Medical Electronics Manufacturing Magazine
MEM Article Index

Medical Electronics Manufacturing Fall 1998

Integrated Circuits

Deciding Whether and How to Use Mixed-Signal ASICs

Jim Gentile

One of the most significant challenges currently facing all segments of the medical industry today is the need to reduce costs by moving patients from hospital to ambulatory care as rapidly as possible. For designers and manufacturers, this need has meant an increased demand for portable medical devices that incorporate semiconductor components capable of meeting new technical and economic requirements.

Compared to nonportable systems, portable medical devices should ideally consume less energy and take up less space while at the same time delivering enhanced performance and improved reliability, all at a lower cost. Depending on the application, these portable devices may also have other requirements, such as remote or wireless communication capability.

Standard and semicustom semiconductor components are often unable to meet the requirements of these devices. When these components are not suitable, the best solution for designers may be to use mixed-signal ASICs.

To decide whether ASICs are appropriate, designers need to consider 10 key technical and economic factors: power, size, design flexibility, reliability, functionality, design security, electromagnetic compatibility, cost to design, cost to manufacture, and time to market.

Standard Components

Standard analog and digital semiconductor components include microprocessors, analog-to-digital (A/D) converters, digital-to-analog converters (DACs), and operational amplifiers, as well as memory devices such as RAM, ROM, and EEPROM. With the continual advances in semiconductor technology, particularly in decreased feature size, these components continue to become smaller while offering greater functionality, which has led to a dramatic increase in their use in medical devices.

The chief benefit of standard semiconductor components is that they already exist, providing fast time-to-market and low design cost. But design flexibility is obviously limited by standard components, which may not be able to meet the more rigorous miniaturization, functionality, and power consumption demands of portable medical equipment. Also, devices employing standard components are easy to reverse engineer, which decreases design security.

Semicustom Components

Semicustom ICs address some of the limitations of standard components. Designers can gain flexibility by building required digital functions into a standard gate array, programmable gate array, or digital ASIC. Some limited semicustom analog functions can also be added. The overall power consumption and size of the device in which these components are used can be decreased, while functionality can be increased by custom tailoring the digital portion of the design to more precise requirements.

Time-to-market is typically longer for semicustom ICs than for standard components. However, field-programmable gate arrays can be used to significantly decrease this time. Semicustom ICs also have higher design costs, because they have to be designed and fabricated for a specific application.

Fortunately, improved digital design software enables designers to work at the circuit behavioral level, using high-level hardware description languages, rather than at the schematic level. The software allows designers to simulate the behavior of high-level designs, verifying circuit behavior and performance.

After simulation, designers can use synthesis software to compile behavioral descriptions automatically, producing a list of cells and their interconnections (a netlist) ready for ASIC layout. Some software can even draw a useful schematic from the synthesized netlist. These design tools can shorten the process and decrease cost.

Semicustom ICs offer a solution to many of the standard component limitations, such as device performance, size, design flexibility, and design security problems. For example, because all digital functions are integrated in a semicustom IC, it is more difficult to reverse engineer, which increases design security.

But the benefits of semicustom ICs may not be sufficient to meet all the demands of advanced medical devices, because the ICs are typically limited to digital functions. For some applications, mixed-signal ASICs may be the best answer.

Custom Components

Fully customized ASICs combine desired analog and digital circuits on a single chip. Using ASICs greatly enhances design flexibility, allowing designers to find exactly what they need. For example, a designer can specify an 11-bit A/D converter and not have to settle for an off-the-shelf 12-bit A/D converter. Therefore, ASICs can provide optimum performance in all areas: size, energy efficiency, and functionality. Also, reverse engineering is virtually impossible, because most or all of the circuitry is included in one custom IC.

Mixed-signal ASICs can reduce circuit board component count from 20 or 30 components to as few as five or even one, so there are fewer long-wire connections and board interconnects.

Devices using ASICs are less likely to generate electromagnetic fields that interfere with other equipment, and are less susceptible to ambient electromagnetic fields, resulting in superior electromagnetic compatibility (EMC).

