Blog — Innova-Harmonics

FEB 27 · 2026 DESIGN · SOFTWARE

Weather Update and Huge Steps

Hello readers! Long time no see (okay, so we missed last blog post last week) but we promise today will make it up to you. We have two big things to share: the workflow that got us to the starting line with our first customer, Natural Way Food Group, and also the just-as-big announcement: we have a beta dashboard. Let's get down to the "how we revived some equipment from 2008" and go from there.

Starting, we want to give a hearty thank you, and mini-rant, to our pilot partner Natural Way Food Group. They're a Fayetteville, Arkansas based peanut butter manufacturer, and what they're up against is the ultimate example of why manufacturing in the United States is so difficult. It's also an ultimate example of what Innova-Harmonics can do about it. When we met Austin, NWFG's president, it was clear his goal was to produce his product for as many people as possible — a totally valid dream considering that after having a sample of their peanut butter, multiple of Innova-Harmonics' founder households will now only be having their peanut butter from now on.

This dream however had a clear issue: the obstacles of modern industry in general, and also something that Innova-Harmonics has been seeking to remedy — the old "get old equipment and develop it or retrofit for new equipment" debacle. Each approach has its strengths and weaknesses, but after consulting several engineers, Austin was told a good approach was to retrofit, considering the used equipment he already purchased has outdated controls hardware from around 2003.

This shouldn't have been a big issue. The whole selling point of PLCs and industrial equipment is/was that once you have it, it'll be good for 20–50 years and what works once will always work reliably. This is pretty much true considering our approach — a bit outside of the box, but ultimately effective in reviving the PLC and getting to our next steps, which is tracing I/O and double-checking the program for running the machines it's attached to.

Starting things off, the PLC was running on RSLogix5000 v11.28, a software from 2003. After consult with Allen Bradley, we got the go-ahead to download that software with a license, but with the catch that it physically can't run on modern operating systems. We then decided to run a virtual machine with Windows XP. Yes, that Windows XP — with the 3D pinball and minesweeper. Oddly, after this decision, things moved quite well, because back in the day Allen Bradley would mail you floppy discs or CDs with their software downloadable in parts. In VM software, this was as easy as porting those same downloads from Allen Bradley's website in as virtual discs to then be opened exactly like if you put a physically burned CD into a disc drive back in 2003. That intuition aside, there was some extra stuff like having to download parts of the software in Windows 11 and then porting it through back to Windows XP, but ultimately the worst of it was network bridging to the old hardware.

The panel had two ethernet cards: one for a machine-internal intranet, and another for wider-scope factory networking. That second card, an Allen Bradley 1756-ENBT, still had its IP address. After polling for ARP signal using Linux's built-in networking signals for ARP, it was clear that the ENBT card was wanting to connect to something on the factory's wider network, but nothing else happened. Routing traffic through the bridge was tricky, but after making the host computer's IP address the IP that the ENBT card was looking for, it allowed for CIP interaction — which officially allowed the software RSLinx, and therefore RSLogix, to talk to the hardware. The power of Linux in a nutshell. In the future, Innova-Harmonics hopes that all controls engineers know heads from tails with networking between virtual machines. After this was all said and done, we've officially come online with the PLC to help NWFG on their next steps of getting machines online. For any "higher tech" looks, check out our CEO's LinkedIn posts about each subject.

Allen-Bradley control panel at Natural Way Food Group showing PLC modules and contactors — The Allen-Bradley control panel at NWFG — original 2003-vintage PLC hardware brought back online via a Windows XP virtual machine.

Now, in just-as-exciting news, we've been hard at work developing our big reveal: Innova-Harmonics' digital twin initiative. We have a v0.3 for pilot partners now, with development daily adding in new features and making the platform that interconnects Octopus sensors and their host plants/factories. Breaking into the world of IIoT and predictive/preventative maintenance, Innova-Harmonics wants to demonstrate that we're building a perspective for industry to solidify the power of digital twins. Starting off, our tool is used to dashboard industrial data collected by our sensors for use in facilities, but next on the chopping block is using this tool to predict correlated downtime of machines. Imagining what this can do is limitless — with next initiatives being getting Octopus vibrational data to be part of an in-tool design FEA, or CAD-importable FEA analysis. You could design elements of your machines relative to downtime or flow of other machines. Even with just our correlation models, you could find ways to drive machines after failure to compensate for downtime that has happened or will happen. (We can simulate failure relative to our sensors, after all.)

