Q&A with Scott Reynolds, IBM Research scientist specializing in integrated chip design.
A winner of a Pat Goldberg Best Paper Award for Organic Packages with Embedded Phased-Array Antennas for 60-GHz Wireless Chipsets (pdf), Scott talks about the challenges of designing an organic package capable of transferring data at very high rates.
What problem did you set out to research when you began developing an organic package with antennas and a microchip?
Our integrated circuit team wanted to build a chipset you could use to stream uncompressed video data at five gigabits per second. That's a really high rate, and that's a lot of data.
Almost any data source, whether it's coming from a DVD or streamed from an online source, is compressed. But when it gets to that box in your living room, it's uncompressed. You actually do not want to compress it when you transmit it to your television set: Every time you compress and decompress data, your video image is going to get a little bit worse. That compression-decompression cycle also adds latency, or time delay. The latency involved in these compression-decompression algorithms is hundreds of milliseconds. That's too long for movies and way too long for video gaming.
Remember too that the consumer electronics companies have to pay royalty fees every time they go through a compression-decompression cycle.
Why are you working with organic materials instead of semiconductors?
Basically, we're building radio-frequency systems at really high frequencies that in the past operated only on cumbersome hardware. You needed exotic semiconductors technologies, such as gallium arsenide and Indium phosphide. You needed expensive hardware to run at these frequencies. Now we can build things inexpensively using silicon, circuit board techniques and organic packages (carbon-based conductive materials).
Our organic package is unique because it contains both a chip and antennas. We were looking for an economical way, in terms of materials and funding, to build a radio that operates at very high frequencies -- 60 gigahertz (GHz), or at about 30 times higher frequency than the WiFi card -- capable of transferring data at very high rates. The challenge was to do this inexpensively so we could potentially put it into a consumer product.
We aren't talking here about conventional radio signals. We're talking about signals that best operate in a small space, like a living room, and in a line-of-sight way. When you operate at these very high frequencies, you can't easily separate the antennas from the integrated circuit (IC), or chip. Moreover, all practical packaging materials are really lossy. That is, much of the radio signal generated in your IC can be lost in the package. Not to mention that at these frequencies, the radio signal disappears if it travels too far. Hence, the need for creating a package with a lot of antennas -- the phased array -- so you ultimately can steer the radio signal as a “beam” around the room.
There are also simpler applications of 60GHz that can use simpler packages, such as when you point your mobile device at a video data source and download your movie, music or game in a matter of seconds.
Incidentally, the small-range nature of 60GHz technology can actually be a benefit. You could set up private, "impermeable" networks throughout your house. Sixty gigahertz technology is innately secure.
How did you manufacture the organic package?
My colleague Duixian Lui, as well as our other colleagues, designed the antennas and all the layers inside the circuit board. We did a lot of the material characterization here in Yorktown. When the design was complete, Duixian transmitted a file to the fab where it was manufactured.
A lot of issues came up in manufacturing the whole package. For instance, the early versions would just peel apart. There was a lot of blistering.
One other thing: Even though we talk about the relative low cost of using organic materials, manufacturing this package with organics was still very challenging. We ended up experiencing some manufacturing delays.
How did you test your results?
Basically, we were measuring the antenna patterns -- simultaneously -- of all the antennas in this package. And by measuring those antenna patterns, we could see whether the package was working or not.
The next level of testing was to build a communication link using the chip and the package to see if we could transmit data. That's the final test of whether all these assumptions that went into the chip, package system and design actually worked.
What makes the 60GHz market important?
This 60GHz technology is being considered for a lot of potential applications. At the time that our team developed the chip-antenna package, the market was still immature. Sixty gigahertz in living rooms really hasn't taken off, yet.
But in the not-too-distant-future, 60GHz is a potential technology for increasing the speeds of wireless LAN in your computer.
On a good day today, you can get to 50 megabits with your wireless LAN card. But there are already Broadcom and Intel applications that need higher data rates than that. So, 60GHz is going to appear in the kinds of applications those companies are investing in.
Let's take an example from everyday life. Say you have a smartphone. You might want to synch it with the data on a hard drive. You might want to download a movie. You would have to transmit a lot of data in a short period of time. So, you're going to need high data rates. Enter 60GHz.
What are your next steps?
The system we're working on right now is very interesting because it's for collision-avoidance radar in helicopters.
Helicopters don't have very sophisticated radar systems. When a pilot loses his visual bearings, he is liable to fly into a wire. Sometimes we hear about helicopters plunging into the Hudson or East Rivers. What can happen is that wind and surf conditions are such that they throw up a lot of water into the air, and the pilot loses his bearings.
If you had a high-resolution short-range radar system -- the kind you could build with our organic packages -- every single civilian helicopter could have one. If you tried to do that with conventional technology, it would weigh hundreds of pounds. A helicopter couldn't take off! I think that in 10 years, all helicopters will have them.
Labels: chip design, microprocessor, Pat Goldberg