# Give me a few days

By Warren Pridgeon | Published 2026-07-20 | Category: Maker

Canonical: https://warrenpridgeon.co.za/blog/give-me-a-few-days

---

Here's another one that started the way most of mine do. Someone asking, "hey, do you think you can...?"

And as per classic Warren the answer was "Give me a few days."

([Last time someone asked](https://warrenpridgeon.co.za/blog/the-ugly-macro), it only cost me a few hours. This one needed the full few days.)

A friend came to me and asked if I could find a way to remotely monitor the battery voltages on large gensets. Backup generators, but not little ones you wheel out and start with a pull start to keep the fridge and the TV alive when loadshedding hits. These puppies are the size of a small car and power an entire building. They sit in a basement somewhere with full starter motors and big batteries to start them up.

Now, theoretically, some kippie is supposed to wander down there every so often and check the floating voltage on those batteries, just to make sure they'll actually have the juice to start the generator when the lights go out. And we all know how it goes with those kinds of things... It's like checking your spare wheel to see if it's inflated. The only time you realise your spare wheel is flat... is when you need it.

## Pulling on the gi

So I pulled on my gi and pulled out my GoogleFoo.

At this point my entire experience with microcontrollers was downloading someone else's code and slapping it onto an Arduino Nano to make a midi controller for a planned electronic drumkit ([ancient video evidence exists](https://www.youtube.com/shorts/pBJoAqLUwIU)). Within a couple of days I'd gone far beyond scratching the surface, figured out how to make one talk to a server, how to push that data into a dashboard, and how to fire off a notification the moment a voltage dropped below the safe threshold.

![Close-up of the ESP32 dev board, red power LED lit, jumper wires everywhere](https://warrenpridgeon.co.za/uploads/8368081dfd2b413ea843726ecd5c1643.jpg)
*The mighty ESP32. This thing is very capable for its size. The first computer I owned was a raging beast of a 386 clocking in at 40MHz. The ESP32 lands at 160 to 240MHz for a FRACTION of the size*

Little known fact: All electronics run on smoke and when you let the smoke out they stop functioning. Consequently you cannot force 14 volts into a digital pin on an ESP32 because that's an instant "smoke has escaped" situation. I had to learn how to make a voltage divider to step that down to an acceptable voltage.

![Two buck converter modules side by side, one wearing suspicious solder blobs](https://warrenpridgeon.co.za/uploads/d03094d7a3ec4b7d9e0a0cc1cca3b8cb.jpg)
*Buck converters. The one on the left converted its bucks into smoke.*

If you've never worked with electronics you might not even know what a resistor or capacitor is. There's even less chance you'd know that they are manufactured to different tolerance levels. What this means is that two 100k resistors from the same pack might be 95k or 105k respectively... this compounds when used in a voltage divider. I had to figure out a way to calibrate the device to account for this.

Once the electronics, the software and the dashboard were all playing nicely, I fired up Fusion 360, modeled a housing for the thing, and a few clicks later the 3D printer started its print. After about an hour of it sounding like a transformers battle as my old 3D printer Brrted and chirped I had the housing. I put the prototype into the housing and boom. READY TO RUMBLE!

![The finished monitor in its 3D-printed box, LCD reading Sending 1.37V, croc-clipped to a single AA battery](https://warrenpridgeon.co.za/uploads/59b5a68f1895407cabc187432b41b15b.jpg)
*Bench test: one AA battery, two croc clips, "Sending 1.37V". Working.*

## The demo (and the paperclips)

I headed out to the stakeholder's office to go and demo this awesome thing I'd cobbled together in my workshop. Being that I had been finalising things until sometime after midnight I forgot my adjustable AC/DC adaptor at home. This was my "simulation generator battery" where I'd set the unit to report "low voltage" at anything below 12 volts and I'd initially run it at 12v and show "green dash"... drop the voltage to 10.5 on the adaptor and the dash would show red.

So I MacGyvered a DC power source out of paperclips and my laptop charger. You don't need to be at war and under heavy gunfire to have courage under fire.
Like Michael Jordan in his prime: you cannot stop me, you can only hope to contain me. And an AC/DC adaptor left at home wasn't even close to being able to contain me.

## When the project grew legs

That battery monitor then morphed into something far more ambitious. We pointed it at tracking the running hours on "yellow machines," which is what people in the trade call forklifts and the like. And this was the majority of the project and hard work. The bulk of the development time on this whole project went into the hour monitoring.

