Wide Band Oxygen Sensor

There are ways to guess your actual air:fuel ratio (AFR), but there is only one tool which can give you exact values. That tool is the wide band oxygen sensor. *
The stock oxygen sensor
Nowadays every car has an oxygen sensor, so called lambda sensor, but only a handfull cars came with a wide band one from the factory. What's the difference? The stock 3 wire oxygen sensor as all of the narrow band sensors was designed to to achieve a stoichiometric (14.7:1) mixture under closed loop conditions, as this is the point where optimum catalytic converter efficiency is achieved. Actually the oxygen sensor is a chemical generator. It is constantly making a comparison between the Oxygen inside the exhaust manifold and air outside the engine. If you monitor the stock O2 readings you will find that it has an output ~0-1000 mV. You will read a high voltage with an air/fuel mixture below stoichiometric and a low voltage above this point. Beyond this, the meter has no way of accurately and reliably differentiating between 12:1 and 13:1, or conversely, 15:1 and 18:1. At the stoichiometric point you should get 450 mV, but the readings will never stay there. The O2 sensor is constantly in a state of transition between high and low voltage. Manfucturers call this crossing of the 450 mV mark O2 cross counts. The higher the number of O2 cross counts, the better the sensor and other parts of the computer control system are working. The ECU is only using the O2 sensor in closed loop mode. If you are in open loop mode because the car is not heated up yet, or most importantly in power enrichment mode at WOT, the O2 values are not used by the ECU. Our real quest is to determine the AFR in power enrichment mode... *
The wide band oxygen sensor
As you can see the stock O2 sensor is not an adequate tool to determine our AFR under WOT conditions. To be able to do this we will need a so called wide band osygen sensor, which can tell us the exact AFR not just near to stoichiometric, but in a wider range as well. What are the main components of such a kit?

• wide band oxygen sensor
• an additional mount point in the exhaust for the sensor
• an electronics which will drive the sensor
• some kind of display to show the measured values
• optional, you might need some kind of logging capability

* There are two ways to get these parts. First you can buy them from a bunch of differnet vendors as a kit. Just to name a few, FJO, Motec, FAST etc. If you choose this way you will get a plug and play solution, but prepare yourseld to spend some serious cash. Neither of the above vendors will give you their stuff below ~$800. The second way it to build your own unit using a proved design and low cost eletronic parts. I gonna show that second way in more detail below... *
DIY WB oxygen sensor
First of all, with these kits you still have to buy the sensor itself. The two kits known by me are using an NTK five wire sensor. You can obtain this sensor from several sources for ~$150 It was used in 92-95 lean burn 1.5 Vtec Honda Civic's. The Honda part number is #36531-P07-003. You can also get it from NAPA under #OS791. Or from the The Parts Bin. There are a couple other places as well, but beware because some vendors wanna get $900 for the exact same sensor. This sensor also called as the UEGO sensor. Pls also be aware that there is a widely avaiable Bosch wide band sensor (the LSU4 sensor), but it won't work with these kits.


The sensor has a M18x1.5 mm thread, so you will have to provide such a bung on your exhaust. I've used an M18 nut and welded it up before the catalytic converter. The ideal placement however in the SyTy application would be somewhere on the down pipe near to the turbo. * Also note that the sensor has a special connector which carries the calibration resistance, which's value could change from sensor to sensor. The counter part female connector is made by Sumitomo and you can get it either from TechEdge or from a junkyard. It was used in the headligh harness, near to the brake master cylinder, near to the firewall on the passenger side fender and on 90's Accord on the distributor. You have also the option to pick out the resistance from the sensor's male connector and put it onto the controller. This way you can use any five wires connector which you like. * Back to the DIY kits. There are two kits avaiable which will give you near the same in functionality. The first kit was originaly designed by the members of the DIY-EFI list. A bit later Peter Cargano (also a list member) at TechEdge made his own design and start to sell the PCB's and part kits. Pls note that I've no intention to take part on any side in any license issue between the above parties. I personaly choosed the TechEdge version and has no first hand experience with the other one. I had known the TechEdge version first, but I might still choose it because it's more compact and promptly avaiable. TechEdge even sold built unit for a limited time around $95 and might offer them again. *
Construction
Before we gonna talk about my experience regarding to build the controller, pls keep in mind that the TechEdge unit is not for begginers. You will need pretty solid soldering skill to be able to finish the board. The board is really small.


Maybe the most easy way to get all the parts is to order the parts kit from TechEdge, but you can collect them too. The only part which I had trouble to find was the voltage regulator. With the current kits you will get the LM1086-LT-ADJ low voltage drop out regulator, which allow us to run the unit even around 12.5 V. The older units came with the LM317T voltage regulator. You should not use this regulator unless you can maintain at least 13.4 V all the time to the unit. If you need the part list, you can see it here. I don't want to cover the whole construction here, you can find a quite good construction guide here. Here is a picture showing the controller's outside connections.


