DIYEFI.org Forum

Greetings,

I'm from Turkey. We are trying build custom Bosch LSU 4.x sensor controller which sends data to pc via bluetooth / wifi. Our aim build custom carb tuner, our electronic knlowledge was very limited.

My Questions are

Can arduino / maple or similar class devices has enough cpu power to drive bosch LSU sensor. (we want to use them because of their availability and manufacturing quality)

I see lots of controller designs which have lot of small electonic components other than microcontrollers. Can we emulate their functions via programming ?

My education was junior high scool and I had some self taught english. Can you suggest books / sites / documents to improve myself in electronics are. I'm cabaple self learner.

Best Regards

Delifisek wrote: Can arduino / maple or similar class devices has enough cpu power to drive bosch LSU sensor. (we want to use them because of their availability and manufacturing quality)

Yes, an arduino does have enough processing power if you want to make a wideband sensor

Delifisek wrote: I see lots of controller designs which have lot of small electonic components other than microcontrollers. Can we emulate their functions via programming ?

No you can not, you must use atleast an opamp with some resistors to be able to measure current. You will also need some power device (like MOSFET) to control the lambda heater

The tech edge site has a nice basic explanation on Wideband lambda sensors and the required circuits for it:
http://www.wbo2.com/lsu/lsuworks.htm

If you want to know more about electronics I suggest the following site as it is quite comprehensive in basics
http://www.allaboutcircuits.com/

The wbo2.com site has good information on their DIY unit about analog parts of the design, but it is overly complex IMHO.

There is also this http://www.waltech.com/site/index.php/o ... e-projects

Take a look at the schematics, http://www.waltech.com/wideband-files/S ... 02-1.1.png
It is pretty simple, I would get rid of the pump current measurement part (that is pump current amplifier) and the opamp buffer for vref, just run the ends of the pump current sense resistors directly to the ADC ports. Most arduino boards use the atmega88, I think it maxes out at 10 bit resolution, but you can oversample+decimate or run a digital filter on the samples to increase resolution, the ADC is good for 10-15k sps, so you have plenty of extra samples to sacrifice to the DSP gods to get more resolution.

Some of the higher end arduinos run better AVRs, they probably have differential gains on the ADC pins which you can use to amply the pump current voltage, combo that with some DSP and you should be able to get very good resolution.

Once you build your first controller, you will know what you need to build a competitive controller to go up against the competition.

I would check out www.14point7.com - good cheap modules on there already, I don't see the sense in making one, as there are several good units out there that are reasonably priced and do the job really well.

Dan wrote:I would check out http://www.14point7.com - good cheap modules on there already, I don't see the sense in making one, as there are several good units out there that are reasonably priced and do the job really well.

I totally agree, however I'd like to point out that that's toalan's site ;-) And this is what a good bastard he is, telling people who want to get into his game how to do it! Much respect to Alan, as always! :-)

Many thanks for replys.

We know alans web site. We order that oem kit at Jul 18.... and still waiting. I believe there where problem in customs or postal.

So we decite to do it by our selfs. We choose arduino (actualy olimex's pinguino or similar more powerful units) for better production quality.

Custom production was not feasable, our need was so small, and small quantity does not attract the builders, also in our country electronic manufacturing quality was not so good, plus overseas production generates tons of customs problem which we can't deal with current capital power.

In my web search I found many documantation about LSU driver circutis However, problem is my electonic knowladge was just nothing. I try to read many times of that and still do not understand fully, the schematics of the electronic circuits still loking like black magic.

So I have to start somewhere. When I was searching the net I found lots of debate about LSU controllers in here and I feel this was right place to ask these questions.

Actually I'm looking for startup kit which I can work and debug that LSU sensor

One olimex kit has

STM32F103RBT6
https://www.olimex.com/dev/olimexino-stm32.html

And another was

PIC32MX440F256H 80 Mhz

https://www.olimex.com/dev/pic32-pinguino.html

Please suggest one of these or different unit for the LSU controller project.

My Humble Regards

Can you send me an email, I will get it sorted out and get what you ordered in your hands.

The STM32 and the PIC32 are very very powerful devices, they offer somewhere in the range of 50-100 MIPS of processing power.

The Arduino which uses an AVR has about 16 MIPS of processing power.

If you are designing a lambda controller using typical approach; DAC to drive pump cell, pump Current->amplifier->ADC, AC coupled resistance sensing signal on nermest cell ->amplifier->ADC, PID control of heater, PID control of pump cell, 90% of the controllers take this approach. If you are using the typical approach then I am guessing an 8 bit micro with 5 MIPS is just about enough horsepower.

Most lambda controllers are built with the AVR, it has all the processing power to do the job and is really easy and cheap to develop for. With AVR you can either use arduino or use AVR studio which integrates the free GCC compiler, there is also a fantastic user forum at www.avrfreaks.net. There is another critical reason why AVR is suitable, but I am not telling.

If you already know how the build a lambda controller and the fundamentals then you can use the power of the STM32 or PIC32 to do some exotic methods of controlling the sensor. I am using a 32 bit ARM cortex, 50 MIPS, in my current generation lambda controllers, I have already built lambda controllers using AVR and PSOC, so I decided to build something exotic this time around.

I did my first controller using AVR, it took about 6 months of off-again/on-again work. If you take my advice and use AVR, once you are finished you will realize that AVR is not the best choice for making a Lambda controller, but it is the best choice for making your first Lambda controller.

toalan wrote:There is another critical reason why AVR is suitable, but I am not telling.

ROFL! :-)

Fred wrote:

toalan wrote:There is another critical reason why AVR is suitable, but I am not telling.

ROFL!

Okay okay I will tell.

You want 5v operation, the STM32 and the PIC32 can only do 3.3v or less operation.

In a 5v system, the virtual ground is biased at 2.5v, so you have 2.5-5v to drive current into the pump cell, and you have 0-2.5v to sink current from the pump cell.

In a 3.3v system, you will bias the virtual ground at 1.75v and only have 1.75v in either direction to sink and source current to the pump cell. This will mean your system is significantly slower.

You can rig up a dual voltage system, and have interface circuits and whatnot so the micro is @ 3.3v and the rest of the system is at 5v or higher.

I know of one brand that runs their micro @ 3.3v (might be 5v) and runs the the part that drives the pump cell at ~8v, so they get very good current drive capability and good response times, but that brings with it increased costs and decreased accuracy.

I wasn't asking you to tell. I was just expressing the fact that I nearly fell off of my chair when I read it. :-)

Which brand does the 8V thing? PM/email/tell me to get lost, if it's a trade secret. You've shared enough of those already! :-p

Fred.