WA12 mkII mic preamp modifications

This is something I'd intended to post last year before the fire, but hadn't gotten around to. The WA12 is a pretty well-regarded budget mic preamp, being largely a clone of an API312 topology for under $500. I'm not a deep expert on the history of API or the 312 circuit, but the broad outline is there. Warm Audio opted to go with a Melcor 1731 style opamp here, instead of copying a later API 2520, but it's socketed and you can change it (if you do, see the note down at the bottom!). To their credit, they chose a vintage 1:8 input transformer ratio, rather than the 1:10 you see on modern APIs. It's a weird choice that the 'default' mode on this preamp (tone switch off) is 1:4, whereas 1:8 (tone switch on) is the more vintage-correct configuration, but whatever. This post references the current black-panel mkII version, which seems to have a slightly different PCB and layout than the orange-panel mkII.

Wrapping up loose ends

As most of you have probably guessed, I've moved on from doing Atomium. It's been 5 months since the fire. In that time we emptied the house, sold it, and bought a new one in a different neighborhood. We'll move to the new place later in the fall.

I've increasingly focused on other work in audio production and broadcast for the last 3 years, and that work has carried me through this situation while the gear repair and build work totally stopped. It just doesn't make sense to try to rebuild something that had already become a secondary part of my life before the fire.

This blog will stay up indefinitely, in the hope that people can continue to learn from it, but I've removed my contact and services info. If you're a past client, please feel free to get in touch any time -- I may still take on a passion project now and then. I might even post occasionally, if I learn something worth sharing from a personal project.

I'm not on social media anymore, but you can see what I'm up to with Ghost Lode and Rosetta.

This picture was taken by the Philadelphia Fire Department. My house is not in the frame, and did not burn, but took significant smoke and water damage. We will have to gut the house. My hope is that the majority of the unique music gear inside can be saved, but it's too early to say.

In case someone might need it, here's a schematic for the Supro Blues King 8 combo amp. I was given one of these and traced it (most of it) for fun. Note it's the same amp as the current Delta King 8 model, which just has different cosmetics. I think that for a tiny combo, it has a nice and surprisingly full clean sound. I've been using it for shreddy tiny-amp overdubs on distorted guitar tracks, and for that it works very well.

Some circuit notes down below...

1977 Orange OR120, in immaculate condition. Came in for a power cord swap, and had already been re-capped previously. My favorite Orange amp by a significant margin.

The Black Jupiter flanger/chorus. This is a custom build based on the Boss BF-2 circuit. The 'manual' control on the BF-2 confuses people because of the name, but it sets the pre-delay. Generally, a BBD-based flanger is the same as a BBD-based chorus, but with much shorter pre-delay and the option for positive feedback. For this build, I made the pre-delay range a lot wider than the BF-2's, allowing everything from almost phaser-like tones to super long chorusing, just shy of slapback echo territory. Unlike most of my builds, this one has a custom-fabricated PCB.

In somewhat related news, sourcing parts for custom builds has become difficult and time-consuming. That's mostly because of pandemic-related supply chain shocks, but also because many of the small businesses in the space have scaled up and now have prohibitive minimum quantities. I've never been interested in doing production runs, and the guitar pedal market is far more saturated than it was when I started doing this 11 years ago.

Recently, I've had to decline a lot of build requests (and even many repairs that required specific ICs). Because I'm doing more broadcast and audio production work now, I'm winding down my repair/build operation and will only be taking a few select projects as time (or sourcing) allows.

Quicksilver Mono KT88 hi-fi power amplifier, from the mid 90s. This is a weird one, because of how the power supply is configured. At first I couldn't figure out why the plate voltage was sagging, but the screen node was totally stable. Turns out the supply is silicon-rectified, but the 5AR4 tube rectifier is connected as a series resistance in the plate supply. In other words, it's not rectifying anything. It's just adding sag to the B+ so that the amp compresses at high volume, but the screen supply and downstream phase inverter voltages stay steady. Interesting design -- euphonic perhaps, but not what I'd call "high fidelity."

Framus Cobra. This one had taken a massive impact to the front panel, which needed straightening out. But it had also caused some components on the front PCBs to crack their solder and make all sorts of weird intermittent noises.

