Anyone played with current sensing for the Ip (pump) current using the LSU 4.2 sensor?

I am confused about how the Rcal resistor works. If I have an OP AMP driving a 61.9 ohm resistor sending current to the WBO2 pump circuit I can measure the current by doing the voltage across the 61.9 resistor. The current is [ |V_hi-V_lo| / 61.9 ]. The | is the absolute value function.

Now putting the Rcal resistor (30-300 ohms) in parallel with the 61.9 will change the current measurement. It would be [ |V_hi-V_lo| / (61.9 || Rcal) ]. Since Rcal is assumed to be unknown then how is the current calculated. The || is two resistors in parallel. If the equation [ |V_hi-V_lo| / 61.9 ] is used then the Ip current is off by up to a factor of two or more.

I can do a simple "free-air" calibration and not use the Rcal resistor, but I would like to use the Rcal resistor. It looks like the Rcal should be in parallel with the 61.9 ohm resistor to "leak" some current so that the Ip matches the LSU 4.2 O2 concentration curve in the datasheet. It seems to "leak" a lot of current and therefore the [ |V_hi-V_lo| / 61.9 ] is extremely off by the real Ip current.....and Rcal is unknown for a given WBO2 sensor. I can calibrate each WBO2 in software for the Rcal....but that is supposed to be why the Rcal resistor is there so that a software/elect. cal. is not needed.

 

Have you tried to simulate this circuit in something like Multisim (electronics workbench) or something similar?

We really need a diy lsu4.2 driver. If I had a detailed description of how drive and read the sensor, I could easily produce a microcontroller based equivalent of an LC-1 or LM-1 for the community.

 

Yes.

The datasheet is online. Do a google search for LSU datasheet.

The datasheet says that the Ip current for the O2 curve is measured using a 61.9 ohm resistor. Putting Rcal in parallel with the 61.9 resistor is what confuses me.

 

I just did a quick look around the web and have found some new info that I didnt see a couple years ago when I last thought about making a digital wideband controller. I have several projects going right now, but perhaps sometime this summer I might be able to tackle it.

 

I played around with the resistors a bit. It appears that Rcal is there to leak current so that the current measured through the 62 ohm resistor matches the datasheet Ip curve for lambda.

 

Thats sounds very familiar. I used to have a whole folder full of printouts dating all the way back into the late 1990's (I think it was that old) that were from the diy-efi e-mail list, in regards to figuring out how to drive the NTK sensors. Wonderful stuff.
Have you taken a look at how the VEMS (megasquirt AVR) guys drive the sensor?

 

Something tells me the VEMS folks don't want anyone to know how it works. It appears as if the firmware file is encrypted, it's just garbage when you disassemble it.

 

Oh? My mistake, I thought I saw their code online at some point.

You know, I think I'll attack the LSU4 sometime in mid 2008, I should be done with the 8 coil CNP driver that I'm starting work on right now (the '730 will see it as a 7 pin HEI ) by early summer. I'll post the schematics and code here on TGO (for the digital DIY Wideband) . An AtMega168 should be more than enough.

 

OK, looks like somebody has done basically what I was thinking about making. Difference is, his isn't open source, but its still really inexpensive.

www.14point7.com

version 1 of the "JAW" uses an Atmel AVR (exactly like I was thinking about using) but he has a version 2 in the works that uses a more powerful PSOC from Cypress.
His unassembled kit is in the 40 dollar range. looks like a great deal.

 

The software control isn't bad. A simple PID controller is all that is required for the basic function. If you use an A/D for the Ri measurement just keep track of the pulse output and when you read the Ri voltage.

 

The picture of the kit shows all electrolytic capacitors. The Ri measurement is AC. The out voltage is 0-5v to a 2.5v reference. That means current will sink and source through the capacitor. The capacitor will have a short life. It appears to be a design flaw.

 

So would you suggest a mica or MPF type then?

 

it's possible he's got two caps back to back to make it non-polarized.

 

It could be the case but unlikely. One non-polarized cap would cost the same as the two polarized and yield a better and more simple design. I am guessing it is a design flaw that will only fail after many hours of operation. More hours than it has been tested. The cap will eventually become a short and the DC will pop through and load the Nernst cell.

An X7R cap would be a good replacement for the Ri cap. You would have to trace back the nerst voltage (black wire) to figure out which cap it is.