Electronics Project - Quad CHT Gauge

[sain]

I thought i'd create this thread to document my progress on my current electronics project, which is a low(ish) cost digital 4 channel cht gauge, with over temp alarms. My ballpark guestimate of final unit cost is $80-$100 not including the 4 thermocouples (which will cost an additional $80 for the 4 at current prices). I'll try and post costs as I'm going.

 

Currently I'm in the early stages of designing the gauge, and learning how the electronic dohickeys work, and figuring out what is and isn't a good idea. If anybody has any input, suggestions or criticisms I'd appreciate them. Down the track i'll also be looking for somebody to test it for me (as I don't have my own plane).

 

The design is currently for a single unit, contained fully behind a 2 or 2 1/4 inch front. The gauge will be 4 columns of 3mm leds, each with 8 leds to indicate temperature (4 green, 2 yellow, 2 red). Brightness will be controlled by either a knob or 2 pushbutton switches on the front of the unit. This display has been built, but still needs testing in bright sunlight (works well in ordinary daylight, and dark low light conditions). The display is controlled by a PIC microcontroller (currently a 16f628a, but will be changed to a 16f877), which will also handle the temperature measurements from the 4 thermocouples. The temperature measurement portion has yet to be built and tested, but I'll hopefully be able to do that during the evenings in the next few weeks.

 

The main drawback with that design is that it will require the 4 thermocouple cables to come through the firewall into the behind-dash area. I'm not sure how big a problem this is, so if anybody has any input it'd be welcome.

 

The other alternative is to go with 2 units, one in the engine bay and 1 behind the dash, and a single wire through the firewall connecting them for data transfer. I decided against this approach as it will approximately double the cost of the hardware required (i.e 2 microprocessors, 2 cases, 2 power regulator circuits, etc etc).

 

Schematics and photos to come eventually.

 

Costs:

 

Pic16f877 Microcontroller + 32 Leds + 4 transistors + variable resistor (quick and dirty brightness control): $25.85

 

It should be noted that the costs don't include the circuit board cost, or the case cost at this stage.

 

 

[Yenn]

Interesting. The through firewall connections shouldn't be too hard.

 

There is a site for aero electrical work called I think aeroelectrics.com which you may find of use

 

 

[sain]

Thanks for that site Yenn - wish I'd found their DIY section with its audio isolation amplifier before I designed and built my mp3 mixer dohickey, it would have saved me a load of time. Oh well, it was a learning exercise anyway.

 

I was thinking the through firewall connections would be similair to a cars - basically a hole and grommet arangement. I guess the problem with bringing the 4 wires through would be more with the cost of the thermocouple wire - though that looks fairly minimal.

 

The other issue is with every sensor you start adding to the system its another cable that needs to be routed through. I think if I start a third project along the lines of a engine sensor suite (cht, carby temp, oil temp, exaust sensors etc) I'll probably go with the more expensive processor unit in the engine bay, display unit on the dash option. For a project this simple I think its overkill.

 

Anyway, I managed to get some full direct sunlight testing of the LED array done today, and unfortunately it wasn't that crash hot. Basically I had trouble seeing the greens and yellows unless I was looking straight on at them. I'll have to see if this changes once they are in their mounting. For daylight, but not in direct sun, the display was easily visible.

 

I'm not sure how big a problem this is - I don't think it'll be much of one for high wing aircraft, but I suspect it'll be fairly bad for low wing. Can anybody enlighten me as to how often they have their dash lit up by bright sunlight? I'm afraid I just don't have the experience to know. (damn low hours!)

 

 

[Ross]

Hi Sain

 

Another issue is that a significant percentage of males in the general population have a problem with certain colours of the spectrum.

 

I am not familiar with the details but note that some web designers use colours like a pale yellowy colour that is practically invisible to me (of course it could have been a bad 19" VDU at the time which has since been mothballed. It also used to gradually fade some of the colours which was very hard to pick up).

 

Other colours of course can be quite irritating if you have to look at them for a long time.

 

http://www.btplc.com/age_disability/technology/RandD/colours/background.htm

 

I did a Google search for male colour sensitivity that brought up the above site amongst many others that points out that a significant percentage of men are more prone to partial colour blindness than females.

 

It is probably worth while having an awareness of the information if you are designing screens.

 

Regards

 

 

[sain]

Thanks very much Ross I'll check that one out too.

 

Unfortunately in brighter daylight the yellow is quite pale (though to me still visible). I'll have to think of an alternative, or maybe just make it 5 green, 5 red - which would also be a problem for one of the guys at my work. The available colour ranges for LEDs are quite limited, so I'm not sure how far I can go with this display type...

