Getting into the nitty gritty

[Rx7man]

OK, so I'm going to be going for a design based on the Mega 2560, I foresee the need for the extra IO's in my project, and accompanying memory...

Honestly at this point it's more questions than answers, and not asking them now can only get me into trouble later.  Speaking on my behalf, I'm going to be doing a lot of thinking out loud and listening for the echo.

So looking at the schematics, I need to know what pins are reserved already.. it seems like tx0/rx0 is routed to the USB.. something that could be handy.

Which pins is the CAN interface running over? Seeed's is pin 9 OR 10, and 11 though 13... Is the I2C interface going to be of any use to us?

What should we integrate into that same board?  VR sensor input? Crank sensor? MAP, EGP, EGT, and TPS inputs as well?  those are all pretty critical inputs to running the VGT.. Coolant temp, Jake input, and a bunch more could go on the same board too, space and IO's permitting that aren't quite so critical.

All the controls for my project of the AFC delete could then go on a separate board, most of the inputs needed would already be on the first, leaving more room for options.. I'm thinking the second board would contain PWM drivers, Stepper drivers, and the likes and possibly be mounted stacked like a shield, but separate so as to be able to mount the power transistors on a heatsink nicely... The CAN shield with all it's inputs shouldn't need any heat sinking and stacking it is just fine

[Rx7man]

I'm really pondering what route I should go with the AFC delete side on the hardware....

Here's my options

Stepper motor controlled screw
Benefits, you know exactly where you are at any given time
Problems, if something messes up, you might be in serious limp mode, it's harder and more expensive to control, needs to be initialized against a stop.

DC motor controlled screw
Benefits, Simpler, cheaper than stepper motor
Problems.. same as stepper motor, as well as not having good position control without a feedback loop

electromagnetic
Benefits, cheap and easy to control and program via PWM
Problems.. no position control without feedback, and varying pressure against it will change it's position, though this COULD be a benefit.  May suffer from heat buildup at extended high loads?

Solenoid on existing AFC
Benefits, Cheapest, only need a PWM output with a power transistor and a 4 way solenoid valve like found on most exhaust brakes
Problems, cannot increase or decrease fuel without significant manifold boost first, some of this may be remedied in part if I can make the AFC double-acting and use a lighter spring in it

[hakcenter]

Total redesign!

Then you can just basically remove the AFC, remove the fuel plate, and have this as your total control. I don't think a worm drive would be the way. A simple gear to DC motor drive would IMHO be the best way to go.

Closed Loop Speed and Position Control of DC motors | IKALOGIC

I immediately think, rack and pinion control. The whole area is an oil bath, so you shouldn't have to worry about lubricating moving parts, and if all fails put the sliding gear rack on the driver side, that spot always has oil whenever I take the AFC off, haha!

Remember you'll need feedback since the pedal pressure, presses against the plate(s).

[Rx7man]

unless you want high current through the motor most of the time, you're going to have to gear down the rack drive, which leaves you with the same setup as a worm gear... it basically won't move on it's own if it's geared low enough... I hadn't thought of a rack and pinion drive, it would be another method.  The worm gear system could make use of the encoder as well.

Mouser has some linear stepper motors with a 2" stroke and .004" stroke-per-step, which would eliminate the need for the encoder, but they're not too cheap... They'd be easy to install however since you'd just hack the AFC section and tie the rod into the existing AFC arm.

With the electromagnetic solenoid, you wouldn't care about the absolute position, you'd only care if you need more or less fuel, feedback COULD be used, but wouldn't be essential.  I'd have something reminiscent of a small starter solenoid for it, and the thing about it is it really only needs a power transistor with a flyback diode.

I'm going to look at how much pressure is against the rack on my spare pump to get an idea of what I'll need, and see if it's possible to incorporate a rack position sensor on it.. It looked like if I drilled a hole on the side of the injector pump toward the engine I ought to be able to put a TPS sensor with an arm that feels the rack.. assuming the TPS sensor is spring loaded in the right direction it might just work.. I'm going to go to a wrecker and pick up a bunch of old TPS sensors, Coolant sensors, and the like.

I was also thinking of using an oil pressure sensor for boost and exhaust gas pressure.. I think the Cummins one is too low a pressure, but perhaps an Autometer or other manufacturer has one.. it sounds cheaper than anything else I've looked into and has the benefit of being replaceable by going to your local NAPA dealer.

I woke up to a call from MeaSpec this morning.. he's going to look into everything for me, and probably come up with a $100 sensor for me.