![The starting parts laid out on pallet wood: LCD, microcontroller, cellular module, antennas, resistors](https://warrenpridgeon.co.za/uploads/032aff3bbacc4b889b48772a26e4472f.jpg)
*The starting lineup: an LCD, a microcontroller, a cellular module, and hope.*

And that's where the rabbit hole really opened up. I had to get the device onto the warehouse wi-fi, then fall back to a cellular connection over a SIM card when the wi-fi dropped or if there wasn't wi-fi in the specific warehouse. Then we found out forklifts count their hours differently depending on whether they're electric, diesel or gas, and some are set to only count hours when there's actually someone in the seat. So part of the testing was syncing each new device to the machine's current onboard hours and then watching them check in over time.

![The monitor mounted on a forklift, its LCD reading Hours 23826, right next to the forklift's own dash meter reading 23826](https://warrenpridgeon.co.za/uploads/7a6f9b3caa5b47e1b8ef6699444b005b.jpg)
*My monitor and the forklift's own meter, agreeing on 23826 hours. Few screenshots have ever made me happier.*

My favourite discovery from all that testing? The data showed us the security guards were doing their patrols on the forklifts. HAHAHA.

The whole process was pure iteration. Prototype, test. Add the SD card, iterate. Add the cellular modem, iterate. Add the clock module, iterate. Every version came with challenges and lessons. For example sim cards, you can't just stroll into a Vodacom shop and ask them for 10 prepaid sims... they don't like that. I registered as a RICA agent and sim card distributor. Like life in Jurassic park... I find a way.  

When the design finally settled, I drew up the circuit diagram and went hunting for a manufacturer, because the real thing (custom PCB, the works) is far beyond what I can manufacture myself. We battled to find a supplier for a few months, but eventually found one that could build us a decent device. And every line on the diagram he built from came out of my pile of tangled prototypes. 

![A hand holding a dev board buried in an absolute rat's nest of jumper wires](https://warrenpridgeon.co.za/uploads/8464fce648734f0f9888e2d362966e87.jpg)
*Where it started: held together with hope and if-statements.*

![The final version's internals: a custom black PCB with an ESP32 module, SD card slot, OLED screen and cellular modem](https://warrenpridgeon.co.za/uploads/eed284ef04374c738e56c5497012b3c1.jpg)
*Where it ended up: I prototyped it into existence, they turned my circuit diagram into this.*

The setbacks lined up. After testing, the first supplier couldn't supply us at the scale we needed for the rollout, so the hunt started all over again (months, gone). We did find a new supplier, and that's when COVID hit, with the global chip shortage right behind it. No chips, no rollout, and in the end the plug got pulled on the whole project. It just didn't work out.

![A forklift on the warehouse floor with tools and a component box laid out in front of it](https://warrenpridgeon.co.za/uploads/9c12a7ce66294e5c8432c78fb350a3d4.jpg)
*Install day at one of the sites. Then the chips ran out.*

## The learnings are still there

Just because the project never went live doesn't mean it was a complete failure.

The project never made me a cent in the end. On paper it's a failure. But I walked out of it knowing how to work with microcontrollers, how to build and calibrate a voltage divider, how to get a device talking over wi-fi and cellular, how to model and print a housing, how to take a design far enough that a manufacturer can run with it, all of it. None of that left when the project did.

And those skills didn't sit still while I had them out either. Somewhere in the middle of all this I was playing with crypto arbitrage, so I built a third gadget out of the same parts bin: a little desk box that watched the prices across the input and output markets and told me whether the spread was worth a trade. Favourable? Smiley face. Not favourable? Poop emoji. Peak UI design, I know. That's the other thing about learning something new: you never learn just one use for it. 

There is currently a passion project on the side of the side which is a hardware based stream deck kind of thing using the ESP32, touch screen, some encoders etc... but more flexible (there'll be a blog about this when I get it far enough along).

![A small OLED screen reading Current: 3.00% with a smiley face icon](https://warrenpridgeon.co.za/uploads/0947dfaf9f1e487e93a8eb2f45393446.jpg)
*Conditions favourable.*

![The same screen reading Current: 2.02% with a poop emoji icon](https://warrenpridgeon.co.za/uploads/6ca5b2315ac3457d82b3807df8399e1e.jpg)
*Conditions less favourable. The complete range of human emotion.*

That's the thing about saying "give me a few days" and meaning it. You're not really promising the gadget. You're betting on your own ability to figure the thing out from a standing start, and that bet pays out whether the project lives or dies. The work can get cancelled. The learning doesn't.

So the next time someone asks "do you think you can...?" and you have absolutely no idea how, I'd genuinely encourage you to try the Warren answer. Pull on the gi, pull out your GoogleFoo, and give yourself a few days. Worst case, the project dies and you come out the other side knowing how to do something you couldn't do last week. That's not a bad worst case at all. Failing that... chat to me. Even if you don't commission me for work I'm always happy to chat through options and ideas.