As I mentioned in the introduction, there was one casualty of cramming the evaporator into the space under the cowl, the control cables of the stock air distribution system. I was actually happy to contemplate their removal, since I’ve always found them, not only a pain in the ass to remove and install, but also clumsy and stiff in operation. I remember many years ago when the car was new, I thought the shelf-like lip on the stock dash top pad made it a perfect place to keep the odd pen or pencil so I’d always have one handy until one day I tried to change the air distribution setting from above to below and heard a distinct “crack” as I moved the levers. I discovered a few years later when I was updating the dash to the ’72 and later style with the side air outlets that one of my pens had rolled into the passenger side defrost outlet when I’d braked hard and fallen into the flapper box on that side. When I had moved the control lever, which had so much friction in its action that I didn’t notice a little extra force being required, the plastic flapper broke because it was blocked from moving by the pen. Ah, the little surprises these cars have in store for us. Small molded ribs on the inside of the outlets of the plastic defrost ducts prevent this from happening on the later cars, and I’ve often wondered if those were added when the design was iterated specifically to prevent what I experienced. But no matter, the point is the control cables had to go for this conversion and I’m not sorry.

What made this conversion feasible for me is the availability of relatively inexpensive (~$20 ea.) small electric control motors designed for just such an application from the hot rod A/C outfit in San Antonio, Texas, Vintage Air, and possibly others. They look like this:

Click to view attachment

And here’s what’s inside:

Click to view attachment

Click to view attachment

The motors have internal limit switches and a position-sensing potentiometer, so they can be used for either analog positioning or always simply moving from one extreme to the other. I wound up using seven of them to operate the four flappers in the two air distribution boxes, the two heater butterfly valves, and the outside air damper in the evaporator box. The circuit to drive them is also fairly simple, just a pair of power op amps in a single package and a few resistors:

Click to view attachment

Although the power op amp shown is capable of 1A of output current, the control motors rarely draw more than 100ma in operation. The same outfit that supplies the motors also has controller boards with circuits to drive three motors for ~$35. Rather than using them, however, I wound up re-packaging their drive circuits in order to get seven of them on a 2.5”x3.8” PC board that would fit into the space available in the control module.

In order to hook up to the existing flapper levers that the old control cables used, each control motor mounting was unfortunately different:

Click to view attachment

Click to view attachment

Since the motor/controller combinations are capable of analog positioning, it was tempting to consider preserving the stock versatility in distributing the heated and cooled air by using two slide pots to run the corresponding flapper motors instead of the more modern design with pushbuttons. I elected to go with the latter, not only because I thought it cooler (in the non-literal sense), but also because I couldn’t remember in all the years I’ve owned and driven the car ever having used anything other than the full up, full down, and 50/50 positions of the distribution control levers. I electrically ganged all four flapper motors together, so that heat and cool are always going to the same outlets, the one exception being that the new center outlets under the dash are cooling only and always on when the AC blower is on. I didn’t include a “Max AC” button, as found on most modern cars, because all such buttons typically do is turn on the AC and set the outside air control to recirculation, which in most cars provides a higher air flow than the vent setting. In my system the recirculation input to the AC blower is always on, so shutting off the outside air would only reduce the air flow. The ganged flapper motors mimics modern cars, even though this is still a system in which the heating and cooling of the air come from widely separated sources, not a truly integrated one like most water-cooled vehicles have. In both cases the heat is ultimately generated by the engine, but in the water-cooled case it can be simply piped to a heater core that is physically in contact with or right next to the evaporator coil. In this system, mixing of heat and cold to get some intermediate temperature must happen at the air outlets.

If mixing of heater and AC air is used to control the air temperature, why then do I have a thermostat knob on the front panel? Why indeed. I did it because it was easy to do, and made the system a close analog of the factory AC system for the 911. I speculate that the factory may have done it on the 911 because the hot air supply in the air-cooled cars is too variable with driving conditions to provide an acceptably constant temperature at a fixed setting of the controls. Also, the AC part of the system needs at least a thermostat set to a fixed temperature to prevent the evaporator coil from ever getting so cold that it freezes the condensation water and plugs the air flow through the coil. If you have that much of the stuff required for a thermostat, it complicates things very little to make the set point variable. The required knob for setting also filled out the front panel nicely. Here is the circuit I’m using:

Click to view attachment

This is a simple comparator that compares the resistance of a thermistor buried among the fins of the evaporator coil to a resistance set by the knob on the front panel. The thermistor is about 10-12kohms at normal room temperature, and about 25kohms at 32°F. The values shown allow the temperature at which the circuit will interrupt the compressor to limit the cooling to be varied from about 60 to 38°F. The 1.2Mohm feedback resistor shown is important in that it provides hysteresis, so that the compressor doesn’t cycle on and off rapidly at the set point. The value shown gives a hysteresis of about 4°F. A larger value would reduce this amount, while a smaller one would increase it.

