stuckatcuse

I've got a little something in the works for my real-world signal processing class. Unfortunately, it's grown to incorporate much "feature creep" and uses all SMT and non-hobbyist micros. It therefore, could no longer be considered a DIY project.

When it comes to programming uccontrollers it's hard to find someone willing to do all that work alone for free. Given that we (we meaning EEs that want wideband o2's) represent a relatively small subset of the automotive community, manpower is limited. Now limit those engineers whom are interested to those also having time to work on the project. Now find several of the remaining engineers that all work on the same platform. It's difficult for the few of us EE's to work together because each engineer prefers his/her own equipment. For example, I prefer Freescale products, while others prefer Atmel products. The system architecture and asm are completely different; e.g., “I don't feel like learning another uc architecture for this one project just because another member wants to use the xx micro.”

The only way a true DIY project will work, is if someone takes the lead and specifies the platform and features (with other's input of course) and assigns design tasks (including reverse engineering any necessary components and interpreting product data sheets). A sort of project facilitator, designated to keep things running, but not do all the work.

If one person does it all, it's not a DIY project. It's a gift of intellectual property to the rest of the world, and it's unreasonable for one to expect a single engineer to altruistically give that property away.

Craig Moates

I'm ready to go on this. I figured I'd defer to Bill & others on the hardware layout, at least initially. Will be glad to help on some hardware revision/simplification later on as well. Like Bill, I've got my hands full as well.

Isn't feature creep a trip?

I'm still suggesting that the micro be a Mega8535. It's got all that is needed and is cheap. Can still get it in DIP form, although the SMD version really isn't that difficult.

In terms of DIY, what do you think about making pre-fabbed pieces available? That'd be like a PCB with the micro already mounted up and flashed? It could use bootloader code for field updates. Anyways, just a thought. I have been running my AVR/USB-based stuff through an 'initial' shakedown that isn't too bad on the partial assemblies, and it at least ensures that everything is kosher before assembling the rest of the units.

I think you're right though. I'm a bit disappointed as well. I'd half expected a battery of maybe 5-10 folks with some time on their hands to step to the plate and take off with it. I was just trying to stir up support for such a thing. Then it ended up back in me and Bill's lap. Not sore about that, just a full plate of other stuff. I've got like 5 different prototypes of stuff on my workbench right now (OBD2 datalogger, 16-bit emulator, wireless wideband data transmission, list goes on). There's some room for WB, but space is tight.

Alright, the more I type, the more I convince myself it wouldn't be that bad. Maybe I'll just put something together and at least make it available to folks that don't want to spend a fortune on such a thing. Should be able to fit it in a box the size of the BURN1 programmer I think. Give me a month, and I should have something. I know Bill is pretty busy, but I'm going to lean on him and others for consultation with respect to the circuit design.

My wife is gonna kill me, she's trying to cure me of lying to folks about stuff like this. Ah well, better to be hyper-extended than holed up waiting.

MrBill

I just need to be "motivated" (ie bug the $&!* outta me about it, LOL)

I've also got this Crossfire mofication testing going on...never a dull moment!

jamesbob02

Craig, you're awesome.

I hope you didn't take what I said as an indictment; I wasn't disappointed in you, just that the project hadn't seemed to have gone anywhere. I know, just like Bill, that you have a ton on your plate, so when you're the only ones willing (or especially able) to advance the project, its more than understandable when it hasn't gone very far. If I had the skills, I would be more than willing to help, but alas, all I've had was a required electrical science course.

I'll be the first to volunteer as a field tester whenever a prototype is available.

Craig Moates

Heck, I've got a 680hp motor sitting in my garage and an empty engine bay. What I don't have is time to put it in.

funstick

plane ticket and a case of beer and ill get it in there.

stuckatcuse

Let me know if you need another hand. I've been scouring the LSU datasheets.

KenV

Holy crap. I wish I had your issues.

You rule man. When I get over being a newlywed I hope to start rebuilding the Formula, and this type of pioneering will make the engine buildup more satisfying. Thanks for the effort.

