Impinj lets standard CMOS be used in analogue

US start-up Impinj has revealed details of its ‘self-adaptive silicon’ technology, allowing standard CMOS to be used for analogue and mixed signal circuits.

Instead of using more expensive bipolar or BiCMOS processing, Impinj lets designers use CMOS, which opens up smaller processes to analogue circuitry.

“Our technology allows us to modify individual CMOS transistors after they’ve been manufactured,” said Bill Coleran, CEO of Impinj.


Transistors on standard CMOS wafers from a foundry will have large variations in electrical properties, making them unsuitable for analogue and mixed signal.


Impinj’s self-adaptive silicon allows transistors to be matched after manufacture.

To control the characteristics of a circuit, Impinj adds dual transistors connected by a floating gate to the circuit. Electron injection and tunnelling are used to modify the charge on the floating gate of the pFET device.

“Once we’ve calibrated the devices, the setting are remembered in a non-volatile way,” Coleran said.

The programmed device can be used to set the basic parameters for circuits such as a current source. The charge on the control transistor can alo be reset in the field, should the circuit parameters change or if charge leaks away over time.

Impinj has three phases in its roadmap. This year it plans to show off some demo circuits, including amplifiers, mixers and a 14-bit, 250Msample/s DAC. The latter, a 0.17mm² device, has been fabricated by TSMC on a 0.25?m process and has 80dB dynamic range.

Next year the firm will develop standalone products for wireless applications, including a wideband A/D converter for software radio.

Phase three will see the firm developing complete system-on-chips, integrating analogue with digital on standard CMOS.

Impinj is a spin-off from the California Institute of Technology, and has exclusive rights to some 14 patents filed by the Institute. Moreover, the firm was co-founded by Dr Carver Mead, regarded as one of the most prolific researchers in analogue and VLSI design.

Why integrate the analogue?
Many chip manufacturers contend it is simpler and cheaper to have two devices in a system, one for digital and one for analogue. This is certainly true today for products such as mobile phones.

“In today’s world integration has focused on digital. Typically the analogue is off-chip,” said Coleran. “But, if we can’t get the analogue on-chip with the digital, it doesn’t matter how much you scale down the digital.”

Because Moore’s Law doesn’t apply so well to analogue, products such as mobile phones won’t benefit from process scaling unless the analogue is included, he said.


Leave a Reply

Your email address will not be published. Required fields are marked *

*