PCB Trace width on Power Circuits in Ampeg SVT VR & CL amps

[nevetslab]

While I was looking at the power tube PCB of the Ampeg SVT-VR that I'm must finishing up, and had also looked at the PCB layout of the same board's tube socket wiring on the SVT-CL (while replying to the Ampeg SVT-CL Ticking thread, what caught my eye was the VERY NARROW trace widths used in the layout of the Ampeg SVT-CL Power Tube PCB. Here's both the SVT-CL and the SVT-VR:



The pad size on the tube socket pins is 0.150" dia on both boards. The heaviest traces are, of course, those for the heater circuit, and we see 0.150" wide traces used there on the SVT-VR, while using 0.100" wide on the SVT-CL. On the SVT-VR board, apart from the heater traces, they nearly standardize on 0.10" wide traces. Now, look at the cathode trace on the SVT-CL. Thinnest trace on the PCB! 0.050" wide.

If we were wiring these up point to point with hook-up wire, we'd probably at least be using 20AWG, or 18AWG for the plate & cathode circuit.

What is the wire gauge equivalent of a 0.050" wide trace (call that 50mil width)? 2 oz copper is typically 2.8mil thick (0.0028"). So, in sq mils, 50 x 2.8 = 140 sq mil. Unless I'm mistaken, 140 sq mill is the same as 140 circular mil in cross-sectional area. What is that wire size equivalent? Between 28 & 29AWG! Even the 0.10" wide trace (100mil width) isn't much bigger....that's 280 sq mil, which is equiv to between 25 & 26 AWG wire! Scary ain't it?

Granted, we're not passing Amperes of current thru the plate & cathode, but......the PCB begins life as a solid copper clad sheet of material, before we etch away the copper, leaving our printed wire layout. Having done power amp PCB layouts for decades in product manufacturing, I laid down as much copper in the high current areas as I could, usually on both sides of the board and sizeable component pads, instead of flushing it down the drain. Just how big in trace width with 2 oz copper is the equivalent of an 18AWG wire? 18AWG is 1624 circular mils, so 1624/2.8 = 580mils, or 0.580" width. Why throw the copper away during the etch?

I came from the day of 2X & 4X tape-ups on mylar on a light table, using Bishop Graphics tape-up supplies and Rubylith to lay down a lot of copper area, and also knew how easily tiny solder pads pull away from the PCB material from repeated solder/de-solder cycles, so if I had the space, I didn't throw copper away. With PCB layout people using current PCB layout packages like PADS, Protel, PCAD, etc, and haven't created a library of larger solder pad puppets and haven't gotten a handle on trace width vs current capacity and temp rise on traces, as well as what the wire size they're actually placing when they pack lines in, it just kills me to see their pretty layouts pass thru design reviews and into the products we have to patch back together when fault currents burn them up off the fiberglass.

[loudthud]

There is really no need for most lines to be as wide as to be equavalent to a #22 wire. On a PCB the wires don't need to flex. The more important thing is the spacing between wires. It's interesting to see the little routes between the plate and screen pins on those octal sockets.

Correct me if I'm wrong, but I think a "Circular mil" is smaller that a square mil. A 10 mill wire is said to be 100 circular mills but the actual area is Pi * R^2 or 78.5 square mills.

[Mick Bailey]

Thin traces don't have much of a survival margin in a fault situation and I sometimes wonder what the reason is for minimal width where there's no other obvious reason. When I design my own boards I try to leave as much copper as possible. This makes for a more electrically robust board and has the advantage of a more efficient etch. More often than not I see boards where there's an absolute minimum of copper remaining, resulting in very fragile pads and traces but extensive areas of bare board. I never figured this out in a tube amp. I can understand the reasoning in a mobile phone where there are complexities of capacitance, inductance etc., that get built into the routing model to ensure the thing both works and complies with EMI requirements.

By contrast I've just been working on a Jet City amp and was impressed by the old-school generosity of traces. The boards are tinned as well, increasing current capacity. Also a Hayden MoFO - a very well-designed board with massive pads and traces, plus every cap location with additional pads to take different component sizes. That amp looks like it could take any fault situation.

Andy Marshall was a proponent of the 'more copper' school of thought. The THD Univalve was built along those lines.