SURGEON 25: Mastering EQ Build (Part 4) - Power, Grounding and System Integration
- Eitan Brown
- Feb 3
- 8 min read
Updated: Feb 25
Why Power and Grounding Need Their Own Chapter
Up to this point, most of the work on Surgeon 25 lived in the signal path: filters, switching, control logic, and how audio moved through the unit. Power and grounding sat in the background, present on every schematic, but easy to mentally bracket off for later. Later is now.
Power and grounding are more than just support systems. They set the conditions under which everything else operates. They affect noise floor, stability, safety, and how predictable the system is once everything is connected. Mistakes here not only sound bad, they can be a real safety hazard.
This was also the part of the project that made me uneasy. I don’t come from an electrical engineering background, and mains power demands a different level of respect. I read, re-read, took notes, second-guessed myself, and still found that some concepts took time to land.
Because of that, power and grounding needed space of their own. The goal here was straightforward: build something quiet, stable, safe, and understandable enough that I could trust it.
Choosing a Power Supply
I reached out to my friend, Shun Yoshino, to ask for a recommendation. Shun has a deep understanding of Japanese audio systems through his work as editor-in-chief of one of Japan’s longest-running home-audio magazines, and unique insight into what engineers here are building and using in practice. When I explained the project and the requirements, he pointed me toward a linear power supply kit designed by an independent audio engineer specifically for high-quality audio applications.
The kit met the electrical requirements of the EQ, and just as importantly, it came recommended by someone whose judgment I trust. After that, I checked in with Uroš to confirm that the design would be appropriate for this circuit, and he agreed it was a good fit.
One practical detail also stood out. The power transformer can be configured for different AC voltages with some quick soldering of a jumper wire, which meant the unit can travel overseas if it ever needs to, without changing the power supply.
At this stage, the assumption was still that the PSU (power supply unit) would live inside the main chassis. But that raised a set of questions I couldn’t ignore. How much noise would an internal transformer introduce? Where would the mains power switch go on an already crowded front panel? How comfortable was I running AC voltage wiring alongside dense bundles of audio lines going to and from dozens of switches? AC voltage wires love to bully nearby audio lines into carrying their noise.
Some Sontec-style builds solve this by placing the power switch on the rear panel, which certainly makes the build easier. In my case, the unit was meant to live in a rack, and reaching behind it to turn it on would be so impractical, I’d probably end up never turning it on… or off.
Those questions sat unresolved for a while, and eventually pointed in one direction: the power supply belonged outside the main chassis.
Moving the Power Supply Outside the Main Unit
Separating the PSU addressed several concerns at once. It removed the transformer and mains wiring from the same enclosure as the audio circuitry and dense switch wiring. It eliminated the need to squeeze an AC power switch onto an already crowded front panel. And it avoided running high-voltage AC lines alongside dozens of audio paths, which never sat comfortably with me to begin with.
There’s always a tradeoff. Going with an external PSU meant a second chassis, more metalwork, and a reliable way to get power from one box to the other. It also meant thinking carefully about grounding between the enclosures and how the shared 0V reference would stay stable and quiet.
Very on brand for me, power and grounding were still something I was actively trying to wrap my head around, and I decided to make them more complex instead of less. Adding an umbilical, multiple ground references, and a second enclosure wasn’t the easiest path forward, but it forced me to engage with the problem directly rather than sidestepping it.
That decision also added workload. The PSU enclosure needed ventilation, which meant drilling a lot of holes (far more than I had anticipated). The panel had to be laid out, drilled, and assembled. Nothing here was conceptually interesting, but all of it mattered. It was slow, repetitive work.
Eventually, the external PSU was solid, well-ventilated, and provided stable DC voltage. How to send that power to the main unit, keeping it clean and shielded from surrounding noise was the next puzzle.
The Umbilical: Bringing Power Into the Main Unit
That connection took the form of an umbilical cable carrying the DC rails and ground reference from the PSU enclosure to the main chassis. Rather than using an off-the-shelf solution, I built it myself so I could control conductor count, gauge, shielding, and routing.
The umbilical carries bipolar DC, a shared 0V reference, and a separate chassis ground between the two enclosures. Internally, it consists of two full Mogami 2549 cables, which are normally used as microphone cables and widely regarded for their excellent noise rejection. Each 2549 uses a twisted pair of insulated conductors wrapped by a copper shield. In this application, both conductors of each cable are used to carry one DC rail and its corresponding 0V return. The 0V conductors from both cables are tied together at both ends to form a single unified reference. Those two cables are then twisted together, alongside a separate green/yellow ground wire that bonds the PSU chassis to the main unit chassis. The copper shields are terminated to the PSU chassis, so any noise that does try to find its way into the cable is absorbed by the shield and sent straight to ground. Using two separate 2549 cables here is admittedly overkill, but that excess margin means I never have to worry about noise finding its way into the system through the umbilical.
