1 ... 58 59 60 61 62 ... 64
ShawnG
ShawnG UltimaDork
11/22/20 7:27 p.m.

Put that carpet somewhere that you can get the temperature over 150 degrees F for a few hours and you'll be rid of moths and eggs.

We've fought moths in some of the customers cars. Ozone doesn't work, mothballs just make it smell like a Mexican hotel. Heat always does the trick.

It's the larva that eat the wool.

mke
mke HalfDork
11/24/20 8:58 p.m.

In reply to ShawnG :

I just need the covid crisis to end so Lana goes out and I can sneak the carpet into the oven!

Trying to move through the list.....

The wheel position sensor are wired reading correctly.

The Oil Pressure transducer is wired and reading correctly

The water Temp gauge repondes correctly (same as oil pres, the ECU reads a modern sensor then drives the gauge)

The oil pressure warning light works.

The slow down, aka Malfunction Indicator works.

The ECU is now reading the OEM oil temp sensor which took a little thought as its an 8V analog setup and the ECU is 5V digital but voltage divider made with a couple 2.2K resistors I had did the trick....high enough resistance to be basically invisible to the gauge and low enough resistance to be basically invisible to the ECU which has a 100k 5V pullup on all the analog inputs.....a "5V pull up" is a connection to a 5V supply, usually with big number resistor so it generally won't disturb what its trying to measure but can make it easier to use the input for switches and such...but makes it harder to use the input with old school analog shi...items

I got the ECu to read the 8V analog veglia oil temp sensor and not mess up the gauge reading and not burn out the ecu input pin.......its simple little 2:1 voltage divider that cuts the 0-8V signal to 0-4V and is high enough resistance to be basically invisible to the gauge and low enough resistance to be invisible to the ECU.  Then resistors will break easily so I use the wire and heat shrink for structure......analog shi.......

 

 

 

 

mke
mke HalfDork
11/25/20 2:48 p.m.

Spent a little time staring at the pinout sheet in that last pic I posted....as I run out of pretty much every kind on I/O pin it appears to be time to optimize usage a bit better.  Stuff like some digital input pin have pullups and some don't, the clutch pedal switch would love to have one, but the pin left for it doesn't while the front wheel sensors pins have pullups but those sensors don't need it.  It looks like about 1/2 dozen pins should move.

Also the ECU has 2 built in WBO2 controllers but like everything in this ECU they need a model of some sort to be written to configure them.  Last go I decided to not mess with it and just used a pair of external O2 controllers but now I think I'm trusting what I'm seeing from the NBO2 sensors enough to switch the WB sensors to the internal controllers.  I think this is the last of the wiring items on the to-do list.....I'm going to need the block back soon.

golfduke
golfduke HalfDork
11/25/20 2:59 p.m.

I thought your motor work was a work of art, but your wiring expertise and attention to detail might be even better! 

 

 

mke
mke HalfDork
11/25/20 3:14 p.m.
golfduke said:

I thought your motor work was a work of art, but your wiring expertise and attention to detail might be even better! 

Its a sickness. 

I didn't post this yesterday when I did  because I was a little embarrassed......the new oil pressure transducer came with a mating connector/pigtail to make install easy, so I did the only rational thing, I cut it apart and rewired the connector because why on earth would I put splices out on display in the engine bay?  seriously?!?

And PVC wire on the hot engine?  I think not.

and foil shielding? there's no reliable way to splice into that.....

No, the only answer is cut it apart and try to save the connector....I had to solder the pins as I didn't have replacements but I got......I can't help myself.

mke
mke HalfDork
11/27/20 6:09 p.m.

More slow going getting the ECU's internal WBO2 controllers working.  The guy who owns the ECU company shared a few models and this is one of them but he works in scripting and I don't so I had to figure that out a bit to get the model to built....then try to figure out what it does exactly.  I think I have it it, at least mostly, it seems to be mostly working but I suspect I have something pinned wrong as it settles at .996 lambda sitting in open air which is clearly not right.  