Teeleton

How difficult is it to flash one of these microcontrollers? Can I just use my eprom burner?

Teeleton

Craig Moates

I'll probably set it up with an initial bootloader code so you can reflash it via the serial/USB port.

JPrevost

junkcltr

iTrader: (1)

I would love to help on this project, but this is a tough one. The problem is that every good idea that comes across this board is grabbed by someone to turned it into a cash project for themselves. It is rather unfortunate that people start out good willing to learn and share and then find out they can make a buck of things. Once they see some green then they are more interested in that than the reason they got involved in the first place.
I already see that there are thoughts (questions) of whether a partially built kit should be offered. It should be a completely DIY-kit in my opinion. Yes, the MCU would probably need to be programmed first.

OK, enough rambling. I could help with some of the electrical design/simulation and Microcontroller coding. The MUC chosen doesn't matter to me. I think it would be wise to code in C and compile with GCC and only write assembly code (assembled with GCC AS) where necessary. More people understand C than assembly so this would allow more people to read/learn the code.
I realize some people still love "all assmbler", but eventually they will learn that for much of what they are trying to accomplish it is a waste of time and coding in C is more efficient. Yes, there are a very small amount of things that will need to be coded in assembly....these can be called by the main C code.


For those interested in a DIY-WB LSU kit. Check out the Megasquirt LSU controller in the works. There is good info there on how to control the LSU and they seem like an honest group of people.

J

91L98Z28

neat project.

my time availability is not so great but once upon a time I did know 8051 microcontroller and I can spit out a few opcodes now and then in the psuedo 6811 that GM uses in my '730...

anyhow, i don't think i'd be much help in laying initial code design, but i could probably help in chasing bugs/debugging things, fixing minor problems, source commenting, maybe writing specific purpose subroutines (I imagine we'll need one to calculate the "on" clock cycles vs. "off" clock cycles to do the PWM stuff to heat the heater, will the on-off crossings be interrupt driven? are the new modern micros better than that now?

as long as whomever is in charge would be willing to put up with slowness and learning as I go, i volunteer to pitch in where i can, just don't know if i have the knowledge of current microcontrollers to help much.

Mangus

Have you guys seen Innovate's latest product - the LC-1:

http://www.innovatemotorsports.com/resources/news7.php

$199 with sensor. Simple, cheap, effective. It would be pretty tough to beat that.

stuckatcuse

MAN THAT'S CHEAP. I agree... it will be tough to beat that! I guess they're trying to make all their money with the gauge!

junkcltr

iTrader: (1)

And not to mention that it is "indudstry-leading patented technology". Wow!

A DIY kit is about learning something just as much as it is about the cost. From a cost only perspective, it would be still cheaper to do a DIY version.

Mangus

Cool. I, too, think it'd be worth your time to drive the effort! The project needs a captain. Take the helm!

junkcltr

iTrader: (1)

I might think about taking it on.

I am busy right now doing some code clean-up on a "DIY Emulator" kit. It is all through-hole stuff, doesn't through any engine lights, blah, blah. Lots of features and something someone with average soldering skills could build. Stay tuned. I will be looking for a few beta testers soon.

Anyway, once I get this finished up I might be able to start on the WB thing. I think it is a great idea. Prelim ideas include using an atmel atmega 8 or mega8315, two layer PCB, some voltage regs, etc.

Mangus

Very sweet. Looking forward to your goods! Make sure to save some time for DIY-WB! The townsfolk are getting restless.

DSmith

Is this wb on the market now?

Tony B

Has development of this continued, or has it petered out?

kevindtimm

Can this be 'rebooted'? Did this completely quit or get moved somewhere else? Here are the questions I have if anyone is still involved.

What I'm trying to find is more detail on programming the LSU4.2 sensor. It seems that it's a two step process:

1) get voltage from the sensor
2) send current to the nernst cell

What I can't find enough detail about is how the two interconnect. I know that the diy-wb.org web page had a static system built to do this, but I want to do it with (for example) an 8051.

