Post Go back to editing

LTC2309 : ADC analog inputs sinking current!!

I'm using the  ADC LTC2309 to convert analog signals coming from an opam. The analog voltage to measure ranges from 2 to 3.5 V. The signal has very low fluctuations so we can say that it is almost a DC signal.

In one of my experiments I was measuring 2.4V at input 7  and suddenly  the voltage went up by 0.7 V. Hence the measured voltage became 3.1V. I verified this value either by collecting the data via I2C or by measuring directly on  this input.

I thought it was the opam that was giving the wrong signal. Verifying the circuit, it was effectively the ADC pin that was giving the wrong signal.

The connection of components going from the opam output to the ADC input is a 100 Ohm resistor  with a 47pF capacitor from the ADC input  to ground, as recommended in the last figure in the datasheet.

The Vref pin is connected to ground while the Refcomp pin is connected to +5V. I retreive data from the ADC input via I2C using the command F8 that addresses the 7th channel in unipolar mode and seting the ADC in nap mode. Then leaving the ADC in sleep mode, the result is the same.

Observing this, I realize that the ADC input is sinking current which explains the increase of measured voltage at its input.

Another test that I made was removing the 100 Ohm resistor and connecting the output of the opam directly to the ADC input.

The read value was oscillating around  2.55V . Another observation is that the same value applied on input 8 is also present in other pins of the ADC which are not connected.

The question that arises is why the ADC would start sinking current with no apparent reason. ?

I will really appreciate any hint or advices you can provide.



  • You mention that you have unconnected inputs. This is never a good idea with CMOS circuits. Try connecting all unused analog inputs to ground and see if that solves the problem. If that doesn't solve the problem please provide a schematic of your circuit and I will see what else I can suggest.

  • Hello mr Hoover,

    thank you for your feedback.

    Circuit Configuration until this morning

    Indeed I have unconnected pins. Some other pins are bridged to the input signal and the rest of them are directly connected to ground, as can be observed in the schematics of the figure "current version".

    Whenever the pin connected to the input resistance is not working, I remove the bridge and leave the pin disconnected. At today only  3 or 4 pins provide a correct reading.

    I also remarked that the I2C communication does not work . Do you think this is also produced by leaving unconnected pins?.

    Configuration after setting unconnected pins to ground

    I proceeded to set the unconnected pins to ground, as you suggested, and see if the problem of not correct readings was solved.

    I placed a 100 Ohms resistor between the input pins and ground and left only one pin connected to the input signal.

    The measured voltage in this pin is so far correct. Also the I2C starts working again although I don't know for how long. Nevertheless I'm not sure if the I2C stopped working in the previous configuration because of unconnected inputs.

    Regarding this subject, in the figure below I show a screen shot of the I2C clock signal that is applied to the  LTC2309. The low level of this clock signal is around -0.8V while the high level around 5V. This levels are provided by the I2C isolator. Do you see any issue using these voltage levels?

    I have the interest to use the LTC2309 to perform data acquisition for long periods of time. Do you know if the LTC2309  is resilient enough to support  being accessed for long periods of time.?

    best regards

  • Driving the clock signal 0.8V below ground is definitely a problem. You are probably turning on the ADC's internal protection diode for that pin and dumping current into the LTC2309. If you look at the absolute maximum ratings section of the LTC2309 data sheet you will see that the digital inputs should not be driven more than 0.3V below ground. Turning on the protection diode can cause the LTC2309 to operate unpredictably.

    If driven properly the LTC2309 should be very reliable over long periods of time.

    Can you explain what you mean by the I2C is not working?

  • What I mean with I2C not working is that the ADC was not responding to I2C the commands sent from the microcontroller.  The presence of the ADC was not detected.

    After have connected all the ADC inputs to ground,  the ADC was detected again.

    As you mentioned, the fact of driving the I2C signals with voltage levels 0.8 below ground is producing the problem and will likely push the ADC to operate unpredictably.

    The 3.3V I2C signal provided by the microcontroller is around 0.5V below ground (for the low level). I will need to shift the signal up or clipping the signal below zero to avoid this issue. Do you think I can use a normal diode to clip the signal below zero? or should you recommend something else to do the job?

    Best regards,

  • A Schottky diode would probably work best.

  • Hello  mr. Hoover,

    Following up our last discussion, I installed the schottky diodes on the SDA and SCL signals  so the LTC2309 receive the correct levels on SDA and SCL, as can be observed in the snap shots shown below.

    Nevertheless, after an hour or so, the ADC stops responding to the master board. When this happen, I try scannig for connected devices and the ADC no longer appears in the list.

    Do you have an idea what can go wrong?. So far I don't have any other solution to make the circuit work.



  • What do SCL and SDA on the LTC2309 look like when the ADC stops responding?

  • Unfortunately I don't have snapshots for the SCL and SDA signals after the ADC have stopped working.But looking at the  signals in the scope, they were at high level and oscillate. I also remarked that current consumption goes up for 10mA.

  • I'm not sure what is meant by signals were at high level and oscillating. Please provide an oscilloscope photo. The only thing that I can think of that would make the current consumption increase is if the analog or digital inputs were driven above Vdd or below ground.

  • Here you have a picture of the SDA signal.

    7 of the analog signals provided are grounded to 0 Volts via 100 Ohms resistors.