Now my plan was to roll the push mower over and stick the sensor in the exhaust to test it but the external controller sensor that should be working fine is not giving me a reading, it pins at 1.3 lambda in open air but doesn't change in the mower exhaust....that could be the truth I guess ....the push mower doesn't have a choke I can get to easily....gave up for the night.

mke
mke HalfDork
11/29/20 7:13 p.m.

A couple long days filled with confusion later and the onboard WBO2 controllers are MOSTLY working

The innovate controller I was using lst run wasn't working now because it was mislabeled....there is a good chance all my data from that time is flipped bank to bank.  Easy enough.

I had all kinds of things to sort with the on-board stuff.  I'm pretty sure the model I downloaded was a simulator test only kind of thing and has never actually worked with a real O2 sensor....I need to confirm that but for sure it wasn't behaving in a very useful way for me.  That was for sure confusing me but was not the biggest issue which was a sensor/wiring problem that took a while to find because I was sure the model wasn't doing what it should.  

So, I have a pair of innovate LC-2 WB setups.  They use  a bosch 7057 sensor and for whatever reason I had a 3rd sensor which was also 7057.  When I googled AEM 4.9 sensor I found it was a 17025 which is pinned differently so I adjusted my wiring accordingly....everyone seems to call the pins different names that I'm sure make sense to THEM, but I just went by the pinouts......and 2 days later realized that while all the new AEM stuff is 4.9, they spec a 4.2 sensor for the infinity ECUs so I had it connected wrong.....and the pins of the 7057 4.9 sensor match the 4.2 so it should have been easy.  

That sort I spent today taking reading on my O2 sensor tester......a mt dew bottle, and propane supply.  One sensor is the innovate and one driving on the ECU.

 

My struggle was getting the controller tuned....and I've yet to really succeed.  I get it looking good going lean to rich

but then rch to lean it wants to "ring" or oscillate out of control

but if I tune to never oscillate then it often never matches the properly (cyan well above red)

And that is where the night ended.  I have readings but there is a bit more work to be done.  I suspect that internally the ECU is adjusting the gain on the sensor at lambda=1  from 8 to 17  and then back to improve resolution over the range and that is causing my tuning issue...which brings me back to I don't believe the model I downloaded and have pretty heavily modified at this point ever ran a real sensor.

 

 

Syscrush
Syscrush New Reader
11/30/20 7:45 a.m.

I can just picture the author of that model:

"It works on my machine."

mke
mke HalfDork
11/30/20 9:22 a.m.
Syscrush said:

I can just picture the author of that model:

"It works on my machine."

Those pics are the AFTER I mostly fixed it surprise

In fairness though, it was very helpful to have something to at least look at....and the reason I use this ECU is it lets me set things up the way I want them setup. The downside to that is everything needs to be setup and sometimes things that you'd think are simple are anything but simple.  This is like the CAN stuff, I suspected it was not going to go easily which is why I used the innovate stuff on the first go round as I already had plenty to worry about and I can always give up and go back to the innovate stuff.

....it's close though.  I got good readings across the range.  I think the next step is dig though all the logs I made and see if I can piece together what works in what range then setup tables to select the correct PID values to use instead of continuing to insist 1 set and done its is even going to work well.  Once that's all good I can tie it into the main model and set stuff like heater off when the engine isn't running, some kind of start delay and temp ramp up and such.  Its easy to take all this "little" stuff for granted I guess.......but somebody did a ton of work to get it all setup correctly in products you buy it turns out....which is no doubt why they believe they deserve to be paid laugh

 

Syscrush
Syscrush New Reader
11/30/20 2:13 p.m.
mke said:

I think the next step is dig though all the logs I made and see if I can piece together what works in what range then setup tables to select the correct PID values to use instead of continuing to insist 1 set and done its is even going to work well.

This suggestion is probably obvious to you, but I'm gonna throw it out there just in case it's helpful to you or someone else who undertakes something similar:

Your lookup strategy should probably have some kind of anti-hunting feature. A typical way to do this would be to have different enter & leave thresholds at both ends of the range for a given value.

mke
mke HalfDork
11/30/20 3:24 p.m.
Syscrush said:

This suggestion is probably obvious to you, but I'm gonna throw it out there just in case it's helpful to you or someone else who undertakes something similar:

Your lookup strategy should probably have some kind of anti-hunting feature. A typical way to do this would be to have different enter & leave thresholds at both ends of the range for a given value.

