Problem with sticking lambda value over CAN

Spartan 3
1

Hi,

I’ve been using a Spartan 3 v2 (bought with an LSU 4.9) with a Link ECU over CAN on my Toyota GT86 for a while now.
Everything configured as per the online instructions and working perfectly for many months.
Recently I bought an LSU ADV to try the upgrade (and keep the LSU 4.9 as spare) and after installing and configuring, I noticed a problem: the lambda value transmitted over CAN now sticks for a brief moment (several seconds some times) at apparently random times (but always close to 1.0) and it’s enough to affect my closed loop lambda control with the engine oscillating and even stalling at times.

This is before the problem appeared:

(forum won’t let me post two embedded pictures :roll_eyes:)

Very stable AFR with a smooth CL control and uniterrupted lambda value

This is after the problem appeared:

image
image901×364 12.1 KB

Lambda sticks at fixed value for up to 3 s and of course CL control overshoots and when lambda “unsticks” all hell breaks loose.

The CAN hardware and software configuration is exactly the same. The car did sit stored for a few months though until I topped up the battery and fired up the engine again.

Spartan firmware is v1.10. I have tried to go back to the LSU 4.9 and update the configuration but the problem is still there. Also tried a couple of changes in CAN configuration, resistors, etc but it’s all the same.

Is my Spartan dead? Any test I can make (such us get a direct lambda reading over USB to see if it sticks?

Regards,

Eduardo

2

This is before the problem appeared:

image
image945×346 10.5 KB

Very stable AFR with a smooth CL control and uniterrupted lambda value.

3

Can you check the performance mode, serial command is “GETPERF”

4

It’s in standard performance (0) and always has been since I installed it.

I also tried a MEMRESET and configured everything again but the problem is still there.

Thanks,

Eduardo

5

Set the Performance mode to 1 with the serial command “SETPERF1”.

6

Well this did the trick and not only fixed the sticking value but made CL response better:

image
image1220×441 90.2 KB

Still very weird that after months of working ok with standard performance, it started misbehaving (sticking) on it´s own….

In any case, thanks for the tip! I never enabled high performance because I somehow thought it was innecessary / more show than go / extra wear for sensor or electronics :person_shrugging:

Why doesn´t it come like that by default?

Cheers from Peru,

Eduardo

7

The default was changed to performance mode 1 about 2 years ago. You might have received old stock with the performance mode set to 0.

As the sensor ages, Performance mode 0 runs into issues with response time. I thought I could solve it but could not.