[Rx7man]

just took a closer look... most of what we need is already there in the form of the shutdown solenoid!.. the shutdown solenoid has a strong internal spring that pushes the rack back, without it it's free to move forward.

So, at least to start with, you put a manual shutdown cable on it, and reverse the action on a solenoid similar to the shutdown solenoid, so it pull the rack back when energized, you can control the rack position externally.. the shaft it's mounted to extends through the other side of the IP, meaning you can put a TPS sensor on it there and get your closed loop.  With a bit more work on the linkages, you could keep the existing shutdown solenoid, but you'd have to make it so that it freewheels in a direction (kinda like the cruise control linkage on carbs, where one side activating doesn't force the other side).

The pressure against the rack is roughly a pound of force, or 5 newtons.. this is just what it feels like to me when it presses against my finger.. significant but not heavy pressure.

I think the first version I'll use a pneumatic solenoid to cut the signal to the AFC, Second version will be the electromagnet, since it uses exactly the same Arduino software and hardware (just a PWM), and if I'm not happy with that I'll go the more complicated route of stepper or DC drives with encoders.

[Rx7man]

Just did the math on the wheel speed encoder in the diff (120 tooth), it actually creates a higher frequency than the turbo shaft sensor.. at 160,000 RPM the turbo shaft sensor 2.7khz, at 130 kph (~80mph) the ABS sensor has a 3.3khz signal... Not that terribly different, close enough that the caps can probably stay the same and just toy with the resistor values on the 2907 to keep a 5V output

[Rx7man]

OK, so I've made some blinking lights...

With the CAN shield you're making, it seems like you only have a couple analog inputs left which doesn't leave much room for anything else on that shield anyhow..  If you don't use A3-5, perhaps be sure to leave them somehow accessible for use on a Mega shield.. I'm thinking if I use the Mega I'll make a shield that'll fit beside yours.

[Rx7man]

OK, finally getting a bit of love from Eagle.. getting the hang of it.

Here's a board with 2 VR sensor inputs and a thermocouple input... I'll still have some math to do to figure out the values for the caps and resistors.. but I think it looks OK so far.  I think I'll keep all these analog inputs on a separate board from the digital and PWM outputs for noise.. at least to start with

[hakcenter]

Very cool, need an arduino shield library ?

[Rx7man]

probably wouldn't hurt.

I need to get better at finding the correct type of part, such as resistors in through-hole vs SMD, etc, and connectors too.. as a placeholder i just put jumpers.

From an old VFD I have a bunch of nice little terminal blocks one .200" hole spacing that could come in handy too

I'm thinking I'll design 'my half' of the shield to fit beside yours.. one level for analog inputs, the other for digital inputs and outputs, most of course will have to be driven by power transistors to be of any use to control any significant loads...

I have to do more research into what kind of power transistor is most suited.. FET? (and all their cousins),  IGBT?, darlington pair?

[hakcenter]

Probably find the best heat sinks for all the types, and pick the largest one. haha. Can't really have a case ground rail / heat sink like ecu's soooo gotta get creative.

[Rx7man]

I was going to mount them offboard and use a TO220 package... One side is going to the load, one to ground, leaving only one to go to the board, which makes it super easy to break them out.

I was looking at the STP40NF03L as a candidate.. 30V with internal flyback diode, I think it can be driven directly from the arduino, and it handles plenty of current making it versatile.

pdf.datasheetcatalog.com/datasheet/stmicroelectronics/6794.pdf


Edit: found a cheaper version of it through Fairchild at $1.91 each for 10.. very similar characteristics

   FGP3040G2_F085 Fairchild Semiconductor | Mouser
... put 10 of them on my cart, as if I make an H bridge out of them I may need 4 for that, and then who knows what else they're going to be handy for

[hakcenter]

arduino shield library, Arduino Shield Library Attachment

[Rx7man]

thanks

[Rx7man]

Meanwhile I've gotten into the Eagle peculiarities, and have turned my frustrations onto Pspice.. nightmare of a program.. clutters your taskbar to no end, and doesn't come with any decent libraries, and the ones I've been able to download (apparently for Pspice) don't want to work.. GRRRR

I'm trying to find the biasing I need for the IGBT above, I will probably need a low value resistor on the ground side of it to improve linearity, especially since the PWM outputs aren't that great from what I discovered with an LED...

I once had an old SPICE program that did what I wanted it to.. this one just sucks