Fitting dual wideband 02 sensors
Turbo Mini Forums

There seems to have been a revived interest in fitting dual widebands. I'd like fit them for my own testing, but instead of using sample chambers I want to plug them straight into the manifold.

The sample chamber version seem to target 2 problems - AFR error caused by back pressure and failure/error due to temperature.

To error due to pressure isn't random and Bosch publish a graph showing the variation with pressure. My intention is to measure the manifold pressure, lookup the error and make the correction. Or alternatively, not measure the pressure and accept the error, knowing it's the same for both sensors.

Regarding potential temperature problems, I intend to use longer bungs than normal to keep the sensor out of the flow and reduce the risk of them being damaged - there are examples where people are using pre-turbo widebands without the heat killing them. I believe that although the sensors themselves are temperature dependent, this is managed and accounted for by the controller. Given the wide variety engines and operating conditions, surely this must be the case otherwise the AFR measurements would be next to useless?

I know this goes against the grain of the design that have been already been done, but I'm not one for just copying a design (even if it works perfectly). Id like to get other people's views /ideas /corrections even before I buy another sensor, especially if you already fitted dual widebands.

Edited by PhilR on 10th Oct, 2014.

By all means try it. It would be good to get feedback on alternative methods.

I've used sample chambers twice before and I'm about to build two more sets. The cost of the materials for the compression fittings, the 6mm sample tube and the 16mm connection tube comes to less than £50 for 2 cars.

That £50 ensures that I am protecting 4 sensors, costing over £200. Money well spent in my opinion.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

What I would recommend for such a setup is to use a controller that reports the sensor temperature. You're right that the controller does manage the temperature but there's nothing it can do if the sensor overheats; it can only heat the sensor through the heater.

Having said that, the only controllers I know of that reports temperature are from 14point7. I know there are others but I don't know which ones.

If the sensors overheat, they will not provide a correct AFR and you will not know it until they fail if your controller doesn't report the temperature. And they will fail more or less quickly depending on how much you overheat them.

http://www.jbperf.com/

Edited by Rod S on 10th Oct, 2014.

Schrödinger's cat - so which one am I ???

just to clarify EGT has no effect on the mixture the O2 reads Right? except for the first min or so on start up with a cold sensor

i think there may be a chance of pre turbo O2s but i think if you take to much timing out for on boost or have it retarded a bit much it will end badly really fast

to take your idea of moveing the sensor back from direct flow is good perhaps you could get a heat sink around the bung to help cool the sensor body a bit??

http://www.jbperf.com/

Just to clarify, the reason I mentioned that the EGTs were different wasn't to say the LSU temperatures were being exceeded but more to question whether the pressures where different.
Unless the EGT at the LSU is so high it exceeds the heater temperature, the WB controller will keep the temperature correct - from what I've seen the 14point7 controller does this really well.
What I was suggesting is that the pressures in the 2/3 exhaust may be different to the 1 or 2 exhaust and hence bias the readings if the LSUs are installed in a pressurised environment.

Discus.......

Schrödinger's cat - so which one am I ???

The pressure differential might have an impact. Even if the average pressure is the same, there will be twice as many pressure pulses in the center branch.

One other thing is that from what I got from Alan at 14point7, the pressure affects the current in the cell. So since there is more exhaust flow in the center branch, the cell current won't be the same even with the same AFR so that would mean that the pressure would not affect the AFR reading the same way in both sensors. Which would mean that even a relative reading would not be correct.

I should add that this is speculation based on my understanding of how the sensor and controller work. I may well be wrong. It is however interesting to note that there are no affordable controllers that provide a pressure corrected AFR; there has to be reasons for that.

Jean

http://www.jbperf.com/

Rod / Jean , thanks for the 14point7 mention. I've never hear of them before, but I've just spent an hour on their website and expect to be ordering from them very soon (and at half the price I was going to spend on the cheapest ebay offerings too).

Paul, if I was in your position, there's little doubt, I'd make sample chambers too... but I have a good opportunity to try a different route. If I end up with a working setup that's relatively cheap and easy to implement , then that's 2 potentially options to get people testing dual widebands.

Turbo This.. This is my understanding so far. Guys please correct me if I've not got it right. The controller need to maintain a precise sensor temperature (around 750 C) for the reading to be accurate. Under running conditions the EGT rises and falls, so the internal heater will constantly adjust to balance the sensor's temperature at 750 C. This is fine (and completely transparent) for most set-ups, but put the sensor pre-turbo and EGT could push the sensor over 750 C, losing accuracy and potentially damaging it. I'd assume there's must be some time for a cold sensor to heat up, but I'd guess it was a few seconds not a minute.



I should hopefully have my engine built this weekend. I will modify the manifold for 2 X long o2 bungs (1") and 2 brake pipe fittings for pressure sensors.

Initial tests will probably be :

1) Monitoring the pressures in centre and side branches with 2x MAP sensors (or maybe cheap gauges poking through the scuttle). Work out whether they're the same, completely different, or closely related such that one can be predicted from the other.

