h3rk

Since some of this thread has taken to looking to the future, and since much of making this project all it can be depends on some not yet available capabilities, and because I have to admit I spent a few hours yesterday drooling over that cubloc stuff; I have an idea that I'd like to see for the when we our new brains evolve to the point when they are able to think for themselves, without serial coms with the mothership:

It would be nice if in variables, you can select the ones (checkbox?) that you want defined both in the widows software and on the brain. Then you could click another checkbox to define which functions you want running on the brain. And if several components were set up via I2C bus, choose which device will contain that code ( or maybe that could be transparent to the configurator). I'm not sure if it should be an either/or selection, or and "if serial coms = false then use the local version" but if it were that easy to setup in the config, setting which things you need to function independantly just like setting up the config file... That would rock. Then when saving, the configurator could check for changes and prompt to upload the new file... Maybe?? I bet it'll be a lot of 'behind-the-scenes' code and a ways down the line if at all, but having it that user friendly would make the kind of setup experience that people would tell their friends about.

It would make setup be less like programming PIC's and more like playing 3D chess; knowing where to have functions and variables, how they'll interact if they do, and how transition to serial coms with PC will be like.

That way functions that control things like input>>process>>output could be on chip. And things that effect GUI, and maybe overly advanced features could be in the skin as it exists today.

[/dear Santa]

h3rk

This post is pretty much a non event, but I just wanted everyone to know that I'm still right on track. I have the MOP's in for my dials, I'm experimenting with some clear epoxy resin for modifying the lightpipe type of setup in the middle of the dial, so that I may send the shaft of the motor in far enough to turn the dial (with the aid of another piece i'm making).

And... The reference material I ordered just came in the mail. So I have a lot of reading, pondering, and bouncing off of my own notes to do. This will take some time, as the book is not titled, "How to control any automotive automatic HVAC system with no design specific knowledge of the existing system; A tutorial in adaptive algorithms." I'm still pretty excited about this, so I should get through it asap. From there I will grow what I have to the point that all that is needed is the shell part of the skin.

so more to come...

Dingodongo1

I have embarked on a similar project. After reading this thread I thought this resource may help you with climate control logic once you have straightened out controls and sensors.

http://delphi.com/pdf/techpapers/2007-01-1188.pdf

h3rk

I've been planning this for a while (HVAC) and while there were some outside interferences taking up much of my time, implementing this was not at all what I expected. Mostly because I changed applications, you really can't get 2 more different HVAC control systems than what I had in the chebby and what's in this Nissan. I looked at the manual at first and saw the big difference (serial data coms between the servos' local controllers and the control head), but I figured it would give me a chance to come up with something that was flexible enough to work in many situations. Well this, plus 3 low signal DACs (and at least 2 brains), a few extra sensors,.. should be all of the connections I need to finally get a working model. The method I'm implementing in this application is not direct control of the actuators by means of bypassing the control head. The method consists of interfacing with the control head.

At the current time I have reasoned out motorizing the dials, I found a way to eliminate the need for that, but it's still feasible as a next step, should I feel it's needed.

The 3 dials on the Calsonic Kensai control heads for HVAC are merely potentiometers, each tap is fed to 1 of a 100 pin IC (the indentifying marks on the IC are silkscreened with the manufacturer's logo, Calsonic, so I couldn't find a datasheet for it as it's a proprietary deal). I suspect that those pins are internal ADC inputs. So I intend to place a small circuit in series with that signal path. A 3PDT relay will allow the normal signal flow when the computer is off (or FCC isn't ready) but when the relay is on the pins recieve outputs from 3 small DACs (same ones 2K1Toaster suggested in my DAC thread, since they did work well), but since their max output is 4.2V and I needed a small opamp on the output to get the full 0-5V.
The dial pot's output will be fed to the brain via analog input and the brain will then initialize the DAC's using a bunch of digital outputs. I haven't decided if I want the brain to sample these outputs and compare them to the pot inputs (probably not) yet. Then once the values match, it triggers the "ready" relay and switches over to "brain control", there the brain will mimic my settings unless I move the dials to the AUTO position, then it will decide based on some logic. I've been studying that part for some time now, and I'm getting better at it, that book I ordered is great, all of the stuff I didn't know, but needed to.