Because mixed-signal ASICs allow designers to specify exact pin-out configurations, connections and packaging can be simple and direct. In some cases, ASICs configured as dies or flip chips with gold or solder bumps will offer enhanced assembly. Flip chips provide flexibility in both size and manufacturing, resulting in a small and reliable product.

Mixed-signal ASICs can deliver product reliability and low manufacturing costs. End-product reliability increases as the number of separate components making up the circuitry decreases. When the number of components purchased and inventoried and the number of manufacturing steps are reduced, manufacturing costs will also decrease.

The benefits of ASICs must be considered along with some disadvantages. The design cycle for products using these custom circuits is significantly longer compared to standard or even semicustom ICs. A minimum of one year from the start of design to the receipt of qualified production parts is common, but this process can be much longer, depending on ASIC complexity. Also, engineering costs are high because—unlike all-digital ICs, which can frequently be designed in-house—mixed-signal ASICs require experienced analog designers who are typically available only through outside sources. ASIC complexity and, therefore, development time and cost, increase directly as digital gate count and required analog functions increase. So, in computing time to market for mixed-signal ASIC solutions, design work must be initiated with enough lead time to compensate for these factors.

Matching Capabilities with Needs

An engineering team must carefully consider the technical and economic trade-offs among standard, semicustom, and custom ASICs against the requirements of the medical device. Frequently, the benefits of mixed-signal ASICs will make them the best—and often the only—way to meet portable medical device demands.

Once the team has decided to use an ASIC, the next task is to choose an ASIC vendor or partner. The technical and economic requirements of the medical device must be matched with the capabilities of potential partners and existing technology.

The engineering team can determine ASIC requirements by posing and answering several questions, such as:

• What are the main functional requirements?

• Are there any special testing or screening requirements?

• What are the trade-offs between size and power?

• Will the device benefit from gold or solder bumps for flip-chip assembly?

• Must the die be backlapped to make it thinner, so that it will take up less volume?

• What are the voltage requirements?

• Are there both low and high voltage requirements on the same IC?

• What are the electrical noise requirements?

• What are the key functions, such as A/D converters and memory requirements?

• What passive components, resistors and capacitors, should be integrated into the ASIC?

• What are the electrostatic discharge protection requirements?

• Over what temperature range must the part function?

• How will the ASIC be tested, both at the supplier and in the final product?

When these questions have been answered, the team can prepare the initial technical requirements for the ASIC and use these requirements to start screening potential ASIC vendors. The screening process will yield a list of technologically qualified ASIC vendors.

Budgetary and Technical Issues

When engineering has narrowed the list of vendors to those who are technologically qualified, budgetary factors can be explored.

The team typically requests estimates from the selected vendors. An estimate will be directly related to the quality of the communication between engineering and the ASIC designers. Each potential vendor should be given an in-depth review of the requirements and the end product. This review is crucial in helping the vendor fully understand engineering's needs, and often results in better ASIC design, because designers who have a firm grasp of the goals of the project may be able to offer unexpected and helpful innovations.

The engineering team should compare the manufacturing cost estimates for the ASIC with the cost of standard or semicustom ICs. But higher cost is not always significant. The higher cost for the ASIC is typically offset by reductions in procurement, inventory, manufacturing, and assembly expenses. Also, manufacturing should explore the cost differences among bare die, flip-chip, and packaged parts. ASIC vendors can often supply estimates for all three configurations.

Estimates may not be received from all vendors that are solicited. The low annual volumes and potential liability of medical product ASICs may keep some vendors from responding.

The estimates that are received should be accompanied by a discussion of pertinent technical issues and information on the vendors' application-specific experience. When reviewing these estimates, the engineering team should ask:

• Did the vendor identify any risks?

• Did the vendor discuss whether cells will be reused or new cells will be built?

• Did the vendor delineate responsibilities; for example, will the vendor be performing all design, testing, and development, and, if not, how will these steps be handled?

• Has the vendor produced similar circuits before?

