Cliff
-
Posts
386 -
Joined
-
Last visited
Content Type
Profiles
Forums
Events
Gallery
Community Map
Posts posted by Cliff
-
-
I'm not particularly interested in what the wideband sensors are actually measuring or how. They claim to measure lambda and all of the independant evidence supports that they do a reasonable job under most conditions.
A precise definition for optimal?
Only those you are in very competitive racing environments would require the optimal as you would define it and justify the expense. Those people for one would not be riding Moto Guzzi.
For me optimal is good enough to not require additional expense or effort to improve. If I can be with 5% of the peak I'm happy. That could be an A/F ratio anwhere from 12 to 13.5 for power and anywhere from 13.5 to 15 for cruise. That may seem imprecise to you but for me the results are good enough to keep me pleased with the results. For me efficiency is a higher priority than raw power. The bar for milage used to be 5l/100km. Currently its 4.5, I don't do this by trying to go super lean.
My next project is to go closed loop for spark advance. I'll be hoping to get milage close to 4.
-
I think you might not if you had actually at least read all of the posts, even if you did not understand all of it.
I think Juha is one of the more technically literate folks here and to suggest that he has not read all the artiicles or does not understand is a disservice to him and others here.
In any case, the usefulness of sensors depends entirely upon the relevance of their outputs. How can O2 sensor output be relevant when it can be the same for a 1.9% to 11.4% CO with 1% CO approaching lean misfire, 11% approaching rich misfire and 3-4% being ideal for power, and when it can also read all the way from .2% to 4% O2 at best power.This is where the skewing of facts comes in.
In a rich mixture there is no oxygen. It is a zero quantity. All you measure is noise. This is the same for CO in a lean mixture. Its a noise that is measure.
So yes you are correct in that an oxygen sensor can not give any useful information here and not suprisingly they don't.
However if you look at the quote below from the thread I quoted earlier
First off we have to dismiss the rumour that they measure oxygen. Only when lean. When rich of stoich a HEGO becomes a glorified EGT probe and a wideband measures the content of the other ions in the exhaust gas. Call em lambda sensors and it becomes less confusing. If you wantconfusing go and read up BB's research for the PWC. Needless to say, with a GOOD wideband you can get within 0.2 or so of the correct lambda.
So
fact 1 ) an oxgyen sensor can not be used to determine anything in a rich mixture.
fact 2 ) Wide band sensors can measure mixture at rich mixtures
Conclusion - a wide band sensor is not measuring oxygen when rich.
I don't think anyone here has claimed that a closed loop can produce better results than what can be manually tweaked on a dyno. Even a dynolinked tuned dyno sessionwill fall short.
There are those that will want the absolute maximum power and for them an expensive dyno session is the only option. How much does your tweaking cost?
Most have stock bikes that do not run anywhere like they should or they have made a few common modifications and want to get there bike close to optimal.
For them a dynolink session is all they require. Yes there are quite a few that are happy with the results. A closed loop ECU will give the same result. No it wont give the the absolute maximum available but it will be close enough for most.
-
It seems that FUD is becoming an acceptable practise of corporations. They always seem to provide skewed information to support their product rather than a balanced view.
First we had this supposed open loop vs closed loop thread
http://www.v11lemans.com/forums/index.php?showtopic=4028
A more balanced topic theme would have been "OEM closed loop controllers still not there and closed loop controllers cause additional pain for Dynojet"
Now we have a corporation that in addition to making dynos make their own 4 gas analyser to gain a competitive advantage.
I have no issue with believing their gas sensor is superior to a wideband sensor but I find the hard to believe their line about wideband sensors being useless. Sure they have documented some anomilies with the sensors but there is no investigation as to what has caused them.
So far I've spent a few hours googling for any evidence to support their claims and so far I have found none.
I have even posted a question on the Megaquirt forum http://www.msefi.com/viewtopic.php?p=33486 . There is some interesting information surfacing. There is even an account of a lean reading anomoly from an o2 sensor and an explaination.
