Help Diagnosing Unintended/Uncommanded Acceleration Issue

There is also a thread in lexusownersclub

Sudden Acceleration Resulting In Crash - Lexus Owners Club (USA & Canada)

I have also heard stories on internet about unintended acceleration on Yaris, Tacoma, Camry. But those cars are not hybrid cars. If there is any truth on these incidents then it is certainly not a hybrid related issue.

 

It would be easy to do, and not expensive since the OBDII connection is already there. Make it WiFi enabled and the device could upload its own data. I could build one for Toyota, but I'm sure they already have what they need.

Tom

 

My guess, for all non-drive-by-wire cars, it could be the Cruise Control failure causing sudden acceleration. And even on my Prius, I would probably replace or completely remove the cruise control if I had this issue. Just in case.

 

Yes indeed, something us Macintosh users have been laughing about for over 20 years. I take it you don't have a Windows PC.

There are two folks who oppose such systems:

• drivers who think the speed limit is the minimum
• vehicle manufacturers who want to protect their technology

Could you elaborate on this system a little more? Any especially detailed URLs to take a peek at?

Bob Wilson

 

CANBus also has widespread use in industry, it isn't unique to automotive applications. I've had to use an Agilent 6000 series MSO to patiently track down some odd issues. Without exception, the issues I uncovered were due to a poor connection, intermittent connection, intermittent floating ground, etc

The Agilent MSO has smart software, it will automatically trigger on glitch. With deep memory scan and memory for a lot of events, it will readily capture glitches. Here is an example of a bus glitch I captured



I've highlighted where the glitch occured



This is another glitch the MSO captured, due to random noise caused by an oxidized connection



This is another example of noise causing a momentary but serious glitch



I wouldn't expect a dealership to invest +$50,000 in a sophisticated MSO, plus training on how to use it. A simple black box could be designed, plugged into the DLC3 port, to automatically trigger on glitch

If there is an actual glitch in by-wire throttle, it usually fails closed, not open.

Of course, I'm old enough to remember what happens if you forget to hook up the return spring when you install a brand new high performance carb. Talk about an exciting ride

 

the old toyota scantool can record data and can be purchased outside the dealer (ie, ebay) for a couple grand.

there's really nothing that can be done as far as diagnosis for a problem that is not occurring. how do they know if they fixed it if they can't reproduce the problem to begin with? short of having a tech drive it until it happens again with a scantool plugged in at all times.

 

My thinking is if we record the conditions leading up to the fault, we may gain a clue that allows us to reproduce the problem at will. Then it can be solved.

The one problem I have with monitoring the CAN bus is that it looks like query-response architecture. This means we would only see what is commanded, not necessarily what other states may be going on. Still having recorded data is more than we have today.

Bob Wilson

 

The cruise control in the Prius is not a separate system. It is simply additional code in the ECU.

Tom

 

I would suggest that calling it a bug before the actual problem is known could mislead. What if it is a very rare combination of a intermittently shorted wire and some other condition? "Bug" implies that it is universal in all Pri'i with the same software load. Capturing the internal state of the ECU without capturing the state of the signals feeding in could lead to the wrong diagnosis since only internal information is captured. I fully support the efforts to capture data, but I think it is a little soon to be calling it just a SW issue.

 

iirc cruise is inactive at 25 mph and below. i don't have access to documents to prove that right now, and i don't have the prius right now.

 

On The Bright Side, this is a benign form of unintended acceleration: when he stepped on the brake pedal the car stopped, just as it should. None of the lethal or potentially lethal uncontrollable acceleration that typifies these stories.

 

Excellent!

Rather than reply to the other notes, lets start here:

• low-cost data recorder - ideally we'd like to have a computer that can 'sleep' and 'wake' easily. Some laptops have a magnetic reed switch that when a magnet comes close, the computer 'sleeps' and when the magnet goes away, the computer 'wakes', which means we don't have to reboot it each time.
• some sort of programable batch capability - I prefer to write in Perl or if binary data, say data from a ODB-to-USB interface, I use 'c'.

These features are common with linux and MacOS boxes. My typical practice at work is to take an old (300 mHz) laptop and rebuild it as a linux box. But getting it to 'sleep' and 'wake' is not an area I've spent much time.

Building and integrating the data capture machine is not that difficult but it is hard to describe it over the net. But I need a box like this for another purpose. Let me think about how to approach this for a bit.

I'm really interested in work that others may have done to integrate an OBD-to-USB device. I'll wander around the open source areas to see if something looks promising.

Your observations about the brake are interesting. I was reading about the traction control system for the 2010 Prius and the brake system lets the new "Power Management ECU" (replaces the hybrid vehicle ECU) reduce power when the brakes are applied and traction control events are occurring.

Bob Wilson

 

There are dedicated recorders with embedded RTOS's, such as dedicated power quality monitors or CANBus monitors running on either Wind River VXWorks or Green Hills Integrity

Unfortunately, for the most part, the "low cost" portion is missing from the equation. I prefer a good MSO as it has a lot of other uses for me outside of catching some CANBus glitches. The MSO I mentioned is easily $50,000 with a few options. The dedicated recorders are usually $6,500 and up

 

I think I'll check what TI has been doing with the MSP430 series. If I can get an integrated CANbus and simcard microprocessor for a reasonable cost, it should be fairly easy to hack a 'data recorder.' Then use the JTAG interface to download the data for analysis.

Bob Wilson