I assume it is the bottom circuit that is the one giving you grief. You appear to have a problem with (amongst other things) input bias offset. There will be likely to be some bias current into/outof the input pins of the opamp (unless it;s a fet input type - I haven't checked). So the input resistance seen by the input pins needs to be the same for both the positive & neg inputs. The fix is to simply connect a 50K (or 47k) resistor (calculated from the two 100k resistors on the negative input - effectively in parallel) in series with the positive input to pin 10. This will balance the offset voltage created by the input bias current. If you have significant source impedance in the circuit feeding the positive input, this will need to be accounted for by decreasing the added resistance appropriately. Note: Even if you run a negative supply rail you are likely to get the same effect for low input voltages. However I would reccommend that you try a negative rail also - even if you just patch one up temporarily using a battery. There may be a way you can derive a negative rail from your power supply by using a series diode or two. You shouldn't need much negative volts but most "rail-rail Input" opamps get a bit marginal once they get very close to the rails. Richard P On 07/05/07, Rolf wrote: > This time with the attachment.... ;-) > > Sorry. It really was in the first mail, bust have been because it was > pasted in rather than attached. > > Rolf > > Rolf wrote: > > Sean Breheny wrote: > > > >> Hi Rolf, > >> > >> Could you either post or send to me directly your schematic? It sounds > >> like what you are trying to do should work properly. I've never used > >> the MCP6044 but I took a quick look at the datasheet and it looks > >> fine. > >> > >> For future reference, my favorite op-amp for general purpose stuff is > >> the LMC648x (6482 is a dual, 6484 is a quad) from National. It is not > >> cheap (about $5) so it is not a good choice for most inexpensive > >> commercial applications, but for hobbyist one-offs it can't be beat. > >> It is rail to rail input and output, ultra-low input bias current, max > >> 1mV input offset voltage, pretty low noise, 1 MHz gain-bandwidth > >> product. The MCP6044 seems to trade off GBW product for supply current > >> (it has about 1000 times less supply current than the LMC6482 but also > >> almost 100 times slower). > >> > >> Sean > >> > >> > >> On 5/5/07, Rolf wrote: > >> > >> > > > > > > The top is a simple voltage follower. In my case I am getting erroneous > > output at low input voltages. Specifically, the charger is currently > > charging at 12.5mA as measured with one DMM. Another DMM measures the > > output of the charging chip at 12.7mV (pretty close...). The same DMM > > that measures 12.7mV on the charge chip output (input to the follower), > > reads 1.5mV on the op-amp output. It should be 12.7!!! If I put a load > > on the op-amo out (255R to ground), the opamp out measures 2.1mV. Still > > very wrong. > > > > In the second circuit, I tried to set up a "difference" circuit. The > > theory is that the output should be (4.5V - input) * (100K/200K). Or, an > > "Amplified difference" with a 2x gain. I was figuring I could correct > > the inversion using math on the A/D module. > > > > The problem with the second solution is that the output of the opamp is > > 'pushing' back against the input signal. Currently the charge signal is > > 12.3mV (I type slowly, it has charged a little more). But, when I > > connect it to the op-amp circuit, it rises to 20.2mV, and the out put of > > the opamp, which I expect to be about 24.6mV less than 4.5V, is actually > > 2.36V. So, I am confused ;-) > > > > From the charger's datasheet they make it clear not to load the signal > > in any way.... and the second circuit obviously messes with it. > > > > My original circuit was circuit 3. A simple gain amplifier with gain of > > 2. This woks fine for larger values of the input signal. > > > > I am investigating other options and Jinx's negative rail (but I will > > have to use a different op-amp I think). > > > > Rolf > > > > > -- > 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