-
Does that mean that the earth was not round until everyone agreed upon it? In any case, I'm not the only one who thinks that tuning to an O2 sensor voltage is a bad idea. See http://66.47.68.116/dyno/dyno_error.html and http://66.47.68.116/dyno/4gasEGAvso2sensor.html. For a list of tuners that are likely to agree, see http://66.47.68.116/dyno/dyno_locs.html.
Thanks, that was very interesting even if it was written by people pushing 4 gas analysers. I'll certainly keep that in the back of my head and be on the lookout for any anomalies.
Are there any cheap and simple CO sensors that can be mounted on a bike instead of an O2 sensor?
Whether or not the ECU makes accurate and repeatable pulsewidth adjustments based on the O2 sensor voltage is not in dispute, but whether or not these adjustments are appropriate is. You can say that you have a "monotonic and reproducible relationship between voltage and pulsewidth" and I will not argue one bit. But you can't say that you have a "monotonic and reproducible relationship between voltage and mixture" unless you have some way that is independent of O2 sensor voltage to verify that.Do we have monotonic and reproducible relationship between pulsewidth and mixture? If we do then what I said is correct.
I think what you are saying here is that the relationship between mixture and voltage may vary at different datapoints and what I'm saying it will still be monotonic and reproducible.
So if I wanted to get maximum performance I would need to determine a voltage level for each data point.
If it's possible to have CO readings that vary from 1.9% all the way to 11.4% with a 0.4% O2 content, I can certainly see how your ECU would be prone to making infrequent changes confined to within 1% of a mean. I never suggested that there was some other better O2 sensor available that would solve the problem. Again, the same voltage can exist at many different mixture strengths.Here you are definitely mistaken. Any closed loop system will drift aimlessly with incorrect feedback. If the o2 sensor was not giving me a signal that was related to mixture, my ECU would not produce a stable output.
-
This thread is a spin off of the ECU thread to discuss the various dynotuning techniques that are possible. I'll start with the techniques I have I used.
When I did my first dyno session I only had open loop and the dyno's gas analyser. It was not a particularly successful run but was certainly educational.
The first problem I had was the dyno did not produce any power/torque values that where usable at the low RPMs. The bar graphics and numbers bounced all over the place in response to the twins firing. That left using the gas analyser. This also proved problematic as the sensor was put into the rear of the exhaust and the bend of the sports exhaust prevented this going in sufficiently. MAKE SURE the dyno you use has a sensor that will properly attach to the bike, ideally through the existing screws in the header.
The next time I went to the dyno I had the wideband sensor on the bike and the controller working closed loop. The technique I use runs vertically though the map ( constant speed ). I would have the operator maintain the speed of the dyno and then I would position the throttle to coincide with the throttle breakpoint. I wait for the close loop to settle, this took up to 10 seconds then as the map was way out, and then I would commit the adjustment and move on to the next throttle spot. After finishing all the throttle settings we move onto the next RPM and do the same.
Now my software is a lot smarter and I don't even need to wait for the closed loop to settle. I can set the throttle wait as little as 1 second then move on. The throttle is displayed on the bike's tacho for convenience. Any adjustment is automatocally applied if conditions are right ( a diverging correction factor ). A converging correction can not be used. A few passes are required to catch all the points.
It would be possible to run horizontally across the map. Set the throttle and allow a slow acceleration, then repeat a few times.
One thing to note is you can not get all the data points on a dyno. Only the top left half the map is doable. The bottom right of the map is not doable on a dyno. This is the back off /deacceleration area. I usually fill these on the road. This area might not sound important but it actually is if you want smooth throttle response for things like negotiating round abouts etc.
-
Thats an interesting point. The dyno I used was an eddy current type with attached PC. At the lower RPMs ( probably 2000 was the first one I tried ) it was impossible to determine what the bikes output was, let alone maximise it. The dyno literally showed the power pulses of the twin. It might of smoothed out at higher revs to be usable but adjusting the mixture for a given A/F ( or voltage ) was the only measure I had. I think I'll start a new thread is this one is getting a bit long
-
Thanks for the link Carl. Thats the best info I've seen so far. I've actually been thinking about how to add this in the past week. That and a simple gas sensor controlling circuit. The cheapest at the moment seems to be over $250.
