• Protip: Profile posts are public! Use Conversations to message other members privately. Everyone can see the content of a profile post.

Wide-Band O2 Considerations

Joined
16 July 2002
Messages
5,630
Location
Bay Area CA
Preface: very much a novice, but have garnered a lot of info over the last few months & here are some personal observations on what I have learned.
I welcome your feedback & opinions.


What is closed loop operation anyway? In the perfect control system, there would a real-time control that could adjust primarily A/F ratio & ignition timing to changing load conditions. The critical factor in creating a system that can do that is in the capability of the various feedback sensors accurately reflecting what is going on. (remember the old saying, garbage in = garbage out!) The most limiting of these is the use of O2 sensors to determine the Air/Fuel ratio. Most systems work under open-loop condition, with look-up tables processed for different load conditions: they may have partial closed-loop operation to trim conditions from the table within a certain degree under certain operating conditions (usually not under WOT)
With more engine management systems (EMS or ECU) having capability for closed-loop operation, a dependable O2 sensing system is essential to have it operate effectively. The newly released AEM for example appears to be an affordable package that incorporates that capability. Now, the std. equipment on most vehicles (NSX included) is narrow band, little more than a digital switch around the stochiometric point (~14.7), either "low" for lean, or "high" for rich, with very little descernability over the say, 10-14.7 range; that's exactly where you want to really know what's going on, particularly in a boosted application where you want to run richer than the "ideal" stochiometric 14.7, typically the 12.5 range.
So, enter wide-band onto the scene, which, while not exactly completely linear, gives a predictable curve related to the fuel ratio.

Courtesy of FJO, Inc.

WBO2_d5.gif



Now, you can see from the graph, why you need wide-band to offer any possibility of closed-loop response based on the O2 detector.
As you look around the available(affordable) O2 controller options, there are surprisingly few to choose from: TechEdge & FJO are the popular choices.
I came across one operation, however, EGOR Technology, Inc. & had a very lengthy & most enlightening conversation with its principal, Gar Willis.
I learned much, but one of the most interesting phenomena that is mostly overlooked by the low cost alternatives, is the impact that the operating environment pressure has wrt the O2 sensor calibration: this becomes critical in a turbo application particularly, where the exhaust back pressure can have dramatic effect, in the most damaging direction!

From Egor Technology:
Especially if you are measuring AFR pre-turbo, our exclusive built-in EBP Exhaust Back Pressure correction is a must-have; AFR errors due to exhaust pressure can be as high as 30% of F.S.! depending on how much EBP you're running.

And these errors are in the worst possible direction; they can mislead you into thinking the AFR is much lower/richer than it actually is! Under boost, NO other AFR meter will tell you the true AFR, and may indicate as much as 2+(yes, two+) AFR richer than actual! Ouch.

EGOR has compensation circuitry in its controller that operate in conjunction with O2 sensors that come complete with calibration curve & exhaust pressure sensors that factor in the result automatically! Clever stuff!
Also available is a dual input/output controller, that would be ideal for the TT application.

This would seem to be an ideal complement to the new generation of closed loop engine controllers. It would appear a requirement for all controllers to have whatever corrections are necessary for true closed-loop operation. I would expect the other low cost outfits to incorporate this before long.
On the NSX, typical location for the bung is pre-turbo, either on the header or the collector - w/o any pressure correction however, a much better solution for accurate response would be to locate the sensor after the turbo.

I believe the stand-alone controller boxes will ultimately be replaced by integrated wide-band control w/in the EMS or ECU itself (some may already have this feature)

So, it will be interesting to watch the development of closed loop technology over the coming year or so: with the available processing technology & improvements in the sensing devices, the self-calibrating system is maybe not just a pipe-dream!
Just not quite there yet!



[This message has been edited by D'Ecosse (edited 03 January 2003).]
 
That pretty well sums it up. Full-time closed loop operation is probably the holy grail of tuning, but as you say we're likely not quite there yet. I think the reason that AEM and others don't make it a much bigger part of their marketing is that they too aren't yet comfortable with it at that level. I’m not sure that closed loop under load will produce consistently better results than just a really good map, at least with the technology currently available. I say this because it seems that a good map, which needs to be created either way, can get you pretty close to a target A/F ratio under most conditions, and to improve on that means that your measurements of and adjustments to A/F under load need to be very fast and very accurate, all the time.

If you speak to Gar again ask him if he can supply a graph of typical wideband sensor life and calibration required. It’s my understanding that they don’t age as gracefully as the narrow units, which is one reason that he incorporates a recalibration button on his controllers. They are also very sensitive to being splashed by raw fuel or other contaminates. Even under ideal conditions their life expectancy is not very long (I can’t find my notes on that but I think it was just a few hundred hours in a street car, less that 100 in a race car) and they degrade throughout that time. So, it’s easy to imagine that adjustments made by an aging sensor could prove less accurate than a well designed fuel map. Not that I don’t plan to experiment and find out for myself.