Yes, in enginelab speak that is the "limiter" item

limiter description

I use that for simple stuff like radiator fan on/off and I could use it here...but I try to avoid discontinuities in control logic which this would cause when it switches from 1 set of control number to another.  It would probably be ok because those numbers aren't the actual output and only calculate how the output will change but it still leads to the answer being different in the "dead band" for a rising signal vs falling signal.

I'm not really resource limited so I tend to use lookup tables for stuff like this, probably 1D in this case.

1D look up

With a table it will interpolate between point so its a smooth transition and I can add enough points to approximate the shape of the curve a bit to get good control everywhere.  I could go to a 2d where the other axis would be battery voltage as that will also change how it acts, but for a heater that is simple math I can just include iin the calculation vs needing to lookup. 

There are a couple more tricks I can try like shutting off or resetting the controller when the heater is driven off which would being the heater back on much less aggressively....things like that.  I know its possible to get a good result......I'll spend another week on it before I give up cheeky

 

 

 

mke
mke HalfDork
11/30/20 9:14 p.m.

That all helped quite a bit, its definitely usable now.....I still think it could be a little better  but  overall not bad. Yay.

I pulled a log of the lambda values going full rich to full lean over 10 minutes or so with both the onboard and innovate (that I recalibrate just before the run) and I don't match everywhere.....but that's a simple table to adjust.  Maybe tweak it then pul another log or 2  then switch sensors and do it again and its ready for prime time.

Syscrush
Syscrush New Reader
12/2/20 8:03 a.m.
mke said:

It would probably be ok because those numbers aren't the actual output and only calculate how the output will change but it still leads to the answer being different in the "dead band" for a rising signal vs falling signal.

Gotcha. I misunderstood - I thought that there was feedback between the measured amount and the current sent to the heater, not just signal conditioning on the measured amount. Thanks.

More importantly, I'm glad to hear you're feeling good about how it's working as you continue to spiff it up.

mke
mke HalfDork
12/2/20 8:59 a.m.
Syscrush said:

Gotcha. I misunderstood - I thought that there was feedback between the measured amount and the current sent to the heater, not just signal conditioning on the measured amount. Thanks.

More importantly, I'm glad to hear you're feeling good about how it's working as you continue to spiff it up.

I am anything but an expert on this, h_ll, I very honestly have no idea what the bosch data sheet (bosch writes the worst sheets I've ever seen) is trying to tell me in most cases.  The HW is built so I should just be looking at the application section...but there isn;t one so not being an electronic person who "sees" that the chip block diagram does, I'm left kind of guessing.

So, this is all "as I understand it"  There are 2 main outputs from the chip then ECU also shows me some other info it gets over the SPI buss including a UR_cal signal.  My understanding is I heat the sensor until the UR signal matches the UR_cal signal and that means the cell resistance is correct and the output signal UA mis accurate.  Sounds simple but the UR signal seems to like to pin itself  against the upper limit so heat, heat, heat,...nothing...BAM its on the lower limit.  PID controllers just don't work well on things like this. 

Then add to it that the best I can tell 2 or the 3 sensors I have are bad and by chance the only good one was plugged into the innovate controller so  I'm loking at it's output and wondering how the heck I get that...the answer turned out I don't because the sensor I'm using is broken.  The brand new one, which is actually years old must have gotten dropped as its got an intermittent short...occasionally it works but....the other old one from the #8 cyl bank just outputs the wrong thing, it should be about a 0.5-2.5V signal and ita more like 1.2-3.6 but can be driven to the right range if I move the UR control point to about 3.3....which had me thinking I must have read the bosch sheet wrong and missed the magic offset.....nope, just a bad sensor, put the last one on and magically 2.5V cal point is correct.

