In search of an ideal diode - Using a NI Thermocouple Measurement Kit to monitor the temperature rise in a high current design.

Table of contents

RoadTest: NI Thermocouple Measurement Kit - Industrial Sensing

Author: genebren

Creation date: 25 May 2022

Evaluation Type: Test Equipment

Did you receive all parts the manufacturer stated would be included in the package?: False

What other parts do you consider comparable to this product?: While the full power of the NI Thermocouple Measurement Kit (cDAQ-9171/NI 9210) lies in a combination of Hardware and Software, my temperature measurement needs could have been handled with a simpler and less expensive device. Given the power of this device to be used in more a complex and integrated measurement and control system, it seem unfair to compare it with just a simple temperature datalogger.

What were the biggest problems encountered?: Getting licenses and software downloaded and running took way more time and energy than the testing that I had intended to do.

Detailed Review:

I was extremely happy to have my proposal selected for this Roadtest.  Thank you to element14 and NI for choosing me to be a roadtester for this amazing piece of equipment!

The Plan

My plan for utilizing and evaluating the NI Thermocouple Measurement Kit, was to tackle a real world problem.  This problem was to characterized and test a current monitor and safety circuit that is a part of a product that I have developed.  The circuit consists of a pair of 15.4A P-Channel MOSFETs, configured as a back-to-back pair, with the first MOSFET acting as 'perfect diode' and the second one as a high-side load switch.  The majority of the testing is focused on the reverse voltage protection side of the circuit, while the high-side load switch and associated circuitry will be active in the test.  Here is a schematic (simplified to depict the portion of the circuit to be tested):

The MOSFETs are Vishay Siliconix,SI4403DDY-T1-GE3, P-Channel 20 V 15.4A (Tc) 5W (Tc) Surface Mount 8-SOIC.  The circuit is intended to supply up to 8.5A of current to the load at an operating voltage of ~6V.  Zooming in on the high power section of the schematic (seen below) the input power is routed into J1 and the load is connected to J2 (J2 is a virtual connector, the actual connection is made with short sections of 18GA wire, soldered to the backside of the board). Two wires were soldered to F1 (drain and source) to measure the voltage drop across the MOSFET.  For temperature measurements, a thermocouple was affixed to F1 and F2 (with Silver Thermal Compound) and also to R5 (0.01 Ohm current sense resistor).  A forth thermocouple was used to measure the ambient temperature.

Test Equipment Used:

1. Input power was supplied by a TDK-Lambda, LS50-5 power supply (6V at 10A)
2. Voltage drop across F1 was measured with a muiltcomp PRo, MP30027 DMM
3. Programmable DC Load - BK Precision 8500
4. Thermocouple measurement - NI cDAQ-9171/NI 9210 with 5SRTC-TT-K-24-72 thermocouple probes.

First Look at the NI Thermocouple Measurement Kit

The package from element14/NI arrived in very good shape.

Inside of the box the contents were again well packed and very secure.  Here are the contents:

Here are the thermocouples, more than enough to do some serious testing:

And here are the important parts, the cDAQ-9171 chassis and the NI 9210 Thermocouple measurement unit:

I was very impressed with the quality of the unit, very solid and rugged. I really was looking forward to trying this out.  But, first, I needed to download a bunch of software.  I initially had planned to run this on my desktop system, but Windows 7 was no longer supported in the latest version of the software packages.  I downloaded older version, that were supported under Window 7, just to see how everything functioned.  I also decided that I needed to start looking for a more modern computer system (which I may eventually replace my older desktop with), so I ordered a new laptop with Window 10 Pro.

Installing and utilizing the software

With both my Windows 7 PC and my new Windows 10 Laptop, I was able to eventually get enough software loaded to attempt to communicate with the NI hardware (NI-DAQmx and DAQExpress).  Once all of the hardware was detected and drivers were assigned to the hardware, I started to get excited that I could start using the devices.  My first sign of life, was seeing the NI Device Monitor showing that hardware was attached to my system.

Great to see the device properly identified.  From here I wanted to see what test and configuration information I could obtain, so I click on the 'Go' button for "Configure and test this device". Here is the top of that menu:

Liking a good self-test feature, I clicked on the "Self-Test" icon:

I had expected to get see some data, or at least a series of actions and results, but I did get a confirmation that the self-test had run and was completed successfully:

From here, the 'Run the NI-DAQmax Test Panels' looked like a chance to see data, so I clicked on the 'Test Panels..' icon:

Great, now this is surely getting interesting.  Wiggling temperature lines were very exciting!  And this was the point where I began to struggle to understand where to go and how to log some data.  I soon realized that I needed to study a bit and try and make sense of software and how to best use it to get the data that I really wanted to see.  At this point, I moved on to get a better idea of what I needed to measure and learning how to best get that data.

