This entry looks in more technical detail at the problems that arise when the control probes are disabled in the DME configuration, with the wideband probes used as the regulating probes.
I have heard reports of positive experience, where the engine supposedly runs perfectly after such an “upgrade”. Let’s see what the real situation is.
First source of inaccuracy: the probe control IC
Bosch’s CJ120 and CJ125 (the newer one) chipsets control the wideband probes. They have the following characteristics:
- Input voltage offset of the current-measuring op-amp: ±10 mV.
- Output voltage offset of the current-measuring op-amp: ±3 mV.
- Gain error of the current-measuring op-amp: ±2.2%.
- The reference voltage Uref (1.50 V) is not available, and its offset is not defined. Self-calibration has to be performed regularly, using the built-in “chopper”-type calibration.
To reduce offset drift and gain problems, the chipset is operated in K=8 mode (not K=17).
Second source of inaccuracy: the ADC
The MSD80, for example, uses an Infineon TriCore SAK-TC1796 CPU, which has an ADC with a resolution of up to 12 bits. Its accuracy is limited:
- Statistical error: ±4 LSB, or ±4 mV at Uref = 4.096 V.
- Offset error: 20 mV (!) with self-calibration, or even ±60 mV without it.
- Gain error: ±3% with self-calibration, ±5% without it.
What this means for the Lambda reading
According to the LSU 4.9 datasheet, at K=8 a drift error of ±50 mV corresponds to a Lambda of about 0.95 to 1.05. An error of ±100 mV corresponds to a Lambda of about 0.90 to 1.10.
No complicated calculation is needed to see that, without additional trim, the Lambda measurement is badly inaccurate. When INPA or ISTA shows a correct Lambda of 1.00, the true Lambda could be anywhere from 0.90 to 1.10, or even further out.
This is why the control probes are necessary. They are what calibrate the measuring circuits of the wideband probes. The offset correction (referenced to Lambda 1.00) also has a very wide range, up to ±200 mV.
Instead of the expected homogeneous mode, we get something very approximate. A Lambda of 0.90 or 1.10 means catastrophic CO/HC levels, overloading and damage to the catalytic converters, problems passing the technical inspection (TI), and more frequent misfires.
Why the engine cannot be judged right after the “upgrade”
The second fundamental issue is that engine performance cannot, and must not, be evaluated directly after the “upgrade”. The evaluation has to be done 5,000 to 20,000 km later (5 .. 10 .. 20,000 km in the original). Over that time the previously measured injector data will drift away. No new, correct data (and therefore no correction) will be available, because the engine is running in limp mode, even though the chip “tuners” believe everything is fine. Only after a long while will the effect of such an “upgrade” be felt in full.