The tool is the beginning of what will become the best of our business — giving engineers the tools they need to characterize machines live, while also being able to correlate that designed information with vibrational datasets and corollary datasets of machine failure. If you're interested in seeing what this sort of platform can do for your facility, don't hesitate to reach out to us. We're going to be rolling out this tool and future products in our pilot facilities for massively discounted rates of what would be traditionally seen out of industry. We want to experiment and grow with your team.

Single-asset Reef dashboard view showing Hydraulic Press #2 with health score, sensor traces, and uptime timeline — Single-asset view of the v0.3 Reef dashboard: hydraulic press #2, with health score, sensor traces, and uptime timeline.

There's a lot going on in these screenshots, but we'd like to break down the vibe of the images. Ultimately, this is close to exactly what our customers would be looking at on our software. The goal is to show how vibrational data, acoustic data, temperature data, and magnetic-field data are correlated with specific changes in machine state, then ultimately lead to predicting and preventing machine failure.

Cross-channel correlation dashboard view showing machine traces overlaid against operating state — Cross-channel correlation view: machine traces overlaid against state changes for a single asset.

When the failure is imminent, we are developing those corollary models which let users know what machine failure means in context of their other machines. This can help planners get ahead of knowing how downtime actually impacted their bottom line. Imagine a network of sensors at each machine, networked to other sensors to predict these points for each machine uniquely. Oh wait — you don't have to.

Networked plant-floor view of multiple instrumented machines with correlation graph and downtime predictions — Networked view of multiple instrumented machines on the plant floor, with the correlation graph and downtime predictions for each.

That's right, ladies and gentlemen, we've already taken the liberty of doing this. Of course now we don't have perfect models — ones that are mechanically modeled exactly as they are in real life. Think however where we're going with this: full mechanical/electrical modeling will make stories like Austin at NWFG's a story of the past.

The big-ticket item here is that when we're at full maturity and we've got octopi on machines, we'll have all the mechanical data we need to know how a machine functions and fails in lieu of other machines. Let's take this a step further: what if we could do the same for the electrical signals inside those machines which motivate those machines' states? It no longer matters to have perfect CAD models. It also no longer matters to have perfect controls designs with modern hardware. Any old PLC does the same thing as a new PLC — that is, turn on or off registers of relays for 24VDC signals. Our future plans are to bring to term a digital twin software of the future. One where we completely simulate and optimize existing systems to get ahead of failure causes, and get further ahead of process flow.

It's all possible. You should ask yourself when you want to be part of it.

FEB 7 · 2026 DESIGN · TECHNOLOGIC

Cooking up a report, but first, something fun

To begin this post, it is worth sharing a company update. WE HAVE A FIRST CUSTOMER. More accurately, we should say "user," as we are not asking for money from this group. They have given happy consent to be included in today's blog, and they are called Natural Way Food Group — a Fayetteville-based peanut butter manufacturing company. We will take this opportunity to say that if you have not tried their peanut butter, you should. It is genuinely excellent. Their product is a low-ingredient-list novelty of the modern age. They are a perfect study point for our group as a food-industry manufacturer here in Northwest Arkansas.

Further, because they are a growing factory, they have growing factory problems. These are the same struggles faced by most factories, but amplified by a lack of support from the traditional vectors the industry relies on. This incoming rant will be saved for the next blog post, where we will dive more deeply into what we are actually doing with NWFG, but it is hard not to call out how poorly they have been treated. At Innova-Harmonics, we believe successful manufacturing comes from focused, valuable engineering attention — and it borders on criminal how difficult it is to provide that when starting a factory from scratch.

For next week's update, look forward to how we used a virtual machine to revive a piece of software that is over twenty years old, and hopefully a machine of similar age in their facility. In the meantime, today we are going to talk about a fun dataset we came across while studying vibrational harmonics and their use in machine learning for industry. I, Nathaniel, am going to do my best not to butcher the data science behind the paper and the dataset, while almost certainly earning our CSO Micah, a data scientist, a few extra points on his blood pressure score.

The dataset we are talking about today is "ToyADMOS: A Dataset of Miniature Machine Operating Sounds for Anomalous Sound Detection." It is a genuinely interesting paper, available on arXiv: arxiv.org/pdf/1908.03299.

So as you can see from the title, figures, and abstract, this paper and dataset are focused on compiling information for anomaly detection using toys. This approach strikes us as genuinely clever, especially having seen firsthand the cost associated with analyzing full-scale industrial machines. Innova-Harmonics gets particularly excited when we read even the opening line of the introduction:

"Since anomalies might indicate faults or malicious activities, prompt detection of anomalies may prevent such problems. Microphones have been used as sensors to detect anomalies, referred to as anomaly detection in sounds (ADS) or acoustic condition monitoring, in many applications such as audio surveillance, machine condition inspection, and fault diagnosis."