• Has the vendor produced medical circuits before?

• Does the vendor have reusable blocks and cells already designed?

• Are internal quality systems certified to an internationally recognized standard, such as ISO 9000?

Be wary of an estimate that does not address the technical issues discussed above. Unless the ASIC is extremely simple, this type of information should always be included with the estimate.

Financing and Manufacturing

Each vendor's financial strength and access to manufacturing should also be considered. Questions to be asked at this stage include:

• How long has the company been in business?

• What portion of the business is in ASICs?

• Is the medical market a key one for the vendor?

• Does the vendor have its own fabrication facility (fab)?

• Does it have more than one fab?

• Is the vendor simply a design house that uses outside fabs? If so, how committed will the fab that is used be to this project, now and in the future?

A vendor with excellent technical and design capabilities, but with limited financial or manufacturing resources, is less likely to be a successful partner.

The objective in selecting a vendor is to find a company that will be a successful partner for both current and future projects. This decision must include a full range of considerations: technological ability (both current and future), design skills, experience, financial stability, and manufacturing accessibility.

Price is best factored in during the final stages of evaluation, when only fully qualified candidates are being considered. It may, for example, make sense to pay a bit more for an ASIC if the main cells have already been designed and are proven. Or a higher nonrecurring engineering expense may make sense, because a new architecture will save space and decrease the amount of software code needed. Price is just one of many aspects to consider when deciding on a vendor.

Strategies for Project Success

Once the vendor has been selected, there are several strategies that can help ensure that the design and manufacturing of the ASIC will be a success.

The project should start with firm specifications. Some of these may have changed during the process of evaluating vendors. Before the project starts, the specifications should be reviewed to ensure that they address all requirements. If there are any subjective areas, they should be highlighted to make sure that all parties are in agreement. Specifications that change during the design process can mean disaster.

Adequate engineering resources and technical support should be provided for the project, even if the ASIC partner is completing the entire design.

The client should remain as flexible as possible about noncritical issues. A slight change in a specification may result in significant yield improvement, decreased design time, or decreased risk.

The client should be an integral part of all design reviews and insist on periodic updates of key milestones and development schedules. The best working relationship with ASIC partners involves keeping the communication channels open and keeping everyone aware of project status. To make this possible, both engineering groups must agree on definitions of project stages, from design completion to full production implementation.

The groups should agree on definitions of prototypes or first article parts; on what comprises design validation, preproduction, and production; and on what is required by both parties to release the final ASIC to production. Typically, an ASIC vendor will want to validate the design by fabricating the ASIC at process extremes and running several lots to qualify it before releasing it to full production. Engineering and manufacturing teams should understand this and decide whether they can use devices prior to full production to start manufacturing qualification. Close mutual support and agreement on definitions will significantly help the difficult transition from a completed design to full production.

If the assembly will be produced by a contract manufacturer, that company should be involved in the project, and mechanical samples from the ASIC partner should be given to the contract manufacturer as soon as possible.

During the design process, both engineering groups should keep in mind what tests are required and the testability of both the ASIC and the end product. ASIC design and layout can be changed early in the design process to improve testability (such as by adding a special test pin), but this becomes increasingly difficult if test issues are not addressed until late in the design phase. The client should be prepared to relax testing by decreasing the number of parameters tested or by loosening those parameters if results from production lots determine that doing so is possible.

The need for manufacturers to understand the benefits of ASICs and how best to use them in medical devices continues to grow. After all, the increasing need for portable medical products continues to spur demand for semiconductor components that offer low-power operation, small size, sophisticated performance, reliability, and flexibility. Although mixed-signal ASICs can be more costly than standard or semicustom components, they are often either the best or the only way to meet requirements. Once an engineering team has chosen to use ASICs, it can help to ensure success by finding the most suitable vendor and working in partnership with it.

Jim Gentile is manager, ASIC Products, at Mitel Semiconductor (San Diego) a designer, manufacturer, and supplier of integrated circuits and optoelectronic and analog-line components.