2) Test a 14point7 controller with temperature readout and 'tune' the bung lengths to a safe minimum length that stops overheating. I think I'll put the sensor on cylinders 2/3 with a very long bung, then shorten the bung to the point where the temperature just starts to rise beyond the controllers regulation.

Schrödinger's cat - so which one am I ???

The impression I'm getting is that the sample chambers are not absolutely necessary. Another case of being too precise and building to achieve OEM standards.

So what would be the consequences of fitting 2 widebands upstream of the turbo on an LCB style turbo manifold with the sensors about 50cm from the exhaust valve?

Sounds like we can deal with the temperature issue, there is some data to steer us on correcting the AFR data for pressure. Would sensor life be significantly reduced?

I'm seriously considering trying this on my lads low boost application using a VNT turbo. Exhaust back pressure will be much lower if set up correctly.

EDIT: Data from simulation gives pressure/temperature at collector at 0.8barg boost is 1.7 Barg at 850 Deg C

Edited by Paul S on 11th Oct, 2014.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Gut feeling is 850C could probably be kept to less than 750C at the sensor head with a combination of bung length and heatsink but only just.
Higher output engines should achieve more like 1000C (have you tried the simulation on a higher output ?) which I doubt you could control.

This link is to the TechEdge page (I find TechEdge publish more fact than the others) which also includes a copy of the actual Bosch datasheet. Unfortunately it is quite an old one and I can't find the latest.

http://www.wbo2.com/lsu/lsu4.htm

http://www.wbo2.com/lsu/Y258K01005e03mar21eng.pdf

TechEdge copy the graph and formula from Bosch and as you can see, at 1.7Barg (exhaust backpressure) the error is between 12% and 15% (rich or lean).
That is quite a lot IMO.

However, as there is a nice mathematical formula, if your chosen weapon of ECU logs data in csv format (like the MS2/3) it would be dead easy to just load the file to a spreadsheet and add extra columns to correct the data after the event.

With MS2/3 you should be able to re-write (or write a custom) ini file to display corrected data in real time (waits for Jean to confirm).

What I can't find from the Bosch datasheet is how temperature screws the readings once 750C is exceeded.
The datasheet only seems to give absolute temperature limits (ie, how to kill it) but not how much accuracy is lost >750C, probably because they don't want anyone to run it above 750C (or I may just have missed the data, I did re-read it fairly quickly).

Anyway, quite positive, will be good to see real pressures/temperatures from Phil

The potential weakness is Jean's thoughts that double pulses may further skew the readings - unless the LCB type manifold commons up 1/4 before the collector so that equal pressure samples/pulses can be read.


EDIT - typos plus little bit added later.

Edited by Rod S on 11th Oct, 2014.

Schrödinger's cat - so which one am I ???

fascinating stuff.

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM

Reading the Bosch datasheet.

Cls 2.3 states maximum EGT of 1030 Deg C and maximum body temp of 630 Deg C. After 250 hours at that it may be dead. I don't see a problem with that.

The correction equation uses a factor 'k'. It gives some values for gases used for testing. Should we use 0.39 for the rich condition in our application?

Rod, your calculated error of 12-15% is based on 1.7 bara. I'm expecting 2.7 bara !

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Schrödinger's cat - so which one am I ???

how about putting a washer in the bottom of the sensor boss ,with a 1mm hole in it , then have a brake pipe off the boss leading to underneath the car ?

Medusa + injection = too much torque for the dyno ..https://youtu.be/qg5o0_tJxYM

If you're interested in hearing more on this subject (and other tuning aspects), you can have a look at Scott Clark's presentation at the last Megameet: http://www.youtube.com/watch?v=ueqDFWti47g

The part more related to the current subject starts at around 32:00 but the whole video is interesting.

There are also other presentations at that meet: http://www.msextra.com/forums/viewtopic.php?f=94&t=56179

http://www.jbperf.com/

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Edited by Paul S on 12th Oct, 2014.

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

Very interesting, especially the bit later on talking about very small pulse widths.

The common view on the MS2-E forum seems to be forget anything below 4mS yet I happily run less than 1.2mS.

Yet in the video he seems to support seriously low pulse widths being quite manageable.

I like what he is saying but would be nice to ask a few more questions on the LSU pressure thoughts.

Schrödinger's cat - so which one am I ???

You could PM him on the MS forum, username "dieselgeek"

Saul Bellow - "A great deal of intelligence can be invested in ignorance when the need for illusion is deep."
Stephen Hawking - "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge."

I didn't remember any problems with low pulse widths myself. I'd have to dig out some logs but I'm sure I got that low mS. I also change the number of pulses per cycle and noted that this upset all my smaller VE values. I put this down using very generic dead times (as most people do?). If you can get an engine running "ok" despite incorrect dead times, then you'd surely run into problems at low mS where the error has a greater effect.