Visualizing the heat

Prior to starting the test, I wanted to visualize the temperature of the board under load conditions to better evaluate thermocouple placement.  To that end, I built up a cheap thermal camera, using a Panasonic, (Grid-EYE) AMG8833 sensor.  I mounted the sensor onto some perf-board, and re-purposed data logger board that I designed  for Panasonic Laser PM2.5 (Dust/Smoke) Sensor RoadTest (https://community.element14.com/products/roadtest/rv/roadtest_reviews/1558/particulatesmoke_dat) to collect data. After modifying the firmware of the datalogger and the control panel software, I captured these images of the test board (both sides) at various load current values:

The hot spots in the images mapped out to be the two MOSFETs on the front side of the board and the current sense resistor on the backside.  Here are some images of the setup used to take these thermal images:

Collecting the data

The tests were run in a series of steps, first collecting the forward voltage drop across the F1 and then collecting the temperature data. For the forward voltage drops, the load was stepped at 0.25A steps, from 0 to 8.0A, while the voltage across the MOSFET was measure and recorded.  Here are the resulting charts showing voltage drop across F1 and R5:

Along with the measured voltage drops across F1, I calculated the power dissipated across F1.  Also, using the current step values, I also calculated the voltage drop across the current sense resistor (R5) and the power dissipated across R5.  These two charts proved a couple of things:

1. My search for a perfect diode, did in fact find a very good result. The voltage drop across F1 was less that 100mV at 8.0A.  Yeah! This was result was even better than I would have expected, given a typical Drain to Source On-state resistance of 0.0195 Ω  (0.0195Ω * 8.0A = 0.156mV, so Rds is closer to 0.0984 / 8.0 = 0.0123Ω)
2. The Power rating of the MOSFET and the Current Sense resistor are more than sufficient in my design (F1 showed a max power of 0.787W vs. the specification max of 2.4W and R5 showed a max power of 0.64W vs. a specification of 1W).

After getting at least an evaluation license in place for FlexLogger, I started working my way through the functions, trying to find out how to record and display my temperature data.  It was not easy, and some of the UI functions were less than obvious, but little by little I started to see the light and figure out what I needed to do.

For the temperature tests, the load was stepped at 0.5A steps, from 0 to 8.5A, while the temperatures of F1, F2 and R5 were taken.  As the load current was stepped, it was held to allow the temperatures to settle (at lower currents the hold period was 1 minute, at currents greater than 6.0A the hold was increased to 1.5 minutes).

The temperature rises in the captured data (F1 - 185°F and R5 - 167°F) closely matched the differences in power levels of the two major devices (F1 - 0.7872W and R5 - 0.64W). While the temperature rise is quite high, the good news is that under normal operating conditions, the current sense circuit is measured routinely by the processor, and if the current is above the programmable set point for more that the trip delay time, the processor will turn off the load to protect the board.  But it is comforting to know that there is plenty of time for this to happen before the board is permanently damaged.

Conclusions

1. In the search of a 'Perfect Diode', the Vishay Siliconix,SI4403DDY-T1-GE3 P-Channel MOSFET is a winner! With a maximum voltage drop of 0.0984V at 8.0A, this is a huge improvement over a Schottky Diode, with a best in class of 0.44V at 8A (power dissipation of 787mW versus  ~3.5W).
2. The NI Thermocouple Measurement Kit was a rock solid temperature measurement device.  This was way more functionality than I needed for this test, but I am really looking forward to utilizing some of the power and functionality of NI's LabVIEW software (future purchase) to integrate my existing equipment along side of this device to automate some of my test setups.
3. It was very rewarding to finally have the proper instrumentation to verify the full range of this circuit, with some accurate and reliable temperature measurement equipment. My prior thermal testing was performed with an ETEKCITY Infrared thermometer (lasergrip 774) a bargain at $20, but a painful way to monitor temperature for a multiple step test scenario.
4. Not exactly a conclusion, but once again a huge thank you to element14 and NI for selecting me a roadtester for this great piece of test equipment.  What a great addition to my collection of test equipment!