Presumably the way to approach it is to get the base map as close as possible, fine tuning with the help of closed loop adjustment of the stored map. Then set strict parameters dictating maximum deviation from that map when running in extended closed loop mode for on-the-fly metering, and monitor it very closely with data logging. At least that’s my plan if I ever get the wideband. Speaking of which, I too like the Egor stuff and hope that Gar can work his way out of difficulties and eventually ship mine, especially now that I have an AEM on the way.

Anyway, I’m hoping that full-time closed loop will prove to work very well, but I’m controlling my expectations. With luck perhaps it will at least do a better job of compensating for things like temperature and barometric pressure changes than without it, depending that is on how well the AEM can be programmed to adjust for those itself. I would say that things are definitely looking up!
 
Thanks for your response SJS - I think we are aligned on how we see today & tomorrow!
This is nothing more than a servo control system, albeit with a variety of conditions under which to control. Nevertheless it is done in technology all-over today, with much more challenging circumstances to deal with.
Again, you are correct - the look-up tables are where it all starts, but then the servo's should be able to determine the changes from the "norm" & adjust on the fly, as it were! Sensing accurately of ALL the parameters is the key to good servo control - but they all have to reference some "knowledge" base somewhere.
To your question of sensor life - it is rumoured that wide-bands have limited life: I think that their generally reported expectancy is somewhat pessimistic. I have seen several reports to the contrary; however, that is not to say that "drift" (i.e. a shift from preliminary calibration) is not an issue; The nice thing about EGOR is that there is a free-air calibration cicuit to accommodate; his recommendation would be 6-month interval - of course, sensor should be exposed to environment for this procedure. (Now, these are the very things that make closed loop control viability essential - remember the garbage-in=garbage out theory!)
How the free-air calibration works is that the sensor samples the oustide environment air & measures the appropriate gas components & adjusts for any difference it sees from its previously calibrated setting - effectively, it's using the environment (which is relatively stable) as its calibration std.
I have seen reported quite consistently that Bosch sensors are both less accurate & shorter lived than the NTK or Honda components, but 500 hrs is far short of typical expectancy. Again, I think drift may be a more valid concern; contamination of the head of course will skew the appropriate sensing levels, so conditions that enduce that should be avoided.
 
I beleive the free air calibration resistor is standard on all the sensors, but most controllers don't give you the option to recal at the press of a button. I think the idea is that Gar feels it can be done without removing the sensors if you allow the exhaust to clear and fill with clean air before doing it. He also offers sensors super calibrated with the correction factor stored in the controller. Do you know which sensors he uses for that?
 
Originally posted by sjs:
I beleive the free air calibration resistor is standard on all the sensors,
Yes, original cal when you buy it - does not account for any drift down the road however
I think the idea is that Gar feels it can be done without removing the sensors if you allow the exhaust to clear and fill with clean air before doing it.
Not per our discussion - he specifically said to remove to ensure clean air (think about it - if in turbo application say, with O2 sensor bewteen head & turbo on NSX, probably won't really clear that well anyway) - if only 6 months, not big deal I guess.
I have also learned, EBP cal is important if the pressure is changing - on NSX, make sure you install after the turbo: this will probably eliminate both erroneous readings due to pressure & lessen impact of clearing to free air if you decide not to remove for recal.

He also offers sensors super calibrated with the correction factor stored in the controller. Do you know which sensors he uses for that?
Actually, what he does is provide a calibraton data set on a sensor by using lab stds & that data is flashable into the controller - he can calibrate any sensor by this method I believe.

p.s. of course that only works for EGOR controller

I should probably mention before much further that EGOR is not currently taking orders - for reasons I won't go into here, but unfortunate as this guy has excellent technology at entry-level prices.



[This message has been edited by D'Ecosse (edited 04 January 2003).]
 
I may have misunderstood but in my conversation with him months ago he said that one of the advantages was not needing to remove the sensors for recal. A misunderstanding on my part perhaps.

I was ready to order the EBP sensors but he talked me out of it because he concedes that best results are had if the sensors can be placed after turbo and cat. If that's not possible or if the measured back pressure is still significant then correction will make a big difference. I plan on placing bungs immediately after the turbos and checking BP which should be low.

I think your description of the calibrated sensors is the same as what I was saying. "He also offers sensors super calibrated with the correction factor stored in the controller." In other words, he measures a sensor against a known standard to determine the variance, then stores the calibration data in the controller to be applied to readings made by that sensor. He will also do this to any sensors you send him for a fee, but what I'm wondering is which brand sensors he purchases and resells. Hopefully the same ones he said live longer.
 
Originally posted by sjs:
I may have misunderstood but in my conversation with him months ago he said that one of the advantages was not needing to remove the sensors for recal. A misunderstanding on my part perhaps.