Follow the above construction guide step by step and you will be covered. A few important things:

• The main problem is the TechEdge kit, that it's using T0-92 stuff (three legs quite close too each other), which are pretty though to solder in. Watch these closely.
• The two CalR connections can be connected in any order.
• As you can see I'm using a Pentium heat sink as the heat sink for the voltage regulator, but that's an overkill. A simple aluminum plate should do with the same footprint.
• Do all the final tests before connecting the sensor.
• Remember that you need at least 12.5 V input voltage. If you do the test from a car battery you might need to start the car to get this voltage. Otherwise the test won't work. I'd definately do the sensor connected test with the car running or with 14 V.
• If you have no sensor connected you will get 2.5 V Vout.
• If you have the sensor connected you will get 2.5 V Vout still the led comes up.
• With the sensor connected the led should come up in less than 45 s if you have at least 13 V of Vbat. The led shows you that the sensor had been preheated.
• It's a good idea to mount the controller into such a place when you can keep an eye on its led
• In order to get appropriate Vbat even under WOT, you should connect your Vbat connection to as close to the generator as possible with heavy gauge wire.
• Use a simple automotive relay in your Vbat wire to your controller. You can use a light ignition wire to trigger your relay.
• Leave the sensor on open air when you connect it the first time, you should get around 4.0 V after the led comes up.
• If you get satisfactory result from all the tests you can mount the sensor into the exhaust.
• Never leave the sensor in the exhaust if it's not connected to the controller, or the controller isn't powered up, or if the led on the controller doesn't come up. The sensor should be heated up as soon as possible, because on a cold sensor the exhaust gas will build up a carbon deposit quite fast ruining the sensor.
• If you had the idea to mount the sensor in the place of the stock one just forgett it, unless you use a 3rd party engine management system, or a custom open loop program with the stock ECU.

*
Display
There are quite a few options to display Vout as the AFR ratio. First TechEdge has a display for the kit. It's also a DIY unit with a modified Jaycar 5300 display. It's preaty easy to build, especially compared to the controller.


It will convert Vout to the actual AFR and shows it on a three digit display and on a led bar too. Here is the conversion table.

Vout	AFR	Lambda	AFR target value in DM with Dig's code hack
1.40	10.08	0.69	7.14
1.45	10.23	0.70	7.40
1.50	10.38	0.71	7.65
1.55	10.53	0.72	7.91
1.60	10.69	0.73	8.16
1.65	10.86	0.74	8.42
1.70	11.03	0.75	8.67
1.75	11.20	0.76	8.93
1.80	11.38	0.77	9.18
1.85	11.57	0.79	9.44
1.90	11.76	0.80	9.69
1.95	11.96	0.81	9.95
2.00	12.17	0.83	10.20
2.05	12.38	0.84	10.46
2.10	12.60	0.86	10.71
2.15	12.83	0.87	10.97
2.20	13.07	0.89	11.22
2.25	13.31	0.91	11.48
2.30	13.57	0.92	11.73
2.35	13.84	0.94	11.99
2.40	14.11	0.96	12.24
2.45	14.40	0.98	12.50
2.50	14.70	1.00	12.75
2.55	15.25	1.04	13.01
2.60	15.84	1.08	13.26
2.65	16.48	1.12	13.52
2.70	17.18	1.17	13.77
2.75	17.93	1.22	14.03
2.80	18.76	1.28	14.28
2.85	19.66	1.34	14.54
2.90	20.66	1.41	14.79

The dot display is using the following conversion:


* An other option could be to use a display originaly designed for a narrow band sensor. The units often called as air/fuel meters. I have such a meter from AutoMeter.


It's input voltage range is 0-1000 mV and will light a led at every 50 mV. What I'm trying to do is to scale down the ouput of the wide band controller to this range. I hope I can accomplish this soon with a quite simple eletronics, mainly an opamp and a potmeter.Will update the page as soon as I've results. * An other quite nice option is hacking the stock ECM to log the output of the wide band controller. Dig developped this, Leroy gave a crystal clear explanation to it and I use it. :-) A bunch of thanks to them! The motorollas assembly was dissasembled and an unused A/D channel (the fuel pump's, EMC F14 connector) was used, filtered and put into the ALDL stream so you can check it under "AFR tgt". This way you can follow you actual AFR in DM. Here comes the code itself.

000152: 20
0007C2: 01,A0
0007DC: 01,84
0014C6: 86,50,BD,FB,39
003B39: bd,f4,b3,36,37,4f,bd
003B40: f4,b3,b7,01,a0,fe,01,a0,f6,c1,52,bd,f5,99,fd,01
003B50: a0,f6,01,a0,c0,46,24,01,5f,c1,46,25,02,c6,46,ce
003B60: fb,70,3a,a6,00,b7,01,84,33,32,39,01,01,01,01,01
003B70: 64,65,65,66,66,67,67,68,68,69,6A,6A,6B,6C,6C,6D
003B80: 6D,6E,6F,70,71,71,72,72,73,74,75,76,77,78,78,79
003B90: 7A,7B,7B,7C,7D,7E,7F,80,81,82,82,83,84,85,86,87
003BA0: 89,8A,8B,8C,8D,8E,8F,90,92,93,95,97,98,9B,9D,9F
003BB0: A2,A5,A7,AA,AD,B0,B3,B9


On the left side are the hex addresses, where you should change the code to the one on the right side. This code has the improvement over the original one that it will display the actual AFR in your "AFR tgt" field. Also note that this code is using the ECM's test mode code area, so you won't be able to put the ECM in test mode any more. Not an issue IMHO. * This code will only work with the DIY-WB unit but you can make it work with the TechEdge unit as well. You can check the calibration of your code to compare Vout to the readings in DM. The translation table starts at 003B70s with 0x64, which is decimal 100, which is AFR 10.0. It ends at 003BB0 with 0xB9 which is decimal 185, which is AFR 18.5. Following this logic AFR 14.7 is 0x93. So if your Vout is 2.50V you should see 14.7 in your "AFR tgt" field. If the two values don't mach you can recalibrate the code as you whish. * Pls note that license agreement of the above code. This code is for non-commercial purposes only. License granted free of charge to non-commercial users only. No commercial license is offered.