The other control I wanted to improve compared to the design practice that I’ve seen on the 911 and 944 AC systems is the blower speed control. On both of these systems Porsche used a rotary switch with a small number of positions, three or four, to control the fan speed, directly switching the high current (10-15amps max) drawn by the blower motor through different series resistors to vary the speed. A switch capable of doing this needs to be a heavy-duty device, and the torque required to operate the switch, on the 944 system for example, is high enough that it can be difficult to do, particularly with only the small diameter knob that will fit within the confines of the small control module outline to grab onto. Looking for a better way, one that is not only easier to operate but that also offers more control over blower speed, I first considered a pulse width modulation (PWM) controller, like most light dimmers. These require only a single power FET which is not called upon to dissipate much power as the current control element, a simple potentiometer for the front panel control, and offer smooth continuous variation in the blower speed. Unfortunately, the fans I’m using are very microphonic, and they howl loudly at the PWM chopping frequency, typically 400Hz which is right in the middle of the audible spectrum. I attempted to use slow-down capacitors on the drive pulses to soften the sudden transitions between on and off, and this helped some, but not enough. Another possibility is to change the chopping frequency so it’s not audible. The frequency would have to be increased all the way up to 20kHz or more to be above the range of human hearing, and even then dogs might chase you down the street when you have the blower running. The fans might be much less efficient as loudspeakers at this much higher frequency, but the factor of 50 or more increase in frequency would also increase the power dissipation in the switching FET by the same factor, probably requiring significant active cooling. Lowering the frequency to 10Hz or so would also take it out of the audio range, but when I tried it I found the fans would cog at any frequency much below 20Hz and started to sound like stepper motors. So, back to the drawing board, as they say.

The blower speed control on my Honda sedan is much more like what I would want for the 914. It has 8-10 discrete steps that are selected by a rotary switch with a very light feel. Thinking about how they might have accomplished this for a while, I realized that they most likely are switching just a control voltage with the front panel switch, a voltage which then turns on one of several power FETs that then control the heavy current drawn by the blower. Each FET selects a particular resistor which limits the current the fans can draw, thus limiting their speed. The power that the resistors are required to dissipate can be substantial, 20-25watts for the fans I’m using if hooked in parallel, and this requires inconveniently large resistors. The factory fresh air blower on the 914 has only three speeds, at the highest of which it’s being driven by the full battery voltage, so only two resistors are required. Both are coils of nichrome wire mounted in the air flow to deal with the heat generated from the current required to drive even the relatively wimpy stock blower. I can use the same trick by mounting both the FETs with their drive circuits and the resistors out with the fans, so that the resistors are cooled by the air flow, but, because space is so limited, I cannot deal with a large number of physically large resistors, such as commercial 25watt power resistors. My solution is to run the fans in series instead of parallel for the first six of eight speeds and then energize a relay to automatically re-hook them in parallel for the top two speeds. When hooked in series the fans are much quieter, the current is halved because the fans share the same current, and the voltage drop that must be produced by the resistors is also much less because the second fan doubles the voltage drop of the first alone. Once again, the top speed in each configuration is powered by full battery voltage, so only six resistors are required, and only one of them needs to be rated for 25watts of dissipation. Here is the circuit I’m using for this:

Click to view attachment

I was (barely) able to find room for all of this within the evaporator box, as shown here:

Click to view attachment

Actually, this photo does not show the final version of the fan module. If you will note, there are seven resistors shown, all the same wattage (15watts). This was the system as I initially built it, intending to run the fans in series for all eight speeds. When assembled so I could measure the air flow, I decided it was stupid not to have the option to run the fans at full speed for those occasions on which you don’t care about noise, and I redesigned the drive as described above. Eliminating one resistor allowed enough room to change one other resistor to a 25watt part, and the required relay fit neatly beneath the resistor array. The eight drive FETs are on the small circuit boards on either end of the array.

