In this entry – a bit more technical information regarding problems, which occurs if the control probes are disconnected due to DME config (in case if wideband probes are used as regulating probes).
I’ve heard as if positive experience, when after such an “upgrade” the engine works as if perfectly.
Let’s see, what is the true situation.

The first source of inaccuracy: IC of the control probe.
For the management of wideband probes, chipsets CJ120 and CJ125 (the newest), developed by Bosch, are used.
These chipsets are characterized:

  • input U offset of I (current) measuring OPamp: +/-10mV;
  • output U offset of I (current) measuring OPamp: +/-3mV;
  • gain error of I (current) measuring OPamp: +/-2.2%;
  • Uref (1.50V) is not available (!), its offset(!) is not defined. Self-calibration has to be performed regularly, using “chopper” type built-in calibration.

To reduce the problems with Offset drift and Gain, the chipset is turned on in the K=8 (not K=17) mode.

The second source of inaccuracy: ADC.
For example, MSD80 uses Infineon Tricore SAK-TC1796 CPU. These CPUs have (till) 12bit ADC.
Unfortunately, there are some problems with the accuracy of the ADC:

  • statistic error +/-4 LSB or +/-4mV by Uref=4.096V;
  • offset error 20mV (!), if the self-calibration is performed, or even +/-60mV, if the self-calibration is not performed;
  • Gain error +/-3% with self-calibration; +/-5% without self-calibration.

According to LSU 4.9 Datasheet, by K=8 +/-50mV drift error means Lambda around 0.95 .. 1.05. +/-100mV – Lambda around 0.90 .. 1.10.

No need for complicated mathematical calculations to understand – without additional trim, the measuring of the Lambda is HARSHLY inaccurate! As if measured correct Lambda 1.00 (which will be indicated by INPA or ISTA), the true Lambda could be 0.90 .. 1.10 or even more/less! For this reason, the control probes are necessary. Exactly the control probes are the ones, who calibrate the measuring hubs of the wideband probes! In addition, this Offset (Ref by Lambda 1.00) correction range is very wide: till +/-200mV.

So, instead of the expected Homogeneous mode, we get something very approximate. It is clear, that Lambda 0.9 or 1.10 means catastrophic CO/HC, an overload and damaging of the CO catalytic converters, problems with passing TI, increased density of misfires.

The second fundamental issue: performance of the engine CAN NOT/is not allowed to evaluate directly after the “upgrade”. The evaluation of the engine has to be done 5 .. 10 .. 20.000km after these “upgrades”. During this time, the injector data, measured previously, will be “floated” away. There will be no new/correct data (and – accordingly- correction) because the engine runs in limp mode (although the chip “tuners” believe, that everything is fine). Only after a long while, the effect of such an “upgrade” will be felt in full glory.