 

As a reference the OLED panel I had been thinking of using is $25 + gst in small quantities (and the market isn't big enough to buy large quantities). Not a huge problem, but there is additional complexity in connecting it to the microprocessors. They can certainly display a wider range of colours, and I could allow for some user customisations in the software (so if I get it wrong it could be changed) to get around vision difficulties. If time allows I'll buy a panel and try it out.

 

I'm afraid I'm a guilty party on the pale yellow background for web pages - to me its a a colour which is very easy on the eyes and not at all painful to read for long periods. In future I'll look up information on colour visibility problems before picking them. Sorry about that.

 

 

My intent here, besides teaching myself, is to provide a low cost, "open source" design that can be built by pretty much anybody, and wont require much in the way of electronics skills (ability to solder will be about it). That said I also intend to sell (through Ian, assuming he's willing to have them in his shop) fully built versions for people who can't be bothered (or don't know how) to build it themselves. Hopefully this will help to fund devlopment of other, more complex, projects down the track. If not it'll just have to wait until I've saved the moolah. Waaaaaaay-the-hell-down-the-track goal is an ADSB transponder, but I've got a hell of a lot of learning to go before/if I ever get there. If anybody has suggestions for those "in between" projects they'd be appreciated.

 

Thanks for your input guys, its been very helpful - please keep it coming!

 

 

[sain]

Well looks like the colour visibility problem doesn't have a particularly easy solution.

 

One option is to use tricolour leds, and drive each colour to a user set level (basically gives a user configurable RGB display). This will be more expensive, as each LED costs $0.50 as opposed to $0.14, giving a cost of $16 for the 32 leds, rather than $4.48. There will also be a number of other components requried, costing roughly $12 (i'd need to work the circuit out to be more specific).

 

The other option I've come up with is to use white leds, covoured by a coloured plastic screen (basically the reverse of the current setup - coloured LEDS and a clear screen). Unfortunately as 3mm white LEDs only seem to come in 1000mcd, 3500mcd or 5000mcd versions there may be some risk of girl friends/wives hijacking the aircraft to use it as a tanning bed (the LEDs i've been using are 15mcd). I suspect that bright sunlight visibility would no longer be a problem. Retina damage on the other hand...

 

 

okay, got a little carried away in the sillyness... the brightness would be quite easy to control (and isn't THAT bright). cost for this option would come out at $43.20 for the 32 leds (doesn't include the coloured plastic) but is technically far easy to implement.

 

My preference at the moment is to provide a slightly modified design to support option one, which would cost a bit more than the standard version, but should still be quite cheap. Option 2 just comes out as too expensive (your looking at $123 for the leds and thermocouples alone, let alone housing, circuit boards, plugs, microcontroller, other dohickeys).

 

 

[Yenn]

Sain.

 

Whatever you do please don't dun wires through the firewall with a rubber grommet. You need to prevent the fire getting through the opening and probably the best way is through a 90 deg pipe bend with fireproof insulation filling the unused portion of the pipe.

 

 

[Ross]

Yenn

 

Are the "rubber" grommets supposed to be "fireproof"?

 

I was under the impression that the grommets used were of a material that would withstand quite a bit of flame similar to the hoses fitted over the fuel & oil lines in the engine bay.

 

The stainless steel firewall can withstand a very high temp w/o melting but it does not stop the heat which could set fire to the wood or fibreglass behind it.

 

Could someone comment that knows please.

 

I did not install the original kit cabin heater in my J160 kit because it was manufactured from fibreglass for the housing and an aluminium flap.

 

The cabin heater for the certified version is made wholly from a metal that might be aluminium. I do not know what the melting point is of that particular material.

 

Regards

 

 

[sain]

Good Info - thanks guys. I thought the ones used in the automotive world were fireproof. Easy enough to test I think.

 

 

[sain]

Not much progress on this for a while due to budget issues (i.e I blew my toy budget for the fortnight on other stuff).

 

The change to the PIC16f877 microprocessor means I needed to upgrade my el-dodgo PIC programmer to support the new chip. I finally got around to buying all the bits and pieces yesterday and I hope to have the programmer built tonight (a dinner party got in the way of doing it last night). Hopefully I'll also have time to do some test programs, and get the display stuff working on the new chip (should be straightforward).