Closed loop ignition control. Thats going to spark an interesting debate isn't it.
-
I found optimising the ignition advance very difficult to judge. I only tried at the lower RPMS and the power pulsing on the dyno was hard to interpret. The best method is to use Optimiser at the moment. The Spark Advance menu has a square wave modulation of the spark advance. Sometimes seat of the pants you can feel this and move the advance to the stronger direction
-
From the graphs, the fact that O2 is 0 until the stoiciometric ratio implies that it is not going to be a good measure below that point. Similarly the fact that CO is not produced above that ratio means it is not a good indicator for lean mixtures. The O2 values you quote is just noise around 0.
Clearly the wideband sensors are not O2 sensors in the same sense as the narrow band sensors. Their construction and chemistry is completely different. These wideband sensors are used in many commercial and open project gas analysers. They give coverage from 11 to 1 to well over 15 - 1. If as you suggest these wide band sensors are unusable anywhere rich of stioc, then this fact would be public knowledge. We would hear all over the place of this deficiency.
Please provide some links that say these sensors do not work.
I don't know how these sensors work or what these sensors measure. I just know that I can set a target voltage and my ECU will produce a mixture ( as measured by injector pulse width ) that is stable ( within 1% of a mean). I can reduce that voltage and produce a stable mixture that is leaner. I can increase that voltage and produce a stable mixture that is richer. I have a monotonic and reproducible relationship between voltage and mixture. I'm at a loss to imagine how a "real sensor" would differ from this behaviour or what more it would provide me.
How is you tweaking the mixture for peak power any different to me tweaking the voltage for peak power ( a voltage for each data point not one voltage for all ).
If you dynoed mid winter, six months later is totally different climate wise. You are assuming that the ECU has perfect compensation curves. If you are using a power commander, Wayne Mcdonald has already explained how its calculation is going to be in slight error. Add to that a service were you adjust tappets, balance throttle bodies and adjust TPS. You've pretty much invalidated that dyno session. Unless you're proposing another dyno session.
It is also true that these gas sensors age over time but with a life of more than 100,000km on unleaded fuels I think it is pretty stable over a year or 2.
How can I have all those features I mentioned? Because I can specify that any part of my map have any target voltage. I can make my 25% to 50% throttle region track lean. I can make my 75%+ throttle rich. I can even go open loop at any point if I want to.
As I have mentioned before there is not one type of closed loop controller. As far I know they are all based on an open loop controller and hence have a map that should be adjusted to suit the bike. The gas sensor only gives further hints. Now tts quite possible that you have experience with OEM ECUs that are closed loop. It also would not surprise me to hear that that don't work very well. I however have built my own ECU, if you haven't already picked that up, that has optional closed loop behaviour and it does everything I expect it to do and nothing I don't. Not anymore
BTW I would be pleased if you would answer my question on the other thread http://www.v11lemans.com/forums/index.php?showtopic=4296
-
Couldn't find much on their site. How much???
-
$2000 for a tuning link's O2 sensor?!!!
How much is the Bosch sensor that you use?
Could you bundle it with an accurate* LED meter that would tell me how lean my bike is running and sell it to me for under US $200? (*accurate from 12-15:1 ratio at 0.5 increments. So, it would measure 15 or maybe 7 LEDs or 3LEDs and an adustable target range with 0.5 accuracy)
Does the Bosch thread in to our pipes without modification?
Sorry that should have read $2000 analyser not sensor.
Not sure what the best value WB sensors are nowadays. Wayne Macdonald mentioned a VW 5 wire sensor that was relatively cheap. I'll ask him to post further details.
The bosch cost me about AU$360. Its good in that it doesn't require fancy electronics to drive it like the newer 5 wire ones do.