Also, I see talk about the CJ110/CJ120/CJ125 IC's from bosch, but can't find hardly any info about them.

Craig Moates

Maybe this can help?
http://www.megasquirt.info/PWC/

kevindtimm

That's VERY useful. On the page 'How the sensor works' about 60% of the way down, it says:
The question this begs is, What's the purpose of the pump? I see a little later that they talk about the nernst cell, but I'm still puzzled. Can anyone equate this to how to measure each item to get the true 'linear' AFR?

RBob

iTrader: (1)

Wellll, how much do you want to know?

For now ignore the output that the PWC produces. That value is a created value, not a sensor value.

I'm not sure what you mean by the diy-wb having a static system. The sensor controller is an analog PID loop. Looking at the schematic for the diy-wb you will see 5 wires going to the sensor (ignoring the 2 wires for the calibration resistor).

Two of the wires are for the heater. The heater power needs to be varied in order to maintain the proper sensor temperature. In the NTK sensor this is accomplished via a large temperature coeficient in the resistance of the heater. The Bosch sensor needs to have the impedence of one (or both) of the cells measured. The power up cycle also needs to be controlled. This is required to not crack the sensor by heating it up too fast.

The other three wires are for the sensor cells: one is a pump cell, the other is a sense cell, then ground (a virtual ground). Notice the sense cell output is fed into the front of the PID loop. This forms the feedback from the sensor. This value is compared to the voltage reference to form an error term.

The output of the PID loop pumps the sensor pump cell. This pump can be in the positive or negative direction. The magnitude of the pump along with the direction is dependent upon the forementioned error term.

The idea is to maintain the sensor feedback value at 450mV. This is done by changing the pump current. This is the function of the PID loop (in reality it is better to have the reference voltage a little lower, say 435mV - 440mV, keeps the loop more stable).

Now you ask, how do I get the AFR outa' this thing?

Very easy: it is directly relative to the direction and magnitude of the pump current. Convert that to a voltage and you have the AFR. Invert it to FAR and it is a straight line (linear).

On the diy-wb schematic, this is done by measuring the voltage drop across Rsense. The op-amp is setup as a differential amp. Corrected via the calibration resistor and referenced to the 2.5V offset and output.

To start off I would recommend that the diy-wb schematics be studied until you understand the operation. There are other areas of operation that I didn't mention. But are apparent once it is understood.

RBob.

kevindtimm

Yeah, I misread that. I was thinking that the sensor output was linear.

Yeah, that's the funny part. I'm not an EE, only a Software Engineer. I understand the basics of electronics, but not enough to understand those schematics.

I think this may be more than I can do. I get the uCtrl, I get the DAC and ADC, but I'm major puzzling about how the sensor output and the pump current work together.

The techedge stuff is the nearest to understandable, but (of course) they don't publish their microcode and so I'm having fun with how each of those pieces works together.

RBob

iTrader: (1)

One thing to remember about techedge is that it is a direct rip-off of the diy-wb. That is how they got started. Not a good thing to be mentioning that name in these parts.

Using a uproc you still need to implement a PID loop. Maybe look into some good texts on what they are, how to set them up, and how to tune them. For the Bosch sensor you wil need two such loops: one for the heater and one for the UEGO control.

It isn't all that difficult, just need to do a little reseach, re-read what I previously posted, and experiment.

RBob.

kevindtimm

yeah, I got the techedge/diy-wb tension (here and many other places). Unfortunately, the diy-wb site is gone (though available from the wayback machine) and so I get more info from techedge.

I found a lot of input about the usage of the op-amp and I'll be doing more with experimentation as well as learni ng the PID stuff (also, lots of info on the web).

Thanks, expect to hear more from me

kevindtimm

Disregard the following (I've gotten a better understanding since I wrote it, a lot of it having to do with the penultimate paragraph/sentence).