 

In case anybody is interested I ended up using feng3's multipic programmer circuit (http://feng3.cool.ne.jp/en/pg5v2.html), though I'm building it on a breadboard until I'm sure I'm happy with it. Total cost for the parts on this sucker runs at a few bucks (I think $4.80), unless you buy a ZIF socket (which is a good idea) which bumps the cost up to $30ish. The zif socket makes it easy to swap the chips in and out without damaging them, so its handy if your doing a lot of programming.

 

I've also now got a thermocouple to do testing with. Most of this will be to see if I can get away with out using a seperate cold junction compensator chip. This wouldn't normally bother me, but they just arn't available at jaycar or DSE, which is what i'm trying to use as parts suppliers - so anybody can build one of these.

 

Anyway, the theory is that the thermocouple provides a small voltage which is proportional to the voltage difference at the hot junction (the thermocouple bit, or "sparkplug washer dohickey") and the cold junction. For those of you who got to do experiments in science classes back in high scool (or currently in high school if you happen to still be there), it was the experiment where you had a couple of dissimilar metals joined together in a triangle shape (ish). You stuck one end in the bunsen burner flame, one under the tap and measured the voltage with the millivolt meter. And then promptly evacuated the room, because somebody had accidentally set their school bag on fire with the bunsen burner... or their hair, or the desk, or somebody was holding it at the hot end or... damn science was good fun! although probably not so much for the person with their hair on fire...

 

uhh.. got off topic a bit. Anyway, the cold junction compensator basically allows for the fact that the temperature at the point your measuring the output changes a fair bit, and certainly isn't the ideal 0 degrees. So it measures the temperature at that point, and automatically adjusts the thermocouple output to compensate for whatever

 

the temperature happens to be. My plan is to duplicate this functionality with the PIC, measuring the temperature with a diode (cheap but sneaky trick. transistors also work apparantly) and then adjusting the thermocouple reading by that amount. I'm not sure if this will work, so i'll be doing a fair bit of testing.

 

Feel free to correct my theory if I got it completly wrong.

 

Extra bits bought that are actually needed for the gauge:

 

crystal and other components for oscillator for 16f877: $4.23

 

Op-amp and other components for getting thermocouple output to a readable voltage: $2.73

 

Total so far: $32.81

 

 

[sain]

Well the whole programmer build thing did not go well, though entirely due to me rather than the circuit. I built the thing on friday evening, and gave it a test run. Unfortunately it did not work, so I spent a couple of hours or going over the circuit and making sure I hadn't got anything backwards or forgotten a connection somewhere. Unfortunately I didn't find anything wrong, so I tested it again and it still failed. At this point I gave up in disgust and went off to do other things.

 

On monday I had another look at it, and after going over the whole lot again I picked up the breadboard to get a closer look at a transistor (thinking I may have swapped two inadvertantly) and the PIC fell out - i'd forgotten to push the leaver down on the ZIF, and the chip sitting in it doesn't make the connection properly unless its down.

 

 

So I put the PIC back in, pushed the leaver down, gave it a test program and it worked flawlessly. At that point I decided I was too frustrated at my own stupidity to do any more work on it and went to watch scrapheap challenge.

 

Display circuit and program changeover tonight hopefully (to use the new chip), and maybe some thermocouple testing.

 

 

[sain]

Not much progress due to me being a lazy **** recently.

 

Last night I finally did some more work on it, wiring up the diode temperature sensor and writing the program to measure it. The diode will be used to provide a reference temperature which will be used to perform the cold junction offset compensation for the thermocouples. There are a couple of disadvantages to going this way, mainly that it makes the code more complicated than using a dedicated cold junction compensator chip, but it is a fair bit cheaper.

 

Test of the diode temp sensor should be tonight - I got called away to dinner at my gf's parents before I could test it.

 

I'm also awaiting delivery of a couple of samples - a new PIC microcontroller from microchip and a cold junction compensator chip from analog devices. Both should be arriving this week sometime hopefully.

 

The new PIC is the recommended replacement for the 16f877a (which apparantly should no longer be used for new designs). Hopefully I'll be able to make the circuit and code compatible with both devices.

 

The cold junction compensator will be used to confirm that the diode compensation stuff is working as intended, or if it fails dismally as part of the circuit itself.

 

I've also been having a few issues with the op-amp amplifier circuit for the thermocouple (to bring the output up to something the PIC can read). I think the issue is just that the room temp voltage output is in the noise threshold for the op-amp i was using. I'll try a different one soon.