— Koizumi et al., ToyADMOS

Honestly — that is really cool.

They go on to state that, to the best of their knowledge, no freely available datasets exist for anomaly detection in sounds, and we agree. It is striking just how valuable this line of research could be if paired with the right datasets. I am struggling to remember whether I have said this before here on the blog, but where we find ourselves in industry is at the intersection of "we know what we want" and "we know it is hard to get." Individuals and organizations can often provide electrical hardware, software, machines to pull data from, or the means to compile that data — but very rarely all of those pieces together in one place.

The paper addresses this problem by working with three different toy types and introducing several ways those toys can "malfunction" through intentional damage. The data is then labeled as "normal sound," "anomalous sound," and "environmental noise." The environmental noise samples are included to simulate different factory conditions, with the goal of building simple baseline models that can later be adapted to more realistic use cases. This approach is refreshingly approachable, and we hope it leads to stronger and more accessible research in this space going forward. We certainly found the dataset useful, if only as a motivator for the approaches we are now taking with other data sources and other machines.

ToyADMOS figure showing toy car, inspection rig, toy conveyor, and microphone arrangement — Figure 1 from Koizumi et al., *ToyADMOS* — toy car, inspection rig, toy conveyor, and microphone arrangement used as miniature stand-ins for industrial machinery.

ToyADMOS figure showing recording setup positions for each toy type — Figure 3 from Koizumi et al., *ToyADMOS* — top-view recording setup positions for each toy configuration.

Those reading should check out the paper and further their interest by going the extra mile and downloading the dataset. We believe approaches like this one would go really far in ensuring that industry "catches on" to approaches like this one for characterizing machine failure.

JAN 28 · 2026 DESIGN · SOFTWARE · TECHNOLOGIC

Can You Afford To 'Ignore' Data?

Putting Vibrations In a Silo — When You Can't Afford To: cover collage — Putting vibrations in a silo: a survey of where modern industrial maintenance strategies stand, and the data complexity behind each one.

"Ignorance is bliss," or so the old adage goes. At Innova-Harmonics, our goal is to educate members of industry that ignorance is not bliss. We've spoken with countless engineers in plants who want to know as much as possible so they can properly plan, operate, and improve their facilities. The sad truth is that the pool of people who can reliably turn raw data into actionable insight is shrinking — and worse, much of the data being collected today simply isn't useful in practice.

That belief is what initially motivated the design of Octopus. We knew that machines already contain the data needed to motivate better design and operational decisions. The real question wasn't whether the data existed, but which data actually mattered, and which signals were worth chasing. The answer to that question is what ultimately shaped the sensor layout we've implemented on the device.

What's especially difficult about being an engineer in manufacturing is the constant need to divide your attention into buckets and prioritize only the problems that are easiest to access. All the while, there's an unspoken expectation that when downtime becomes painful enough, it's your fault for not paying close enough attention. We don't believe engineers intentionally ignore their data — but we do believe modern sensors and control implementations make certain datasets incredibly inconvenient to find, correlate, or trust. This frustration isn't unique to industry. It's reflected directly in modern academic research on vibration analysis, where even researchers struggle to extract meaningful information from what is often a chaotic signal landscape.

"Modern predictive maintenance research consistently notes that vibration signals are complex, noisy, and highly sensitive to operating conditions, making single-signal interpretation unreliable even in controlled research environments."

— Gawde, Patil, et al., Multi-Fault Diagnosis Of Industrial Rotating Machines Using Data-Driven Approach: A Review Of Two Decades Of Research

"Reviews of smart manufacturing systems identify fragmentation of sensor data pipelines and limited fusion strategies as a major barrier to extracting meaningful insight from industrial data."

— Tsanousa, Bektsis, Kyriakopoulos, et al., A Review of Multisensor Data Fusion Solutions in Smart Manufacturing: Systems and Trends

Vibration on its own isn't enough anymore. And yet, plants are still accustomed to making critical decisions using narrow, single-line data pathways. Engineers worth their salt know how to make calls that save plants real money, and they know how to justify those decisions with rigorous data. The problem isn't the engineers. The problem is that data collection is often constrained by legacy control systems and outdated standards. Wouldn't it be better if system maintenance — and the prevention of downtime — were driven by higher-quality, higher-context data from the start? We think so.