I've dug out my MS2, making some changes, and will fire it up shortly to try and refresh my memory on how everything works. Do any of you have experience with the CAN bus? I intend to put an Arduino with CAN capabilities in the case. What I want are analogue inputs that I can manipulate before passing to MS2. I assume I can just use or log these like any other sensors plugged into the MS board? With my inexperience, this sounds too easy; do you think this idea is flawed or unlikely to work?

I will probably buy 2 of the 14point7 SLC Free kits and keep my current, downstream 02 sensor. If I can make all 3 sensor readings add up this should help validate the results. Assuming no one knows any relevant differences between the LSU 4.2 and 4.9 I'ff probably buy the kits with the 4.9 sensors. There's some hype for the 4.9s that's been cut and paste across the web, but sound very much like a sales pitch: So far I'm assuming neither will be more appropriate than the other for this application.

http://www.ecotrons.com/technology/bosch_l...lsu_42_sensors/


I'm still set on this idea. If I get this far, I feel I could collect data without melting anything and without buying much that I wouldn't need for a sample chamber setup. If it's a failure, I can then revert to sample chambers, sell my AEM gauge for more than the cost of 1 SLC Free and reuse any logging hardware I've put in. The only real investment is time, but I feel there's also a lot to learn whether it fails or succeeds.

Lots of ideas and information to soak in, so thanks for all the discussion and enthusiasm so far.

Edited by PhilR on 12th Oct, 2014.

You would need to implement the Megasquirt CAN protocol on your Arduino. You might want to have a look at my TinyIOx

http://www.jbperf.com/

I do like the look of TinyIOx. It looks reasonably priced and no doubt, it will just work with little effort. For the Arduino solution there is a risk that it may be harder to implement, but I only need to spend £8 on the Arduino CAN hardware to try it.

What ever I choose, I want the most flexible solution I can. With the Arduino, as well as signal conditioning with discrete components, I could have the added option to digitally pre-process inputs and later try out some other projects unrelated to this topic.

example:

Imagine I could pass both AFR and exhaust pressure readings to MS and have it calculate a corrected AFR using a formula. I anticipate that the pressure sensor may resolve all the individual exhaust pulses, I could end up with a jittery AFR, or worse still, a skewed AFR if MS happens to sample the data at the same time each engine cycle, in time with the pulses.

This is a hypothetical problem, but with previous IO projects, I've always run into weird and unanticipated issues.

Re. the first question myself and Graham both use CAN on an MS2 for additional I/O (just talking Minis on this forum, lots more use it on other projects). Not sure whether Paul still does as he's moved to MS3 which has more I/O in standard form.
If you use any of the IOx variants it is pretty straightforward, I'm not sure how easy arduino would be.
The SLC Free, like the AEM, only provide a 0-5V analouge output but if you want to future proof yourself for the ability to read the highly accurate digital data that the SLC-OEMs and most of the Innovate range provide, Jean's IOX code handles that directly.

Re. the LSUs, personally I would ignore the hype about the 4.9s and stick with 4.2s if only for consistensy /comparison against the rest of us (I think we are all on 4.2s)

As for the comparison on flexibility, I've not really done much with arduino so I assume you put the signal conditioning components on breadboard or veroboard or similar. Whilst the Tiny IOx is a bit limited on space for that. the full size IOx board has all the conditioning locations available (just choose the correct value components and solder in) or the somewhat more extreme solution I changed to which was my own custom PCB with all I/O hardware predefined and an IOx-OEM plugged in.

But I can see where you are coming from with the arduinos's ability to pre-process.
Your hypothetical problem is quite interesting. With the sample chamber arrangement I have always assumed any pulses would be damped out physically by the nature of small bore sample tube and large exit to exhaust after the turbo (where the pressure will be stable as the turbine will have damped out the pulses). But I've never measured it to see.

So far as I'm aware the CAN data isn't tied to engine cycle because of the latency in a CAN bus. And the display/logging data sample rate (EDIT - through CAN) isn't very high on an MS2 anyway.
But if you did want to pre-process the data to damp out pulses, ie, not rely on the signal conditioning hardware, it begs questions like do I pick the peak or the average or some sort of time weighted average. And it begs the question how does the wideband controller deal with it ?
Whilst the vast majority of LSUs are in areas with stable pressures (ie, after turbos or after N/A manifolds have commoned up), now that the wideband hardware has started becoming affordable, and quite a few EMS/ECUs have the ability to trim individual cylinder pulsewidths, there are quite a few V8 owners putting them, or intending to put them, in the individual manifold runners.

Interesting thoughts.
I did actually add enough pressure input circuits on my IOx-OEM board to allow for this, but for a different reason - I considered it might need different bore sample tubes to ensure the chamber pressures are equal as one sees twice as many pulses - but never got around to installing the plumbing. Maybe I shall re-visit it when the engine is next out.

EDIT - little bit added 5th para

Edited by Rod S on 13th Oct, 2014.

Schrödinger's cat - so which one am I ???