Electrinically Challenged! Need advice!

I am excited to have discovered this forum/website but I am a bit overwhelmed at the variety of products. I have a project I am trying to complete that requires a microprocessor to monitor two analog inputs and compares them. Then outputs the higher of the two. The inputs are millivolt sources that vary between 3.8 and 40mV. I need to resolve .25mV steps. If my calculations are correct I will need at least 12bit resolution to do this.

Since this is a one off or at least very low quantity project, I think a suitable reference board is the way to go rather than have to contract a board design to handle power and signal conditioning.

In addition to the comparison function, I would like to incorporate a memory and replay function to replay a minute or two of signal acquisition and have an LED that indicates which channel is being displayed.

Can anyone recommend a product for me to evaluate that would meet these goals? I am continuing to browse for possibilities but could really use a suggestion or two.

Thank you,

KKrofft

You need an op amp to amplify the signal for input to an ADC. 12 bit ADCs aren’t very common - the PIC18F6393 has one. Careful PCB design will be required if you want all 12 bits. Most MCUs with 12 bit ADCs, like the aforementioned one, have a lot of other stuff on them that you don’t need, it might be cheaper to interface a 12-bit ADC to a simpler device.

Leon

You are using only 145 steps so even 8-bit ADC will suffice.

I remember now… 8 bit will work but I was hoping to have a little headroom in case I wanted to try a finer resolution. If it comes down to finding an inexpensive development board that has 8-bit AD and DA over having to create a special board to do 12 bit I will start out that route.

Recommendations?

Most PICs today have 10-bit ADCs so that should give you about 7x headroom (.036mV steps).

A PicKIT2 Starter Kit ($50 including programmer, 20% discount may be available) has a PIC16F690 with 10-bit ADC.

http://www.microchip.com/stellent/idcpl … e=en023805

You say you “would like” to incorporate memory replay. Does that mean it’s not a hard requirement and just a “nice to have?”

If so, then for the first part you don’t even need a processor: you just need a precision comparator and an analog switch. It could be built for less than $1.00 in parts.

In any event, if you need to maintain 0.25mV resolution, this is not going to be trivial: circuit board layout will be critical. Since it’s a one-off I’d recommend building the analog portion of the circuit in dead-bug fashion. You can generally get excellent signal quality that way.

Good insight. I considered the comparator approach but would like the added functionality of the memory replay. have the poor background in electronics that I do, I am stuck with having to do lots of research to devise the simplest of circuits but with help from a few friends I usually do OK.

The application is for combining output from a pair of EGT probes on a car. due to space constraints I would like to use just one gauge and have it read the hotter of the two banks when in default mode. It would be desirable to

be able to force it to read the cooler bank manually as a sanity check that both sensors are functional.

The sensors are K thermocouples and as such output a mV signal. In my simple concept I made the assumption that it would not be too difficult to simply compare the inputs and duplicate the output with a micro to drive the gauge. The gauge accepts input directly from the TC so as long as the output matches the input, it should read correctly. The gauge I am using is divided into 25 degree segments so if it resolves 12.5 degrees it should be adequate.

As for the replay function, during a spirited bit of driving, it is not a good idea to be staring at the gauge so it would be nice to “replay” the last 2 minutes or so to look for spikes.

Lastly, can you explain “dead bug fashion”?

Thank you,

KK

Dead Bug prototyping is when you create circuit with all the components lying on their backs with their pins facing upward (like a dead bug) .

Why does the orientation of the components improve the signal quality?

kkrofft:
Why does the orientation of the components improve the signal quality?

It’s not usually the orientation that matters, it’s how short the interconnects are to various components… :wink:

(making it less likely for the circuit to pick up external interference or adding any unnessersary resistance in PCB traces)

Thank you for the explanation… Makes sense now.

I prefer “live-bug” construction - shorter connections to the ground plane, the chip names are visible and the connections don’t have to be reversed.

Leon

So do I, but it’s easier to say dead bug since more people are familiar with the term :slight_smile:

I normally will try to straighten out the DIP pins horizontally as much as I can and support the chip by soldering its ground pin to the board. That way there is easy access to the pins and you can read the chip names like you said. I’m less likely to make a mistake than when the chip’s flipped on its back.

kkroft:

You have a bit of an uphill battle, but it can be a fun one. Low level signal instrumentation is an interesting challenge in my book. My favorite reference is an Analog Devices publication called “An I.C. Amplifier User’s Guide to Decoupling, Grounding and Making Things Go Right for a Change. (AN-202)” I have the dead-tree version, but you should be able to find it on their site by searching for AN-202. Also this page has some useful information: http://www.analog.com/library/analogDia … ry/12.html

Enjoy :slight_smile:

leon_heller:
I prefer “live-bug” construction - shorter connections to the ground plane, the chip names are visible and the connections don’t have to be reversed.

Leon