In linear tape drives, the read/write transducers in the head are connected to the main electronics card of the drive using a 10–20-cm-long flex cable. A 16-channel tape drive typically has two head modules with 16 write transducers and 16 read transducers in each module, in order to provide “read while write” verification. Although only half the total number of readers and writers are used simultaneously, the flex cable(s) must still provide routing for all of the transducers for a total of about 144 lines including servo readers and shielding. The flex cable has a thickness of about 80–100 μm and is composed of a thin film of base material such as polyimide, with patterned copper layers on either side which are protected by thin cover layers glued to the base film. The relatively long cable length is necessary to provide a sufficient range of motion for the head to access the full width of the tape while providing a low stiffness such that the cable does not interfere with the mechanical response of the track-following actuator.
The 2012 INSIC Tape Roadmap forecasts a fifty percent increase in data rates every two years or essentially with each new generation of tape drive. This is expected to be achieved through a combination of an increase in the number of parallel channels as well as through increases in linear density and tape speed that provide an increase in the data rate of an individual channel. Both of these strategies give rise to specific challenges for the design of the head flex cable. First, increasing the number of transducers that operate in parallel leads to an increase in the number of copper traces that must be routed along the cable, resulting in an increase in cable cost, size, mass and stiffness. The increase in mass and stiffness can cause a degradation in the performance of the track-following control system. Second, increasing the per channel data rate can be difficult due to the challenge of matching the impedance of the readers and writers to that of the cable and the analog front end electronics on the electronics card. These impedance matching issues can limit the maximum bandwidth of the read-back process as well as the current switching time in the write elements which in turn limits the maximum write speed as well as impacting the quality of the written transitions and hence the SNR of the read-back signal.
Both of these challenges can be addressed by the integration of some of the front-end electronics directly on the flex cable adjacent to the head. Currently, we are investigating:
- the design and integration of a CMOS multiplexer directly adjacent to the head to select the active channels and reduce the number of conductors that must be routed along the cable and,
- the design and integration of a write driver circuit adjacent to the head in order to achieve faster write current switching times and at the same time reduce power consumption and heat dissipation, see Figure 1.
The cost of the chip and the assembly process onto flex cables are critical to the viability of this approach. We are currently pursuing a flip-chip assembly technology, which can reduce the assembly costs and chip area, see Figure 2. Flip-chip bonding also provides good heat sinking of the chip, another critical aspect that we are investigating using finite element modeling, see Figure 3.
The close integration of the front-end electronics with the read/write transducers will potentially become one of the key enabling technologies to facilitate the continued scaling of magnetic tape-recording to higher areal densities and data rates.
- Figure 1. A prototype write driver chip with C4 solder balls for flip chip attachment to flex cable.
- Figure 2. Front end electronics mounted on a head flex-cable. The chip is mounted using a flip-chip technology.
- Figure 3. Finite element model of heat dissipation in a chip mounted adjacent to a head module on a flex cable.