others to collect that data for me... i.e. they have their angle kit installed along with my converter device set to "stock", and then they measure wheel angle with their cell phone (same way as I am) and compare that to the SWPS readings reported by the converter.
I could then send them a file to flash onto the converter device and then they could verify angle measurements.

That's a very real possibility considering all the other similarities between late C5 and at least the early C6.

 

FDF Mini Mantis kit arrived!! Have rod end rubber boots and new inner tie rods on order.

Made a lot of headway over the weekend. For a functional prototype it is nearly complete. Just needs a few more items:

1. wheel angle vs SW position measurements, and then implement those into the code (basically recreate post # 46 with real data)
   1. repeat with angle kit installed
2. detection of SWPS errors
3. code that verifies SWPS values at center when the device starts up (this kinda ties into error detection)
4. see how the EBCM handles a voltage signal from my device... should be quick but I've been busy with other Corvette projects (starter replacement/upgrade, DIY hot balance clutch/flywheel)

There's now a fully featured menu system and I took a video of it as well. These are the highlights:

1. Start screen / status screen: says what the device is and provides status (i.e. 'needs calibration', 'okay', or 'error')
   1. Prior to calibration, button long-press sets the center calibration from this screen
   2. After calibration has been set/stored, button long-press resets the device and restore it to a default configuration
2. Center Calibration screen: provides status of calibration: yes/no.
   1. Button long-press sets the current SWPS position as the new center calibration position and stores it into memory. The calibration data is briefly shown.
3. SWPS angle and Angle Kit selection: SWPS angle reading is in degrees: prior to calibration it shows an uncalibrated angle. After calibration this value is automatically offset by the calibrated center position (and therefore will show a zero angle value right after calibration).
   1. Button long-press changes Angle Kit selection. Starts off with "stock", cycles through Mini Mantis, SLR Speed, and Vettenuts right now. This selection is written to memory.
4. Raw Data: Provides raw voltage readings of SWPS in 10-bit format (0=0V, 1023=5V). Also shows which zone the device is interpreting the SWPS to be in... this is just part of my process for correlating SWPS voltage to SWPS angle.
   1. Button long-press shows center calibration values
   2. The point of this menu in general is to provide info for troubleshooting

 

Last night I finally set out to measure wheel angle vs SW position. I shouldn't have expected it to be easy! Ran into some issues/challenges with my car's SWPS:

1. At the SWPS's Phase B zero position, Phase B sits at 0V while Phase A traverses 3.7 degrees of rotation. Also Phase A's electrical center (relative to Phase B) is about 2.65V. I expected this to be nearly 2.50V like the spare SWPS that I've been playing with.
2. SWPS center is further off from SW center than I would have expected, considering SWPS gets installed with an alignment tool. Phase A calibration value was 38, which translates to (38/(1024*2)*360) = 6.7 degrees. Although, this is well within the range that my current code can tolerate which is about 40 degrees.
3. The phone app I planned to use to measure wheel angle, well, isn't working right. If I put my phone flat on a surface, calibrate it, and leave it alone, the reported angle starts drifting... over just a few minutes it will report 15 degrees despite not moving. So I started looking for alternative phone apps, and after trying maybe 10 of them I think one will be suitable.

Where does this leave things?
My approach to calibration has to change. Phase A can still be calibrated with the current center calibration technique but it doesn't work for Phase B.
I'm thinking to incorporate a second calibration method where the SW is turned to at least +/- 45 degrees and the device collects data that it will use to generate a Phase B calibration value.

Previously I wasn't planning to share this... but it's now relevant. The graph below represents how I use both Phase A & B to determine SWPS position/angle. The gist is that there are pre-defined "zones" which each last for 90 degrees of rotation, and the device transitions through these zones as A & B reach certain voltage values.

Zone 0 is for the SWPS center region, and it uses Phase A. Once SW rotation reaches 45 degrees then it transitions to zone 1 and switches to Phase B, which is used until (45+90)= 135 degrees, and then flips to zone 2 / Phase A, etc.
At the zone 0-->1 transition I was seeing the reported SWPS angle jump from 45 degrees to 60 degrees because Phase B's calibration was so far out of whack... hence why Phase B needs to be properly calibrated (especially compared to Phase A).