Other than that there's http://www.wbo2.com or the Wideband commander is not a bad looking package. Both depressingly expensive. Hopefully in a few more years they'll start charging realistic prices.
Edit: An ebay search on Wide-band Sensor shows some cheaper options including kits.
Here's some interesting info about cheap sensors http://www.megasquirt2.com/PWC/lsupur.htm
The standard thread size is 18mm. There are a few smaller sizes , 12mm I think ) but there would be less choice of sensor.
-
Well those gas charts certainly explain why the common narrow band gas sensor will not work well with A/F below 14/1. This limitation though should not come as a surprise
To dismiss all o2 gas sensors as not workable in the richer range though is to ignore the fact that wide band sensors like the Bosch 0258 104 002, that I use, and the newer 5 wire sensors do exist and work very well in that area. They are the basis of many commercial analysers including RacerX's US$2000 sensor.
It may also be true that a single A/F target will not give maximum power across the board. However just as you could tweak each map point to with 1HP by adjusting the mixture, what is to stop one similarly tweaking each points A/F ratio to do likewise. No I don't have that capability today. I could have it tomorrow. One thing I would be sure of though. Which system would still be within 1HP six months down the track.
WRT to what do we tune for.
Would I be happy with a 2% drop in power for 5% better economy. You bet.
Would I be happy with a 5% drop in power for 20% better economy. You bet.
Do I care that I get 95HP rather than a 100. No
Do I care that the throttle is responsive and there are no misses at. Yes
I'm considering dual plugging in the future but not because of the HP increase.
The good thing is I can have all of those features above, including the 100HP if I want.
-
I'd have to say my curiosity is peaking. Or is it paranoia.
Recent member to v11lemans with not too much said about himself.
Doesn't have a guzzi.
Appears in the middle of a hot ECU thread, shortly after Tuneboy announcement.
Does produce a lot of numbers for an aprilia and has an expensive gas analyser.
Has some negative sentiment about closed loop and/or at least its feasibility.
My guess is he's a power commander guy.
So please Moto, tell us a bit more about yourself.
-
How does the ECU decide not to respond to sensor inputs? Does it not respond if they are transient? I thought the whole advantage of using O2 was to be able to do just that. Does it not respond if they are outside of certain parameters? If so, it also won't respond when they are legitemately there.
In my case I think it comes from mixture update rate and how I respond to the o2 inputs. My update rate is based on engine speed from about 1 per second at idle to about 10 per second at redline. Thats fairly slow in engine terms. There's also they way I respond to the o2 value. If I happen to sample in a transient and see 0V I don't go "give me full fuel Scotty". Instead I bump my mixture by 1%(my maximum increment). Chances are next time the sample is slightly over voltage and the mixture is pulled back again. If I happen to see 2 transients it will go another 1%. Chances of see 3 transients are fairly remote. Its a statistical averaging.
What tends to happen in practise is my mixture slowly fluctuates by about 1%. Of course this assumes the underlying map has been previously flattened. I consider that to be pretty tight. I think that equates to about a .1 change in A/F. So if my target voltage was set for 14to1 my engine running is between 13.9 and 14.1.
-
I think you guys are over complicating it with regards to gas analysis. Sure O2 is an indirect measurement and can be polluted by air leaks and even the valve overlap could cause air to pass straight from inlet to exhaust. The whole thing is like analysing a fractal image, always more detail the closer you look.
The important thing though is the bottom line and I will outline the results I have got on my two bikes.
With my sensor, if I set .6V as a target a short ride will give me a light grey coating on the sensor ( it shows better than the spark plug ). If I set .8V a short ride gives me a sooty black coating. Of coarse I can set any value in between and I expect a usable range of 0.5V to 0.9V. I couldn't tell you what the corresponding A/Fs are. I don't particuarly care. I can make and adjustment on the fly. Say changing the target voltage from 0.6 to 0.62. Doesn't sound like much does it but I will see my closed loop adjust mixture by a small percentage.
