ADC circuit help " is this possible "

[cncmachineguy]

Simple answer: If full scale =0 - 3.3V, and the adc is 10 bit, LSB = 3.3/1024 = 3.22265mV. I would think you should plan for 1LSB error so you could detect ~6mV worst case. Now there are plenty of debate around here about 1024 vs 1023, but I live in 1024 world

[Charles Linquis]

Even though the A/D is 10 bits, you won't get a lot of accuracy unless you have a dedicated Vref and you average a lot of readings. You might get 9 bits of accuracy in that case.

You can do a lot better if you amplify the signal with a rail-to-rail input / rail-to-rail output op amp.

If you don't want to go that route, you are going to have to get a bigger voltage change than you will get with your setup. First, R1 will have to be smaller than 10K. I don't know how many "ematches" you plan to have, but 330 ohms would seem to be a good value. Of course, the resistor would have to be a 5W type. This will give you 72mA through the resistor, and .218V across 3 ohms. You will also need a 2K 1/2W resistor between the junction of R1 and R2 and the PIC input to protect it when the switches are open. That should give you enough signal to work with.

[cncmachineguy]

Charles Are you saying it won't detect 6mV changes?

[Charles Linquis]

I'm saying that it won't detect 6mV changes reliably. If you can't tolerate the current draw of a smaller R1, you should use an op amp with a gain of 10 or so.
You CAN detect 60mV changes reliably.

[HankMcSpank]

I'm with charles...what made you choose 10k for the upper resistor?<br>
<br>
If you fed say a fixed voltage 1.024V to the top of the resistor chain, and brought R1 down top something more appropriate like 100 Ohms. (the chain would be a draw about 10mA so would need external power)<br>
<br>
If you then go with a PIC12lf1822...you can set its ADC poistive reference to its internal vref of 1.024V (even if it's supply is 3.3V)<br>
<br>
if you used 10 bits, this in theory would yield 1mV resolution (can't predict how noise will start impacting here though!) ....which ought to be sufficient.<br><br>Another idea would be to have the 3 ohm device turn on/off a mosfet, which IN TURN&nbsp;then kicKs in higher value resistors in your divider chain ...which would make the whole thing a lot easier.<br><br>Without seeing more details, I'm not sure ADC is the solution ....can you ground these 3 ohm devices so they could for example&nbsp;pull a PIC IO pin down to ground - this would be a far more robust method of establishing which device is on or off....simply&nbsp;use an&nbsp;IOC interrupt.<br><br>

[pyrogeek]

well, i picked the 10k for a couples reasons -

1. i wanted the least amount of current going thru the circuit, there is 10 different circuits that will be tested. Also the device will be powered by a 4 AA battery pack most of the time.

2. you can only put so much current thru an E-match before you have a violent reaction, , NFPA 1123 regulates me to 30ma.
But 30ma will kill some AA batts pretty quick.

I was thinking about using a 1k, but then i would want the device to be able to have some way of connecting each circuit one at a time for test and then grounding it while it's not tested, i was thinking of using a shift reg, or maybe even using 10 pins on the pic. I was thinking that this would help with not wasting battery power and maybe even keep th RF interference down from the MRF89XAM9A-I/RM transceiver.