Part 3 of R&S NGP814 Power Supply - Review
Two chassis mount resistive loads from TE were used for these tests. They are wirewound resistors in a large ceramic core, they have many uses, I have a few that I use as a dump load while testing capacitor banks, rail-gap switches and inverters.
The first resistor is supposed to be 1ohm with 5% tolerence but it is measured at 1.3 ohm, the second is 47 ohm 5% measured at 48.05765 ohm as seen in the photos below.
Part A: Series Combination
Resistor used: 48.05765 ohm
Current = 2.644265 A
I ripple p-p = 2.647849 - 2.640457 = 7.392 mA
Voltage = 127.9272 V
V ripple p-p = 127.9281 - 127.9262 = 1.9 mV
The video below shows the test set up, the touch screen is very useful to change various parameters to cut down testing time.
The delayed output feature can help to monitoring performance and efficiency of voltage regulators and DC-DC converters at different voltage windows.
Part B: Parallel Combination
Resistor used: 1.3 ohm
Voltage = 19.00135 V
V ripple p-p = 19.00181 - 19.00117 = 0.64 mV
In the video above all channels are connected in parallel, I wanted to monitor the cutoff point at which the current limit is exceeded, in this case the ramp function came in handy. Voltage can be seen rising to reach steady state, and graphical view can also help when testing multiple devices to compare and observe their behaviour.
I noticed an issue with channel 3, no matter what voltage and current limit I set it does not output the same power as the other channels, so I tried a few other combinations and changed the limits to stay within operating range as seen in the photos below.
The photos above shows an unexpected behaviour, when channel 3 is switched off all other channels have a similar output. Whereas the current output of channel 3 is less than 50% of the other channels.
This will need further investigation, I need to look into all my cables that I am using and test again before contacting R&S support.
Dynamic Load and Logging Features:
The NGP800 series was designed with a 125 Samples/sec aquisition rate, which can be considered very limited compared to most mid/high range oscilloscopes/ DMM/ or data loggers, however it can be very helpful when a high sampling rate is not required. The example below is for a circuit used in monitoring industrial washers used to clean engine parts.
Considering that the device powered by the NGP814 is a low power device it might not be a good example to demonstrate and test the logging features, but the same concept and steps used can be applied to any type of dynamic loads that can be more suitable to the rated power and sampling rate of the NGP814. I agree with fellow raodtester gpolder that the "Graphical view" is limited, it will help to be able to zoom in and out or focus on a certain region and also the PC software does not support realtime viewing which can be helpful to pin down analysis to certain windows instead of having to save the whole session as a CSV file then analyse. One way around this will be to write a script or a labview code that will fetch data from the CSV to use for plots and further calculations.
Power measurements are also very useful for a system designer that wants to calculate energy usage, one integration formula for power measurements will show the energy consumption figure.
Update: 10 Jan 2021
Spent some more time trying different combinations for parallel connection. As lui_gough mentioned in his comment and the screenshot he shared from a different series (R&S HMP4040) slight differences in voltage should distribute the load between channels more evenly.
I tried to shorten the jumper cables and make sure they are all same length to mitigate mismatch as much as possible then tried different combinations between channels to see changes in how the output power is distributed.
A- Variation of 1 mV
The channels were set at 20V with 1mV steps. 20.001, 20.000, 19.999, 19.998
The current output of every channel was not fully stable and kept going up and down by about 500mA.
Positive -> Ch1+
Positive -> Ch4+
Positive -> Ch2+
A- Variation of 100 mV
A stable output was achieved when the voltage difference changed by 100mV. It was not a perfect balance between the channels, but at least all channels were within the limits.
after about 10 seconds the output changed from the first photo below to the one shown in the second photo and then remained constant.
Positive -> Ch2+
- As R&S mentioned, the load might not be evenly distributed between channels due to tolerences.
- What might be a useful feature is internal connections (a series/parallel mode like the Keysight E36300 like the photo below) to avoid using a lot of external cables and to minimize mismatch due to voltage drop in cables.