And for reference again the map of voltage vs SW rotation:

 

Had some success last night on the new calibration process

Compared to my recent video, calibration is now much more involved but the screen does a good job of guiding the user through the process.

• When you enter calibration (same way as before: hold button), you get into a menu which says: "CENTER ST. WHEEL AND PRESS BUTTON". Afterward it displays "CENTER IS CALIBRATED". Phase A is now calibrated.

• Next up is "TURN WHEEL RIGHT.. ....SLOWLY...." As soon as the SW reaches 45 degrees clockwise the screen says "STOP" and automatically records the 1st Phase B calibration/offset value which it shows.

• Then "TURN WHEEL LEFT.. ....SLOWLY....", and same thing it says STOP @ 45deg CCW and records/shows a 2nd value.

Calibration is now complete, and Phase B calibration/offset value is calculated as an average of the 1st/2nd measured values.

There's still more testing to do, but the goal is for a nearly-seamless transition between zones so that the reported SWPS angle value doesn't jump during each zone transition.
(Previously it would jump from 45 deg to 50 deg because of wrong Phase B offset value; obviously I want it to smoothly move from 45 to 45.1)

 

The difficulty calibrating could be why GM used an extra linear pot in the early SWPS.

I'm not positive I agree with only using 1 signal at a time. It seems to me that could make it more susceptible to noise or failing in a bad way due to a dirty SWPS that could be ridden through using both signals.

 

FWIW, I agree that using both signals simultaneously is a better solution. If you or anyone else have a good idea on making that happen I'm all ears

That said, I do have plans for error checking that ought to weed out most SWPS issues:

• During calibration: [these are already in-progress]
  • check if SWPS's electrical center is reasonably near SW center, which is done using Phase A
  • check that the offset between A & B is as-expected. i.e. at 0-90 degrees CW rotation, B should be ~2.5V less than A
• At startup:
  • Compare A & B and verify their offsets, potentially using stored calibration values
• During regular use:
  • Periodically, probably at zone transitions, repeat 'At Startup'

Thus far I haven't determined how to test out these error conditions. Perhaps it might involve another Arduino with a DAC that simulates erroneous SWPS signals.

Noise entering the system would do so through the SWPS +5v / ground lines, I think... hopefully these are well filtered at the source. It might not be a bad idea to add an input power capacitor on my board, though. Not being an EE I don't have any initial ideas on how to size one but am open to input.
It may be worth noting that the +5v line is Vref for both the ADC (input) and DAC (output), so fluctuations in voltage should be a non-issue. (assuming Vin doesn't drop below ~3.3v causing the Arduino to shut down)

Another idea would be sort of a SWPS health check. By rotating the SWPS (using steering wheel) through a full 360 deg rotation, A & B could be compared to each other throughout the whole range to verify there aren't any bad/dirty spots on the resistive tracks.
I don't have any plans for this currently... but such things could be just a few lines of code away.

 

Progress has slowed lately due to challenges...
Originally I assumed that for either Phase A or B, when the SWPS moved from 0-->5V this correlated to a steering wheel travel of 0-->180 degrees (seemed to be the case for my spare SWPS). I still think this is true in principle. However, the SWPS in my car is not at all behaving like that (as I alluded to before when I commented how Phase B seems to "sit" at 0v).

Yesterday I came up with a dynamic calibration technique so the handoff during zone transitions is now smooth and the reported SWPS angle doesn't jump at transitions. But this is where my current obstacle arose: with the assumption that 0-->5V = 0-->180deg, turning the steering wheel 90 degrees (for example) does not correlate to a SWPS reading of 90deg., but something like 85deg. Then at SW 180deg. SWPS showed 168deg.

What's going on?? If we go back to my zone transition map:

It shows a certain slope for each line (related to 0-5v = 0-180 deg). In reality my car's SWPS angle vs voltage map doesn't match this very well, and I have to assume other C5 SWPSs could exhibit comparable behavior.