I think we're maybe debating the extent of "removing" - ideally they just need to be unscrewed from the bung, not removed to bench for set-up or anything like it.
Now, if you do place it after the turbo,(he also suggested to me that after eliminates issues with EBP) then it probably will ultimately clear - you could help purge I suppose by flowing some low pressure air from a compressor into the tail pipe! If it's accessible though, shouldn't be big deal to remove it.
Back to location again briefly - you mention after turbo & CAT - I would have thought that it would need to be before the CAT surely for accurate O2 measurement? (that's me - I'll check with Gar)

from SJS
I'm wondering is which brand sensors he purchases and resells. Hopefully the same ones he said live longer.

from EGORproducts page:
Honda/NTK UEGO WideBand O2 sensor, w/full sweep Heywood-Eltinge gas calibration (mixture) AND our exclusive EBP (exhaust back-pressure) calibration, used in EGOR's EBP Error Correction. [OEM Honda, NAPA, & Bosch off-the-shelf part nos. on customer request, but they won't include the gas & EBP calibrations, of course]

This states he is using the Honda/NTK sensors - these are the ones widely reported as better compared to Bosch.


[This message has been edited by D'Ecosse (edited 04 January 2003).]
 
On the Bell setup the turbos come immediately after the cats with just a very short adapter pipe in between. That sounds horrible if you know much about turbos, but it makes sense for several reasons. First is the fact that they don't cool the exhaust ahead of the cats which would make those less effective and possibly compromise emissions testing. Secondly that places them in the right position physically given the limited available space, with a minimum of additional tube bends. And thirdly it's not as far from the engine as it sounds since this is a mid-engine car. It's also one of the reasons that the Aerodyne turbos were such a good choice since their variable vane design compensate for the losses.

As for accuracy of measurement, contrary to what I've read from others here and elsewhere, Gar is quite adamant that after the cats is fine. According to him the cats are very effective at what they do but only over a narrow range of A/F. He claims that by the time you get to typical turbo mixtures they actually do very little at all and won't compromise readings. I hesitate to rely too heavily on one guy for important technical knowledge, but Gar seems to know his stuff and generally backs it up with enough explanation to be believable.
 
In response to a question about full time closed loop operation, a tech support guy at AEM responded:

There is a certain amount of delay all Wideband sensors have, depending on the type you use will determine how effective it is under high RPM/boost. The NGK/NTK units have a delay of around 15-20 milliseconds with the sensor located around 30 inches from the exhaust valve.

Placement of the sensor, type of sensor, and engine speed has a definite effect on how well the O2 feedback can correct.

It sounds like the delay is largely the time it takes for the exhaust to reach the sensor, which makes sense. So if the EBP correction of the Egor systems works really well it might be worth placing the sensors ahead of cats & turbos to minimize that delay. I wonder what the total lag time is before the A/F ratio actually changes if you start with that 20 milliseconds then add time for processing and output to the injectors. Ho much lag can be tolerated before active correction is less accurate than just a good map?


[This message has been edited by sjs (edited 05 January 2003).]
 
@8000rpm
ignition/Injections /sec = 8000*3/60
= 400 per sec
= 1 per 2.5ms
= 10 per 25ms.

I would imagine that the delay he is quoting would be the time to correct (including processing time?) - even then, we're talking about 10 ignitions with potentially bad A/F?
I'm not sure how bad that would be - just wanted to do the math out of curiousity to see what it meant in number of ignitions under this false input.
That means in the power range, between 4k-8K, it is always going to lag by at least 5 ignition cycles. Does that matter? After 25ms, the engine speed is probably not discernibly faster; it may be marginally weaker initially, but enough to cause serious problem? I would doubt it .....?
Interesting though - a fuller answer from the tech would be appreciated - what is it actually doing in reality because of that delay? You could possibly run a few points richer to your resolving point
(i.e say 12.25 instead of 12.5 for example) to be safe I would think ......

p.s. I've mentioned "resolving point" in a couple of posts - that is my own term, not sure what the techies call it - need to read the programming guide more. What I mean is that set-point that you elect you want the AF to strive for - it's 14.7 on a stock NSX ECU, but you probably want it to be mid- 12's on FI. I would imagine that value must be selectable for closed loop operation - you certainly don't want it to immediately try to be correcting (or resolving) to 14.7!

[This message has been edited by D'Ecosse (edited 05 January 2003).]
 
Regarding WBO2 sensor life, the NTK/NGK sensor used on the Egor & FJO meters was original equipment on one of the 1990's Honda Civic. So it would appear that sensor life should be quite long (10K's miles). Granted, the Honda Civic is making much less power than the NSX. But unless you are at full throttle all day long, the sensor will see similar exhaust conditions.
Bryan Zublin
 
Back
Top