As per Bob/Forrest/Joe - MAXIMUM recommended input Z is 10K. So you can use 10k to ground and whatever is needed to input and end up slightly under 10k. (Z ~+ Rin x 10k/(Rin + 10k)) As Joe says, one factor is leakage current http://ww1.microchip.com/downloads/en/DeviceDoc/41291F.pdf fig 8.8 page 90 shows ADC block diagram. 500 nA nominal leakage in 10k gives 500E-9 x E4 = 5 mV offset or about 1 bit at 5V full range with 10 bit converter. YCMV. Table 17.5, page 251 shows leakage currents of 0.1 uA typical and +/- 1 uA max so the figures above could be 2 bits worst case and 0.1 bit typical across full rated temperature range. ie at a typical temperature range a typical IC would typically be well under 1 bit error in 10 bits with 10k input. ADC discussion around page 100 gives acceptable timings - without digging further 1.6 uS looks like minimum ADC cycle time. If you cycled ADC REALLY fast you'd start to get ADC holding capacitor charge time issues if you exceeded 10k input Z but odds are this will be secondary to bias offset as above in real cases. As Joe says - try it and see. Take a power supply of known voltage, connect a say 100k pot in series to ADS pin and measure voltage with PIC. Try this at say 1 25 50 75 99 % of full scale voltage. That plus a range of Zins and you'd get a fair feel of how it behaved. BUT - 10k to ground + any input R should be OK :-) Russell McMahon On 14 June 2010 12:33, Forrest W Christian wrote: > I have a space and cost contrained application where I'm using a > 16f886. Among other functions, I need to be able to read the 'raw' > input voltage to the circuit. Today I do this with a 10K/1K 1% voltage > divider into an ADC pin (with appropriate protection diodes). With the > 3.3V Vdd I run the circuit at, this gives me something like 36V max vin, > and a maximum power consumption of just under 1/8 watt 'burnt' across > both resistors. Not ideal, but it works. > > I am now having to modify this design to operate on up to just under > 60VDC. This means I have to adjust a voltage divider. using 10K and > 560 ohms would put me in the right range... but because of the E*E part > of ohms law, I'm now burning just over 1/3W across the resistors... not > acceptable. > > In an ideal world, I'd just do a 100K and 5.6K... unfortunately the > 'minimum' input impedance is 10K on the 'f886. > > I would prefer to avoid adding a opamp or other buffer. I remember some > discussion that you can actually get away with violating the input > impedance spec, especially if you add a input capacitor. Since this is > a DC source, can I just do the 100K and 5.6K and say put a .1uF cap on > the input pin? What am I traiding off. Other ideas? > > And no, I don't have the budget/board real-estate for a opamp... even > though that's probably the right answer. > > -forrest > -- > http://www.piclist.com PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist