Arduino - Digital AC control system for '80 911

[Dr J]

Here is the wired up system:



Here it is after warming one sensor with a blow dryer:



Current draw with the relay activated about 100 ma

[Dr J]

Here is a schematic of the system wiring. The software tool (Fritzing) did not have the DHT21s so I picked the DHT family it had in its place. Also, no Nextion screen so I just used another one for illustration purposes. I colored the wires to match the wired-up breadboard to make it easy for Karl.



So... after exactly one week, I gave Karl his parts, files and the working program.

Now I'll sit down like everyone else and watch Karl do his stuff. Go, Karl, go!

[Discseven]

^^^ 1 week = JohnThebulletTrain!

... Had I been at IT helm, would have been in Arduinoville well into 2017. Kudos once again for stepping into this with me---is one thing to say "Yeah, I'll do that." Is another thing to genuinely do it. Thank you John---excellent piece of work here! (Tech support into perpetuity... what a wonderful thing! )

My go-go for today...

Going ahead with gel epoxy plan. Leather situation… there are variations to back side’s texture in hides I have---some areas are smooth, other areas are rough. Managed to cut out a very smooth area. (Tested applying gel to back side as a preliminary smoothing-coating. Works but some flexibility is lost and stretch is gone. Would work for certain applications---not here as stretch is absolutely needed.)




Rather than attach all three inside curves to panel at once, will connect them individually. Top first…




Synthetic---as in fine grain---sponge is used to push leather into curves. Just enough gel is applied to soak into leather and… to spread out no further than planned connection area. All gel areas that transition to none gel areas cannot have a lump of gel that cures---if this happens… lump of gel translates into bulge in leather. (...is from: what-not-to-do from my dashboard adventures.)

After gel sets, leather is pulled back to check contact area. This looks as it should.




Second and third epoxy segments. (Doing small sections like this seems silly… until you’re actually setting a cover into multiple inside curves with epoxy. If things are not going right, it becomes a panic to make adjustments before epoxy rules you---which is at around 3 ½ minutes. Would rather take a little extra time and deal calmly with smaller sections than risk screwing something up by loading up on connection points.)

[kuehl]

Originally Posted by Discseven

Fully agree on eyes focused on what's coming next as opposed to interior distractions.

Karl,

My interest in watching you and DJ having fun here is not so much as an AC application, rather for "other things"; I sense your curiosity with what was happening with the evap core was quite a motivator !

Visually, if the display will hold up to the environment (light wise), and other possibilities I foresee (although the display location limits a few).

First, and probably DJ can address this, is voice recognition. With the current Arduino (gee I wish they found a better bar name to use) I've seen a few apps, Bitvoicer with the limited Uspeech library. Any thoughts on using a mic and the app's to direct the system or inquire status?

Second, while Karl is cutting cloth in the hood (keep the lad busy; keeps um out of trouble), for those with aftermarket engine management, a touch screen would be sweet to access and make tweeks without fussing with a laptop or tablet. Maybe the crew at Adafruit might have some ready made toys?

[Dr J]

There are still quite a few pins available in the Arduino so more functions could certainly be added. For voice Recognition I would explore EasyVR speech recognition system.

I told Karl to let me know if he wants to put a 2 axis accelerometer and we could put a page in the display to show how many gs he's pulling!

[Discseven]

Originally Posted by kuehl

Karl,

I sense your curiosity with what was happening with the evap core was quite a motivator !

Indeed CG!

Originally Posted by Dr J

I told Karl to let me know if he wants to put a 2 axis accelerometer and we could put a page in the display to show how many gs he's pulling!

... There's also electrolysis manufacturing control and automated hydrogen manifold injection. And minifridge. Possibilities are infinite. (Will draw line when Arduino can drive car.)

[techman1]

Even a rudimentary "Kill Switch" that enables a relay that grounds coil and prevents a start. Code needed to unlock.

[Discseven]

^^^ TechM... that's interesting direction. Potential for Ard solutions begs a thread to stimulate & file ideas. Suggest you run with it.

[Discseven]

Could be something.




Was planning to finish attaching leather to "box" today and later drill/cut holes for rotary switches (and possibly 2 paddles.) Seems good idea to drill & cut panel before leather attaches to top area. Paddle switches arrive Thursday. Scale drawing says everything fits... will do mockup to make sure.

Display screen fits panel window but two minor grinds are needed. First, notch is needed for wiring connector entering panel. Next, need to move display slightly forward in window so screen's face flushes with leather. This will allow bezel to sit on screen's surface. Will grind down panel's mounting platforms or display's mounting wings. Display seems easier... but perhaps more risky.

.

