Robot grippers need touch to catch a slip. Microchip's new chip reads tactile skins up to 800 times a second

Robot tactile sensing got a dedicated chip on November 10, 2026, when Microchip Technology introduced the ATFCD52, which reads dense capacitive sensor grids on grippers and humanoid hands at up to 800 frames a second. Microchip says that speed lets a robot notice a slip starting and correct its grip before the object drops or is crushed.

Key points

• Microchip Technology (Nasdaq: MCHP) introduced the ATFCD52 high-speed capacitance digitizer on November 10, 2026, and sells it directly and through its distributors.
• The chip supports up to 52 channels, wired as up to 32 transmit and 20 receive lines, with 20 analog front ends sampling every receive line at once.
• Microchip says it reaches frame rates of up to 800 Hz, depending on array size, and resolves capacitance changes smaller than 1 femtofarad.
• A human fingertip carries an estimated 241 touch receptor units per square centimetre, against 58 in the palm, according to a 1979 study in The Journal of Physiology.
• People correct their grip 60 to 80 milliseconds after a small slip begins, a 1984 study in Experimental Brain Research found.

Why do robot grippers need tactile sensing?

Because a camera can find an object but cannot tell a robot how firmly it is holding it, or whether it has started to slide. Human hands solve this with dense touch sensing at the fingertips: Johansson and Vallbo estimated 241 low-threshold mechanoreceptive units per square centimetre at the fingertip and 58 in the palm, an estimate built from median-nerve recordings in conscious volunteers and counts of the nerve's fibres. In a later study of people lifting objects in a pinch grip, Johansson and Westling found that grip adjustments followed the onset of small slips by 0.06 to 0.08 seconds, and that matching grip force to how slippery an object was depended on input from the skin. A robot hand that is meant to handle soft, fragile or oddly shaped things needs the same two ingredients: many sensing points, and electronics fast enough to read them before the grip goes wrong.

How does the ATFCD52 read a tactile sensor array?

It turns the readings from a fine two-dimensional grid of capacitive sensors into digital data, which a separate processor uses to work out the normal and shear forces at the contact surface. Microchip says the chip drives up to 32 transmit lines and listens on up to 20 receive lines, and its 20 on-chip analog front ends acquire all receive lines in parallel; that parallel capture is what allows frame rates of up to 800 Hz, depending on the size of the array.

ATFCD52 at a glance	What Microchip says
Channels	Up to 52
Sensor lines	Up to 32 transmit (TX) by 20 receive (RX)
Analog front ends	20, acquiring all receive lines in parallel
Frame rate	Up to 800 Hz, depending on array size
Sensitivity	Capacitance changes below 1 femtofarad
Noise handling	Frequency hopping and oversampling
Paired processors	Microchip MCUs, MPUs or FPGAs
Development tools	EV-ATFCD52 board and ATFCD52 Configurator software

Light touches produce very small changes in capacitance, and inside a robot those signals sit next to noise from servo motors, power electronics and switching regulators. The company says the ATFCD52 is designed to resolve changes below 1 femtofarad and uses frequency hopping and oversampling to keep that noise out of the data.

"Robots have become increasingly capable of seeing and understanding their surroundings, but the ability to feel and respond to physical contact remains a critical challenge."

— Giovanni Fontana, Senior Director, Human Machine Interface Division, Microchip Technology

What does robot tactile sensing at 800 Hz mean for hardware designers?

Our analysis: at its top rate of 800 frames a second, the ATFCD52 delivers dozens of tactile frames within the 60 to 80 milliseconds people take to correct a slipping grip. Because the chip digitizes and leaves the force calculation to an external processor, Microchip says designers can pair it with its MCUs, MPUs or FPGAs, picking an architecture for localized control or for higher-throughput computation. Teams building hands for humanoid robots can evaluate their sensor materials and layer stack-ups on the EV-ATFCD52 development board, with configurator software that logs data in real time. Humanoids from makers including Unitree and UBTECH are due at GITEX World RobotiX 2026 in Dubai in December.

The same front end is pitched beyond robots: Microchip names AR and gaming accessories, sensing gloves and floor mats that collect data, and electronic skin worn on the body, part of the flexible and wearable electronics field.

Frequently asked questions

What is a capacitance digitizer?

It is a chip that measures small changes in capacitance across a sensor's electrodes and outputs them as digital values. In a tactile skin, pressing or shearing the surface changes those capacitances, so the readings map where and how hard something is touching.

Can I buy the ATFCD52 now?

Yes. Microchip says the ATFCD52 can be purchased directly from Microchip or through a Microchip sales representative or authorized distributor.

What development tools does Microchip offer for the ATFCD52?

The EV-ATFCD52 development board, for testing capacitive sensor designs, tactile materials and stack-ups, and the ATFCD52 Configurator software, which sets parameters and logs data in real time.

Who makes the ATFCD52?

Microchip Technology Inc., a semiconductor supplier headquartered in Chandler, Arizona, whose shares trade on Nasdaq under the symbol MCHP.

Sources

• Microchip Enhances Robotic Tactile Sensing with High-Speed Capacitance Digitizer: press release from Microchip Technology via GlobeNewswire, 10 November 2026.
• Tactile sensibility in the human hand: relative and absolute densities of four types of mechanoreceptive units in glabrous skin: Johansson and Vallbo, The Journal of Physiology, 1979, via PubMed.
• Roles of glabrous skin receptors and sensorimotor memory in automatic control of precision grip when lifting rougher or more slippery objects: Johansson and Westling, Experimental Brain Research, 1984, via PubMed.
• Microchip Technology: the company's website.