transistor measurement using 4135 SMUs

[4135SMU]

Hi, 

we are using our PIXe system equipped with two 4135SMUs to measure a MOSFET.  the vi we use is the example one from NIDCPOWER package,called " hardware-timed two channel voltage sweep". A transistor has three terminals, "source" , "drain" and "gate". but this example vi only used 2 SMUs. so we connect the "output LO" from one SMU( randomly select) to "source" terminal of our FET, and connect the output "Hi" of the 2 SMUs to "gate" and "drain" terminal respectively. what we found out is we could get I-V curves like Id (drain current) vs. Vds. however, it is quite different from what we got from other measurement platform like Keithley 4200,see the attached. We wonder if the difference is due to the "output LO". maybe the electric potential of "output LO" of the 2 SMUs are different, like one is like 0V, but relative to the other SMU, it may be 0.1V or dynamic changing. Can anyone shine some light on this issue?

regards,

ted 

[Brad_O]

Hi Ted,

That looks like a self-heating effect.  Were the tests run at different speeds?  

The two LOs of the SMUs should also be tied together, but I wouldn't expect to see a curve like that from not sharing a common LO.

[4135SMU]

Brad,

  after I tied  "LO"s of my two SMU together, my transistor behaviors are like what I got from Keithley 4200. so I think it is due to the LOs. Funny thing is if the two "LO"s has a fixed potential difference, even they are not tied together, I should still be able to see the normal transistor behavior just with some shift in Vds.  since it didn't happen, it suggets the potential difference between two "LO"s  is not constant over the measurement time, which is odd. what do you think of it? 

regards,

[Brad_O]

Glad it was something so simple!  That is interesting... since the SMUs are floating, their Lo's could very well be drifting from each other.  Probably the SMU without a Lo connection drifted up in Voltage over the course of the measurement, trying to reach the voltage it was told to reach (the Lo path is vital in an SMU's feedback).  I think your curves could've been created either from a drifting Vg or a drifting Vd.  Either the plotted Vds is not the actual Vds, or the Vgs of each line is changing over time.

[尧尧]

Hi Ted, 

    I have two 4135 SMUs, but I found the test method of  "hardware-timed two channel voltage sweep" is different  Keithley 4200, that is when I set the current limit is 10 mA, the off-state of current is 220nA.  when I set the current limit is 10 nA, the off-state of current is pA-level, but I could not get the entire the Id-Vg curve if I set this current limit. 

    Keithley 4200 could auto adjust the output resolution of current, from low to high, like I set the current limit is 100 mA, the off-state of current is pA-level, and the test value reach the max resolution of current, step to the higher resolution, until the current limit.

     Do you have the same problem, and how do you solve this problem.

regards,

Zhang

[4135SMU]

Hi,Zhang,

  sorry to reply late. I have not login into the community forum for a while. actually the problem you run into are a general (hardware) issue of National Instruments' SMUs. their SMUs can not automatic change the measurement range. BTW, the range and resolution are related, for a 100mA range, the resolution may be 10nA, but for a 1uA range, the resolution may be pA.  

   so basically the answer is "No". you can NOT use a NI SMU to ACCURATELY measure the full range of  transistor or diode. it is pretty much a "resistor" measurement tool because the current range would be fixed. A possible solution is try to do a "software" adjustment. you can first set the range to the maximum and get the current reading , then change the range setting to the closet one which fits the reading. In other words, you set the range for each reading. the programming would be tedious.  however, NI has no such example program.  

then my final advice is stick to Keithley 4200 and stay away from NI SMUs before they make a huge improvement.

Regards,

Ted

[尧尧]

Hi Ted

     Thank you for your advise, I will try to "software" adjustment, and I will ask you some questions if you have time.

Regards,

Zhang