While wrapping my head around what I would need to rebuild a new amp system from the left-over parts of my Exodus III itteration and 26 preamps, I was left with what to do for the filament supplies. The signal section does not contain many parts, but all the B+, bias and filament supplies for the Usagi and Yojimbo amps would be a boatload of iron and caps. This would make a single chassis for the preamp or the L/R poweramp completely impractical.
The filaments will all be fed with current regulators, in specific Rod Coleman's DHT regulator boards. These would be located near the tubes which puts them inside the signal chassis and would put the transformers and rectifiers well away from any tube that could pick up hum from them (looking at you, 26!!!!).
In the old days, ahem, I would use a experimental euro-pcb and lay down fugly solder tracks because I am too lazy to cut wires. In the end it is a lot of work, completely mind-numbing, and lets be honest; looks like shit. So in light of turning a new leaf, I decided to try out a PCB design program I had bought in combination with SPLAN (that I use for the schematics). Spent hours and hours on this program making a decent layout and creating a bunch of custom parts for future use. Lots of fun, and the preview looks awesome. Now lets hope they actually work; that would be nice.
The schematic is a little over-engineered, but the small increase in size has almost no impact on the price per PCB, so might as well have some options. Emphasis was made on decoupling and snubbing, as well as six positions to hook up external chokes, ferrites, drop resistors or simply bridge them if not used. The diode positions are for TO220 packages as I have a bunch of SSTH802D ultra-fast recovery diodes (200V/8A) that should handle any job up to GM70 levels. All film and ceramic caps are 100V, all electrlytics are 35V. I will be hooking up an 18V/4A transformer for each board.
The first capacitor is split into two units so as to split the ripple current between them. I've used the Nichicon UBY with can already handle a higher ripple current, but decided to split them anyway to allow caps with lower ratings as well. For a 1.25A output these caps need to handle about 5A spikes, or about 2.2A rms. The BY caps handle 6A each, so should handle quite nicely here as single or double.
Below is a preview of the PCB, showing the bottom traces in a darker shade. I'm impressed by my own work. This will be the first decent PCB I have ever had made :) The masks will be black to get some extra contrast with the gold plated pads, and I have my site name in gold as well. Gotz to haz my bling! The large caps received some extra pads to allow caps with 10-15mm lead spacing, should allow for a wider range of choices between manufacturer. Dito with the fuse holder that will hold two clips or a single pcb fuse holder. Options are good! Both side have ground planes connected to the 0V output terminal.
2.0mm thick, double-sided, 70µm copper, gold plated pads, black masks, white silkscreen. Dimensions 170x55mm
NB: I've ordered 20 of these puppies as I need 6-8 for myself and know someone that wanted a couple as well. If anyone is interested, drop me a line. I can also supply the common mode chokes if you can not get the right size locally.
4x Ultra-fast recovery Diode 200V/5A (TO220 package)
1x LED (power ON)
2x 4700µF/35V Nichicon BY (high temp., high ripple current)
2x 10000µF/35V Nichicon FW
4x 100nF/63V MKT
5x 10nF/63V MKT
1x 220pF/63V X7R ceramic
1x 1K5/0.25W (for LED)
1x Common-mode choke ±10mH, minimum 3A
1x Fuse holder 5x20mm or 2x fuse clips
1x Fuse 5x20mm, slow-blow, chose value to suit your filament
2x Terminal (5mm pitch)
2-6x drop resistors of choice, 5-10W (preferably mounted underneath to keep heat away from caps)