These amps are another design very much in the lineage of the SLO, though my impression of them is that they're much "squishier" than the SLO. The power supply is pretty saggy (the power transformer causes this, it's not something you can mod out), and actual clean headroom is about 75-80W with 4 EL34s. They also have a tendency toward bias shift, which makes a prominent crossover notch appear on the output at about 60% of full power. Still, at moderate volumes, a really good sounding high-gain circuit, with lots of flexibility.

Another build of my Hotcake recipe, with bass/guitar mode toggle. I have a couple of these in the pipeline at the moment -- for whatever reason, it seems like there's been more interest in this circuit recently (though it's been around since the late 70s).

One recent change I've made to the circuit is reducing the size of the input cap (to 6.8nF or 5.6nF), which helps to control 'farting' (blocking distortion, from bias shift) with really big input signals from low-impedance sources (active pickups, booster pedals, etc.).

PRS Archon. I don't see too many of these, but they share a common circuit lineage with the SLO/5150/Dual Rectifier etc. They're nicely built amps. This one just had some shipping damage and a bulging capacitor in the DC filament supply, which was probably undersized to begin with. Not super fun to work on, because replacing components usually means desoldering individual wires from the PCBs in order to flip them over and access the underside.

My new solo record is out today! This is a two-song exploration of human and animal physiology bending under the weight of increasingly hostile natural environments. For fans of sad guy power drone.

Here are the broadband noise curves for six different opamps in the Boss SD-1: Yellow: Stock JRC4558DD Orange: RC4580P (TI) Red: JRC4580D (NJM) Green: LF353 Blue: OPA2134 & OP275 (identical traces, same performance in this circuit)

I get asked a lot about how to reduce the noise in this pedal. Replacing the opamp can play a part in that, but it’s not a big improvement -- as you can see from the graphs above, no more than about 4-5dB, max. That’s on a pedal with over 45dB of gain, not including the tone stack boost. The reality is that most of the circuit’s noise comes from outside -- it has high input impedance for guitar, making it susceptible to EMI, and it has a lot of gain, so any tiny signal on the input is amplified enormously (45dB is a factor of roughly 200x), while of course higher-level signals are clipped. Any device with deliberate massive clipping inherently raises the noise floor; there’s no way around that.

The JRC4580D does the best in this circuit. Note that the LF353, being a FET-input amp, is actually much worse. The fancy/expensive OPA2134 and OP275 do as well as the 4580D at high frequencies but have noticeably worse noise than stock at low frequencies. TI’s RC4580P also has worse noise than stock at low frequencies.

Obviously bipolar-input opamps do better here. But I’ve tried some of the popular modern low-noise bipolar amps (NE5532, LM4562, OPA1612) and all of them had stability problems while not improving the noise floor over the 4580D.

Test methodology: Tested into an interface DI input w/ -125dBu EIN, all knobs cranked and the effect engaged, with the input shorted to ground. This is a modified SD-1 where all the signal path resistors have been changed to metal film, so that resistor noise is as low as possible already.

Music Man 112 RD (chassis # 2100-RD). This is a later one with two 6L6GCs in the power stage, driven from their cathodes by NPN transistors (grids are at AC ground).

The design claims to put out 100W from just a pair of tubes, by running them in almost entirely class B. But in reality, the output stage clips at around 65W. You can actually run EL34s in this amp (as the earlier models had) but the sound and bias point hardly change at all. It’s still unclear to me why MM made the change from EL34/6CA7s to 6L6GCs -- in general, 6L6s are less tolerant of super high plate voltage. May have just been a supply-chain thing. Today, EL34s are significantly cheaper, and there’s really no reason not to use them in this amp if it suits you.

The output driver transistors short all the time in these amps, and if one shorts, they should both be replaced, as a pair, from the same batch (matching is important for the tube bias to balance properly). The schematic part number is JE1692, but this was an internal MM part reference. They were actually 2N6292s. You can sub modern 2N6488s just as easily, and they should be more rugged.

Also important to keep your fingers out of these things. B+ is 700V, with 400V on the screens.

Blasting away at some more sad-guy power drone, to tickle your eardrums later this spring. Keep an eye on ghostlode.bandcamp.com for more on this.