Powering a Homecoming in 90 Days: Part 2

This is the second post in a series that describes how Vicor Custom Power developed a solution for the Navy to replace an obsolete power supply, despite having no specifications for the part that needed replacement. If you’ve not read it yet, read the first post here.

Our task began with an initial architecture that would rectify the input AC power and develop the necessary voltage and power to the output.  We knew right away that this would take a fair amount of physical space, plus we had an estimate as to how power would have to be dissipated as heat.  The heat sink was going to be fairly large.

We were not disappointed once the unit arrived.  Somebody failed to mention that the part of the ship where this is mounted is kind of dirty – oily, greasy, and fairly dusty.  Nobody wanted to be the first to handle it once we opened the box.  Eventually it found its way to the bench and we crowded around it.  It was big.  It utilized forced air for cooling and had a heat sink with plenty of surface area.  Measuring some 20 inches in length and roughly 6.3 inches in width and 12 in height, it delivered nearly 2,000 W (approximately 160 A) of power.  Its fan was mounted in the middle of the heat sink, blowing down and out either end.

The first task was to examine the power supply itself, and establish what failed and why.  This became obvious quickly: oily dust had clogged the heat sink in between the fins to a point where little if any air flowed along its length.  We decided that this aspect of the design would have to change if we were going to deliver a reliable solution.

The team returned to the whiteboard to fine tune the architecture and started the task of selecting the power conversion modules they would design in.  The team chose Vicor’s Brick Family of DC-DC converters and filters.  Our estimate called for four M-FIAM3 and four V300 500 W MAXI DC-DC converters in parallel.  Together with this would be circuitry to support the modules, and protection (over temperature, under voltage, etc.) circuits.

The mechanical team took on the tasks of re-designing the heat sink to prevent future failures, replace the heavy metal cover with a new 3D printed plastic version, and worked in concert with the electrical team for the placement of components which would lead to the design of the printed circuit boards (PCBs).

The schedule told us that there were seventy days left.  The mechanical team had their concept for a design that would be significantly more efficient design for the heat sink – a fold-fin design.  This approach, together with moving the fan to the end of the heat sink, would provide us with about double the surface area for cooling.  By moving the fan to one end and making it removable, we would allow the crew to easily clean out the heat sink by placing a hose where the fan mounted.  Now all we had to do was find somebody to make this in time.

65 days were left and we held one more design review to see how this would all go together.  Within a day the custom materials would be ordered, and the PCBs would move into layout.  The item that worried the program manger the most was the custom heat sink.  This would make or break the schedule.  A vendor was found who stepped up to help with the thermal modeling and said they would be able to deliver this complicated piece in time.  In the end, this one piece of hardware would make all of us lose sleep for the next fifty or so days.

We had selected Vicor Brick products and had ordered four sets.  One set for the actual deliverable unit, and the second set for a spare unit should something happen to the first.  And there would be two full extra sets of Bricks, which were included to allow us to validate the electrical functionality of the design before it all came together.

At 60 days left, material started to arrive.  Electrical components, connectors and some of the mechanical hardware were received.  In another few days the PCBs would arrive.  If there was a problem with the layout of the PCBs, we would have time for a re-spin but barely.

55  days and counting, the PCBs were in assembly.  The DC-DC converters and filters were on hand now and with a bit of luck, within a day or so the team would begin powering up the power conversion end of the design.

20 days left and the heat sink arrived.  It had a problem.  Along one end, the welds that held the heat sink to the baseplate were wrong, preventing the heat sink cover (which forms the channel forcing the air to flow down it length) from fitting.  It was back to the vendor, this time hand carried.  They promised all-nighters to make a new one if they could not fix it.

With about 15 days left, all custom parts were in and the prototype-build PCB was being assembled.  The mechanical team, with toolboxes in hand, moved out onto the production floor as it was up to them to put it all together.  One way or another, what would not fit would be ground down, drilled out, hammered, re-welded, or whatever else it would take to assemble it.

With 10 days left we moved into test, as the home stretch was in sight.  The team had planned on fully powering up the new supply, running it at a full load of 1,950 W.  We also planned on testing for EMI, specifically MIL-STD-461 CE101 and CE102, even though the prototype unit did not have to meet the requirements.  From there it would be subject to extended high temperature soaks in our chamber to simulate the normal operating condition the design would see while in port.

There was a happy ending to the story.  The unit shipped out with in time, making it onboard the ship a few days out from docking.  It fit perfectly and met the electrical specifications we were given.  Once installed, it was turned on.  It worked.  Well, mostly it did.  It turned out that somewhere along the way somebody had changed the wiring to the unit and certain signaling functions kept telling the control panel that it was not working but it indeed powered the system as required.  The ship put into to port despite the false faulty signal.  It was a very happy and on-time homecoming for the crew.

Within a week, Vicor Custom Power had our mechanical designer and then later our electrical designer onboard investigating the problem.  To their surprise they had to descend nearly a dozen stories, all by stairs, to get to the installation.  These two visits resulted in a simplified method of installation allowing only one crew member to install the unit by having all of the mounting hardware as a one-piece assembly.  The faulty signaling was corrected.  Today the custom power supply has completed EMI testing along with shock and vibration testing and has been assigned national stocking part number, replacing the obsolete one.

Tags: , , , ,

Comments are closed.

Find out more about our Cool-Power Buck Regulators subscribe to vicor newsletter Contact Us

Get Connected