MPGuino release one workspace

[dcb]

Originally Posted by ttoyoda

I plan to use optoisolators.

That is a fine plan The supermid uses a couple transistors FYI. I started with the couple transistors approach but must have messed it up (it is more complicated to assemble than just a zener), and swapped in the zener and resistor, and it worked and so called it good.

After researching it a bit more (hint: we could use more researchers), I did see that some of the arduino crowd add a 12v regulator prior to powering the duino, i.e. http://www.arduino.cc/cgi-bin/yabb2/...num=1191452682 , the claim is that bits were randomly inverting without it?

And they have a document describing vehicle interfacing, with both an opto-isolator and a transistor based (similiar to the supermid).
http://www.arduino.cc/playground/upl..._4_arduino.pdf


But FYI, there are a number of already built models with the zeners (since the beginning of may), and I need to know for certain that we have a problem and exactly what the problem is before I ask folks to disassemble their boards and add stuff to them. It also tends to result in more complicated than necessary circuits if each addition goes unchallenged. The zeners may very well be inadequate protection, but I need to see it before complicating the assembly and price and affecting the installed base. It may be that we add opto-isolators and a 12v regulator ahead of the arduino before we call it version 1.0, hopefully the o-scope will make that decision a no-brainer.

re: production, I'm not really in this project for the money, really it is just to learn (plenty of that going on) and have some fun It's just a hobby at this point, though I'd be a fool to turn down some small kickbacks on a reasonably priced kit or pre-assembled unit. But not even thinking about manufacturing these at the present, as that would take a lot of the fun out of it. I *might* have it together enough to put a PCB and small pile of parts in a baggie, but it isn't really my thing.

I think about the work other folks are doing though, a lot. The project doesn't exactly have enough funds to call it a budget at the present. But I do keep the other contributors in mind, folks willing to buy some hardware and see if it works, and run tests, and do research, and keep the project towards the goal. I would like some ideas about how to identify and compensate, if even in token only, the folks that are helping to make this happen.

[dcb]

Originally Posted by monroe74

You probably realize that you can also look at the injector waveform if you send it to your PC through the audio input.

But of course you can , and I would reccomend havin at least that much in ones electronic toolkit. But it isn't nearly as accurate in terms of time resolution and actual peak voltage identification. Automatic Gain Control will mess with the signal values.

[MilesPerTank]

Originally Posted by dcb

After researching it a bit more (hint: we could use more researchers), I did see that some of the arduino crowd add a 12v regulator prior to powering the duino, i.e. http://www.arduino.cc/cgi-bin/yabb2/...num=1191452682 , the claim is that bits were randomly inverting without it?

I run a 6.5V-40V to 5V DC/DC converter on my front end now. I also put TVS's on the input to protect against surges. I installed the reverse polarity diode on the input of the DC/DC with surge clamp and gigantic bulk capacitance in an attempt to handle cranking brownouts. I think if a surge doesnt kill the 7805, your problems are going to be due to a brownout. Mine runs great but I fully admit it is complete overkill.

I wouldnt use an optoisolator. There are multiple other ways to level clamp a voltage that dont involve opto's. In my experience an optoisolator is used to establish / maintain isolated grounds. We're obviously sharing the same ground. They also arnt the cheapest things in the world.

I still maintain a simple diode bridge will be adequate. And in DCB's defense the zeners although low on the knee and not fully in breakdown are working on my and many others cars.

Hopefully with the scope you will be able to track down the IC count problem. I know the base pulse of the backlight PWM is running at 500Hz if you want to wire off of that for counting.

Unfortunatly I took the fun car out of town this weekend so I didnt get to do any decent calibration of the MPGuino.

[ttoyoda]

I wouldnt use an optoisolator. There are multiple other ways to level clamp a voltage that dont involve opto's. In my experience an optoisolator is used to establish / maintain isolated grounds. We're obviously sharing the same ground. They also arnt the cheapest things in the world.

I have a lot of optos around, somewhere, if I ever find them. Here is a new one for $1.89,
http://www.alliedelec.com/Search/Pro...2&DESC=NTE3041
so it IS a lot more than zeners. Since I was assuming dcb was making this design for production, that added cost WOULD make a big difference, so I figured that was why it had not been chosen.

I agree with you there are other ways to do it, I don't have your knowledge about all the options. My training is in mechanical engineering and heat transfer, so I tend to go for overkill in electronics that I design. You have to admit that an opto is a simple and SURE way. Nothing is going to get thru that opto to upset the guino.
Also, I was also thinking in terms of powering the duino with a dc to dc converter I have somewhere, I would not connect any grounds to the car body under that condition.

And again, it was NOT my intention to attack or criticize dcb personally.

Laboring under the assumption that dcb was developing this as a product for sale, I was role-playing the part of a paying customer- what would that customer expect in a device that cost real money? So dcb is (rightly) annoyed because he assumes I am whining about something I am getting for free.

Meanwhile, I am figuring dcb has put this out publicly here as a product development, and that he WANTS to hear about all the things that could POSSIBLY BE PROBLEMS under any conditions (because that is a brilliant thing to do- put out your design for review by thousands of design people, many of whom HAVE shipped real consumer or industrial product) before he has 2000 of these made in China, has sold them all, and has put in an order for 5000 more units.