 

Extra-bits to be used in the project:

 

1x 4k7 resistor, 1 x 1n4148 small signal diode: $0.70

 

Total so far: $33.51

 

 

[sain]

The diode thermometer worked suprisingly well (first attempt even), and once calibrated was accurate to approx +-2deg C, which I think will be "close enough" for the cold junction calcs, especially as the inaccuracy is consistant.

 

For those who want to build a simple and really cheap digital thermometer (with reasonableish accuracy) the site I used to get the idea for is here http://www.micro-examples.com/public/microex-navig/doc/098-temperature-sensor

 

the sensor itself is consists of a 4k7 resistor connected between +5v and a 1n4148 diode, with the cathode (striped) end of the diode being connected to GND. The junction between the resistor and the diode is then connected to one of the AN pins on a PIC microcontroller (schematic to follow)

 

The code for the PIC was a partial cut and paste job from the site listed above (the sensor read and conversion to deg C portions), and was then just a case of displaying the result on my already built test display. I chose to display it as a binary number, which was a simple as outputting the number to the display port, and turning on one of the display channels. I also added code to log the first 16 reads to the PICs internal eeprom (to be read later by the computer).

 

Testing was done by placing a k-type thermocouple (connected to my multimeter set to display deg C) next to the diode and then reading the temps on both. The initial test run (no calibration) had the diode sensor showing consistantly 5 deg lower.

 

I'm planning on doing some further work on the calibration code tonight, with the calibration changing the results of the deg C conversion, rather than the initially read value - I believe this will make the calibration easier and quicker. If that works I'll then test storing the calibration value in the eeprom and loading this at startup. If that works I'll be good to move on to the thermocouple sensor portion..

 

 

[sain]

schematic for the diode temp sensor

 

Here is the schematic for the diode based temp sensor i'll hopefully be using to do the cold junction offset calcs.

 

[ATTACH]5425.vB[/ATTACH]

 

 

 

[sain]

Last night's testing turned into an absolute shocker. The initial test using a post conversion offset worked fine, but the subsequent power off and back on testings (checking that the calibration remained valid after power off) didn't work as expected.

 

I then reverted to the pre-conversion offset for calibration method and changed the code to store and load the calibrated value. This somehow resulted in the calibrated value being changed continually (lots of pretty lights on the display in other words).

 

Unfortunately when I pulled the PIC and stuck it back in the programmer I broke the programmer, which took me a good while to realise, and several code revisions.

 

A couple of hours later I had it all working again, and stable code (without the store and save bit, or the calibration adjustment code). I then added my new and improved calibration code (without the store and save) and it all went horribly wrong again. At this point I gave up and went to bed.

 

I'll give it another crack tonight.

 

 

[sain]

The upgraded microcontroller (a PIC16f887) sample has arrived, and I've decided to use it for the project instead of the PIC16f877a I've been testing with currently. The new controller is largely backwards compatible with the older one, but includes a handy internal oscillator (meaning the crystal and associated capacitor arn't needed). Anyway, hopefully it'll work fine with minimal changes. I was able to get the current display code working without any trouble, and I'll be testing the diode temp sensor tonight.

 

I've been thinking more about the design for the thermocouple to pic interface. I think i'm just going to go with a quad op amp, with each op-amp output driving a different A/D channel on the pic - it has plenty to spare, so why not - this will be about the cheapest way to do it I think (assuming it works). It'll also make expanding to an 8 channel version very simple (just add another op-amp).

 

At this stage I've also been thinking about what features to add to the system, besides just displaying the CHT temp. I would also welcome thoughts from others on what would be considered must-haves, should-haves, and would-be-nice-to-have. Should and would be nice options will be considered, and added if they don't increase the cost/difficulty of the unit to far. Must haves will be done if its achievable.

 

This is what I have so far:

 

Must Haves:

 

Visual Alert approaching over temp. - i.e channel flashes until cleared. (mostly done)

 

Visual Alert on over temp. - i.e channel double flashes until cleared. (mostly done)

 

Variable display brightness (already done).

 

Should Have:

 

Audible alert on approaching over temp and over temp (i.e intermittent beeps until cleared)

 

Would be Nice to have:

 

Voice alert on approaching over temp and over temp (i.e speaks until cleared "CHT warning" or "CHT Alert")

 

Data logging facility to SD Card or MMC card.

 

Visual + audible alert on power supply problems (i.e warning when supplied power begins to drop - indicates charging/alternator fail)

 

Option for user-definable colour settings for LEDs - for colour blind support.

 

Suggestions welcome - even on prioritisation.

