Getting started with Maxim EE-Sim® OASIS Part 3: PSU Assembly & TestingFollow article
Assembling and testing our Raspberry Pi Industrial PSU PCB
Part one of this series showed the schematic circuit creation using the online Maxim EE-Sim DC-DC Converter Tool around the switching controller MAX17576 , then carrying on from where we left off in part two, we now have a PCB layout ready to send off for manufacture. We chose to use Eurocircuits for this, as they’re reasonably priced, good quality boards and have a quick turnaround.
As we have already generated Gerber manufacturing files in our previous post, we now need to drop these into a ZIP archive for ease of uploading to our PCB manufacturer’s website. With Eurocircuits, this is as easy as dragging and dropping the file - a series of choices is then presented to us. We chose to skip uploading assembly data as we are assembling the PCB in-house, and then we jumped straight into the “PCB Configurator”.
The PCB Configurator gives us a nice render of our board and auto-populates the correct board dimensions based upon the outline layer generated from DesignSpark PCB. It is worth noting that stencils can also be ordered along with the PCB in the same order, by scrolling down and picking the combination of top and bottom stencils. As we are assembling, a top stencil was ordered and the default stencil options were left as-is.
There is also an embedded PCB Viewer that provides a render of the Gerber files, which can be accessed by clicking the image of your PCB. This proves useful for checking that the uploaded Gerber files were processed correctly, and provides a last-minute sanity check that everything that should be included on the PCB is there!
After this, all that is left to do is add the order to the basket, then check out the PCB order and wait for them to arrive.
The bill of materials (BoM) needs to be generated from DesignSpark PCB so that all the components can be ordered - this is easily done by heading to the “Tools” menu, then “Reports” and picking the “rsbom” report option.
Once done, a CSV file with the project name and “(rsbom)” will be generated in the same folder location as the project file.
From the BoM, all the components were ordered ready to populate the board.
Populating the board
Now that all the components, PCBs and the stencil have arrived, it is time to paste the board. We have a solder paste stencil jig which makes holding the stencil and aligning it with the PCB an easy task.
Once the solder paste has been applied with a squeegee blade, the stencil can be removed and the PCB is ready to have the components placed - this is a delicate job involving a steady hand and a pair of tweezers.
The fully populated board then went through a reflow oven, which requires some setup. Reflow profiles are a bit of an art and the process requires some careful monitoring - an ideal starting point is the reflow profile provided by the solder paste manufacturer. In our case, we used Chip-Quiksolder paste and entered a reflow profile in our oven that matched that. We found that the first test board that went through looked a bit sad and overheated, so the profile was adjusted accordingly.
This can be a bit of a trial-and-error process, and it is worth keeping some spare boards to hand to help with setting up a reflow profile. In our case, we reflowed a test board that had a similar copper distribution and tuned the process based around this.
Finally, all the through-hole components can be soldered onto the PCB - this is the polyfuse, input connector, output header to plug into the Pi and the LED indicators.
For testing our Raspberry Pi Industrial PSU, we have used a variable, current-limited bench power supply — the RS Pro RS-3005D— together with the RS Pro RS-KEL103 electronic load , and an oscilloscope to inspect the quality of the output. The electronic load is connected to a computer to allow for easy setup of the step load changes so we can observe the output stability.
Now that it has been assembled, it is time to bring up the board and do testing. The first step is to check for continuity between the power and ground inputs, highlighting any shorts that need to be fixed before the board can be powered up.
If the continuity test passes with no shorts, the next step is to apply a current-limited source of power - the “smoke test”. We applied 12VDC with a current limit of 100mA to the board, and observed for any odd smells, signs of things getting hot, or smoke - none of which happened. Both indicator LEDs illuminated on the board, one which is connected to the input after the fuse, and the other is the “PGOOD” output from the MAX17576, which indicates the output is valid.
The next test is to check the output voltage, to ensure the output voltage is in the range that we specified in part one - we designed the converter to provide an output of 5.1Vdc as the permitted USB supply voltage range is 4.75-5.25Vdc. The screenshot below is the voltage indication from the electronic load, showing we are well within the 4.75-5.25V range, and only 40mV away from our target voltage of 5.1V.
Now we know that the output is in range, we can check for ripple on the power supply. We set a target of 2% output voltage ripple, which gives us a ripple voltage of 0.1V. To measure this, we connected the oscilloscope to the output, and set the input probe coupling to AC - this removes the DC component of the input and shows only the ripple and any transients. We then set an appropriate vertical scale to show the ripple, and add a peak-to-peak measurement to the waveform. This gives us a figure of around 40mV of ripple.
We can also perform a step load test on the output to check for overshoot and undershoot. We set a maximum overshoot/undershoot percentage of 2%, meaning the output voltage should not overshoot/undershoot by more than 0.1V on a step load change. To test this, a two-step program was entered into the electronic load software.
Another two measurements were added to the oscilloscope, maximum and minimum, which will measure the highest and lowest values of the trace currently displayed. The test programme on the electronic load was then set going, and the oscilloscope instructed to do a single-shot capture.
We can see that the overshoot and undershoot falls within the 0.1V limits, with the maximum overshoot being 56mV, and the maximum undershoot being 43mV. This is the extent of the testing that we can do, without more specialist equipment to produce a Bode plot.
The below video illustrates our test setup, and shows the overshoot and undershoot in real-time.
In this article series, we have taken a look at Maxim EE-Sim OASIS, used it to design and simulate a switch-mode power supply suitable for powering a Raspberry Pi, designed a suitable PCB, assembled and tested the completed PSU.
The schematic and PCB layout can be downloaded from the Raspberry Pi Industrial PSU GitHub repository, along with a bill-of-materials.