[Discseven]

Attaching bottom section of leather to panel. Process: Gel is applied to base and leather is laid over and pushed as evenly down across surface as possible---goal being to ensure good/even transfer of gel to leather. This “push” can result in a very lumpy surface---all depends on leather’s back side. (Squeegee technique does not work AT ALL to make surface appearance corrections. And, the more gel that’s applied… the more leather or vinyl tends to “float” …which becomes more problematic.) Leather is pulled away from adhesion area. This allows visual inspection of adhesive transfer and serves to raise & texture the gel. While I have no scientific data, I believe this step is beneficial to final outcome. Leather is then stretched slightly as it is laid down again. No pressure is applied down on the leather’s surface after this lay down. Stretched tension is maintained until well beyond the 5 minute cure time. As gel begins to stiffen, adjustments are made to fine tune the last bit of panel’s inside curve.




May seem like an abundant waste of leather here. Actually this is purposeful---it provides material to grab and manipulate… or clamp… or do unusual things with in order to achieve desired forming. This panel is not a complicated form requiring odd-ball leather torque’ing---Targa roofs, dashboards, seats… those each have unique cover torque’ing requirements where abundant perimeter material is very helpful. Point = better have too much material around perimeter than otherwise.




Display’s electrical connection… neglected during panel’s mold making.








4D System's 2.8 display bezel. Nicely made. Might be better were it not a solid covering but was simpler course getting this than making bezel in either 3-d printing or CNC machining. Challenge on hand with this is fact that bezel's window is only slightly smaller than live area of Nextion 2.8 screen. Could make bezel’s window area smaller but doing so = risky---might screw up bezel. Alternative path taken is to embrace given tight tolerances… and get on with it.

In this case, bezel has first-priority in terms of positioning. To fit display according to bezel… had to grind panel’s window slightly.




Air-gap between screen’s face and solid-face of bezel is minimized. As material is removed from ledge… display moves closer to bezel. Once screen’s face is flush with leather’s surface… that’s it. Ledge is then drilled & tapped to secure display to panel. Fasteners = 4-40 screws w/double washers. (Leather thickness affects final thickness of ledge and so the number of threads provided for fastening display to the panel. Another reason thin leather was used here.)






There are two “sources” that enable info to be displayed on screen. 1. Info-content comes from data loaded into Arduino. No console disassembly is required to update Ard’s programming (given USB’s terminal being at console’s front access hole.) 2. On-screen structure of displayed content comes from display’s onboard memory. While attending ledge work, it occurs to me that uploading to the display occurs via a chip that inserts the display's top side. This means---as the config now stands---every time a design refinement is made to displayed content… the display needs to be removed from the panel. Could be worse---could have epoxied everything together.

I’ve yet to mount BOTH the Ard and display to check existing clearances. Certainly not the end of world as everything---when you think it over---exists in a state of perpetual flexibility. Do wish I had been smart enough to NOT assume everything for the screen came from the Ard.

Plan is to leave config as is. When first return trip comes to reprogram screen, will decide then if a mod for SMART access is something worth attending.

Sensors, screen contrast dimmer, and relay wire terminal. Disconnecting this is made as box is being pulled from cubby---extra wire length on female side needs remembering. One extra pin will exist here when all wires are soldered in. (Does not include paddle switch wires.)

[Discseven]

Paddle switches. Scale drawing proof’d fit between switches and AC knobs and rear-side hardware. Given proximity of controls, scale drawing alone is dangerous… a mockup is the way to think this through. As this config stands, paddle switches are as low and as close together as possible. Position of knobs is flexible. Pic approximates driver’s angle to console. Looking at this, I’d move knobs up very slightly. (To move paddle switches down would require grinding shelf out.) Am still undecided as to whether to install these extra switches or not. One voice calls them “idiotic control clutter.” Other says, “They’ll be valued downstream.” Discussions are in mull mode.






Got: RR3402A. Is ON-OFF.
RR3402D is ON-OFF-ON. This model would allow centering paddle between knobs. Wanted to order 2 “Ds” for test but they’re not easily available. (1000 units could be had in several weeks.)

[Discseven]

Stock fan switch and potentiometer (for "Set Temp.") Notch surface on pot switch shaft needs to be cut back 12 thousandths" in order to match stock shaft's notch---this mod will center knob's rotation on shaft.




Almost attached leather to knob area before cutting switch holes. Doing so would have been a mistake.




Making sure switches fit holes before attaching leather.










Tilt. Console = 38 degrees. Display = 5.7 degrees. Total display tilt = 43.7. Provided display is tilted correctly, reflection (when installed) will be black headliner. ("Propeller" knob insert will be switched out for "temp" insert.)



