I have a project on which I used two mcp617's for the purpose of buffering a signal. Inputs An1, An2, and An4 work as expected, but An3(J7 connector) does not. An incoming signal level of 4.4V on pin 3 of U22(tested with 3.3v as well) is outputting 2.8V on pins 1-2. I have tested this across numerous different boards and chips with the same result. I also have double checked that there is no load on the output

At this point I am unable to find the difference in layout or schematic between these circuits. The MCP617 is listed as unity gain stable as well.

MUCH larger version of circuit diagram here - Imgur saves uploaded files at larger sizes than it displays.

Update: After going through all boards, it seems this is indeed happening to all devices somewhat randomly, probably due to running these outside their characterized parameters.

Schematic enter image description here

  • \$\begingroup\$ The schematic show a power supply of 3.3V to the opamps, is this correct? \$\endgroup\$ Jul 19 '14 at 12:55
  • \$\begingroup\$ Yes, the supply is 3.3v. The adc input is 3.3v. \$\endgroup\$ Jul 19 '14 at 13:18
  • \$\begingroup\$ As well, the datasheet spec's a max of 2ma, if input higher than 3.3, so 13V input = max possible of .13ma(13/100k) \$\endgroup\$ Jul 19 '14 at 13:19
  • 1
    \$\begingroup\$ Grin*, Well you have to stay within all the maximum ratings, not just a subset. So Vin can only go 0.3 V above or below your supply rails, (I have no idea how you were getting 4.4V out of the other amps.. poor babies.) (And as @WhatRoughBeast observes the common mode input range only goes to Vdd-0.9.) *I grin because I feel your pain, I learn the most from my own electronics blunders. \$\endgroup\$ Jul 19 '14 at 13:39

The problem is the CM input voltage range- for a voltage follower you need rail-to-rail input and output if you want the output to be able to swing rail-to-rail.

enter image description here

Note that the maximum input voltage of Vdd-0.9V (for functionality) is only guaranteed at 25°C, so caution would suggest you allow more headroom that than.

Also, you should not assume that under the conditions allowed in the datasheet (less than 2mA flowing into the protection network) that the other op-amp in the pair is functioning normally. That may not be a problem for you, but it should be considered.

  • 1
    \$\begingroup\$ I agree with this assessment as well, I will be trying a different op amp here. \$\endgroup\$ Jul 19 '14 at 17:58

The opamp is only rail-to-rail for outputs. It is not rail-to-rail for inputs. Its' common-mode input range is Vdd - .9

Your schematic is completely unreadable, so I'm guessing you have a 5-volt supply. The results of overdriving the input are unpredictable. Why you would get the same result with 3.3 volts I don't know, but the fact that you seem to get adequate results from the other chips (and I suggest you check their accuracy for 4.4 volt inputs very carefully) suggests that your problem child is just a weak sister.

Note, for instance, that the data sheet figure for a gain of 1, large pulse response, used an input level of 3.9 volts in order to keep within CMR limits.

  • \$\begingroup\$ He's actually powering the circuit with 3V3 that makes things a little stranger. \$\endgroup\$ Jul 19 '14 at 13:11
  • \$\begingroup\$ Not sure whats unreadable about 3.3V? See Vdd pins. No, its not a 5v supply. \$\endgroup\$ Jul 19 '14 at 13:24
  • \$\begingroup\$ Your answer is good as well, thanks for this input. \$\endgroup\$ Jul 19 '14 at 17:59
  • \$\begingroup\$ I have always been confused by this terminology. What exactly do you mean by "common-mode" input range? \$\endgroup\$
    – sherrellbc
    Mar 1 '15 at 21:43
  • \$\begingroup\$ An op amp, of course, has 2 inputs. The difference between them (the midpoint, if you will) is called the common mode voltage. For an inverting configuration, this is normally held at ground or a virtual ground. For a non-inverting configuration the inputs can swing around. The output voltage is fed back to the inputs and tries to keep them very close together, but this only works for a certain range of voltages, and outside this range the input circuits start having problems. This voltage range is called the common mode input range. \$\endgroup\$ Mar 1 '15 at 22:09

Your schematic shows the op-amps being powered from a 3.3V supply. The most you'll get at the op-amp outputs is 0 to 3.3V. (Even the data sheet confirms this on line 3 of the features list - "Rail-to-Rail Swing at Output").

With the op-amp configured as a unity gain block this means that the input cannot range outside of 0 to 3.3V.

If you try to drive the inputs above the power supply level you may damage the parts. In the same data sheet they specify that the maximum rated input is at VDD + 0.3V. That is not a recommended operating condition but one above which you could cause serious damage to the op-amp.

It is possible that the internal design of the op-amp is such that inputs above the supply rail may try to act as a power source for the circuit and cause the outputs to do strange things. This could include clamping to a lower level output or to rise above VDD as well.

  • \$\begingroup\$ I am quite aware I can't get more than 3.3. I want 3.3 when the input is more than that. \$\endgroup\$ Jul 19 '14 at 13:22
  • \$\begingroup\$ Damage the parts? Whatever, see datasheet, it is well under the 2ma they spec. \$\endgroup\$ Jul 19 '14 at 13:22
  • 1
    \$\begingroup\$ @ErikFriesen - You cannot pick and choose which of the Maximum Ratings you want to abide by. For specified behavior you need to stay within the bounds of ALL of the Maximum Ratings - Including inputs within the range of VSS-0.3V to VDD+0.3V. \$\endgroup\$ Jul 19 '14 at 13:29
  • \$\begingroup\$ Page 11 datasheet. \$\endgroup\$ Jul 19 '14 at 13:31
  • \$\begingroup\$ Sounds like you have something to learn here. The resistor in series with the input is simply being put there to prevent serious damage to the part. However for the part to be operational within specifications you MUST abide by the conditions specified in the first two sentences of Page 11 - which is exactly what I and others here have said. \$\endgroup\$ Jul 19 '14 at 13:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.