I find that the raw sensor voltage can vary substantially especially when tracking the leaner voltages. This is with static conditions - constant throttle, constant RPM, constant ECU output. Tracking to 0.6V I will see many "wild" excursions to 0V or .75V. I can understand that someone trying to be too smart and tracking it closely could get into trouble. I wont brag at being able to do something super smart that the OEMS can't do. I just did the simplest thing I could think of. Perhaps thats the secret. A good engineer will simplify, a team of engineers will complicate. The OEM ECUs have about 2-3 times more components than mine.
The raw sensor inputs are not the important thing, its the system's response to the inputsthat matters. Even with this seemingly erroneous data, I find that the tracking is stable. The closed loop adjustments will stay within 1% or 2% of the mean value.
Is 1-2% good?? I'll let someone else pursue that line.
-
A voltmeter would be easier to install and give you similar information.
-
Welcome Pexi
Another thread you may be interested in
http://www.v11lemans.com/forums/index.php?showtopic=4028&hl=
I'm about to restate roughly what I said then. This thread is one of those arguments where everyone is agreeing with each other but everyone is arguing different points.
The main argument seems to be what is a "closed loop controller" or what is a "full closed loop controller". Perhaps the marketing droids have pencilled in some definitions for us. Lets hope they do better than the "1 bit DAC" boast used on audio CD players.
So lets get really technical.
In a classical engineering sense a closed loop system is quite simple. One can be easily constructed from an OP amp ( operational amplifier ) and a few resistors. Basically the system has an input and produces an output. The output is fed back as an input in a negative sense ( negative feedback). This negative feedback is used to correct the system ( reduce distortion and non linearities ).
Closed loop systems can suffer many problems ( reponse times, stability ) that must be engineered out. - applying the correct amount of feedback and dampening. ( Suspension system has these same issues )
In the case of an closed loop ECU the input is the A/F ratio or O2 sensor voltage. There would be none of the current inputs of air temp/pressure etc. There would not even be a throttle input or a map or memory. The ECU would correct the injector pulse based purely on the O2 sensor.
Such a "full closed loop" ECU would be difficult if not impossible to achieve today and why RacerX is correct in saying that there a no "full closed loop ECUs" out there.
However thats not to say an ECU can not make use of the o2 sensor to improve its output. Probably all closed loop ECUs out there are open loop ECUs that can take hints from the o2sensor. Heck the market droids may call it "fuzzy logic". I've never understood that term fully. Is that the logic you use when you don't know what logic to use?
So all closed loop ECUs currently are next generation open loop ECUs. They have a map, they adjust it for barometric variables AND they adjust for O2 feedback.
Of course the actual implementation is completely different from one ECU to another. How quickly they respond to barometric and O2 inputs, how much correction is allowed whether that information is reapplied to the map ( short term or long ). You DEFINITELY can't take one and assume it is typical of others.
I heard of one car where if you drove up a mountain you had to turn off the engine and restart as that was the only time the pressure sensor was sampled.
In summary RacerX may be arguing whether a controller is full closed loop or not, whereas I'm arguing whether the use of a gas sensor can improve a bikes performance ( not necessarily its max power). Definitely not the same argument and they should not be confused.
-
Sorry Cliff, I wasn't trying to suggest any problem with your system running in closed loop.
You're alright, I was primary replying to another post.
Just starting to get a little annoyed that some one's supposed argument as to whether closed loop works or doesn't, is based on the assumption that what the OEMs produce dictates the state of the art, or the fact that closed loop gets in the way of what a power commander would like to do and must be disabled, therefore its no good.
Realistically how many race teams would have the resourses to pursue this to the level of producing custom ECUs. The majority will just use whatever off the shelf components are available and for most that may mean strapping on a power commander because the OEM ECUs give them little other option. You don't need closed loop to produce maximum power, you just need cheap dyno time.
That may work better for race conditions where you can dyno the bike every time you rebuild the motor.
For me closed loop means
1) my bike runs reasonably well even if the map is off.
