Perhaps a bit anachronistically, only *after* getting my hands on the previously-torn-down 2i4 gen2, i also came into the possession of a 2i4 gen1, that was no longer being detected completely / correctly by the computer. But hey, better late than never, right? I also grabbed a working one (because the price was near-irresistible), to have a known-good source for the firmware, in case that was the issue.
You'll be forgiven if you struggle to see vast differences between this and the later version. The rear panel is "cleaner", due to a slightly different mechanical construction (no visible screws anymore), and the silver knobs on the front have a metallic finish, but that's about it.
Not exactly what you'd call a world of difference on the inside either. But then again, why would you expect that, anyway?
Due to my unwillingness to remove and resolder the two (relatively stiff) ribbon cables, the overall internal view is a two-parter, so, umm... Deal with it?
As in the 2i2 gen1, the virtually identical power supply section is made up of two Intersil / Renesas ISL97519A for the phantom power rail and the analog rails, as well as two OnSemi NCP1521B for the 3.3V (digital) and 1V (Xmos core) rails.
And yet another familiar sight - just as in the 2i2 gen1, the same Xmos XS1-L8A-64-TQ128 processor is running the show, with the firmware being stored in a Winbond 25X40CL 4Mbit / 512Kbyte flash memory, an SMSC / Microchip USB3343 interface and a Microchip PL611 clock generator.
Audio conversion is performed by a Cirrus Logic CS4272 (where isn't it used?), with an additional AKM AK4384ET stereo DAC for the additional output pair.
The preamps also look somewhat familiar, and yet they don't. The preamps themselves are again the Focusrite "trademark" JRC NJM2122's and NJM4565's. On the other hand, unlike the 2i2 gen1, we have two OnSemi MC74HC4066 analog switches, instead of just one. But then again, that makes a certain amount of sense, since the 2i4 has switchable attenuator pads on the two inputs; the first 4066 handing the switching between the microphone and line input, depending on which connector is inserted..
But that, apart from the updated Xmos controller, is where the bulk of the 1st gen vs. 2nd gen differences lie - where only NJM4565's were used before, in select locations rail-to-rail opamps are used in the latter version, which allows for a noticeable roughly 2Vpp extra headroom before clipping.
The outputs are driven by even more NJM4565's.
And last but not least, a(nother) lone NJM4565 drives the headphone output.
Now, on to the repair. Windows says "USB device unrecognized", and in Device Manager it mentions something about "reset failed". Flashing a known-good firmware didn't help, so my next suspect is the USB3343 chip. Well, part of that is not wanting the Xmos chip to be dead, but i'll cross that bridge when i get to it.
I must admit, i guess i got relatively lucky this time. True, the new USB3343 did cost me almost 9eu, but on one hand, it wasn't from China (but from an italian component distributor on eBay), and on the other hand, delivery didn't take a month or more (but under two weeks, or perhaps even only about one). After removing the stock one, cleaning the pads, fluxing them thoroughly and applying the new chip... I was greeted by an "Unrecognized USB device" in Device Manager. After a quick visual inspection, i noticed the little common-mode choke on the pair of USB signal lines had gotten shifted over a bit, during the hot-air soldering, thus disconnecting one of the two data lines. Once that was moved back into place, my test-rig detected it straight away, the USB-link-valid LED on the front panel lit up, and the thing works!
Hi, cant thank you enough for the teardown.
ReplyDeleteI saw your sketch of the preamp section in Google Drive.
Is the xlr mic input connected directly to the 2122?
No transistor whatsoever?
Not really, no. The only(?) transistor there is used with the switch built into the combo jack, to activate or deactivate the padding for the line input (which gets fed into the mic preamp).
DeleteOh! Where can I find the sketch? Thanks.
DeleteHi!
ReplyDeleteThank you very much for the teardown and detailed description.
I too, got a 2i4 gen1 for cheap and could not resist buying it.
Mine even worked out-of-the-box. At least to some extend.
It gets recognized and I can output the PC audio through headphones and the main outputs.
However the inputs behave very weird:
Both input channels work with direct monitoring. But in the DAW (Audactiy, Ableton) I get the maximum input level of 0dB, even with nothing connected and gain at zero. Same on both channels.
If I try to record my guitar and set the gain circa midway, I can see something in the waveform but it is all shifted to the bottom and very distorted.
Maybe you have an idea where I can start investigating.
Thanks and BR, Gerald!
If the direct audio is fine, that should rule out most of the analog circuitry, but there's gonna be at least one more opamp between the point where the direct signal is taken, and the inputs of the analog-to-digital converter.
DeleteI don't suppose you have access to an oscilloscope, do you?
Thanks for the quick reply!
DeleteActually I do. Also a Salea Logic Analyzer.
I work as an embedded dev but don't have experience with audio stuff.
If you point me to a few interesting points I can take some measurements tomorrow.
Thanks, Gerald
Ok, that's great then. What i would do is to feed a sinewave into one of the inputs (even from the Scarlett itself, just no direct monitoring so you don't get feedback). Then scope out each opamp's output and see if and where the signal goes wonky or disappears.
DeleteMight be worth either to feed that test signal to both inputs at once, or just keep notes of which opamp (half) carries each signal if you only use one input at a time.
Worst case, the CS4272 might need replacing, but maybe just cross that bridge when you get to it (if at all). But since the outputs are working, that rules out any missing digital clocks there.
What a great post! I wonder where I can find the 2i4 schematics (1st gen.). I have a short circuit in the 10 V. I took off the four 4066s but still I have no 10V and also no 48 V (Phantom). Thanks!
ReplyDeleteWhich "10V" do you have a short-circuit in? The NJM2122's are rated for max +/-7V, so there should(?) be nowhere to measure 10V on this board.
DeletePS: Googling for "2i4 schematic" showed this as the first image result:
https://elektrotanya.com/focusrite_scarlett_solo_2i4_audio_interface_main_board_schematic.pdf/download.html
Thanks a lot!
Delete