PDA

View Full Version : Electrical engineers anyone?



Alvin
October 30th, 2006, 02:43 PM
I'm looking for some help..

We have a 1500 CFM throttle body that takes very little pedal to start getting the air moving.. Consequently the car drives around barely reading more than 1% TPS.

I'm looking to make a combination of a wheatstone bridge and a low gain op-amp so that we can boost the resolution down low.

Being a ME i've always sucked at selecting parts. I have a good idea what i want but no confidence in picking the right op-amp for the job and such.

Anyone care to help?

joecar
October 30th, 2006, 02:58 PM
I've used the LM324, LM741, LM748, LM709...

Radio Shack has various cheap and available op amps...
http://froogle.google.com/froogle?q=LM741+radio+shack&hl=en&lr=&sa=X&oi=froogle&ct=title
http://www.radioshack.com/product/index.jsp?productId=2062592
http://www.radioshack.com/product/index.jsp?productId=2062594

Alvin
October 30th, 2006, 03:32 PM
Yeah but the deal is I don't know which to use... Woudl I need to run this thru a FET.. I wouldn't imagine so but the uncertainty kills me.

joecar
October 31st, 2006, 05:51 AM
You don't need a FET on the input since op amps have high impedance inputs (FET op amps 10^12 Ω, BJT op amps 10^7 Ω);

You only need a transistor on the output if you're driving a load;
you don't need one if you're driving an input (such as TPS input to PCM) which is typically high impedance.

i.e. you need an output transistor(FET, BJT, Darlington) only if there will be output current flow greater than several 10's of mA.

I would use the 741 or 741-like...
it's proven, it's easy to use (it's hard to make it oscillate by accident) and it's cheap, and comes in dual and quad packages if you need more than one, and has alot of appnote/reference documentation.

joecar
November 1st, 2006, 05:26 AM
For anyone reading along, the LM124/LM324 op amp's datasheet has pretty good "application hints":

http://www.national.com/ds/LM/LM124.pdf

Ira
November 3rd, 2006, 06:50 AM
I'm looking to make a combination of a wheatstone bridge and a low gain op-amp so that we can boost the resolution down low.

Let's start by saying that I've pondered this problem more than I should care to admit and I've never come up with an answer which means something between I'm completely clueless and it's not possible! I'd prefer to discover I'm clueless than it's not possible, but I'm open to either. Times like this I wish my dad was still alive, he was a kick-ass analog designer.

I don't see what a wheatstone bridge and an op-amp is going to do. What you need is a device whose output looks like something like:

out = 5 * sin( in / 5 )

Don't know if that's the actual formula you need, but you need an output that rises faster that the input at low throttle openings and slower than the output at high throttle openings. I personally have no idea how you would make a DC amp that would fit an equation like that but if someone does, I'd love to see an example.

Ira

joecar
November 3rd, 2006, 08:02 AM
You need a log amplifier which can be made using a diode and an opamp (and a few other resistors/capacitors).

joecar
November 3rd, 2006, 09:24 AM
Here what Wikipeadia briefly says: http://en.wikipedia.org/wiki/Operational_amplifier_applications#Logarithmic_out put

Almost all books on op amps cover this also.

Edit: further references to log amp:
http://www.answers.com/topic/operational-amplifier-applications

Ira
November 3rd, 2006, 09:37 AM
Interesting idea. I've spent some time searching the web and analog.com to learn about log amps and it's hard for me to see how one might build one of these with the adjustability he will need until the proper curves are determined. It's also not clear to me that the log slope devices aren't way too steep.

joecar
November 3rd, 2006, 11:25 AM
TOL (thinking out loud)...

couldn't any/all of the following be scaled by appropriate placement of resistor(s):
input
output
feedback (path with diode)

Ira
November 3rd, 2006, 05:27 PM
couldn't any/all of the following be scaled by appropriate placement of resistor(s):

Well, the problem is you need to start with 0=0 and end with 5V=5V. Ratios are easy, curves are a wee bit harder. What you really want is the ability to draw bezier curve between the 2 endpoints with the endpoints staying fixed. Something like what you can do in Adobe Acrobat or most any other vector drawing program.

Ira

wait4me
November 7th, 2006, 05:42 AM
Why dont you just ask Ross to see if there is a Table in the pcm that you can just set for voltages to mean a certain TPS% That would be the simplest solution.

joecar
November 7th, 2006, 06:00 AM
Why dont you just ask Ross to see if there is a Table in the pcm that you can just set for voltages to mean a certain TPS% That would be the simplest solution.That would be the cleanest solution. :cheers:

wait4me
November 7th, 2006, 06:09 AM
I have had that one mapped out for a while. Ill Email Ross with some addys so he can put it into the software for everyone.