Accurate Lambda Meter - Bosch LSU4.9 and CJ125 driver chip

[Matt Lee]

We noticed INNOVATE LM-2 did not have an "instant mode". LC-1 seems to have it. We did the comparison again with LC-1. Again, ALM beats LC-1 in response time and accuracy. Here is a Scope-shot. More pictures can be found at our website: http://www.ecotrons.com/ALM_VS_LC-1.html

[Matt Lee]

First, we did the comparison between ALM and LM-2 (more expensive one than LC-1), and found out that LM-2 did not even have an "instant mode", which is way too slow for any engine tuning. Then we compared the LC-1 and ALM. LC-1 has an "instant mode" and it is faster than LM-2, but it has a lot of noises. ALM beats LC-1 in response time, too. ALM is >50ms faster than LC-1 during rich to lean transitions, and it is repeatable. During lean to rich transitions, sometime LC-1 is ~40ms slower than ALM; some other time it can catch with ALM; but, again, it has a lot of noises. Noises mean you have to use filter(s). Filters mean further slow-down. As for the accuracy, we do not have an absolute reference yet. By measuring the oscillations of the analog signals from both ALM and LC-1, we can show ALM has much better relative accuracy than LC-1. For measurable lambda range, LC-1 only allows 0.7 to 1.5 lambda conversion. ALM allows 0.65 to infinity. Practically user should define the ALM lambda range to their applications. It's not uncommon, for diesel/CNG engine applications, that the lambda upper limit can easily go above 1.5.
The data and pictures can be found at our websites. Here I can not upload the pictures for some reason.
http://www.ecotrons.com/ALM_VS_LC-1.html

[Matt Lee]

To clarify the noise comparison tests: when we did the tests, we tied the LC-1's system ground and heater ground together (no analog ground available) and bolted it to the 12V battery negative terminal. And we did the SAME to ALM. Just to double check, we did the ALM vs LC-1 noise comparison tests again, and followed LC-1's manual: "the best is to solder the system ground and heater ground to a lug and bolt it to the engine block". It did NOT improve LC-1's noise level. The noise level was same as before ( there were always noises > +/- 0.2V, which was equivalent to 0.04 lambda error).

Then just for test purpose, to get rid of any noise source from the ground, we used a 12V DC power supply to power the LC-1 and LC-1 only, then the LC-1's noise DID reduce some; and in many cases the noise level was about +/- 0.1V (with some spikes igored), equivalent to 0.02 lambda.

ALM behaved same no matter where the common ground point was: battery negative, engine block, or 12V DC power supply.

Automotive electronics must stand the severe surrounding noise and that is common understanding. Be sensitive to ground noise itself means it is a poor design.

[Matt Lee]

LSU 4.9 is superior to LSU 4.2.
The major difference between LSU 4.9 and 4.2 is that LSU 4.9 uses the reference pumping-current, while LSU 4.2 uses the reference air. What does this mean? Let's read this true story from the auto industries: when Bosch first designed a wideband oxygen sensor, a reference air cell was used to provide a reference of stoic AFR. The technology was to keep the pumping cell balanced with the reference air cell, by pumping the oxygen out of the pumping cell. The pumping current was the indication of the actual AFR in the exhaust gas. The bigger the pumping current, the more the oxygen in the exhaust, and vice versa. Therefore the reference air was vital to the accuracy of the sensor, because it was THE reference. It worked well in the lab, but not so good in the real life. Because the enviroment around the sensor on a car was much worse. The reference air cell was susceptible to be contaminated by the exhaust gas, and / or other surrounding pollutions. Once the reference air was contaminated, the whole characteristics of the sensor were shifted to the low side. It was called "Characteristic Shifted Down", or CSD, in the industries. This was the biggest problem of LSU 4.2 that was used in some early OEM applications. And it caused the big warranty issue to Bosch. To fix this problem, Bosch redesigned the LSU sensor, and came up with LSU 4.9 version. LSU 4.9 sensor completely got rid of the reference air. Instead, it used a reference pumping current which was equivalent to the stoic reference air, but without having any physical air in the cell. So the technology became: the actual pumping current was compared to the reference pumping current to maintain the balance. The actual pumping current was still the indication of the actaul AFR, but the reference was a calibrated electrical signal, and stayed same all the time, all the situations.

