#74 Focusrite Saffire 6 USB 2.0 modding (for stand-alone operation)

First of all, the main issue with using this stand-alone (without a computer) is that the Xmos controller only enables the analog power supplies (the two MIC2505's and the LM3488) once it's done whatever handshaking it needs to, with the computer. But since i'd only need / want it to act as a rudimentary preamp / mixer / headphone amplifier, i kinda-sorta need the analog supplies working.

Now, i had previously found a documented mod precisely for this purpose. Trouble is though, that it pertains to what i'm guessing is the first (older) revision of this model, mine being the newer "USB 2.0" one. Despite the mod apparently functioning as intended, there was nothing specified about "how" the mod achieves what it does, or why etc. I'm a bit skeptical about the barebones "run a wire from here to here, job done" sort of approach, but i guess i'm just weird like that.

Fortunately though, we can glean some information from the datasheets of the chips in use. The MIC2505's have a dedicated "enable" pin, which is active-high (ie. at least 2V); the LM3488 has a "shutdown" pin that is also active-high, with an inverting transistor next to it. Near the chips, there's a 4.7k resistor, with one end connected to the input voltage rail (the 5V coming from the USB connector, and going to the input pins of the regulators), and the other end to one of the three pins of a transistor-looking device (the "TR2" designator is a bit of a hint), as well as to the enable pins of the regulators. One other pin of the transistor goes to ground, while the remaining one connects to a long trace that heads vaguely in the direction of the Xmos controller.

The "WCM" marking on the transistor indicates it to be an Nexperia NX7002AK N-channel MOSFET. Its drain (regulator enable signal) is pulled up to the USB 5V through the aforementioned 4.7k resistor, its source is connected to the ground plane, and its gate is controlled likely by an output pin on the Xmos.

The mod article linked above only instructs connecting a wire from the drain of the NX7002AK to one of the pads of an unpopulated DIP-8 footprint. That turns out to be connected to the 3.3V feeding the digital circuitry. Now, if while powered but not connected to a computer, the Xmos keeps the regulators disabled, that would require the MOSFET to be conducting, which would pull its drain pretty much to ground (disregarding the 3-5ohms of on-state resistance). If said wire was connected to the 3.3V rail, that would virtually short-circuit it to ground so... I can't help but question the validity of that advice.

Regardless, this will be done "right"(?) and even without any additional components or wires. All that's required is disconnecting the MOSFET's drain pin from the pad carrying the enable signal. Once that's done, it can no longer pull said signal to ground, so that will be hard-wired (via that 4.7k resistor) to the 5V rail, so all the regulators get enabled as soon as the board is powered up.

And there's the whole mod. Lifted the MOSFET drain off the pad, solderwicked the solder, and slipped in a piece of Kapton tape to ensure the isolation between the drain and the pad. And as no surprise, things work precisely as intended.

Since i had the mainboard out, i went ahead with the re-cap as well. I decided to replace the 10uF and 22uF capacitors that were pulling power filtering duties, with single or pairs of 10uF/16V ceramic capacitors. On one hand, to keep impedances nice and low, and on the other hand, because i had a whole bunch at hand. The rest of the electrolytics were a mix of Panasonic (M, SU series), Rubycon (MCZ, YXG, YXJ), ChemiCon (KZE) and Nichicon (SR), pretty much whatever "close enough" but purposeful values that i had in my parts collection, that fit the respective footprints. I didn't bother with the ones on the rear outputs, because i'll likely never use this for anything particularly critical, especially on the move, so there's that.

For now, i think i'll be content with running this off a powerbank, but i may well return to this later. I'll need to cook up a circuit with an IP5306 powerbank chip and a couple transistors, so stay tuned...