Widebands That You Would or Wouldn't Buy

[Fred]

All posts regarding MWP's planned purchase and onward discussions, unedited, unmodified, have been moved here: viewtopic.php?f=6&t=2551

Let's keep this extremely high quality thread exactly that way with no more light-weight content.

Cheers!

[dr.mike]

Looks like AEM has followed suit, after PLX, and, updated their 30-4100 ( now 30-4110 ) AFR gauge to use the LSU4.9 sensor. Seems to be a lot of that going around.

[JaniM]

Innovate has LSU4.9 in all kits now.

[Daemos]

Innovate has LSU4.9 in all kits now.

The question for me is why do they still require 're calibration' on innovates...

[bmotorsports]

Innovate has LSU4.9 in all kits now.

The question for me is why do they still require 're calibration' on innovates...

The sensors drift, using the cal resistor indefinitely doesn't compensate unless you know how to.

[HelmutVonAutobahn]

Innovate has LSU4.9 in all kits now.

Well, that makes everybody

I see they are selling both the LSU4.2 and LSU4.9 sensors. And that the LC-2 manual mentions the LSU4.9. But, I don't see any announcement, etc. Or, any notation of which kits have LSU4.2 vs. LSU4.9. I guess I will have to order one to play with.

The manual, mentions a firmware update ( 1.01 ) is required to run the LSU4.9 sensor. But, this seems a bit sketchy; as the LSU4.9 requires additional hardware, in the form of a bias current source for the Nernst cell reference. ( maybe it was already in there ? ). Without the bias current, extended time on the rich side of Lambda 1.0 will deplete the reference cell, and cause errant readings until the sensor has been "lean" long enough to replenish the reference cell.

Are they saying that their existing hardware supports the LSU4.9 with the firmware update ?

The question for me is why do they still require 're calibration' on innovates...

The manual says yes. Since they do not use the sensor's factory calibration resistor, the "free-air calibration" is the only way to get the sensor characterized for use. To be fair, it is not necessary to be constantly calibrating them. Sensor drift is usually only a big issue if there is something systematically wrong. Such as lead or silicone poisoning of the sensor, overheating, extended use in very rich conditioning ( carbon fouling ), etc.

[JaniM]

All kits has automatically LSU4.9 in box, what I got week a go.

[Daemos]

Are they saying that their existing hardware supports the LSU4.9 with the firmware update ?

Yes, apparently so see:
http://www.innovatemotorsports.com/support/49.php

That was their 'announcement'

[HelmutVonAutobahn]

Seems like something is wrong. I went ahead and downloaded the firmware upgrade tool and followed the directions. I tried sever times. But no luck. The program recognized the LC-2 ok. But, no matter what I tried, it would always said that the programming failed, after it finished.

I'll try with an MTX-L later ( when I find it ).

[HelmutVonAutobahn]

Update... the firmware process did work with the MTX-L. I tried again with the LC-2. But, got the same result. Now to try the LSU4.9.

more...

Ok... a mixed bag. The MTX-L will run the LSU4.9 sensor; at least on the bench. I tested it in free air, and, at lambda 0.88 1.00 and 1.22 ( The gasses that I had handy ). The accuracy was pretty decent. It was within about lambda +/- 0.015 ( They only show 3 digits ).

They still have the same problem with the heater. But, it is much less pronounced with the LSU4.9 than it was with the LSU4.2. This is due to the LSU4.9's higher Nernst cell impedance ( 300ohm vs. 80ohm ) at operating temperature. And, a couple of other sensor changes. So that part is good.

The bad part is the measurement cycle time. That is, the rate of lambda measurements. In free air, it is about 18Hz. In the lambda 0.88 gas it got up to almost 40hz. That means that, even if they re-enabled "instant mode", you could only get about 20 samples per second, reliably. Really, only 1/2 of that rate ( Nyquist ) in transition. This works out to about ~12 samples per second. Which matches the serial data rate and the maximum analog output rate. Which, again, is probably not an accident.

Even assuming a mid-point between 18hz and 40hz ( 29Hz = 34ms ) This yields a t63 time of 170ms and a t90 time of 193ms at the 12 samples second data rate.

For comparison, the LC-1 would run at about 120Hz in free air and ~200Hz near lambda 1.0, in "instant mode". t90 was ~16ms-20ms

I would be a bit worried about the heater control in actual use. The LSU4.9 spec requires a heater PID control loop rate of 100Hz or higher to deal with the much lower thermal mass of the LSU4.9 sensor element vs. the LSU4.2 element. The minimum control loop rate for the LSU4.2 was only 2Hz ( Yes... only 2Hz ). The LSU4.9 requires a control loop 50x faster to maintain operating temperature in actual use.

Of course, just sitting on the bench, it will find an equilibrium point and stay there; even at slower control rates.

Since The MTX-L uses the same measurement cycle to measure the sensor temperature, they are only getting 18 to 40 temperature samples per second. Much less than the required rate. This could lead to wide swings in sensor temperature ( and possible damage or premature aging ) before the control loop has a chance to catch up to a hot or cold spike.

For example, decel fuel injector cutoff causes a cold spike and they only get 18 samples/second for heater control in free air.

Maybe it's OK. But, I don't think Bosh would have upped the control loop speed requirement 50X if there was not a reason ?