Liquid Instruments has introduced generative instrumentation to the test and measurement industry with the launch of its Moku:Delta 2-GHz test platform. Claiming infinite instrument configurations, generative instrumentation enables engineers to create custom instruments and configure complex systems to meet their exact requirements using AI and natural language prompts.

Representing a significant leap beyond traditional instrument design, agentic AI enables engineers to customize and optimize their test setup for specialized applications, automate complex test scenarios, and respond to evolving test landscapes, according to Liquid Instruments.

“We see a future where engineers and scientists describe what they want, and the instrument configures itself,” said Rory Smith, lead AI instrumentation engineer at Liquid Instruments, in a statement. “Generative instrumentation won’t just simplify everyday tasks; it will enable the creation of entirely new capabilities on demand, unlocking test setups that were previously impossible or impractical.”

The new AI capability, supported in an upcoming software release, is available in the new Moku:Delta, the company’s fourth-generation hardware platform, joining the Moku:Go, Moku:Lab, and Moku:Pro platforms. The Moku platform offers a reconfigurable suite of test and measurement instruments in a single device.

At the heart of every Moku device is an FPGA, which can be configured to run in a variety of instrument modes, Danielle Wuchenich, COO and co-founder of Liquid Instruments, said during a keynote address at BWM Welt. “At one minute, you can run it as an oscilloscope and look at your signals in the time domain, then switch it, and within a matter of seconds, the FPGA will be reconfigured and you can look at it in the frequency domain using a spectrum analyzer.”

“As FPGAs are getting bigger all the time, the Moku OS is an absolute ninja at slicing and dicing the FPGA, which lets you run multiple instruments on the same hardware,” she continued. “We can run multiple instruments in different regions on the chip, and this has a few great implications for what you can do. You can wire up multiple instruments together, and all of them are on the same physical chip, getting these amazingly high-speed, lossless digital interconnects running at 20 Gbits/s.”

Two key features of the Moku platform are Multi-Instrument Mode, enabling users to configure and run multiple instruments simultaneously without additional hardware, and Moku Cloud Compile, which enables users to write and deploy custom code on the FPGA. Together, they enable engineers to create custom test systems in a single device.

Moku:Delta 2-GHz test platform with Multi-Instrument Mode (Source: Liquid Instruments)

AI capabilities

Before launching its generative instrumentation, Liquid Instruments integrated AI help, trained on its internal documentation, and the Moku Neural Network into its test platform.

“One of the most transformational technologies that has emerged in recent days is neural networks and how they can be used to solve a whole range of problem classes,” Wuchenich said. “Our users are really interested in this technology, but implementing them in the real world has been a challenge. A lot of times they’re confined to being run on a computer, so inferences are often done on CPUs or GPUs and connected over a network with hundreds of milliseconds of latency.”

With the Moku Neural Network, it runs directly on the Moku instrument, enabling new capabilities such as adaptive denoising, advanced signal analysis, and inline intelligence for real-time decision-making. “It lets us connect up signals using Moku’s I/O, so we can get physical signals in and out with sub-millisecond latency,” Wuchenich said.

Bringing agentic AI to test instrumentation, the company calls the latest platform a launchpad for the future of instrumentation, combining hardware, software, and AI. “We’ve made the first steps incorporating elements of AI into our products with AI help and then with the Moku Neural Network,” Daniel Shaddock, CEO and co-founder of Liquid Instruments, said during the keynote presentation. “This is a big leap forward where for the first time, we’re bringing agentic AI to the test and measurement world.”

At a very high level, this is how generative instrumentation works in the Moku:Delta platform. At the prompt, the user tells the Moku agent what instrument to build with key specifications. The user then clicks to build, which takes the agent between three to 15 minutes, depending on the platform and complexity, compiling the system so it can run on the FPGA.

When it’s finished, the user can verify the generative instrument in three ways: asking the agent to describe the instrument (what it does, how it works, and how to use the instrument), which is provided in a plain language version; opening the VHDL code in the Moku Cloud Compile to review the code and modify if necessary; and generating test configurations, such as debugging the generative instrumentation with a waveform generator and an oscilloscope.

“To do what we just did a few years ago without using this technology would have required me to tell an engineer ‘I’d like you to build this system,’” Shaddock said. “It would require that engineer to collaborate, build a design to identify components, fabricate circuit boards, manage the supply chain, build it in a factory, and then ship it across the other side of the world.”

Shaddock said this process would take months at best, and with generative instrumentation, it can take less than five minutes. “I think it’s one of the first examples that I’ve seen in our industry of generative AI being used to produce what previously was a physical artifact, and I think that’s quite interesting.”

AI configuration can also extend to non-generative instruments, making it easier for users to reconfigure the Moku test system.

At the same time, the company is bringing more powerful hardware, offering higher bandwidth, more channels, digital I/O, and lower noise. The Moku:Delta offers 3× higher bandwidth and one-third of the noise compared with the Moku:Pro, double the channels, and higher speed. It also features the first digital I/O in the flagship model, and the most digital I/Os in any of the systems.

Claiming the highest-resolution 2-GHz oscilloscope and the only spectrum analyzer that provides full 2-GHz bandwidth down to 0 Hz for 1/f noise measurements, along with a flexible eight-channel arbitrary waveform generator, the Moku:Delta platform provides more than 2 billion possible instrument configurations with up to eight instrument slots and 15 supported instruments.

Key specifications include <10 nV/√Hz at 100-Hz noise,  5-GS/s sampling rate on analog inputs, 10-GS/s sampling rate for analog outputs, 1-ppb clock stability, high-speed streaming with 100-Gbits/s QSFP, and 32 digital I/O channels. Connectivity interfaces include Ethernet, SFP, QSFP, and USB-C.

Future-proofing the Moku platform, the company expects continual AI-powered enhancements, including the capability to deploy custom algorithms to the on-board FPGA. In addition, the platform offers integration with Python, MATLAB, and LabVIEW.

Moku:Delta, unveiled at BMW Welt in Munich, is available to order now. First shipments are expected in August 2025.

The post Liquid Instruments unveils generative instrumentation appeared first on Electronic Products.