This is the fundamental difference between the LSU 4.2 and LSU 4.9.

LSU 4.9 gets rid of the reference air, and therefore gets rid of the biggest failure mode. As a result, LSU 4.9 has a long life and can maintain the accuracy throughout the life. Only since then, Bosch LSU sensors have been used widely in the auto industires.

Nowadays, all OEMs who use Bosch O2 sensors are using LSU 4.9. GM, Ford, and Chrysler all use LSU 4.9 now. If you check out the O2 sensors on your recently bought vehicles, cars/SUVs/Pickups, (since 2007 or later), on the exhaust manifolds, you will find out that they are all exclusively LSU 4.9. No more 4.2 sensors can you find on OEM cars.

Most aftermarket wideband controllers are still using LSU 4.2, mainly for low cost reasons. Bosch sells the LSU 4.2 to the aftermarket at a much lower price than LSU 4.9. Plus, many of those manufacturers do not want to or are not able to adapt the new LSU 4.9 sensors. There is a big mis-understanding that LSU 4.9 is only for diesel engines, because it can measure very lean AFRs. That's not true. There is a diesel version of LSU 4.9, called LSU4.9D, mainly because of fuel and temperature difference. LSU 4.9 has been widely used with the gasonline engines. In fact , it is the most popular gasoline engine O2 sensor now, not only because it measures wide range of AFR, but also because it has the very good reliability, and high accuracy.

There are a few wideband controller companies in the aftermarket using LSU 4.9 already. But that does not mean all controllers using LSU 4.9 are equal. Even with the same LSU 4.9 sensor, the controller can make a big difference. Some wideband controlers are designed for AFR display only. You can imagine that they may not have good accuracy and fast response rate because they are not designed for those purposes. Those gauges are more for good looking than for engine tuning purposes. For engine controls, the accuracy and reponse rate are the most critical characteristics of a wideband controller. In fact, one way to tell whether a wideband controller is good or not, is to see whether it can be used as a feedback device for the ECU. A feedback device must provide a real-time signal in the fast rate and high accuracy, even under dynamic situations. The requirements for a feedback device are much much more than those for a gauge.

Even with a LSU 4.2, the controller makes a big difference. Bosch sensors are not easy to fail even with a LSU 4.2, if controlled appropriately. Especially, LSU 4.9 is designed for more than 10 year life because it has to, for the vehicle life. It should not fail in short time, like a year. Many OEM cars have been running with LSU 4.9 for years. Why so many aftermarket wideband systems have failed LSU sensors? Because many of them don't have a good heating control strategy. The number 1 failure mode of a LSU sensor is being heated up too fast or too earlier. O2 sensors are made of ceramic materials, which can be damaged by severe thermal shocks, like condensations, liquid residuals, or just high heating power when it's still cold. A very careful heating strategy to detect the dew point and a close-loop sensor temperature control are vital for the life of the sensors. That's why the LSU sensor must be controlled in the context of engine controls. You may say, only those know engine controls can design a good wideband controller.

Furthermore, the accuracy of LSU sensors is highly dependent on the operating temperature of the sensing element. The sensor reading can be very different if the temperature of the sensing element is different. LSU sensors must work at the vicinity of certain temperuatures for the good accuracy.

Bosch CJ125 chip is designed for this task. The heating strategy is a close loop control based on the measured sensor temperature. LSU 4.9 has a much higher sensor temperature resolution because of the resistance characteristics, so the heater controls are much better than LSU 4.2. Therefore, 4.9 has a longer life and better accuracy.

In short, not only the sensor LSU 4.9 is superior to 4.2; but also the controller with a CJ125 chip makes it an OE equivalent system.

[Mbirt]

Awesome, the MotoTron 128-pin ECU has a CJ125 onboard. I'll have to get my hands on an LSU 4.9 to give it a try.