2) it can be used to improve the map - as the map improves so does the closed loop
3) adds an extra diagnostic - if I see large corrections I know I should start looking for problems. If its the o2 sensor I can flip a switch and go open loop.
-
I can assure you that no current bikes use full closed loop.
It was only a few years ago you could say no current jap bike uses EFI. Those high revving/high performance engines must have carbs. Those Jap engineers most know something heh.
Clearly those running My16M closed loop including myself have been deluding ourselves. Hell I better double check my code maybe its not actually running closed loop at all.
-
-Cliffs My15M is his version of a Marelli unit that basically just blows open the code so you guys can access it. His unit will set you back about $290 US according to his latest info for a 'full kit' that you must be slapped into YOUR ECU. That means, unless you have a spare ECU ($100 on Ebay maybe??), then there is probably no going back after you gut yours. Also, he gives no price on his optimizer. How much for that??
The latest prices since 1st Jan have changed. The AU$700 for the assembled and tested ECU now includes the OEM ECU ( subject to availability ) and postage. The Optimiser price is hidden on the bottom of the Optimiser page. I've forgotten to update to the latest which will be AU$300 assembled inc postage and the full kit being AU$150 inc postage
-
Well I think the Tuneboy is a great solution for the 15M and neatly solves most if not all of the 15M problem/limitations. Probably would of gone that way myself except 10yrs of working with microsoft software had giving me a gut full of trying to work out how the other guy does things.
There's still some hope for My15M though. I'm sure all My16M users will agree, there's more to a sweet running bike than mapping. Most were running better with my Sport Corsa map on My16M than they were with their weber. That includes Centauros, Daytonas and even a Cagiva Eliphant? I think only 3 or 4 have actually dynoed their bike. So maybe My15M will succeed simply because it works better.
To summarise My15M benefits over WM15 + Tuneboy.
- Throttle pump emulation
- Optional closed loop implemention that does not have any problems that I know of.
- On road adjustability/diagnosis ( unless you ride with you laptop )
- Automatic map improvement ( currently with optional Optimiser only )
- On going development
To summarise PC benefits over WM15 + Tuneboy.
Throttle pump
-
If Cliff's My15M ECU would allow Tuning Link mapping input (and I see this one brewing at the local pub already)... you'd be getting somewhere.
If you have the details I'm sure I could accomodate.
I did the last complete dyno run in under 2 hours at AU$110 an hour. That was the first time I'd done it in closed loop mode so there was a bit of farting around and I'd just changed map formats so it was out by more than 10%. I reckon I could do it in under an hour nowadays with the optimiser.
I tell the dyno operator to hold the revs and I then open the throttle to the desired spot, wait about 10 seconds, then move to the next throttle setting etc. The updates are automatically applied. Then move onto the next RPM.
I prefer doing it on a quiet road nowadays, at least the main data points. Not to keen on standing next to a screaming motor on a dyno.
And my exhaust goes all sorts of colours.
-
Cliff.
Have you thought of expanding your product to the new Ducati's (59M ECU) ?
Is that the one in the monsters that looks like 2 15M together?
Anyway good on you for doing a version for the 15M and for doing a great job on it. It is a must have for anyone with a 15M. In fact I had a query for My15M this morning and I told him to make sure to have a look at tune boy also.
Fortunately you can't do anything for the older controllers that are EPROM based.
Maybe its time to look at knock detection.
Couple of questions for you.
Does the 15M only have one o2 target voltage?
Does it work with the newer sensors that give inverted voltages?
If you're near Sydney at any time we should get together.
-
- a bit pricey. compared to My15M where you get both H/W and S/W for less money,
but maybe it just us old hackers who likes the openess of My15M.
I'm glad that My15M is still an attractive option for some. I was thinking it was sunk as a product.
Guess I'll just need to add more features.
ECU
in Technical Topics
Posted
Thats an interesting point about static testing Carl. Most ECUs have an enrichment for acceleration. If you maximise the static case, how much do you lose on the road due to the now over rich mixture.