I still have a full working HVAC system with vaccuum solenoids and temp-servo, and 4 speed fan (with the power transistor) that I plan to accomodate as well (sitting on a bench). I'll be bench testing that one in the next couple of weeks as well. I'm thinking I may make a few small multi-application HVAC interface kits for making connecting brains to HVAC systems. But as of yet I need to get one working.

Controlling the actuators/motors directly is the best bet for creating more universal applications, but there are some...like mine where it's just not practical....


More to come soon, progress is being made!

h3rk

Alrighty, I've got some progress to report on, for anyone left following this:

To recap, this thread has taken some twists and turns. Mostly because I changed vehicles. But My goal has been generally the same. Control HVAC with my PC. It was not as easy as I had hoped, but throughout this process I've learned a lot.

There are a few main ways to attack this, and the way I finally decided on was to retain the original controls, this provided me with a few important advantages:

1. A means to control HVAC, changing settings (hopefully only desired temperature) has been accomplished by taking inputs to the brain from the control head. I was concerned about having to select a screen on the touchscreen every time I wanted to change the temperature, or something else.

2. Fully functional HVAC with the PC/FB off. The brain has an output which, once everything is initialized and current settings on the control head have been read into the software and it's own outputs are matched, it turns on.

3. This allows for the functionality I desire, where with the brain can control HVAC unattended (in conjuction with my remote start interface), regardless of where I left my settings when I got out of the car. I really don't see a need now to motorize the dials, but I have most of that work done and parts here if I decide to.

4. Many Nissan cars (present case included) have a LIN bus setup for the servos where each servo has it's own onboard controller and the control signal from the control head is not PWM or analog of any sort. The control signal is serial, and I have very little data on the data being passed, I'm not going to go through all that would be included with decoding it. Interfacing with the control head allowed me to accomplish the same thing directly, as the dials are all 0-10K pots, working off of 5V. I have to say I was pretty happy when I found that out.


So interfacing my HVAC took a long time of reverse engineering (of sorts) the control head, I have most all of the PCB converted to schematic. This is not an option I recommend for those without some electronics experience. I suppose the major factor involved with doing it this way was my desire not to change the servos, and to have a way to control HVAC with PC off.

Here's a picture of the dial interface I just finished testing:


It uses 3 DACs and a 4 channel OPAMP, and a 4PDT relay. It reads the dial position and either passes the voltage through passively (relay) or bypasses the pots (pots are still active as an input to a brain). This is output intensive though, since to get the same number of discrete settings (detents on pots) I needed 6 bits for 2 of the DACs and only5 bits for the 'mode' dial. That's a total of 17 outputs... oh well, 2 brains it is.

All that is left is:

install the new factory sunload sensor, and the aspirated in-car sensor, a modified Mass Airflow sensor out of a Hyundai, and a 0-100% humidity sensor. I hope when all testing is done I'll find a way to get the performance I want without having to use the MAF.

finish the interface circuits for both of the above and interface both the refigerant pressure sensor, and the outside air sensor, already working in the vehicle.

And finally hook it all up, and complete the software side.

Dan2008

WooHooo!!! Glad your making good progress.

I've been following your thread as I'm doing something very similar. The advantage I have is that it will be a completely new HVAC system, designed for a street rod. It only has three controls: push button compressor on/off, temp pot and vent selector pot. My heat valve (hot water) is servo controlled so that makes that pretty easy to deal with. Finding the ranges for the pots (and what happens in that range) should be pretty easy, especially since I can put everything together on the bench for testing.

When I get closer to putting everything together I'm sure I will be begging you for info.

__________________
Thanks,

Dan

h3rk

I"m always willing to help, I like to talk about this stuff, and I try to help with all of the PM's this thread has brought. I really hope to have something completed one day that will be versatile enough to work with most any application, and researching and helping other applications doesn't hurt that cause, one bit. I know stuff like that dial interface is a major turn off to most of the people interested in this. That's why it should be just one part of a total, black-box product that has enough different output types, and configurable input channels, so that hooking this up is almost as easy as a car alarm. That's my hope, at least. But that kind of flexibility will cost $$$.

h3rk

Well I'm done with the interface board, now all I have to do is put together all of the connecting cables. The new part consists of analog input channels from left to right they are:
Humidity/temperature
Interior temperature
Sunload
Intake temperature (at evap)
Outside temperature
Air velocity (inside airbox)

well, when you add inputs for defog/AC compressor/recirc on signals from car, and Fan/Temp/Mode dial positions, that adds up to

13 Analog inputs to Brain.


There are fourteen wires going to the control head through that Dsub connection.

