Awesome! Sorry for sending you down a wrong path, it makes perfect sense now

Wow. I have never felt dumber. There are some smart folks on this site!

Cool stuff!!

Not that it matters now, since you have been experiencing success with your interface using the Arduino, but what I was referring to with the use of the OE gauge guts was to place the guts behind the gauge face in place of the original, put the Boxter needle on the shaft of the OE unit, and then use the original tank sender. I see this as simple, but not nearly as much fun, nor elegant The problems I foresee doing this are possible length of shaft of OE too short, needle wont work on shaft, sweep ark of OE not same as Boxter and thus gauge face will not be correct measurement/needle deflection will be incorrect, OE guts will not fit.

With your initial start up causing needle bounce, is it possible to have a value programmed in to make the needle hold a value consistent with the E at start up? How bout Full and then is drops to actual? Is this bounce of the needle stop due to poor or no signal? Does the stock setup in a Boxter do this, or is this just due to your interface? How about a delay in the interface that allows it to power up before sending a signal to the gauge so that there is no "seizure" on start up?

Nice work Skeates!

The Boxter cluster up against the OE bezel looks great. The cluster looks nice and small, the biggest problem may be the depth. I don't recall how much room you have to work with, the Boxter looks small but deep.
For the bottom you could trim some off the lower dash sub frame and remove or remake the top loop. I removed my top loop and the cluster mounts to the bottom of the dash.
It seems like, with a custom bezel, it would look great in the OE dash.
What motor is going in this car?

So, until someone tells me this is cluttering up the web I'm going to keep posting my progress . Made a bit more progress on the boxster gauge cluster this weekend. First things first; even though I'm making a custom dash cap I figured I'd check to see how close this thing fits within the stock set-up. I figured if it fits that would open up a world of opportunities to those wanting to update their gauges, and stick with the Porsche feel without having to make a custom dash. So...I dug out the mounting plate for the stock gauges just to see how close the fit was (see first pic). It's hard to see in the picture, but there are two issues as I can tell which would make it difficult (though not necessarily impossible) to use this with the stock dash cap.

1) The toggles on the top of the cluster (for setting the clock, adjusting brightness, etc.) line up just about perfectly with the two top mounting screw holes. This wouldn't be too hard to work around since you'd be fabricating a custom surround to mount the gauges. At that time just move/make new tabs on the dash frame that line up. Done.

2) The second issue may be somewhat more difficult to work around. Looking at the bottom of the picture you can see that the pod housing the warning lights sticks out slightly below the bottom of the mounting plate. The boxster cluster is just a little bit too tall from what I can tell to fit within the stock dash cap - but it's such a close fit! So close that without an actual dash cap to play with I can't make any final conclusions on the possibility of making this thing fit with the stock dash cap. (if anyone wants to head up to Truckee, Ca with a stock cap to test this out with feel free to swing by )

Click to view attachment

So, onto the fun stuff - electricity! I've spent more time then I care to think about tracing the boxster wiring diagrams and trying to piece together what's needed and what's not. There are several power inputs, (2) which are constant +12V and (3) more which are powered when the key is in the ACC/Drive position in the ignition. Also, there are (2) grounds for the cluster and an additional "ground" for the sender units (fuel, coolant temp, etc). In order to keep strait the mess of wires I manually pinned out and labeled each wire through the connectors. While doing so I found that many of the expected wires/inputs were "missing" from my pins. Some of these are for inputs from the tiptroic transmission which I expect to be missing on a manual cluster. Others though I think are related to the on-board computer option. At this point I'm just worried about basic functionality, so I'm not paying a whole lot of attention to the missing inputs, but they may become important later if the cluster is expecting an input that I'm not delivering. Note my pig-tails are out of a different '97 manual boxster than the gauge cluster, so there could be some miss-match happening if the options on the cars were different. Below there's a picture of my test-rig (basically just a bread board to facilitate quick & easy circuits and and arduino board that I'm using as a signal generator to simulate speedo and tach signals.

Click to view attachment

With the gauge cluster powered I then went through at tested the voltage at each (well most anyways) wire to see if any were powered. After doing this I found out something very interesting - most of the indicator lights, the speedo signal, and the tach signal are all powered and expect a ground as their "input". In the case of the speedo and tach they need to be pulled down to ground at a certain frequency, like a flickering light switch, which corresponds to the speed of the car/engine. I'm not sure whether or not this is "normal" for these types of signals, but I was expecting to have to produce a +12V signal to each, not a ground. It actually made it easier in a sense since all I needed to do was use a general purpose transistor as a switch and feed it a PWM signal from the arduino. I addition to those signals, I also got some potentiometers out to simulate the coolant temp and fuel senders. The pic below shows everything running...sort of...

Click to view attachment

So, the moral of this story is that I've been able to get all the basic functionality working! I've got an Excel spreadsheet I use to calculate at what frequency the PWM signal needs to operate to simulate a specific MPH or RPM. I've also got the coolant temp gauge sender resistance values figured out, the gauge back-lighting/dimming is working, indicator lights are all functional, as well as the turn-signals/high beam indicators, etc.. But....and so far it's been a big pain in the butt....the damn fuel gauge is only sort-of working and represents the final piece to this puzzle.

The top 3/4 of the fuel gauge range works flawlessly with a 0 - 400 Ohm potentiomter. However, as soon as you get to ~ 400 Ohms (just less than 1/4 tank) the fuel gauge drops to empty and the low fuel indicator light starts flashing. Doing some reading I found that on the 996 C4's the gas tanks were designed such that a sender wouldn't fit all the way down and then the gauge had to switch over to some sort of internal calculation approach using fuel consumption data from the ECU to estimate fuel tank levels below 12 litres. I didn't get the sense though that this was true of the C2's or the Boxters. Something definitely happens at the 1/4 tank mark where the gauge stops listening to the sender signal. In fact, once it hits that mark I can no longer get the needle to move at all by changing the resistance values. I have to do a full reset on the cluster (un-plug and plug back in) to get it to listen again. I figure it has to be looking for a signal that I'm not giving it right now, and I think I've isolated it to one or two wires:

1) There is a fuel tank status wire going to the DME (ECU) which I thought was an output, but now I'm wondering if it's an input. It has no voltage when the cluster is powered.
2) There is a wire labeled "KVA DME" on the cluster diagrams, but on the DME wiring diagram it's labeled "fuel consumption indicator". I think this one may be the missing piece, but I have no idea what signal it's sending. On the cluster this wire measures +10 V (same as the speedo and tach wires) which makes me think its looking for a similar signal to ground. My current guess it that it will somehow correspond to the fuel injector pulse signal.

If anyone has any ideas (or information) on how to get the last quarter of the fuel gauge working I'd LOVE the input. It's the last 10% that takes 90% of the time/effort!