 

 

[sain]

Some progress over the weekend. The diode temp sensor is now working quite well, and I think is "close enough" to be useable for the thermocouple temp offset.

 

The new 16F887 processor also worked very well, though I did need to upgrade my programmer sofware (from ic-prog to picpgm). The new software is much faster, as it only writes and verifies the data you actually use, instead of the full memory range. Certainly cuts down the time spent waiting for the chip to be programmed to a couple of seconds. The processor is pretty much compatible with the old 16f877a, so hopefully only minimal changes will be required if people don't want to use the new processor. Certainly the code i've written so far runs fine on both.

 

From here I move on to the thermocouple to pic interface, which will be the therd last part of the project (the second being the power supply and final will be the circuit board and caseing). At this stage I'm looking at using a quad op-amp (tl074 - cost $2), with one portion of the op-amp in use per thermocouple. Each ouput will feed directly to a ADC port on the PIC microcontroller. Initial simulations last night looked promising, so I'll move on to tests tonight.

 

Power supply should be fairly simple too, basically it'll just be a simple rectifier+capacitors+regulator and should run to about $5-10.

 

 

[sain]

The thermocouple interface to the PIC went very well last night. It was about the first time i've managed to design what I thought the circuit should look like, build it, and have it work straight off. Got to be happy with that.

 

The thermocouples connect to a quad op-amp, in this case a tl074, configured as a non-inverting amplifier with 100x gain. or in other words you hook the thermocouple to an input on the op amp, and two resistors (100k and 1k) between the other input, ouput and gnd. Each output of the op-amp will then be connected to an A/D (analog to digital converter) port on the pic.

 

Next stage is to write the software to handle the conversion to degC, cold junction compensation (from the diode sensor) and display the output.

 

Unfortunately i'll be very busy due to work for the next week or so and wont get a chance to do so.

 

 

[sain]

I'm back from the trips away for work (at least for a couple of weeks) and so I'm back into it.

 

The power supply portion is now done, but I need to do a quick calculation on load (in case I picked the wrong regulator) before I post the pricing. The power supply is just 4 1n4001 diodes (configured as a bridge rectifier), a couple of electrolytic capacitors and a LM7805 (TO92 package, 100mA max) regulator. Current draw should be fine, i'm just not entirely sure about the how much current the thermocouple op-amp is drawing.

 

Next step will be to put it all the portions together and test that, followed by the circuit board design and build, then the final step of putting it in a case.

 

I've been meaning to make a circuit board for the feng3 pic programmer for a while (its on a prototyping breadboard like everything else at the moment) so I spent the last couple of nights stuffing around with that. Unfortunately its been quite a while since i've done this stuff, and my old favourite method is causing some problems. I print the circuit board pattern onto an overhead transparency, then coat a blank copper board with positive photoresist spray and stick them both under a UV lamp for a while, develop and then etch. However the photoresist spray that I have expired in 1999 (it really has been a while) and is not giving usable results. None of the local electronics stores seem to stock photoresist spray either, so i may have to find a different method.

 

 

[sain]

I spent the last week and a bit trialling different methods of PCB (printed circit board) manufacturing, without much success, right up until I caved and tried the very expensive (at least if you buy it from jaycar) press-n-peel method.

 

This worked fantastically, with the resist-coated board only needing a touch up in one corner where I didn't get the iron onto it well. After being etched the finished board was very high quality (at least until I picked it up with the circuit board laquer still wet and left a finger print on it). If anybody else is making some prototype boards then I recommend this method - soooo much easier than anything I've done before, and far less mucking around with nasty chemicals.

 

Anyway, with the feng3 programmer now on a PCB and working properly (first go - woo!) I'll be moving on to getting the code portion of the thermocouple reader sorted out, as well as designing the PCBs for the display and main boards. The display board is looking like a dual layer board (because its easier that way and i'm lazy), and the main board a single layer.

 

Hopefully this week I'll also be putting in an order for 4 CHT senders. Anybody got any recommendations for a cheap supplier?

 

*edit*

 

oh yeah, if your going to use the press n peel stuff buy online direct from the manufacturer - much cheeper.

 

 

[sain]

I've been busy with work for the last few weeks, so there hasn't been much in the way of progress.

 

I have been gradually sorting out the microcontrollers program, and while going over the "preset calibrations" portion of the code I finally figured out that the +-2 degrees accuracy of my diode temp sensor doesn't really cut the mustard, as the operating range of some of the engines (rotax in particular) isn't all that large.