Seam line between newly made "AC control box" and edge of cubby... the way it is now, am confident Porsche designers would never have done a seam like this---it's "unfinished." Am going to look at eliminating this seam for "exploration purposes" only. (A display---slightly larger than this one---will eventually be positioned up near steering wheel (TY Tim) in order to delete entire console... so am not looking to start over here! Primary goal now is to configure digital system that works. To each his own as to where it then goes.)


Locating sensors in the car is almost at hand. Exterior temp & humidity is planned to mount behind front bumper ("A"). Am OK with this---if you have better location, bring it on! Cabin temp & humidity is planned to mount on ceiling behind seats ("D"). This location continues to be a questionable---seems too high...

Where do you like sensor "D?" (Dimensions in next post.) An explanation as to "why" you're choosing an alternate location will help make sense of it---TY

[Discseven]

[Discseven]

Before disassembling John's fine bench model, will put sensors to mercurial test (to reconfirm his findings.)





Hanging wire is for relay.






Startup screen… displays while system collects first data. Nice touch John.




Bezel laid on. Only evap temp has been set to read decimals. Purpose being to cut power to compressor 0.1 degree above evap freezing. (Sensors pass my mercurial test to confirm John's tests.) 1 degree difference between ambient and cabin sensor readings (given close proximity of sensors at this moment) is not an anomaly---sensors are very sensitive. John mentioned black background as seen displayed here having illuminated pixels that can be turned OFF. Doing so would further improve contrast.




This bezel came attached to 4D System display---that display purchased only for this part. Prying it away from 4D display was not easy. 3M adhesive applied here is VERY sticky.




Arduino's now installed in panel-box. Display's screws can be easily accessed to remove display for downstream updates. Ard does not have to come out to remove display.

[Discseven]

Display's data upload chip and docking---dock being at display’s top edge. (Was asleep at helm where this chip's existence/docking location is concerned! Is what it is now.)





Battery in display is not needed in this case.

Display updates… Only one set was done after John’s system completion. We improved color balance of on-screen items and refined the “fade” when adjusting screen contrast. Display shows that upload in progress here... and reports when 100% done & successful status.




Will begin hard-wiring later today---need to pick up solder. Could perhaps leave pin-ends in Ard but will solder wires to these insert racks instead.

[Discseven]

Wiring progress...

Using 26 gauge stranded.




Datak mini panel. Will distribute power & ground to system components from this.




Nylon posts.






Positioning terminals. Then soldering in from behind. (Will bridge soldered side later when terminals are all in position.)






Power / ground enters via small terminal on right side. All other terminals service system parts. 4.7k resistor is on evaporator sensor signal line---it connects to power (for some reason... not a clue why. Since John likes it... so do I.) Power is blue. Ground is grey. Signal wires are purple or yellow. (Exception is display wires. Power is red. Ground, black. Signal wires are blue and yellow.)

[Discseven]

Wiring quick disconnect. Simplifies removing box from console. (Am making mistake here. Center row of wires should be soldered before these.)








Arduino was positioned at an angle inside box to accommodate position of power cable entering Ard. By chance... happens to be good position where removing display is concerned (necessary when reprogramming display.)




"Set Temp" wires... feeding to Ard and power / ground panel.




Next... need to sort out exactly where sensors, display's contrast dimmer, and relay will position---in order to make cables for those runs. (Could use phone cable or CAT 5E. I considered both and decided against them... mainly because their terminal configs get more elegantly complicated than is necessary for this. Pin ports are simpler solution.)

.

[Discseven]

CAT 5e parts. Am not using but system may be right for someone else here. Cables---8 wire---are dirt cheap. Receptacles---high end ones---are not. Think each terminal is around $10. Believe there are also white and blue terminals. Phone wire would also work with Arduino and related hardware but cables would need to be “corrected” (to align the wires at both ends… so I’m told.)












CAT system would have a nice look here but pin port is much simpler. ("Main terminal" will allow console to disconnect from car.)




Terminals near sensors, compressor relay, dimmer switch.

[Discseven]

Cable from AC control box running out of console. Am now on female (disconnect) side of terminal that's mounted on box. 14 of 15 pins are used to get info for: evaporator core temp, cabin temp & humidity, ambient temp & humidity… And to operate display dimmer switch and compressor relay. (Relay is oddball---requires power from Arduino, additional power not from Arduino, and 12v for compressor. Only “from Ard” power is run through this terminal. Non-Ard power and 12v for compressor will come from outside of AC control box.)






This terminal allows AC control box to make easy in/out console trips.


14 wires (26 gauge stranded) fed through ¼” heat shrink tube. 3’ length. (Would be difficult to push through longer length---would have to rig a pull.) Will complete system and bench test before shrinking.

[Rawknees'Turbo]

WOW, Karl, your fine detail work looks seriously badass - love it . . . long time!!!