Well, I let this go too long without replying. Sorry about that. Ok so here is everything in one post.
Vroom,
Very cool little board. I like the layout from what I can see of it.
ShokWaveRider,
Potentiometers can drift, especially poor ones. The biggest problem I forsee is actually getting better than 1% settability and so this drives the demand for precision multi-turn potentiometers. But this isn't really a big deal as 12-turn 200k pots run about $1 in single unit quantities. If desired, once set properly, a drop of epoxy on the lead screw will lock the thing pretty well.
As far as the brightness, yes, if the REFOUT pin on the LM391x is used to bias the comparator ladder and to adjust brightness, one will most certainly affect the other due to non-zero output resistance. But in my design, I've used an external LED to generate a stable voltage reference that drives the comparator ladder and this way REFOUT can be used to control brighness without affecting linearity. Furthermore, the LED could be replaced with an actual compensated voltage reference IC, but I foresee no real issues there. (BTW, the LED won't really light up all that much off a 3mA current)
CoolWht,
Absolutely, Ian's work is absolutely top-notch. I would have no reservations having him do the work. But it doesn't have the project value. I have this problem of making all tasks as difficult as I can, even when it isn't necessary. Hmm, yea, that's me.
BrentB,
It would definitely be cool to have all those other signals measured. The LM3914 does pretty well for such a simple linear signal from the boost gauge I would imagine, but for temperature and other things a more complex signal processing would be needed (assuming a standard double log thermistor). In reality, I think that it makes a lot of sense to go the MCU (microcontroller unit) route and ADC the signals.
The
design that I am going with includes a decaying peak indicator, which somewhat increases the complexity, but I think for a turbo gauge it would be very useful. But then again, it begs for digital rather than analog processing. Why analog, well, I think really just for coolness. Who wants to really design an ADC to a MCU and then do all the processing in software for coolness?
The big project that I am trying to work on is an integrated car-computer. When I did the 5-light brake mod tonight, I pulled the CD-changer mounting bracket so I can modify it to accept a Mini-ITX motherboard. When I get around to putting the whole thing in I'll build/buy (depending on cost) a multi-input ADC board so I can feed these other non-time critical signals in [oil pressure, exhaust temperature, intake temperature, etc.] to the computer and have virtual graphs displayed on the screen in real time, along with logging.
Of course, I am trying to design a DC-DC power supply for the computer too, but that is much more complex than this little meter.
I think the secondary sensor will give a more reliable and calibratable signal, but
Tech4TDI has a good graph of voltage vs. pressure for the OEM MAP sensor. I'll probably start out using this as well, but arrow.com has the MPX5700GP for $15US.
I definitely like SMT over through hole. Very cool and thank you about the CNC.
Oldpoopie,
Yea, I can't wait to see how this turns out. All my prototype parts are ordered so I can breadboard this thing.
Everyone,
So I did some redesign taking into consideration a lot of the advice you have all suggested as well as some own personal decisions. I pulled off the top LM3914 making this a 20 output max instead of a 30 output max. I also switched over to all surface mount components to shrink the size down since at this point, I figure not many people will actually want to build this on their own.
I updated the PDF on my site to include the revisions of both the circuit and the board, as well as a pretty complete description of the operation. The schematic includes most of the formulas solved for determining the correct calibrations given any desired input and output swing and or offset. As stated earlier, you could use this peak/average reading graph to read any linear analog input between 0 and 10v without any change in components.
Anyway, I'll keep everyone posted. Thank you all for all your help