Mechanical hearts tick with chips

The microprocessors that go into implantable cardiac pacemakers and defibrillators aren't very fast. Their technical features are scant. Their processing power is unimpressive. And their designers are proud to have made them that way, because stripping a processor down to the bare essentials enables it to do exactly what it needs to do-operate inside the human body for years on only a single, small battery.

Who deserves the credit for designing such power-miserly processors? Not microprocessor vendors. They tend to design general-purpose chips with features that make them versatile, but power hungry. In addition, many commercial processors don't meet government regulations for use in life-support systems.

That leaves manufacturers of implantable cardiac devices to design their own processors. Engineers at companies such as Medtronic (Minneapolis, MN), for example, have created processors patterned after the RCA 1802 and Motorola's 6805, but implemented only with features that are essential to a cardiac device. The Medtronic philosophy is explained by Jack Keimel, vice president, for research and advanced concepts, cardiac rhythm management. "We don't need state-of-the-art speed or performance in a processor," Keimel says, "but we need very low current drain and the ability to operate on every heartbeat."

Indeed, the computational load placed on a processor in a pacemaker or a defibrillator is not at all heavy. Essentially, all a pacemaker has to do is to be on the alert for a too-slow heartbeat and, if necessary, stimulate the heart with appropriately spaced mild shocks to make heartbeats occur at proper intervals. A defibrillator watches for the opposite condition, an abnormally rapid heartbeat that could lead to cardiac arrest. If necessary, the defibrillator provides a high-energy shock to restore a normal heart rhythm. Neither a pacemaker's nor a defibrillator's task is computationally intensive. Monitoring a heart that beats at 70 or so times a minute is no big deal for a processor that ticks at several million times a second.

Long battery life, on the other hand, is essential to a pacemaker or defibrillator. An implanted cardiac device must be surgically removed and replaced when its battery nears the end of its operating life, and heart patients are loathe to undergo surgery more often than necessary. Fortunately, the battery in a modern pacemaker or defibrillator typically lasts from five to nine years, a period that would be much shorter if the device's processor didn't operate so efficiently.

When designers were still using commercial processors, that kind of power efficiency was hard to come by. A few years back, for instance, Minnetonka, MN-based engineering consultant Scott Armitage examined about a hundred potential processors for a pacemaker design and quickly narrowed the list to two devices, neither totally satisfactory. "You need extremely low current, like a microamp," Armitage says, "and low voltage. Even 3.3 volts might be too high." Thus was born the minimal, custom-designed processor.

The trick to using a minimal processor, says Medtronic's Keimel, is to minimize the amount of work the processor does. For example, Keimel notes, the processors in Medtronic's implantable pacemakers and defibrillators operate only a small percentage of the time-typically about 10 milliseconds out of each second. The processor is in "sleep" mode until a sensor in the implanted device recognizes a heart beat and wakes the processor up. Upon awakening, the processor performs its necessary tasks and then goes back to sleep until the next heartbeat.

Modern cardiac devices save battery power in other ways, too. Currently available pacemakers, for example, apply an electrical stimulus to induce a heartbeat only if the natural heartbeat falls below a certain count, which is specified and programmed by a physician. Earlier pacemakers provided stimuli at preset rates whether needed or not, thus using much more battery power.

Minimizing the program code in cardiac devices also saves power, by reducing the amount of power-consuming memory needed to store the code and by reducing the processing power required to run it. In 1987, Keimel says, a pacemaker typically used only 4 kbytes of ROM. Today, he says, memory capacity is an order of magnitude higher, but still minuscule compared to a personal computer, for example.

Code reduction extracts a price in programming effort, however, because it often means resorting to assembly language instead of an easier-to-use high-level language. Compiled high-level languages may be suitable in some cases, but some subroutines almost always require hand coding for minimal size and execution speed.

Nor is writing application code by hand the only difficulty in producing software for pacemakers and defibrillators. In most cases, developers of cardiac devices have to create their own operating system software. Although fast, compact operating systems exist for even simple microprocessors, they often are not small enough for systems with extremely limited memory. Plus, says Armitage, government regulations for life-support equipment require the availability of source code for all software, and most commercial OS developers provide only compiled and assembled object code in order to protect their intellectual property.

Custom hardware and software development is what embedded-system development is all about, of course, but perhaps no other embedded systems require as much custom development as do implantable cardiac devices. Where small size is essential, extremely low power consumption is crucial, and reliability can never be compromised, design "from the ground up" is almost a given. Such are the challenges of developing systems that aren't just embedded, but implanted in the human body.

Please contact Gary with ideas on issues design engineers face with embedded systems. He can be reached at [email protected].