Audio playback solutions for medical and healthcare equipment

In order to indicate unsatisfactory physiological patient states, alarm signals have been integrated into the monitoring solutions, the broad variety of default alarm settings did not provide a unified or easily manageable scenario. As patient safety depends on the operator's ability to understand the alarm signals, the International Electrotechnical Commission (IEC) has provided the first focused standard on audible and visual alarms for medical devices. The standard – IEC60601-1-8 – defines a specific melody that corresponds with a specific physiological function. This ensures that alarms are homogenised, instead of varying between manufacturers. It also limits the number of alarms to eight, with a cautionary and an emergency version of each. The IEC also requires the note must consist of the fundamental tone and at least four harmonics. Audible alarms with richness in harmonics make them easy to localise and more resistant to masking. The alarm's priority determines the number of notes and the dynamic characteristics of the note sequence. The fundamental and four harmonics of alarm sound defined by IEC60601-1-8 must not differ by more than 15dB in amplitude. With a conventional architecture, it is sometimes difficult for sounds to comply with the standard: complex knowledge of audio design and acoustics is mandatory. Rohm's audio playback ics offer standalone solutions with high audio quality that are easy to integrate. With memory and speaker amplifier integrated, there is nearly little analogue board layout needed. By selecting multichannel playback ics, several sounds can be played simultaneously (up to four channels are mixed). For example, voice guidance can tell the user how to use blood pressure equipment, then give the measurement results. At the same time, another channel can output alarm sounds, button confirmation sounds or even music. The ic has direct access to internal or external memory to read previously recorded and compressed audio content. This data is decoded and converted by a d/a converter into its original analogue format. The signal is then amplified and played back by a speaker. No data processing is required by the host mcu, which only controls the ic by sending such commands as play, stop and volume control. With such systems, the mcu's resources are kept free for other processing requirements, such as calculation, motor control and data display. ROHM/OKI's portfolio consists of four main categories, with combinations of external or internal memory and external or internal speaker amplifier. In addition, versions with an integrated mcu are available. For those chips with internal memory, users can choose between flash, where voice data can be rewritten after mounting, and mask rom or production programmed rom. In order to reduce package size and board space, versions with external memory offer a serial interface. External memory is more flexible if more than 10Mbit of memory is required. Parallel interface flash requires a 16 or 18bit address bus, plus 8bit address bus and a control signal. This requires a high pin count package. The spi interface flash is limited to four signals (CS, Serial Clock, Serial Data in, Serial Data out), allowing the device to be supplied in an sop8 package. While standard 16bit PCM audio is not optimised for this kind of application, other compression methods, such as ADPCM, are a good compromise between complexity and compression rate. In order to achieve a higher compression rate with a high sound quality, Oki and the Kyushu Institute of Technology developed the High Quality ADPCM (HQ-ADPCM) method, which guarantees a further 20% reduction, compared to 4bit ADPCM. Playback time at a sampling frequency of 8kHz and with 1Mbit memory is 40s when using HQ-ADPCM (compared to 8s with 16bit PCM and 32s with 4bit ADMC). Also, Rohm/Oki use a unique compression technology for better sound quality with minimal file size, making them compatible with music and voice guidance applications. For a sampling rate of 16 or 32kHz, the algorithm can be set up with compression rates ranging from 16 to 192kbit/s. While the 16 kbit/s option will generate bigger files with high quality, the 192kbit/s option produces small files, but with low quality. Typically, 32kbit/s is a good compromise. In order to generate rom code, the designer can use a software tool to convert .wav files into the compression format. D-Waver or Speech-Utility software can generate a rom image as a binary file, which can be then loaded into memory. Each file can be encoded with a different sampling frequency and compression rate. The mcu can play a specific phrase or sound by sending a simple command to the playback ic via i2c or spi bus. Alarm sounds or voice guidance do not require very high power and 8? speakers can be driven with 0.5W at 3.6V and 1.5W at 5V. The required output power will depend on the application. For example, a syringe pump will need to be louder so medical staff can hear alarms from an adjacent room. A diabetes meter alarm, however, may not need to be as loud. Medical applications can be split into two main product families: mains driven medical equipment, such as syringe pumps and patient monitors; and battery driven portable applications. To meet rising demand for miniaturisation and power reduction, Rohm offers a range of speaker amplifiers. These 2.5W Class D devices come in a wafer level chip scale package measuring 1.6 ×1.6 ×0.55mm. An efficiency of 90% contributes to low heat generation and energy conservation. In addition, only three external parts are required, making them suitable for compact electronics of all types. The ML610(Q)34x series of low power mcus have a hardware audio decoder with integrated flash or mask rom. The cpu core is a high performance cmos 8bit microcontroller, which can execute one instruction per clock by parallel processing with a three step pipeline architecture. The chip integrates program memory,ram, peripherals, a 12bit a/d converter, uart, spi, a timer, a general purpose port, audio decoder, digital filter, 16bit audio d/a converter and a speaker amplifier. Because it can operate in slow mode and low power consumption mode, it is suitable for battery driven applications and its current consumption is such that it can operate from batteries for 10 years. The ML610(Q)34x series achieves low current consumption at sleep mode (0.5µA in STOP, 1.5µA in HALT) and the device will maintain operation until the voltage descends to 2.3V. The Bill of Materials is reduced by implementing all necessary functions for audio output (except the speaker) into a single chip. In addition, the integrated mcu can handle other functions, such as led or sensor control. Author profile: Finn Lange is a product marketing manager with Rohm Semiconductor.