Closed loop lambda problem

Give input on tuning as well as any tips and tricks you may have. Also feel free to share base mapping files for various engine types.
Airspeed
Posts: 37
Joined: Tue Aug 02, 2011 2:05 am
ECU Model: S60 Pro
Firmware Version: 63
Location: The Netherlands

Closed loop lambda problem

Post by Airspeed »

Hi,

After many years I have the car running again, but when I have closed loop fuelling on, the lambda values are starting to rise and fall dramatically! Like from 10 to 15,5 and then over and over again. I feel it clearly under driving, so these high - alternating low values are really present and the ecu is obviously overcorrecting. This never happened all those years ago.
When closed loop is tuned off, lambda values are very stable and normal...

Lambda device is an Innovate MTX-L and I recently replaced the sensor with a brand new LSU 4.2 unit.

Could it be a software setting in lambda control I accidently touched?
Or is the MTX-L unit the cause?
Experiences? thoughts what could cause this?

tnx!
Walter
11.06 @ 124.5 mph
'74 VW Beetle turbo S60pro
iceman_n
Posts: 141
Joined: Mon Jan 16, 2012 1:39 pm
ECU Model: S40 Pro
Firmware Version: 58

Re: Closed loop lambda problem

Post by iceman_n »

i think PID factors need adjustment...
Bollah
Posts: 130
Joined: Tue Jul 26, 2011 10:15 am
ECU Model: S60 Pro
Firmware Version: 71

Re: Closed loop lambda problem

Post by Bollah »

Plus what is the target lambda setting as well as how much correction it should make when it hits a certain value
Airspeed
Posts: 37
Joined: Tue Aug 02, 2011 2:05 am
ECU Model: S60 Pro
Firmware Version: 63
Location: The Netherlands

Re: Closed loop lambda problem

Post by Airspeed »

Hi,

The PID values are:
- Prop: 500
- Int: 20
- Diff: 10

See for the other values enclosed picture. Does this look ok or should I try other values?
Anything strange you guys notice?

Image

Thanks in advance,
Walter
11.06 @ 124.5 mph
'74 VW Beetle turbo S60pro
stevieturbo
Posts: 3577
Joined: Tue Aug 02, 2011 12:08 pm
ECU Model: No ECU
Location: Norn Iron

Re: Closed loop lambda problem

Post by stevieturbo »

First off, according to that screen you have closed loop lambda turned off.

Re-calibrate the wideband to a more sensible range to improve resolution and also confirm it actually outputs correct voltages when requested.

0.5 to 1.5 Lambda is a massive range and you would never ever need either extreme ( unless it's running methanol or something )

Even 0.65 to 1.3 would be a large range that could be trimmed down for the 0-5v

When at idle with closed loop turned off, is the engine tuned correctly with stable AFR's ?

Why do you need to allow it a swing of +30% and -20% ? That's huge.

Does it idle anywhere near actual target ? All tuning should be good regardless of closed loop, which should only be for small corrections.

As for PID's, if they're all zero or very low, no corrections will take place. So if there are swing issues after all the above are looked at, start low and then try larger numbers to see what impact it has.
Alex DTA
Posts: 1622
Joined: Fri Feb 28, 2014 1:03 pm
ECU Model: S40 Pro
Distributor: DTA
Firmware Version: 79

Re: Closed loop lambda problem

Post by Alex DTA »

Before trying to use closed loop, make sure the injector dead times are correct. use the test in the Display and Test functions.

Once that's done, use the fast logging and the Periodically Perturb Fueling option to see how quickly the system responds. Use that to set the main sampling rate.

Only then can you start looking at altering the PID factors.
Airspeed
Posts: 37
Joined: Tue Aug 02, 2011 2:05 am
ECU Model: S60 Pro
Firmware Version: 63
Location: The Netherlands

Re: Closed loop lambda problem

Post by Airspeed »

Excellent info guys!
I'll get started then ;-)

TNX!
Walter
11.06 @ 124.5 mph
'74 VW Beetle turbo S60pro
Airspeed
Posts: 37
Joined: Tue Aug 02, 2011 2:05 am
ECU Model: S60 Pro
Firmware Version: 63
Location: The Netherlands

Re: Closed loop lambda problem

Post by Airspeed »

Injector dead time of these ID1000 worked out to be 1,08 seconds, which is about what it was set on (1,05) and I reduced the max gain and reduction of lambda correction.

I in the PID values set to 10 and D set to 10 as well
Reduced the P in the PID values from 500 to 350 and problem seems gone!


Tnx!
11.06 @ 124.5 mph
'74 VW Beetle turbo S60pro
iceman_n
Posts: 141
Joined: Mon Jan 16, 2012 1:39 pm
ECU Model: S40 Pro
Firmware Version: 58

Re: Closed loop lambda problem

Post by iceman_n »

Airspeed wrote:Injector dead time of these ID1000 worked out to be 1,08 seconds, which is about what it was set on (1,05) and I reduced the max gain and reduction of lambda correction.

I in the PID values set to 10 and D set to 10 as well
Reduced the P in the PID values from 500 to 350 and problem seems gone!


Tnx!
as for the injectors if it is injector dynamics brand they provide a full table of dead times for 3 - 4 or 5 bars...so email them and ask about the dead times for your fuel pressure and use the full table of dead times in Other Map Settings -> Battery Compensations...
Alex DTA
Posts: 1622
Joined: Fri Feb 28, 2014 1:03 pm
ECU Model: S40 Pro
Distributor: DTA
Firmware Version: 79

Re: Closed loop lambda problem

Post by Alex DTA »

iceman_n wrote:
Airspeed wrote:Injector dead time of these ID1000 worked out to be 1,08 seconds, which is about what it was set on (1,05) and I reduced the max gain and reduction of lambda correction.

I in the PID values set to 10 and D set to 10 as well
Reduced the P in the PID values from 500 to 350 and problem seems gone!


Tnx!
as for the injectors if it is injector dynamics brand they provide a full table of dead times for 3 - 4 or 5 bars...so email them and ask about the dead times for your fuel pressure and use the full table of dead times in Other Map Settings -> Battery Compensations...
Injector Dynamics dead times and slope values are calculated using MoTeC hardware. If you are using a different ECU, you will need to check the values are accurate for your ECU. Different injector drivers will have different dead times.
The values provided should be close though.
Post Reply