My simplistic understanding (of the diagram of) goes like this:

Stoich is 450mv
The amount of current (negative or positive) used to bring the sensor back to stoich tells me what the real AFR is.

So:

If (Vs == 450) do nothing
else If (Vs > 450) measure negative current
else if (Vs < 450) measure positive current

That's how it looks in the analog PID above anyway.

My question now is, can't I just monitor at Vout, with the value there giving me 'a value' that equates to AFR?

Put another way, if I do it the analog way, won't that give me most of the info? Or do I need to measure the time and amplitude of the correction to get a real value?

BTW, wikipedia has quite a bit of info about PID, so I'm out there reading that, but embedded.com has a 'PID without a PhD' article -- but that's a lie

RBob

iTrader: (1)

Yes, Vout is a voltage that is directly relative to the AFR. Below 2.5 v is rich, at 2.5v is stoich, above 2.5 v is lean. Free air comes in about 4.0 volts.

RBob.

kevindtimm

?

So this is as simple as it looks? And the need for a 'complicated' software loop that reads Vref and Vs and then sends voltage to the Ip circuit is overkill?

Essentially, it's just a Wb02, an op amp and a resistor (as in the diagram above)?

It seems far more complex than that. What (if anything) am I missing?

RBob

iTrader: (1)

Somewhat, no, no, missing the PID control.

The key is that the Vs signal from the sensor needs to be kept at 450 mV. The pump current is changed to maintain that voltage from the sensor. If the Vs signal from the sensor strays too far from 450 mV, then the pump current is incorrect. If that is incorrect then the AFR reading is incorrect. And PID lock has been lost.

For testing purposes I always data logged the Vs signal. That in itself is important. If the controller circuit loses lock, then the reported AFR is incorrect. It is a test of the circuit response and capability to control the sensor (even now I data log the Vs signal).

I guess at this point in time I need to ask you, what is your goal: To design your own controller and release it? Or, design your own controller and sell it?

RBob.

kevindtimm

Ah ha, that's what I thought

I'm back to where I was before (which is good).

450mv is ALWAYS the goal from Vs, but the delta between Vref and VS is not AFR. At least, not exactly, as this is a Narrowband Sensor, with it's attendant inadequacies).

The goal, when there is a delta, is to get Vs back to Vref (when there is no delta, AFR is 14.7:1).

Now, back to the PID. I get that voltage has to be applied to the Ip circuit (OK, current, but my little brain only gets voltage). I even get that there should be a table (of sorts) that determines how much voltage to apply, dependent upon the delta. What I don't get is how I can determine how much current has been absorbed in returning the system back to Vs == Vref. Is it a function of voltage * time? This is where I get completely lost.

In regards to your final question, I have a hard time seeing where I can make any money with this. However, it seems the previously mentioned purveyors had the same thought originally and look where they are now. Where can I draw the line? I'm doing this without the benefit of diy-wb (besides the fact that they did it first, I won't bear any similarity to that system -- except functionality-- nor will I use any thing that they did. I love to get help (and appreciate it all) and so love the idea of sharing what I've done. Confused yet? I know I am...

junkcltr

iTrader: (1)

It sounds like you are doing it with the benefit of the DIY-WB. Have you looked at the schematics?

So are you doing this to sell it or release it?

kevindtimm

While I have seen their schematics, all I can tell you after looking at them is that they have resistors, capacitors, diodes and some other electronic components represented therein. 30 years ago I took a year of electronics in HS. At that time I could read (from memory) the color code on a resistor, now I would have to make sure they are left in their packages so I could tell them apart. To summarize, I can't learn anything from their website (unless I wanted to try and build their kit - which I would have to do fast, as the website availability diminishes every day).

I don't know? I'm not being a PITA, I just don't know. If I invest a bunch of time (and money) and find out that it's pretty dang cool, I'd be foolish not to try to recoup. However, I'm pretty involved in other open source products (SW engineer, ya know) and so that's very interesting (and attractive) too. Somewhere in the middle seems the best. I'd like to do it, somehow release it to sourceforge and still retain the ability to build and sell them myself. Does that make sense?

