minimum input voltage on TLV2472

Simon,

TLV2472 has a rail-to-rail input common-mode range so the problem you see must be caused by your circuit configuration;  for example, if you use the part in a follower configuration, the circuit may in fact stop working properly when the input voltage is less than 0.5V above the negative rail, not because of the input range problem but because you may be exceeding the low-level output voltage (see below). 

If you have additional questions, please provide the full schematic including the supplies/input voltages used in your application.

Well, I don't think that this is the problem. The scematic used is the one below (without the current generator, that is a mistake). As you can see there is a gain between the minus input and the output so the voltage should not go below the low level. Do you have any other idea?

Thank you

Simon,

You say you don't use the current input source shown in your schematic but then what is the actual VSRC input voltage?  If it is 5V then this is exactly the problem with your circuit that you drive the output into rail - just the positive one.  You have an attenuator of 1.2k/7.4k (0.162162) at the input and then a gain of 7.4k/1.2k (6.166667) at the output so that Vout/VSRC = 1;  R6 place no role in the overall gain and it acts only as a resistive load so I do not know the reason for including it.  If you use VSRC=2.5V, the output should follow at 2.5V and there should be no problem.

Yes the VSRC Voltage is at 2.5V. The current generator is actually in parallel with R12 so that we can have a voltage depending on the current.

The problem is, when ISRC=0 (whatever the vaue in fact) we have theorically V+= 2.5*1.2/7.4=0.4V.

This value should be found on V- too, but we read 0.1V. After some test we found that if the value in the +input of the amplifier is below 0.5V we don't have the same on the -input (the amplifier is not working). And I really don't know why.

We are thinking of reducing the atenuator to have a greater value but in this case we lose in precision for the current reading.

If your VSRC is in fact 2.5V and the circuit loses its virtual short between the input terminals, one other possibility is that the supply voltage is 2.5V and not 5V as shown on the schematic.  Also, perhaps you should take TLV2472 out of its current configuration and test it as a follower sweeping its input voltage across the entire supply range while monitoring the output voltage.