This has had me tearing my hair out honestly and I've completely given up on smooth, continuous, elegant solutions.  Right now the heater just comes on for 5 seconds, but on a limiter function as you suggested so on above x and off  and below/y just to put some heat in it as it acts very different cold than warm.  Then it goes to PID, but if it gets above 4.8V the limiter kicks back on and drives down off the upper limit what the PID can see it moving.  Then I have 2d PID table that look at both UR volts and the slope to decide how aggressive to act.  This is for sure a brute force approach.....I just got it going like this last last not but it seemed to be working better than previous attempts but needs a bit more tuning.

Contrast this with a guy making his own controller from scratch over the the rusEFI forum who said on his a simple PID is all that was needed.....that is what I was expecting then all these issues, including the ECU company owner saying "why not just buy an external controller and use your time on more important things?" had me believing all the issues I'm seeing are normal and why most just use an external dedicated controller......then rusEFI guy says its simple and plans to sell the units that cost about $8-$10 to make but I don't know the final price point.....that just doesn't match what I'm seeing.

Rant over...the battle continues.....hoping the blcok comes back soon so I do have better things to do and can justify setting this aside.....

woodbutcher
woodbutcher
12/2/20 10:27 a.m.

smiley Great thread.Great thinking out of the box.Thanks for posting.

Good luck.Have fun.Be safe.

 

mke
mke HalfDork
12/2/20 10:29 a.m.

hmmmm.....the 3 sensors I have (well 1 now) are all 7057, short for 0258007057 which has the standard bosch 4.9 lsu connector but is in fact a 4.2 sensor.  reading the LC-2 instructions it says it works with both.....mine came with the 4.2 sensors.  This knowledge may change things a bit.  

mke
mke HalfDork
12/2/20 1:17 p.m.

A pair of actual 4.9 sensors will be here Friday, hopefully that sorts out these control issues.  I've been kind of hacking stuff into and out of the model....maybe use the time to pretty it up a bit so I'm restarting the work in an orderly state. 

Syscrush
Syscrush New Reader
12/3/20 10:07 a.m.
mke said:

I've been kind of hacking stuff into and out of the model....maybe use the time to pretty it up a bit so I'm restarting the work in an orderly state. 

Years ago some colleagues and I joked that source repos would be doing us a favor if they just nuked files that had frequent small updates so they'd have to be re-written fresh.

mke
mke HalfDork
12/3/20 2:05 p.m.

In reply to Syscrush :

Same is true with mechanical stuff, you want to find the issues just look at the rev numbers and go stay to the parts with lots of changes. 

On the other side my bother-in-law used to work at a place that made a lot of stuff for the rail industry.  About 15 years ago he got an order for part where the last rev was in 1906, it was hand drawn, never converted, just make a copy and order more for a 100 year.

Amazon just dropped off my sensors so I know what I'm going to do tonight!  They've got a cool new tracking, it pops up a map that shows where the truck is and how many stops before me.....looked out the window and sure enough, there was the truck on the other side of the neighborhood.  My how times have changed

mke
mke HalfDork
12/3/20 5:34 p.m.

sensors arrived...but it turns out that just because the connectors LOOK the same doesn't mean they are the same.  The 4.9 is the same shape but much smaller.  There will be a slight delay on this part of the project while I again wait for parts. 

mke
mke HalfDork
12/9/20 8:02 p.m.

Ok then.  Now that I have an  actual 4.9 sensor connected instead of a 4.2 I thought was a 4.9 the thing is dead stable with a simple, normal PID setup...DOH!

Speaking of dead, the one 4.2 I thought was working is really just spitting out gibberish....it moves up and down in response to a change in O2 but not in the way it should.  I'm not certain the cal is right on the new sensor but I am certain then lawn mower isn't >1.3 and that and that I can get a bottle filled with pretty much straight propane.  The new sensor put the mower at 0.8 which seems rich but....maybe.  Just need to wire up the 2nd sensor and merge this new stuff into the main model and O2 sensors are done.

That means I really had no idea at all which mixture I was running when last the engine ran.....

mke
mke HalfDork
12/14/20 8:25 a.m.

Fun with O2 sensors continues...but I think is mostly sorted.  I'm pretty sure all the sensors that were installed when this engine was last running are trashed.  Also one of my new genuine imitation bosch 4.9 sensors died....mostly died, its intermittent.  I also learned to keep the charger on the battery....system voltage changes heater output it turns out.

