10% Ethanol, Narrow and Wide Band Tuning

[wurk_truk]

That ECM 5220 is $10k!!!!!

I think for most of us here, the very first thing we need to concentrate on is sensor placement.  After seeing that pipe/bung at the show...  I am convinced.  Nothing to do with CL or OL... but accuracy for whatever tools we use.  I AM going to beat on this drum for a bit, but I can't think of ANY problems, involving sampling, if we use the Herko Blocks correctly.  Does anyone think this is not so?

Also, on a CL bike, wouldn't a GOOD test of the bung being used by the NBs be this:  Pull the NBs, install BBs like Twin Scan, and test in a map area that that is open loop.  Then jump over to the Herko blocks and test the AFR using them.  If the two tests fairly matched, the bung placement would seem to be OK, but if there is a noticeable difference, reworking the bungs would be in order.  This sound right?

[rbabos]

That ECM 5220 is $10k!!!!!

I think for most of us here, the very first thing we need to concentrate on is sensor placement.  After seeing that pipe/bung at the show...  I am convinced.  Nothing to do with CL or OL... but accuracy for whatever tools we use.  I AM going to beat on this drum for a bit, but I can't think of ANY problems, involving sampling, if we use the Herko Blocks correctly.  Does anyone think this is not so?

Also, on a CL bike, wouldn't a GOOD test of the bung being used by the NBs be this:  Pull the NBs, install BBs like Twin Scan, and test in a map area that that is open loop.  Then jump over to the Herko blocks and test the AFR using them.  If the two tests fairly matched, the bung placement would seem to be OK, but if there is a noticeable difference, reworking the bungs would be in order.  This sound right?

Sounds good to me.
Ron

[wurk_truk]

I am quoting Glens from this thread...

    Member
    Posts: 3094
        View Profile
        Email
        Personal Message (Offline)

Re: O2 Sensors, why do they do that
« Reply #23 on: Tuesday, October 25, 2011. 01:28:27 AM. »


I like the mental image of the sensor essentially being a little battery.  When it's up to temperature and the oxygen differential across (inside/outside) the element is sufficiently within range, the sensor will make DC voltage, varying in proportion to the oxygen differential.

One thing that maybe belongs here in this thread is a general overview of the broadband sensor systems.  The sensing elements themselves are the same as our stock narrowbands, so they have the same voltage output characteristics.  Their innate switching speed will be generally the same (switching speed is the rate at which they are capable of changing output voltage in response to changing O2 differentials).  The thing about these sensor systems that few folk seem to appreciate is that the sensors have an additional "mechanism" which allows air to be "pumped" into/out_of the back (outside the "pipe") of the sensing element, altering the O2 differential, and this aspect is manipulated by their own closed-loop controller which is what puts out the 0-5V signal.  We never see the 0-1V sensing element signal which that controller uses internally.

In exactly the same way as our ECM varies the injector duty cycles to cause the O2 sensor to "switch" back and forth across the set output voltage, the broadband controller varies the amount of O2 being "pumped" in/out the backside of the element, causing the element itself to "switch" back and forth across a set output voltage.  The controller derives its own output voltage (0-5V) based on what it needed to do with the "outside/backside air pump" to get the sensing element itself to act like it was, on average, immersed in the result of a stoichiometric fuel-burn exhaust stream with equal pressure both sides of the element.

That all takes time, and if it's chasing a moving target, like the output of our ECM-controlled injectors provide, it's likely-as-not going to take even more time.

But time itself isn't the deal-killer.  What is is the unknown pressure differential between the two sides of the sensing element.  If there's a differential pressure of anything other than zero, the controller cannot know that.  But it would care, because it's measuring the current it uses in the "pump" to cause the sensing element to switch, and this is its main criteria for putting out a signal (0-5v).  If there's an excess of oxygen (excess from what would be present in the result of a stoich burn) inside the pipe, the controller has to cause excess oxygen to be pumped into the backside of the element in order to reach the proper equilibrium.  If there's an oxygen deficiency (from what would be present in the result of a stoich burn) inside the pipe, the controller has to cause excess oxygen to be pumped out of the backside of the element in order to reach the proper equilibrium.

What happens when the pressures are different on the two sides of the element?  The controller is going to have to pump oxygen one way or the other to make up for that, but it won't know it's doing so for that reason.  It'll think the exhaust is the result of either a richer or leaner burn instead.  Now if there were pressure transducers both inside and outside the pipe, right at the sensor, the controller could poll them and take the readings into consideration when it's developing its output signal (0-5V), and that output signal would be much more pertinent to the task at hand.

Obviously, the temperature of the sensing element would be valuable information for the controller, as well.  But between the two, pressure and temperature, the pressure is going to be the larger error-causing factor when its differential value is only assumed by the controller.