91L98Z28

Our community is very small, and there are very few people with the technical expertise to put together excellent projects such as the DIY-WB.

The community you have and see here today is because others before you have given all they had to give, and then gave some more.

Your very interest in in WBO2 is driven and influenced by those before you.

Please give whatever you create back to the community so that the next generation may have something to encourage their creations and reason to give back.

We as a community have lost several projects due to profiteering off the backs of those who gave till it hurt. One such example is the ion knock detection system. This would have been an awesome gain for us, lightyears beyond the silly knock sensors we have to run. who needs an ear to the block when you'd almost have an eye in the combustion chamber? sadly, this (and other) efforts were lost because others decided to profit off of derivates of others' works. people don't like working for free to make others rich, and so the community as a whole lost out.

if you want to innovate a new idea/project and sell it for profit - great, do the research yourself and discover the principles and methods yourself, or hire researchers to do it for you, whatever. there is plenty of room for genuine innovation and new products. look at what craig has done. innovation after innovation and he's practically a one stop shop for all the cool toys you could want.

but please don't borrow public domain ideas from others who have given freely, and then profit from it. the others who have given freely will cease to do so, (they have no interest in providing their time so that someone else can profit) and then we ALL lose out in many ways:

* we lose the contributions of those who would otherwise have given freely

* when you lose the present generation of whizbang doodads, there is less inspiration/building blocks for tomorrows' whizbang doodads.

Whether you choose to go commercial and profit, or DIY community and give freely, I encourage you. Just don't involve the community in a project and derive work from others and then take it and profit.

kevindtimm

I understand your points and agree with them. I don't want help for free (although, interestingly, most forums excel in providing just that). This was the group that appeared to have the best knowledge of how these sensors worked and I was (and still am) having a hard time understanding exactly how to interface with them. I asked the questions and received a lot of very good answers for which I am very thankful. Hopefully I can figure this out and get something together. As I progress, I will keep this thread alive.

Craig Moates

Here's a good tutorial on PID controls:
http://www.engin.umich.edu/group/ctm/PID/PID.html

It's really not that complicated once you think about it. Just like most of the car control stuff, but a little more regimented and less open-ended. The key is having that feedback so you know what you're controlling.

For instance, think of controlling an analog circuit using an RC network and an OP-amp to generate an output voltage. You could excite the input to the OP-amp using a PWM output from a microcontroller, and look at a measurement of the output voltage from the OP-amp using a feedback A/D conversion back to the micro. Then just adjust your output PWM duty cycle until you reach the 'setpoint' voltage.

Keep in mind with these heated elements that they have some thermal 'inertia' and that as you start applying heat, it's easy enough to have 'overshoot' if you get too aggressive. Too conservative, and they'll take forever to heat up and won't respond well to changes. So you've gotta get it to where it'll track pretty tight.

SuperRx7

I just stumbleon these WBo2 sensor forum. And this is first time that actually I am in a forum that probably has been closed for many years by looking at some dates here.
Anyway I just got a question:
Everyone is talking about Voltage or Delta voltage to correlate it to AFR.
So 2.5V (stoichmetric) is equivalent to 14.7 AFR or if the gas used is diferente then gasoline for example a forklift that uses propane AFR would not be 14.7 How can one make it read the correct AFR for a particular combustible?
wouldn't it be just easier to read % of O2 and that would work for every engine.

RBob

iTrader: (1)

Welcome aboard TGO, the main facilitators of the DIY_WB do and used (RIP) to reside here.

The reason that the term 'AFR' is used here is that most of us use AFR when tuning. The WB sensor actually reports Lambda, with the AFR derived from that value.

With the DIY_WB controller the conversion is a table of output voltages along with the AFR that they represent.

WB controllers that use a uProc do the conversion internally. Either through look up tables or doing the math. In this case the controller can also be told what the stoich AFR is, so that it can do the conversion correctly.

RBob.