TS3A24159: High leakage current on the out pin that is not selected

Part Number: TS3A24159

Hi,

I am using the TS3A24159 to switch an IO pin that comes from an MCU.

The device is connected as follows: (drawing one of the two switches for simplicity)

The issue is that when the SELECT pin is HIGH, and the IO is HIGH I see voltage on the output (which should be disconnected from COM at this state)

At first the output was not connected to anything but the voltmeter and I measured up to 600mV (at 25C).

I tried adding a pull-down resistor (100KOhm) to ground, and I measured 10mV (which might be considered fair),

but when testing in high temp. (80C) the output voltage with the 100kOhm PD was measured up to 130mV.

Quick calculation shows that it means 1.3uA is flowing through the NC pin which is much more than anything stated in the datasheet.

(I got the same results on 3 similar circuits)

What might cause this issue?

Thank you.

  • Hello,

    Please make note of the logic table below for this device:

    When the IN pin is low (VIL less than 0.5V), the NC to COM path will be connected, in this case, connected to the output. Now, the way you are currently using the device is setting the IN pin high and trying to measure the NC pin ("output"), when it is actually connected to the NO pin.

    Just need to reverse your logic and this application should work properly.

    Hope this helps!

    Bryan

  • I am aware that when IN is high the NC pin is disconnected.

    The issue is that even though it's disconnected I measure relatively high leakage current from this pin (up to 1.3uA) which does not match the datasheet specifications.

  • Hi,

    Understood, thanks for clarifying! Definitely seems a bit higher than I would expect. Do you have the schematic for what the outputs (NO/NC) are connected to? I know you mentioned that you tested it with just a voltmeter and the output floating (did you lift the pin from the board? How did you make sure the pin was floating?), but would still like to see the whole picture to see if anything else could cause this increase.

    Thanks!

    Bryan

  • I have the device connected with 0Ohm resistors on most of its pins so I disconnected them and connected wires directly to the pins while the device was still on the board.

    The V+ pin is connected to 1.8V supply with a 100nF capacitor, and the SELECT is connected with a 10K pull-up to the supply.
    Other pins were connected directly either to a voltage supply or a voltmeter.

    I actually retested one of the boards again with a precise ammeter and got these results:
    each measurement was taken while SELECT and COM are 1.8V and the NC pin is connected to ground through a 10KOhm resistor,
    current is measured through this resistor.

    Temp NC1 NC2
    25C 95nA 97nA
    40C 270nA 275nA
    60C 830nA 844nA
    80C 2257nA 2283nA

    I got similar results for the symmetric case measuring the NO1, NO2 pins. 

    Thank you.

  • Got it, again, appreciate giving all these details! So essentially you are doing what we call an "IOFF" test where you are trying to measure the current going through the unselected switch path. In our datasheet, we outline the test conditions below:

    Note that we are specifying the IOFF leakage at around 10nA when the COM voltage is 1.65V and the NC voltage is 0.3V. Your test setup is when the COM voltage is 1.8V and NC voltage is grounded, which will make a difference, but it is a data point we do not currently have for you to reference. Could you try to replicate the above conditions and see if you match up with the datasheet?

    Thanks!

    Bryan

  • I did the test again with the same conditions as on the IOFF test. (V+ = 1.95V, COM=1.65V, NC=0.3V)
    Under those conditions the leakage current was under 10nA even at 80C.
    Connecting NC to ground results in 800nA at 80C.
    And finally setting COM to V+ results in about 2uA at 80C.

    I understand now that the configuration that I tested originally was not tested on the Datasheet,
    but it seems that the conditions tested on the Datasheet were chosen to show nice numbers instead of real useful data.

    Thank you for your time,
    Guy.