 

In order to have better accuracy, i'll use a lm335 ($3.30 at DSE or Jaycar) temp sensor. Fortunately the circuit doesn't change all that much from the diode temp sensor (I think, anyway), it'll just require one additional trimpot (for the initial calibration).

 

Hopefully I'll have some progress tonight.

 

 

[sain]

Again, not much progress. I ended up ordering the cht senders i wanted directly from wicks aircraft (sorry Ian), and they finally arrived last night.

 

Unfortunately the leads on them are very short, so now i need some thermocouple wire too. It shouldn't hold me back from doing some testing on the household iron though (c:

 

The test platform has been selected, and its that well known aircraft - a ford laser. still, it'll do until i've ironed out the bugs and its ready to go in a real aircraft. As soon as i've got the thermocouple wire issue sorted it'll go into the car.

 

I'm working on the initial prototype schematic and pcb at the moment, which I'll post when its done. Should be sometime this week hopefully.

 

 

[sain]

well, its been a while since i've worked on this - sorry to anybody waiting for it, i've been a touch busy.

 

Anyway, the circuit design is pretty much complete (I think), so I'm posting the schematic in case for comment etc.

 

Please note that this has not been tested, so if you build it, use at your own risk. That said it wont do anything at all until you have the program to load onto the PIC microcontroller. Code will follow when I have it working as well as I would like.

 

Circuit explanation (for the curious) will be in the next post.

 

 

 

[sain]

Circuit Explanation

 

The top part of the circuit, comprising the 4 diodes, the two capacitors and the 78L05 voltage regulator are the power supply for the rest of the system. The 4 diodes are configured as a bridge rectifier, so that the gauge will be immune to somebody hooking it up to the power the wrong way around. The two capacitors are just providing filtering for the 7805 regulator, to smooth out any supply/demand ripples.

 

The second part of the circuit is R1 and the LM335 (the weird 2 circle looking thing). This is the temperature sensor that will be reading the temperature where the thermocouples connect to the circuit. This is so the PIC can use the temperature there to do a cold junction offset calculation, and get an accurate reading on the CHTs. Cold junctions for thermocouples are normally at 0 degrees C, so if its not you need to allow for the temperature difference. The output from the LM335 is connected to AN0 (the first Analog to Digital converter on the PIC).

 

The third part of the circuit is the 4 op amps (actually 1 TL074 chip) to the left of the PIC, which take their input from the 4 thermocouples (the other side of the thermocouple is connected to circuit gnd). These op-amps are configured to provide a voltage gain of 243 (roughly), which is enough to bump the output of the thermocouple up to about 1V at 100 degrees C (and 2V at 200 C etc). The 4 op-amps have their output connected to AN1-4, which are Analog to Digital converter pins for the PIC.

 

The big boxy thing in the middle is the PIC16F887, which is the microcontroller which will do the temperature reading from the LM335 and the thermocouples (via the op-amps). It will then display the output (scaled to predefined settings) on the LEDs connected to RB0-7. These will start switching on when the CHTs get up to operating temperature, with the 6th (the last green) coming on a bit before the top of the engine's operating range. The first red (7th) will come on right at the top of the operating range. If the operating range is exceeded the whole bank will flash on and off.

 

Each bank of LEDs (on the right hand side) is controlled by a transistor, which is switched on by the PIC when its displaying that particular bank. Each bank is switched on and off very rapidly in turn (thousands of times a second), which gives the appearance of it being on continually. This allows the PIC to control the 32 LEDs with only 12 pins. The LEDs are connected in turn through a variable resistor, which is used to control the brightness.

 

The remaining pins on the PIC are left empty for future expansion/features (such as additional display channels, a SD or MMC card for logging etc). Note that in its current state, the PIC could be replaced with a smaller version, such as a PIC16f886.

 

Corrections to schematic:

 

while writing this i noticed that IC2D (the 4th, or bottom right op-amp) is connected to RA4 - this should be RA5, which is the next pin down.

 

 

[jcamp]

Interesting project, looking at something similar

 

Couple of points on the circuit:

 

Power Supply:

 

LM78xx normally have 35v max. This can be problematic with a voltage surge on shutdown, would be inclined to make the cap say 63v and stick a 20v zener in there.

 

Op amps

 

A DC gain of 273 might be stretching it, I would be inclined to use an instrumentation amp in there to avoid hassles like device selection. Probably doesn't really matter what, IIRC last time there were some INA 118's floating around and they worked OK.

 

(Not an electronic engineer but employ some.)

 

Also may have to go to shielded stuff on the thermocouple wire if there are too many problems with a noisy environment.