03-10-2012 11:48 AM - edited 03-10-2012 11:49 AM
I am using PXI-4132 DC SMU to make some high-speed DC measurements. I just took a 1K ohm resistor, forced 1mA and measuring 1V. I have this setup done and code completed. I get correct measurements. But, my problem is with the measurement time. The total time is about 45ms to complete the setup and measurement excluding the "Initialize and Close".
I checked the time at each stage and I noticed that the maximum time is taken during the "Voltage Read". Out of the total 45ms, it takes about 34ms just to make a measurement. That is, just one VI "niDC Power Measure" takes about 34ms, which is ridiculous compared to the Kiethleys. I tried varying the "Aperture time" and turned the "Auto Zero" off. But the measurement time still is high. Can anyone help me and let me know what is the fastest measurement time acheived using 413x series SMUs?
Solved! Go to Solution.
03-10-2012 12:16 PM
I just found out one thing. When I set the Auto Zero to "OFF", I could reduce the measurement time from 34ms to 17ms. But the first measurement will still be 34 ms since the autozero will be OFF from the second measurement. So, that means, If I am taking 100 different voltage measurements with Auto Zero OFF, the first measurement will take 34ms and the remaining 99 measurements will be at 17ms. Now my question is: will I be able to reduce the measurement further?
03-12-2012 08:47 AM
What do you have your source delay set to? The default is 16.66 ms or 1 PLC (60Hz). You can reduce this time to something much lower, but you will want to make sure that your source has settled to a steady enough value before moving forward with taking the measurement. Are you sourcing a different voltage/current with each measurement or would you like to set your voltage/current to one level and then take 100 measurements afterward? You may want to consider using NIDCPower Fetch Multiple instead of niDCPower Measure, as this will reduce the software latency with making 100 individual calls to the hardware. NI-DCPower Measure Record.vi in the Example Finder would be a good reference on how to use Fetch Multiple. If you can give us a little background on how you would like to set up your measurements we should be able to make suggestions to optimize your measurement time.
03-19-2012 12:22 PM
Thank you for the valuable inputs. I have set my source delay to 10ms and nplc as 1 and the test time now is 33ms. This is what I would be doing on my device. I will be setting
the voltage to a particular value say 1V and I will test around 1000 devices. We will be measuring leakage current in the range of 3-10nA. How accurate will my readings be if I reduce my source delay and also nplc ?
Another observation I made was the very first reading the I make is slightly higher than the other readings . There is a 5% variation in the very first reading. Is this a normal thing or do I have to make some changes to the settings ?
Right now I have set the autozero off.Thank you .
03-19-2012 07:52 PM
The amount you reduce your source delay really depends a lot on your DUT and how steady you need your voltage level to be before you start taking your measurement. If your DUT is reactive, you may see overshoot or ringing when transitioning your output level and therefore would want to allow enough settling time before taking your measurement. The best thing to do is experiment with various source delays and see how much delay you need to get a consistent and repeatable measurement. If you have a scope available, this will also show you the step response of the SMU with your DUT. If your first reading is higher than your other readings this could very well be an issue with not enough settling time. What sort of DUT are you testing?
From page 4 of the spec you can see that at 1 PLC, we can expect approximately 1 ppm RMS noise of the range at 1 PLC. This means in the 10 uA range our measurement noise would be 10 pA RMS or 60 pA pk-pk. However, your DUT + cables will pick up additional noise and you should consider using twisted shield pair wires to reduce the noise picked up in your system. The shield can be terminated to chassis ground on the SMU side of the cable to help reduce noise appearing in your measurement.
Let us know what you find after experimenting with you DUT.