#69 Focusrite Scarlett 2i4 (2nd gen) teardown & repair

For a little while now, i've been thinking of "upgrading" from my old-but-modded M-Audio Fast Track to use as an in-ear monitor amplifier, and the Scarlett 2i4 caught my eye. The mono/stereo direct-monitor selection, as well as the hardware-switchable pads on the inputs made up just the right feature set, in addition to it being usable as a mobile recording interface. As it happens, just the other week, mere days after having saved an eBay search for 2i4's, an email notification revealed a 2nd gen allegedly-faulty unit being up for sale from Italy. I quickly asked the seller about shipping costs, and ended up sealing the deal for an even 50e. 

The description said it was no longer being detected by the computer. From the photos on the listing page, the USB connector still looked to be in one piece, which still left several possibilities on the table. Power supply issues, firmware problems, or worst case, a dead controller. But all in due time; first thing's first though, namely a closer look, followed by a close(-r) inspection of the insides. Perhaps referring back to the 1st gen Scarlett 2i2 teardown from earlier may be appropriate, especially since the clipping-too-soon issues on the instrument inputs of the first iteration were reportedly solved. 

After removing two screws hidden under the rubber feet under the rear of the unit, the whole guts slide out all in one piece. I guess they wanted to give the rear panel a "cleaner", sleeker look, with no visible screws or anything. Fair enough, i suppose. Also, in the name of mechanical robustness (i'm guessing), the rear connectors are fastened to a metal rear plate, on top of which the plastic rear panel is mounted. At least they didn't skimp on such details.

Barring the mechanical details and the extra RCA connectors on the back, the insides look almost identical to the earlier 2i2 i had dug into, a while ago. But delving deeper now...

Starting off with the power supply section, the beefiest-looking chip is a Monolithic Power Systems MPQ3425 boost converter. This, if my cursory trace-following and assumptions hold, may well be in charge of the 48V phantom power. In the vicinity, two MP2159 step-down converters handle the 3.3V for the digital logic, and the 1V rail for the Xmos core, respectively.

Speaking of which - i must've spent a good few hours already digging through the internet for "XFR002C", but for the life of me i couldn't figure out which damn Xmos chip this is. At least nowadays, they don't have too many TQFP-64 / LQFP-64 packaged chips, and even fewer with a built-in USB PHY, so that should narrow it down to not too many. I may well need to start comparing the apparent pinouts with whatever datasheets are available... Regardless, the firmware is stored in a Macronix MX25L8006E 8Mbit / 1Mbyte flash memory, while the clocking is provided by a Cirrus Logic CS2100.

Later edit (and spoilers): Firmware swap from a known-good one didn't help, so i contacted the Focusrite spares department yet again, and once more they very much did NOT disappoint. In addition to confirming the chip is an Xmos XU208-256-TQ64-C10, they also exceedingly generously offered to send me one, for free! 

PS: Quite officially, screw Xmos and their cryptic markings, despite what the datasheet (on page 28) might claim... 

PS2: I kinda wonder if the full 256K of RAM is required or used in this unit; the 128K RAM version is much more plentiful on AliExpress, than the 256K one. Just a thought...

While we're around the Xmos chip... Maybe i'm paranoid, but i could swear i see quite unmistakeable remnants of flux around the chip, and not only in the immediate vicinity. And the soldering on its pins doesn't exactly scream "factory" either. Was this chip reflowed / resoldered previously, if not even replaced? Regardless, moving on...

Resuming the power supply side analysis - an MP1542 boost converter is pulling double duty, creating both the +6V as well as the -6V rails, powering the opamps and the rest of the analog circuitry. 

Now that i've mentioned it, might as well go through it. No huge shock to find a very familiar-looking setup, what with the NJM2122's, NJM4565's and CMOS switches (HEF4053 and HEF4066) in the analog input area. One "novelty" is the smaller-package NJM2746, though.

A couple more NJM2746's act as output drivers for the unbalanced RCA connections, while more NJM4565's drive the ADC inputs, as well as the balanced pair of outputs.

A lone NJM4565 is in charge of the headphone output.

The conversion is handled by a now "classic" CS4272 as well as an additional AKM AK4384ET for the second stereo output pair.

Now, as already revealed earlier, the first attempt at reviving this involved cloning the firmware from a working one, but to no avail. On the bright side though, under two weeks from my last contact with the Focusrite spares department, an envelope was delivered, containing an Xmos XU208-256-TQ64-C10. In less than half an hour, the suspect stock(?) chip was removed from the board, pads cleaned (despite having lifted one, but without having broken it off), and the new chip installed and... No huge surprise (albeit major relief), the thing works!

 I also took the liberty to reverse engineer (most of) the preamp side. Component designators match the ones pertaining to input 1; JP2 is the line/inst switch, while JP3 is the pad switch. One of the major differences from the preamps in at least the 2i2 gen1 is the usage of the NJM2746 rail-to-rail opamps. That alone allows an extra 1-3Vpp of headroom as opposed to the NJM4565's. For all i know, this may or may not be the "silver bullet" that nullified the low-headroom issues, but... If you'll excuse the foreshadowing, we'll have that cleared up in the near future.

Now, as for the usage as a headphone amplifier... As it turns out, the Xmos controls the enable signal for the analog supplies, and i couldn't be bothered to try to decipher what mods would be required, or to start cutting traces and whatnot. But in the mean time, i stumbled upon a potentially-faulty older Focusrite Saffire 6 USB, with a documented mod to enable stand-alone usage, but... I'm getting ahead of myself, you'll have to keep an eye out for further developments in this area. Wink-wink...