Take a look below. This figure comes from Multi-Fault Diagnosis Of Industrial Rotating Machines Using Data-Driven Approach by Gawde, Patil, et al.:

P-F curve diagram showing machine condition over time, with detection points labeled by maintenance strategy and a rising cost-to-repair curve below — The P-F curve from Gawde, Patil et al. — machine condition over time, with detection points (vibration, oil, temperature, noise, heat) tied to the maintenance strategy that catches each one, and the cost-to-repair curve climbing underneath. Used with permission of the author.

The goal, of course, is to identify the true "start of failure" within a machine, using techniques that are still largely novel to industry. Vibration analysis is only the beginning. A broader machine view offers far greater clarity. Now imagine going even further: incorporating richer operating context into models that improve over time as they learn from real operation. Industry will adopt these technologies. The real questions are when, and how effective the implementations will be when they do.

Shouldering the responsibility of making decisions in a factory shouldn't be a solitary burden. You can always hire more talent, but what if the mechanical problem you're chasing is riddled with blind spots you can't even see because the data context is too narrow? Would you knowingly take that risk?

In future blog posts, we'll explore the evolution of maintenance strategies and the experiments that led to the academic discoveries discussed here. Stay tuned, leave a comment, or send us an email — we'd love to hear from you.

In conclusion: this is a tough job, and often an even tougher call. We want to hear from those facing challenges like these. If you're interested in sharing your story, or discussing it on our blog or LinkedIn page, don't hesitate to reach out: nathaniel@innovaharmonics.com.

JAN 21 · 2026 DESIGN · GADGETS · YAPPING

We Are Innova-Harmonics

Close-up render of an Octopus prototype board — Close-up render of an early Octopus board — the head of the device that hosts the sensing elements and signal chain.

Welcome to Innova-Harmonics.

As a first blog post, it's hard to explain not just what we are, but what we could become. So rather than oversell a finished vision, we'll start with something simpler: intent.

Let's talk etymology.

Innova-Harmonics comes from innovation and harmonics. We want industry to think more broadly about vibration analysis, and beyond that, about machines as harmonic systems. Machines don't just run — they resonate, interact, drift, and respond. Understanding that is where better control, maintenance, and optimization begin.

That idea led us to our flagship platform, Octopus (more on the name in a moment). Octopus can operate as part of a controls system or independently, equipping operators, managers, and maintainers with data that actually reflects the machine's operational reality.

How exactly does it do that? Honestly — we're still discovering the best answers.

We've only been operating since October 2025, and as of writing this, we're not even officially incorporated yet (that changes tomorrow, hopefully). But in that short time, we've designed an industry-compliant, robust industrial electronics platform that can attach to almost anything that vibrates — which turns out to be most machines. That data feeds machine-learning models we build in-house to deliver plain-English insights via an HMI, email, or even a group chat.

The road so far has been encouraging. Partners and interested parties, simply by understanding where this technology could go, have wanted to keep tabs on us. We don't claim to know everything. What we do claim is a commitment to improving industry by approaching machines differently than what's typical for our region of the U.S.

Maybe we're destined to classify shaft eccentricity on a conveyor.
Maybe we're just here to tell you a "clunk" clunked a little too hard that day.
Or maybe we're at the beginning of machines talking to each other to optimize entire processes.

We don't know our limits yet, and that's the exciting part. We're finding them, testing them, and learning our market, product, and application space along the way.

So why the blog? Why the website? Why incorporation?

Because Octopus is special.

It's compelling enough that our data scientist is basing his master's capstone on it. It takes an amalgam of vibrational, acoustic, magnetic, and temperature data and runs it through some genuinely fun math to answer questions operators already ask every day:

Is it broken?
Is it behaving strangely?
Is that sound something to worry about?

It's called Octopus because it has many arms and one head. Each "arm" senses a different physical signal and routes that data back to a central processor for machine-learning analysis — classifying, correlating, and communicating the machine's state.

You know that maintenance tech who can walk into a plant, listen for ten seconds, and tell you exactly what's wrong? Imagine having that insight available all the time, across many machines.

That's what's cooking at Innova-Harmonics.

Follow along. Read future posts. Ask questions. We want to learn about your systems, your process flow, your maintenance headaches, and your best machine stories.

Field notes from building Innova-Harmonics.

Weather Update and Huge Steps

Cooking up a report, but first, something fun

Can You Afford To 'Ignore' Data?

We Are Innova-Harmonics

Weather Update and Huge Steps

Cooking up a report, but first, something fun

Can You Afford To 'Ignore' Data?

We Are Innova-Harmonics

Want to hear about the next one?