So when dcb says "$3.00 chip" I am concerned he is not taking the (potential) issue seriously and his PRODUCT is in jeapordy. I say (role-playing the part of a future, hypothetical, paying CUSTOMER) "I have built this in and it is hard for me to take it out of my car and find the cables and learn how to re-program it, I don't have the experience." Because I feel that for a PRODUCT, the customer cannot be expected to change out chips.

Anyway, sorry for the unintentional increase of peoples blood pressures.

[monroe74]

Originally Posted by dcb

But it isn't nearly as accurate in terms of time resolution and actual peak voltage identification. Automatic Gain Control will mess with the signal values.

Thanks for the quick response. As usual, I'm learning a lot from you.

I understand that AGC could be a big problem, but I think there are ways of turning it off (depending on the HW and OS). But I hadn't given much thought to the issue of horizontal bandwidth, and I didn't realize that old analog scopes have a ton of it. Thanks for pointing that out.

I also stumbled across a very helpful article here: http://en.wikipedia.org/wiki/Oscilloscope.

[Yoshi]

Originally Posted by monroe74

Your equation is correct. And when you apply it to the first injector on the list, it does show that the proper correction factor is 0.504mS. Trouble is, the other injectors on that table seem to behave differently. When you apply your equation to the other injectors, they all have correction factors higher than that. The other correction factors range from 0.550mS to as high as 0.805mS. That latter number applies to the last injector at the bottom of the second table.

(I guess I was doing this at the same time dcb's wife was. And it's nice to see that we found the same maximum!)

And it turns out that using 0.500mS instead of 0.805mS (for that particular injector) will lead to an overall error of almost 18%. The error is large because the absolute difference between those two numbers represents a large portion of the pulse width (2.500mS).

I also looked at other injector specs (like here: http://www.racetronix.com/), and they support my finding, that the discrepancy between static and dynamic flow cannot be handled by a single, universal correction factor.

I think this is ultimately a very manageable problem, because there are various ways an end-user could determine the proper correction factor to apply for his vehicle. But I think it's probably a good idea to make this a parameter that can somehow be altered by the user.

Hello,
Thank you for your comments.

Again, my data was Toyota Prius's case and we saw very good real world results.
I'm not saying this project should use the 0.5msec delay, but my recommendation is to include some adjustment into the program.
Do something is better than do nothing.

Yoshi

[Yoshi]

Originally Posted by monroe74

I hope you don't mind if I ask you something about this. Fuel-use calibration seems like a very troublesome problem, because it seems to me that we are relying on the pump nozzle to shut off in a reliable, consistent manner. I have a feeling this can produce large errors, especially when buying a small amount of fuel. On the other hand, the errors will tend to equalize, over time.

Anyway, I wonder what technique you have been using in this regard, as far as how you gather your fuel consumption data. You might have some tips that could be valuable for others to hear.

We use total fuel usage of three tanks, approx 120 liters, to adjust the SuperMID's fuel parameter among five testers.
We could adjust the parameter within 1% error using the 0.5msec fix amount of delay.

Yoshi

[monroe74]

Originally Posted by Yoshi

I'm not saying this project should use the 0.5msec delay, but my recommendation is to include some adjustment into the program.

Good point. I agree. I'm glad you convinced me I was wrong.

We use total fuel usage of three tanks, approx 120 liters, to adjust the SuperMID's fuel parameter among five testers

This is helpful to know. Thanks for explaining.

[timgray]

I jsut read all posts about his and I want to say, Good job and great effort! I'll be buying the parts to join in on the development of this great device. I do have a couple of hints though, and it comes from my EE education and background. the inputs from the VSS and injectors needs to be built up. opto-isolators would be a good idea as they will protect the processor and they can take way more punishment plus will help to shape the signal to be more of a square wave.

just my 3.25 US cents... adjusted for inflation.

NOTE: the reccomended Freeduino clone is no longer available. they now sell the following....

http://www.nkcelectronics.com/freedu...ompatib20.html

is that one 100% compatable with the older one the prototype and development is based on or will I have to tweak the code for it.

again, THANKS!

[dcb]

Anything running an atmega168 @ 16mhz is code compatable. I am currently favoring the iduino for $18 with onboard USB, but the diagrams are for the good old arduino/freeduino platform.

re inputs, can you explain why you believe the Zeners (which is a common form of circuit protection) cannot protect in this instance? Have you looked at the signal reaching the chip in a vehicle with an oscope? Have you compared the chip specs and the zener specs and applied the anticipated vehicular signals? What is that statement based on again?

The problem I have with optos is that they do add to the cost/complexity, and a zener is incredibly easy to assemble for the Do-It-Yourselfer. I'd like these to work the first time people put them together. I started with a transistor isolator myself, but managed to mess it up somehow, so I would imagine that every complication we add to the circuit will mean more folks who assemble their own will also encounter complications and frustration. This is a DIY that is trying to be as simple as possible for the largest diy audience as possible (no code changes just cut and paste the code, minimal circuitry and expense). So I request actual proof behind the statements requesting a change.