Power modules vs. discretes: White paper quantifies design cycle times
In recent history the march of power design technology has been towards the use of web-based simulation to ease the design chores of putting together power conversion circuits with discrete components (discrete in the sense that power converter ICs are surrounded with passive components and then placed on a pc board.)
There is another trend in power circuits, however, and that is towards modularized PSiPs (power systems in package) PSiPs integrate the power converter MOSFET switches, control, compensation, and the inductor all into one IC-like package. You don’t need to go through the simulation step to select external components because they’re already inside the package, selected, matched, assembled, tested out, and verified as to noise immunity and interference. Companies that make PSIPs are Enpirion, Intersil, Linear Technology, Rohm, TI, Torex, Vicor, and Vishay.
Proponents of both approaches say theirs enables the fastest time-to-market. Mark Cieri, director of business development at of Enpirion thinks it’s time to quantize the time involved in the steps. He says, “I think guys that don’t make modules are trying to make up the difference [in design time for designing with discrete parts] with web-based simulation tools and say that web tools are the answer. The question is, will web-based simulation provide a tailwind to allow discrete solutions to stay around forever? Or, is the more compelling movement going to be the integration of components and the elimination of design decisions? Because even after a simulation of a design, few engineers would commit to a system design without still breadboarding their circuit to verify actual EMI, component, and board layout gotchas.”
Ciery thinks that if the topic is time to market, it should be able to be measured. Power industry analysts at the Darnell Group performed a commissioned study for Enpirion looking at just that: The study’s goal was to quantify the relative ease of design for a PSiP design flow used in a 5A dc-dc regulator for an FPGA or similar load vs. a typical dc-dc regulator solution.
The Darnell Group qualified and validated the design steps for a general PSiP design configuration by conducting a series of surveys comparing a “typical” DC-DC regulator design flow with a comparable PSiP design flow. Survey respondents were engineers at a variety of leading server, storage, data communications, personal computer and telecommunications industries in the US, China and Europe. For details on the methodology and survey’s design-flow steps, you can download the white paper here. (It’s a .pdf on the Darnell web site.)
Basically, the design flow is the same for the two processes with the significant exception that there are no simulation steps in the PSiP design flow, and the prototype testing step is significantly reduced. To cut to the chase: Even with the use of web-based simulation tools, it still takes an average of 464 man-hours to complete a dc-dc regulator design flow vs. only 254 man-hours to complete a PSiP design. This is for a simple 5A regulator design flow, for a more complex design, the process could take 20-40% longer.
(This is independent of the PSiP vendor: regardless of whether you’re using a PSiP from Linear Tech, TI, Enpirion, etc, the results would be the same.)
I didn’t find the results surprising: After all, the history of electronics says that the trend is towards integration because it is a space, design-time, and cost saver. Was the study really necessary? Cieri emphatically said yes: He said that his big challenge in explaining PSiPs to power system designers is to challenge the sovereignty of web-based web-based tools in dc-dc converter design.
I suggested that a bigger barrier to the adoption of PSiPs is the belief that PSiPs command a price premium in a market where pennies count. Cieri replied, “This [whitepaper on quantifying time-to-market] is just the first step.” It sounds like the players in the PSiP industry will be tackling the total BOM costs next.