All of that wiring had to land on a single 4-pin XLR connector at each end, which is not what those connectors are usually designed for. I also hadn’t accounted for the XLR boot, which assumes a single cable rather than a bundled set of three wrapped in mesh sleeving. It didn’t fit. I ended up cutting slits into the boot to widen it, and even then tightening it closed was hard enough to give me blisters. It worked in the end, but if I build another one (which I will, if only to have a spare), I’ll be using a wider boot.
Power had finally reached the main unit. Grounding, which was where I felt the least grounded, was what I had to face next.
Grounding: AGND, PGND, Chassis, and Star Ground
If power distribution raised questions, grounding forced me to slow down and be precise. The schematic already distinguished between AGND (audio ground) and PGND (power ground). On top of that came chassis ground and protective earth, along with the question of how many ground references were actually present and where they should meet.
Each concept made sense on its own, but applying them across multiple boards and two enclosures required more care than I expected. What made this difficult wasn’t a lack of information, but keeping the entire system in view at once.
Ground paths don’t announce themselves, and it wasn’t always obvious where current would return or how noise might move from one part of the system into another. I spent a lot of time tracing connections, revisiting the schematic, and checking my assumptions against how the unit was actually wired.
One piece of that puzzle was pin 1 on the balanced I/O connectors. Understanding how the cable shield is terminated, how that connects to chassis ground, and how this relates to the AES-48 (Audio Engineering Society) convention helped clarify the separation between signal ground, chassis ground, and protective earth. It wasn’t a singular insight, just one of several concepts that eventually settled into place.
I finally had a handle on how the grounds related to each other. From there, it was about picking a single star point where they’d meet.
Loop Breaker and Ground Strategy
Multiple ground references were unavoidable, which meant figuring out how to tie them together without creating ground loops. This is where the idea of a loop breaker entered the picture, drawing directly on Rod Elliott’s 1999 article on grounding and loop breaker design.
In this system, the star ground feeds into the loop breaker before bonding to the chassis and protective earth. In normal operation, it effectively sits in the background, allowing the system to reference ground safely without carrying current. If a fault, surge, or ground loop tries to force current through that connection, the loop breaker blocks it. In practical terms, it preserves safety while reducing the chance of hum or noise.
DC Power Delivery and Distribution
Power was reaching the main chassis. Distribution inside the unit still needed a plan. Power had to be distributed to four boards: two EQ boards, the mid-side board, and the relay board. Each needed the same bipolar rails and a solid 0 V reference.
I wanted a layout that was easy to follow, easy to disconnect for maintenance, and easy to spot when something went wrong. Trying to solve that purely with point-to-point wiring started to feel fragile. Power distribution and grounding were closely linked, and handling them together made more sense.
This was starting to feel familiar. Looks like another PCB.
DC Distribution and the Second PCB
A few closely related pieces needed to live in the same place: DC distribution, grounding references, and the loop breaker. They were already tightly coupled conceptually, and keeping them physically close made the most sense. Rather than spreading that logic across wires and terminal blocks, I decided to put it all on a single board.
This became my second attempt at designing both a schematic and a PCB. I was more comfortable in KiCad this time, but still very much learning as I went. The overall structure came together quickly, but it wasn’t mistake-free.
One schematic error didn’t become obvious until much later, when the entire unit was assembled and powered on. Fixing it meant lifting parts, rerouting connections directly to capacitor legs, and carefully insulating everything to avoid accidental contact with the original pads. That’s why the bulk capacitors in the photos sit slightly raised and tilted rather than flush to the board.
It isn’t elegant, but it’s solid, safe, and easy to verify, which mattered more at that stage than perfection.
This time, I ordered the PCB in purple. Between the green main and mid-side boards, the red relay board and the purple distribution board, the inside of Surgeon 25 was getting colorful.
Closing
Power and grounding finally had a concrete form. There was a physical layout I could point to, a grounding scheme I could trace, and a system I could reason about without constantly second-guessing where things were connected or why.
This part of the build wasn’t about refining sound or shaping tone. It was about building trust in the foundation underneath everything else. Quiet operation, predictable behavior, and basic safety aren’t exciting features, but they’re what allow the rest of the unit to exist without compromise.
With power and grounding finally sorted, attention could turn to something far more fun. The next part is about panel design: making aesthetic and ergonomic choices, and figuring out how to fit a lot of intention into a very finite piece of metal.
Comments