Kind of have to go back to the drawing board a little bit... I definitely need to map my car's SWPS angle vs voltage to see exactly what's going on. Then maybe work on a multi-point calibration that's not too burdensome for the user (like calibrate at 0deg, and +/- 90deg) and have it generate a slope (voltage vs angle) for each phase & location/zone on the map.


12/24 EDIT: Lately I've been rethinking things, doing more research on encoders and signal processing. For starters I came across an article that essentially describes our SWPS as a "dual wiper quadrature endless potentiometer": https://www.edn.com/rotary-encoder-w...-and-low-cost/ I haven't found any good info on actual signal processing, but that link and this patent shed a little more light: https://patents.google.com/patent/US4978954

 

Happy New Year, guys! Hope you'll forgive the less frequent updates, been busy with the holidays/family and other car projects. Also still running into a bit of a mental block because of how my car's SWPS is behaving... this has actually led to me consider replacing it, and to that end I bought a new-old-stock SWPS off ebay for only $60.

For starters, I finally figured out a good method to measure steering wheel angle using my Android phone and an app called Protractor Tool:


(sorry for crappy pic, lighting was bad) It's very rudimentary but worked quite well.

Then I came up with a modest way to log Phase A & B voltages; I'm not thrilled with that so I've added a micro SD card reader to my Arduino setup for collecting/logging better/higher-speed data... but it enabled me to capture measurements that went into the following graph:


What is this?

• Blue dots are the Phases A + B voltage (in 10-bit format) measurements. Remember that 0=0v, 1023=5V, 511=2.5V.
• Red lines represent what I expected the blue dots to look like, such as in the graph in this post.
  • For reference, the bottom would be 0 degrees, right 90 degrees, top 180 degrees, left 270 degrees, and then back down to bottom which is 360 degrees.
• xx deg SWPS means that those points are where I would interpret each of the angles to be based on the SWPS voltages
• xx deg act show where I measured each of the SW angle values at (using the SW angle measurement seen in my first picture above)

Breaking this down... it's obvious that the SWPS in my car isn't centered to the SW, and is off by a few degrees. That's not a big deal and would be overcome by a center-calibration technique.

However, it's important to note that Phase A & B are not offset by 90 degrees of rotation like they are supposed to be.
i.e. at zero degrees, Phase A should be 511 and Phase B is 0, which correlates to 90 degrees phase offset between A & B.
In the case of my car's SWPS, when Phase B is 0 [0V], Phase A is 538 [2.63V], which translates to ~4.7 degrees offset between A & B. For reference both my spare SWPS and new SWPS show 510-512 for Phase A when Phase B is 0. But it's apparent I have to assume all SWPS could have some offset angle which is quite unnerving.

Also I don't like is how both phases "sit" at 1023 [5V] for a decent amount of the rotation range. I've been trying to determine if that's just an issue with my voltage measurement method [Arduino 10-bit DAC] or a problem with the SWPS itself. In testing with a Tech 2, I don't see the voltages sit at 5V like my Arduino test shows. But on the flip side my spare SWPS doesn't sit at 1023 like my car's SWPS... this (and the previous paragraph) are why I bought a new SWPS and am considering swapping it in.

Also with the graph above I can't really generate a good SWPS angle calculation which is what's really throwing me off. I mean, I can manually develop equations after determining the slope of each of the lines, but that doesn't lend itself well to being automated in Arduino code. (which is the goal... to be automated with minimal calibration required)
There are some ideas I'm pondering to overcome this but I haven't gotten far enough to put it into words.


Beyond all that I built a prototype Arduino board that fits into the ashtray location. This one has a Micro SD card slot as mentioned previously. Pictures are below.
I still have my breadboard setup so I can continue to write & test Arduino code on the bench, and its wiring/schematic matches the prototype so that any code changes will apply to both setups.
Furthermore my C5 is now wired to accept the prototype board. Currently there is just a jumper plug connected since the car is still stock and I'm not ready to plug in the prototype.