For Digital outputs I have 6 for the fan DAC, 6 for the temp DAC, and 5 for the mode DAC. then 1 for each of: defog on output, recirc output, compressor output. And one more to turn on the relay, giving HVAC control to FCC.

So, 1 more day for making cables, another day to test and see how many mistakes I need to fix, I have to make some jumper stule resistors for those pins up top (some use just jumpers), I tried to make the circuits flexible that way.

I could have this in my car for testing as early as this weekend's end. I know I'll be back, heavy into the software by then.

2k1Toaster

awesome! Cant wait to see how this turns out!

__________________

sporty_drew

Wow looking great!

__________________

h3rk

Ugh, I don't know how I missed this. When making up the cables for the DACs I realized something; for the DACs I need +5 (on)and 0(off), and I can get this with the FB's switched-low digital outputs and a pull-up. BUT... In order to get +5 to the DAC, the output needs to be off (no current through the pull-up resistor connected to +5V), with the Digital Output on, full corrent flows and the output drops to 0V. This works, I think, but it's not ideal. With +5V to each DAC's input, the DAC's output becomes maximum (4.2V) which, through the opamp and feedback circuit, creates the maximum output. I did not invert things. What this means is, the default or failed position of the interface's outputs which replace the dials is FAN -100%, Max Heat, Defrost. Not the most ideal failure mode, especially in LV. But, if the outputs fail, then the relay drops out, removing PC control. Which means Fan-Minimum, Heat (wherever it's set), Mode- Max AC. Ok not so bad, but I just don't like the default-on outputs here.

I wonder if sending the DAC output to the inverting terminal of the opamps instead of the non inverting would make me happier. At a minimum, I have to add another small board with another 17 pull-up resistors to +5V, and more headers. I have a little room left. I have to be kind of careful here, I don't have a lot of room between the halves for cables/wires. I'm probably going to have to make the box a little taller (good thing it's made out of ABS).

I'm open to suggestion.

h3rk

After completing all of the connections from the Brains to the Interface board. I just can't look at it without telling myself, "Don't cut the red wire!..."

This is taking a long time, my crimping tool is cheap and I don't like it.

Since this is my first run, and I plan on upgrading a few more times until I have one that is versatile enough to work with different setups, I'm trying not to get carried away with my retentive tendancies. I do have to modify the box a little, raising the roof so the jumpers don't hit eachother (top-bottom). I'm putting a 12V plug on the outside, holes for the USB (decided not to do extensions from the brain out to the same style female plugs on the box exterior), and one more D-sub connector on the box.

I can't decide if I just want to make holes for the sensor cables to go through where they can plug directly to the interface board, or do some sort of molex plug on the outside. (6 external sensor cables, 16 pins total).........


I guess I'm still on track for plugging into the car tomorrow night.

h3rk

I just finished putting this thing together, and I started to test the inputs and outputs, but my test skin is freezing up. I noticed with the virtual brains that one of the brains is hanging up after a few seconds. It's reminiscant of problems I had several months back. That brain hangs if it's the only brain attached, or in a pair. The other brain works great. So instead of making and breaking connections to test one part at a time, I ordered my new brain, and I'll just wait. I have 3 more cables to make for my sensors anyway.



And it's HUUGE! Oh well, it's only a start. Still, I'm going to clean up the box a little, prior to the brain swap.

Oh, and the used MAF sensor I had bit the dust, so I'm looking at making a homemade heated thermistor sensor. Good thing I made the input channels on the interface board configurable. I'm also looking at adjusting some of the loops so that air velocity measurements (other than initial with a hand held anemometer) won't be needed.

Punky

Ohh man...
When you suggested this idea for my final year project, I'm quite glad i didn't take up this project. Not that its a good project, its a great project! But damn its complicated. Mainly, because i don't have enough programming background to come close to finishing something like this. Anyway, just wanna say, great work, and goodluck with the rest of the project!

h3rk

I remember seeing someone recommend that for your project in some thread, somewhere. But it wasn't me. Hopefully once it is up and running and functional, I can find ways to simplify it. I still need to put in support for direct servo, vacuum solenoid, and fan control. And then I can work on simplifying it.