$BTC Tech Nerd Army, it's a bitter end to another wild week – but what a ride 2 x NerdQaxe+s and 1x NerdQaxe++ RIP this week !
My latest victim? The beastly NerdAxe ++ pushed to its absolute limits in the TNA R&D cave in the name of Firmware Science. 💥🔥
I cranked it to the edge and beyond:
- Hashrate:10.6 TH/s
- Clock:1350 MHz
- Vcore: 1.58V
- Amps to ASICs: A scorching 165A!
That's when it sparked out HARD – but hey, all safely contained in the immersion fluid, as designed. 😎 Wasn't expecting the board and caps to tap out first... Those CSD95472Q5MC Synchronous Buck NexFETs (VR) are rated for 60A continuous / 90A peak per unit, but we hit ~55A and boom!
Thought the TPS53647RTAR or ASICs would cry uncle first. Lesson learned! 📊
On the next TNA Tech hall this Monday we will discuss the current TNA-v3.5 firmware and its uncontained capabilities.
Tech Nerds, dive deeper in the TNA Discord for full breakdowns, pics, and next steps. attached screen shot of last moment.. and first look into tank seconds after blow out.
View quoted note →
Watch Videos on Nostr with Plebs
Plebs is a censorship-resistant, decentralized video platform powered by the Nostr social protocol
Watch Videos on Nostr with Plebs
Plebs is a censorship-resistant, decentralized video platform powered by the Nostr social protocol
This week alone, I've wrecked a second miner. Yeah, you read that right—second. Earlier this week, I blew the power regulator on a board while writing new firmware for the voltage mapping. And it's not from some catastrophic failure or external sabotage. No, this latest one stems from a deceptively tricky task: attempting to remove a heatsink after a botched thermal epoxy application. Unlike thermal paste, which is forgiving and removable, thermal epoxy is designed for permanent bonding—it's not meant to come off without a fight.
All of this is happening as I'm building an immersion cooling system for the miners. The goal? To cool them more effectively, drive up efficiency, or even overclock for higher hashrate. But to get there, I need reliable thermal interfaces, which led to this epoxy experiment. The setup involves SHA256 ASICs—the workhorses behind Bitcoin mining—bonded to a heatsink with thermal epoxy. The aim? Efficient heat dissipation to keep everything running cool and stable. But things went south during the application. Out of the four chips on the board, three were humming along at a comfortable 40°C. The fourth? A scorching 80°C. That's a red flag screaming inefficiency, potential damage, or outright failure if left unchecked.
In my attempt to salvage it, I tried peeling back the heatsink. Spoiler: it didn't go as planned. The hot chip didn't rip off entirely, which is a small mercy, but the damage was done. If you look closely (imagine a photo here with a green circle highlighting the culprit), you'll see the thermal epoxy layer. Right in the center of that problematic chip, there's a glaring gap—a missing patch where the epoxy failed to adhere properly. That void was the root cause of the temperature spike, trapping heat instead of wicking it away. It's a classic case of "now we know," but hindsight comes at a steep price in hardware land.
This isn't just about one ruined miner; it's emblematic of the broader challenges in self-funded projects like this. I've built a million things in BTC hardware over the years, pushing boundaries in every direction. But funding your own R&D means every dollar counts, and mistakes like this can set you back weeks or months. BTC mining hardware is unforgiving and expensive—pushing the limits of efficiency, power consumption, and durability in a decentralized ecosystem.
Why share this? Because the Bitcoin and Nostr and hardware communities often glorify the wins—the successful prototypes, the efficient rigs, the innovative integrations. But the path is littered with failures like this. It's demoralizing, sure, but it's also fuel. Each wrecked board teaches something invaluable: better techniques, more precise methods, or even redesigning interfaces altogether. If you're out there grinding on similar projects, know you're not alone. Press on, iterate (carefully), and remember: in hardware, the "recompile" button is your next prototype.
It's been five years of this solo grind, pouring a fortune into it all while navigating setbacks that leave me feeling utterly drained and overwhelmed on days like today. The truth is, building at this level can't remain a one-person effort forever—the vision is simply too expansive, and the challenges too daunting to shoulder alone. Yet onward we go, trying to find spare parts for this one or eventually sourcing new ones, because in the world of BTC, true advancement requires more than solitary determination; it truly flourishes when compassionate supporters step in to share the load and help turn these visions into reality.
On a lighter note, does anyone have 8 spare Bitmain BM1368 ASIC chips laying about?
Unfortunately Its not always good news in the BTC R&D cave I'm afraid. This mornings job was to address a temperature concern, 1 asic (no1) was 15c above the rest.
Im afraid in the process of today's FAFO to get a better thermal mount of the PCB with asic chips.. it resulted in a poor mounting between the asics using indium heat springs and the heatsink.
The Result was 2 ASIC fried!
I saw a 200C reading on one...
I was expecting some shit like this to happen with what I am trying to develop for the overclocking Bitcoin mining community. Unlocking firmware comes with great risks when your task is to remove limits and un leash the hardware's true potential.
The next firmware release and Overclocking Dev work will be parked for the time being as this was my only NerdQaxe miner.