One final thing I wanted to be sure to “improve” about the stock control module was its ease of removal and installation. Granted, most of the hassle associated with removing the stock item was due to the very stiff control cables, which I’ve now eliminated. However, anything with the complexity of my electronic replacement and the probable reliability problems that will only surface over time, needs to be easy to remove and fix because it will no doubt need to be done several times. My approach to making the module easy to remove and replace was to make it fit into a tray that mounts permanently in the space formerly occupied by the stock module, handling the make and break of the electrical connections via a 50-pin “D” connector. The removable module fastens to the tray only at the very front by means of two captive screws that protrude through the front panel:

Click to view attachment

The tray, shown at the back on the right, attaches to the car at the same rear mounting point as the factory module, and to the part of the dash sheet metal beneath the knee pad via two adjustable length studs that are not shown. The module itself slides in and out carrying the chrome trim frame with it. The interior of the module is shown below. This is actually my third version of the controller, the first not having worked at all, the second having had problems with the thermostat circuit and utterly chaotic wiring, and the third finally organized and working as it should.

Click to view attachment

Next steps? I’m not sure what, if anything, I’d want to change about the controls, but it should be possible to introduce climate control, now that I have, in principle, electronic control over both heating and cooling. That, however, would be another project.

................I was just about to write this same thread

Mad-skills being demonstrated here, sir. NICE

Awesome job!! Did you source the climate control buttons or did you also make that from scratch?

Very nice mechanical and electrical work. Excellent!

I like how the power resistors are placed in the path of airflow for cooling. Very smart.

What schematic capture program are you using? Where did you get your boards fabbed?

I couldn't be an online ass and post whore without critiquing your work. Everything looks great. Components I would add are snubber diodes across the relay coils. Maybe even on your current switches.

Spoke,

Thanks, coming from you that's high praise indeed. I did consider adding snubber diodes on the relay coils, but the parts I used, the ZVN4206AV FETs, are specifically designed to drive automotive relays and supposedly do not require them. There's a nice app note that goes into detail about it here: http://application-notes.digchip.com/040/40-33513.pdf.

I use ExpressPCB's layout and schematic capture programs, and have them fab my boards. I've stuck to thru-hole parts for ease of loading, since I'm not tooled up for nor do I have much experience doing surface mount by hand.

Steve,

Oh, so that's what you mean by "climate control", what I would call the radio buttons that control the flapper positions to direct the conditioned air flow. I did implement those myself. I would've included the circuit, but that part of the circuitry is fairly complicated, and it was going to push the size of the post beyond the limit. However, since you asked, I'll attach it here:

Click to view attachment

Two of the five pushbuttons need to be alternate action, push to turn on and push again to turn off, while the other three are radio buttons, which cancel others in the group when selected in addition to selecting their associated function, although all of them are actually momentary pushbuttons. Their contacts are connected to logic on the controller board via the 10-pin connector shown, while signals to light confirming LEDs on each switch are fed back to the switches via the 5-pin connector.

Click to view attachment

On the controller board, the signals from the three radio buttons are first de-bounced by the CD4044 IC and then fed to the group of OR and NOR gates (the CD4071 and CD4001 parts) that implement the radio button function. Its outputs drive relays that send the appropriate signals to the control motors via the connector marked J8, and confirming signals to the button LEDs via the J7 connector. The two other buttons each trigger one of the two timer circuits in the LM556 IC hooked to implement an alternate action function. The "recirc" output from these drives a relay that sends its signal to the corresponding control motor also via the J8 connector, while the "AC" output controls a relay that is in series with both the "blower on" signal from the AC blower switch and the thermostat circuit, and ultimately controls both the compressor clutch and the condenser fan. Both relays also send confirming signals to the corresponding button LEDs via the J7 connector. The parts shown powered by "+12V mem" are hooked to a direct battery connection so that they will retain their last state when the ignition is switched off. The CMOS parts shown draw very little current, so this doesn't constitute a significant drain on the battery.

Holy cow, this is amazing!!

My own personal quibble is that the controller looks a little too modern for the 914's cockpit.

I bet that wouldn't be that hard to change, though.

--DD

I agree, but still looked very German to me. Maybe it's the knobs. That's why I thought he might of sourced it from a vw or Audi

They're from a 944.

How well does the AC work? Gotten warm enough to test it?