At the moment I have a NB and a 4.9 connected and have plenty usable.  The ECU has 100k 5v pullups on all AN pins which is not right to read a NBO2 sensor, but it seems work and with a new sensors is basically 0.2-1.2V output, so a 0.2V offset putting 1.0 lambda at 0.650V.  

Then, and I'm trying to confirm this, the cj125 chip on the board appears setup for a 4.2 sensor....which I have little interest in using, but seem ok with the 4.9 but the calibration is off.  I get an output of 0.04-4.03V  full right (shrouded in propane) to full lean (open air) and  with the NB sensor telling me where 1.0 lambda is at about 1.1V (I may see if I can get a scope on the 4.9 sensor nerst pin just to be sure).  I plotted the cal info from the bosch data sheet with V on the x axis and its linear (r^2>.99) though the rich region which is what I care about so I adjusted the calibration table to put 1.1V at 1.0 lambda and rescaled to put my .04V at there min output of .192V, then same on the other side of 1.0 knowing that isn't linear so anything leaner than about 1.5L is not actually calibrated but also not important to me. 

All that done I'm pretty confident the mower I've been using as my sanity check is 1.1 lambda and covering the carb with my hand produces the expected responses.  The screen shots are the more "controlled" mt dew bottle "test station", going propane ot air and back where I can make the change happen slowly.

Now I need my block back, called the machine shop this morning, waiting for a call back.

 

Syscrush
Syscrush New Reader
12/14/20 10:39 a.m.

I haven't done any driver development, but I'd have to imagine that correctly working sensors makes a big difference. laugh

Good work getting it sorted out.

mke
mke HalfDork
12/14/20 11:04 a.m.

Without O2 sensors its plugs in, plugs out, plugs in, plugs out all day and trying to take good notes....really slow going.  I remember having a NBO2 sensor for the first time....just wonderful compared to nothing.....but nothing compared to WB data.  My how things have changed.

The other bit is being able to look back at logs....there is always so muh going on with a brand new everything custom setup it can be more than a little overwhelmingly, at least for me ans that makes good data logs a godsend. One thing I love about this ecu is I don't need to pick what gets logged.....all data that I have displayed on any of the tunings/setup screens is logged so I can go back after the fact, have a drink and look over everything.  I can't count the number of times in the past  I realized something wasn't quite right but I only had...30 data channels? and didn't have what I needed because I'd been working on something else....that's no longer an issue with modern ecus.  This is also why I'm spending so much time now getting all the sensors right and the dash gauges matching the logged data, it will make everything so much faster later.

I probably shouldn't be, but THIS time around I'm feeling pretty confident I'm going to put this all back together and its just going to work.....pretty confident because it just seems like anything that could possibly go wrong already has blush

mke
mke HalfDork
12/16/20 5:56 a.m.

The machine shop foreman's been out a couple days, they should have an update for me today.  I would really like to put the engine back together......a sink arrived the other day so not sure how much longer I can stay out of the basement.

The replacement WBO2 sensor came so both are working now.  I couldn't help but play with the settings a little more so now they are working more better. I just need to spend a little time looking at the full model to be sure everything that was looking for the external innovate sensors is now looking at internal sensors.  Also remembered I never calibrated the water temp gauge which is now controlled by the ECU so I took care of that.

I think this now just leave the exhaust valve controller to setup but I'm not completely sure where I want to mount it.  I think I'll run wires long enough to reach any reasonable location so I can get on with zip tying the harness and mounting the ecu, relay and fuss blocks.  Then the center console is in place but not actually installed and few other things to deal with up there....I guess at least a week checking and fixing everything cabin related....all important I guess but irrelevant when there is no engine so hopefully there is good news from the shop soon, the guy I talked to when I called sounded completely overwhelmed  but hopefully I'm near the front of the line at this point.

1 ... 58 59 60 61 62 ... 64
Our Preferred Partners
XZJbuj2RGw2KnJe3drgTuXvWqxRbDsk5gEYiCqH32VcYesl1Jk2HsdW4TmnKWMIE