And don't forget that the controller is going to need at least a couple of "switches" resulting from its reference-oxygen pumping action before it can draw any conclusions, and this takes time.  The time factor could be minimized if the sensor's closed-loop controller had intimate communication with the engine's injector controller and knew what was going on in that respect at every instance.  Without that information (and, of course, a bigger, faster "brain"!) you could wind up with a situation anywhere between the two systems being in perfect sync or perfectly out of sync.  And this will prove, in the long run, to be the underlying reason the broadband sensor systems are "slower" than the stock sensors.  It's not that either sensing element is faster than the other (when new, and within design limitations, naturally),  its that the one is standalone and the other is part of a complete closed-loop system of its own.

So what would happen if a dyno operator was using a broadband [system] along with a device to vacuum-pull a sample from the pipe to and across the sensor tip, but he did not include the sensor itself in the same amount of clean-air vacuum, leaving it instead exposed to "outside" air pressure?  Would his readings have a "built in" skew one way or the other?

Steve, would this be true?  On all dynos, vacuum is employed.

[glens]

Steve may well choose to leave that question unanswered.  Though I'd like to see what he has to say I'll not be disappointed if he foregoes comment, either.

I must state that I'd had the "pumping" direction wrong.  The "pump" evidently works on the sample side of the sensing element, not the reference side.  Apart from that glaring error and the necessary descriptive swaps of "rich"/"lean" that result from it, the information should be entirely pertinent.

It would be a rather "simple" test for you to undertake with the equipment you have on hand if you've got the time:  Use one of the broadbands in one of your fancy new bungs and the other in one of your shiny new Herko blocks on the same pipe.  "Free air" calibrate both sensors the same immediately before installation, take note of the readings at various loads/speeds, pull the sensors and check the "free air" calibration then swap them (maintaining sensor/controller pairing).  Repeat the readings at (hopefully the same) various loads, and maybe even check the "free air" calibration one last time after pulling the sensors.  To be very complete, naturally, you'd repeat the test with the sensor/controller pairing swapped to cover all bases.

Of course you'll have to assume that both your custom pipe bung/location and Herko block installation are providing for "good" samples otherwise.  A comparison of the data should indicate either a trend or the lack of one.  At some loading conditions the pipe should be under greater internal pressure than external at your custom bung location, at some greater external pressure than internal.  And I'd assume your Herko block sample would always be at nearly the same pressure differential across its sensor.

You may be able to readily see the effects of pressure differential changes in some of the readings but determine that they're inconsequential at the "AFR" the sensors are "reading" (hahaha).  By the way, it may be best to use a straight "13.5" or so across the board in your fuel table for this, to eliminate that variable.  It wouldn't be so important for this test that the VEs in use were precisely correct because you're merely comparing sensor readings, though it'd be nice if that could be another variable removed from the "equation".

[wurk_truk]

You just gave the 'bung test'  any difference would be showing whether a welded bung is properly functioning.  I was planning on this test using independent LC1s to test bung location.  Throw one LC1 in each hole and they better match, or bung is worthless.

MY thought is since reference air has NO contact with exhaust air (inside of the O2 sensor), and simply reading oxygen content that it doesn't matter.  EVERY DYNO on the face of the earth uses vacuum to pull across the sensor.  One would have to have sufficient vacuum and not too much, to mimic what happens when the O2 is under pressure from exhaust.   Pressure, vacuum... same difference.  Moving oxygen molecules flowing past the sensor head.

DJ used to use a diaphram pump that pulled 15 in of mercury.  Now they use an air operated pump that pulls 35liters per minute.

As long as the reference molecules are in the 'free world' I feel everything is doing its job.  After all, these WERE designed to read FLOWING oxygen molecules -vs- a more 'stationary' set of oxygen molecules for reference, right?

[glens]

I'm not so sure that any difference would show the bung location to be problematic.  The bung location might be perfect but you're pulling too much vacuum on the sample side, as opposed to the reference side, of the "reference" sensor.  And both the bung and the vacuum setup could be wrong...

Not that a "free air calibration" indicates how well the sensor/controller will operate nearer stoich samples anyway (all else being equal), but the "free air" is at exactly the same pressure both sides of the sensing element at the time the reading is taken while that being the case in "usual" operation is probably unlikely.  So there's that discrepancy already.

I think that, ideally, if you're going to "pull" sample gas past the sensor you should "pull" exactly the same amount of reference air past it, too.  Either that or measure the pressures both sides of the sensing element and factor the delta into the controller's reporting scheme.  That sounds about like what the "true" widebands do, and in part would account for the greater cost of those systems.

If you perform the test I outlined above, I'd be grateful if you'd share the results.

[hrdtail78]

Would closing the vents on my lowers cause a pressure drop on my front O2 outside pressure reading?

[wurk_truk]

True wide band, like ECM 5220 has only one pressure transducer.

[glens]

True wide band, like ECM 5220 has only one pressure transducer.

That's all that's needed to determine the pressure delta across the sensor.  Think "gauge" pressure, not absolute pressure.

Would closing the vents on my lowers cause a pressure drop on my front O2 outside pressure reading?

Maybe.  Maybe it's the other way around.  Can you make a 60 MPH wind in your dyno room to check it out?

[wurk_truk]

Well, Glen...  I surely will try this out and report back.  Always GOOD to learn things.