Why milliwatts count in kilowatt designs

3/27/2010 4:34 PM EDT

But I have also seen low-power AFEs for high channel-count imaging systems, where the MRI or CT snanner is connected to a pretty hefty AC line and dissipates hundreds of kilowatts. Why would a few milliamps matter in such a design? After all, the power dissipated by the analog channels is dwarfed by the needs of the rest of the system.

It has to do with both the cost of delivering those clean power rails to the ICs, as well as local heat dissipation on the circuit boards. There's a real "ripple effect" where extra milliwatts at the load translate into larger supplies and losses at the intermediate converter stages, as well as heftier supply leads. When you reduce consumption, you simplify the design of the entire power-delivery subsystem, while enabling higher channel counts due to reduced thermal load as a consequence of lower dissipation. Lower amperage means that fewer, smaller, and thinner power rails and PCB traces are sufficient to deliver the regulated and unregulated rails with low IR drop.

In addition, low power allows IC vendors to pack more channels into a single IC, further increasing front-end density. A designer has to watch the local heating on a PCB, as well as the overall heat dissipated by the system. A box which is cool enough "on average" may still have internal localized hot spots which exceed thermal limits for individual ICs, necessitating extra passive or active heat sinking strategies.

Long story short: even if you have plenty of AC line power and thermal budget for your overall system, saving a little bit of power at the AFE (or anywhere) can pay big dividends in many ways, some imoedtaly obvious, l and some not-so-obvious. The good part is that all the implications of lower-power AFEs can be assessed and calculated as part of the up-front system-level analysis, to determine if the cost versus benefit ratio is favorable. ♦