#15 Alesis io|26 repair & modding

So, i got my hands on one of these Alesis io|26 interfaces for pretty cheap (50 bucks delivered). "Of course", i bought it as faulty (can you see a pattern yet?), with the already "classic" issue of dead Firewire chip (the ubiquitous Texas Instruments TSB41AB2).

I went for it mainly for two reasons. First, because i knew it was based on the TC Applied Technologies Dice II chip, and as such, should have flexible enough mixing / routing options; second, due to the ADAT inputs available. The sort of "added bonus", one might say, were the gain adjustments for the 8 analog (mic/line) inputs.

Before we "get into the meat of'er" (sorry, just can't shake AvE's influence...), how's about a bit of a going-over, hmm? Yes? I thought you'd agree.


The unit's made up of several smaller circuit boards. Fair enough, i suppose - you need at least four layers to fan out all the signals from the Dice II chip (200-some-odd balls), but the rest of the circuitry can do just fine on cheaper two-layer boards. And besides, smaller boards are less costly (you can fit more copies onto the same panel), and if anything goes wrong during assembly or whatever, it won't affect the overall yield that badly - you only need to replace a smaller sub-board, instead of the entire all-encompassing board.

The unit can be either bus-powered, or from the included plug-pack (or wall-wart, or whatever your preferred term is). Out of that single DC input voltage, you need to source several (other) voltages - +/-10-15v for the opamps and whatever else is in the analog stages; 5v / 3.3v / 2.5v (more rarely) / 1.8v for the various digital components. In this case, it's a daughterboard, piggy-backed on top of the main digital / processor board.

No real rocket-science going on there - a multi-secondary flyback converter. Couldn't be bothered to identify the PWM controller, but that'll be the MSOP8-packaged chip surrounded by that myriad of passives. Nice beefy DPAK N-channel MOSFET, and an assortment of diodes, capacitors and inductors.

On to the digital board - nothing terribly shocking here either. Dice II is "the brains of the operation" (oh, AvE...); Spansion flash chip containing the firmware; Samsung RAM chip; and surprisingly enough, a Freescale Semiconductor (now NXP) DSPB56374AFC. I have literally no idea what that's doing on the board - this io|26 isn't advertised to have any DSP capabilities available, and the Dice II itself is very much capable of signal mixing & routing on its own. I know for a fact my old Saffire Pro 40 contained no extra DSP chips of any sort, and the hardware 18x16 matrix mixer within the Dice II worked plenty fine. Go figure...

I was surprised to see that extra Wavefront Semiconductor AL1402 in the corner there. That's a "naughty" (or "novel"?) way of bringing in that second ADAT input. That being said, it's a way LESS over-engineered way to achieve that - they used an entirely separate Dice Jr. in my StudioKonnekt 48, for the second set of ADAT in/outs (which are only used at 96kHz anyway).

Output board - nothing to write home about here either, just "business as usual". There's the Cirrus Logic CS4382ACQZ 8-channel DAC, and a handful of MC33078 dual opamps, and the associated passives. Although granted, this also is only an "after (the mods)" photo, so that's enough of a sneak preview by now.

Not all that much to see on the input side either - your regular assortment of connectors and passive components.

That being said, i WAS quite surprised to see the preamps were developed around the THAT 1510 chips. I could be mistaken, but i think those go for a good few bucks each, so... Take from that what you will. And the DC blocking caps in the gain control section were Nichicon HM, no less. "Downstream" from those there's a handful of MC33079 quad opamps, a couple of TL072, and two Cirrus Logic CS<didn't-write-the-part-number-down-and-i-can't-read-it-in-the-photos> 2-channel ADCs.

First step was, obviously, swapping out the Firewire interface chip. Sadly, i had run out of brand new ones (although there's a few in my Mouser shopping cart with my name on'em). On the bright side though, that was a bit of a non-issue, in fact - since the Konnekt 8 i had scavenged the Dice II chip from, as described in an earlier blog post here, i figured i wouldn't be needing the services of the TSB41AB2 (which i knew to be good) that was on that particular board, so... Why the hell not?

The "transplant" succeeded uneventfully, and after installing the drivers, as expected (or at least hoped), the thing works just fine. That being said, the control GUI looks oh-so-very-90's-ish... I mean, see for yourselves...

And now we come to the part of the <ahem> project that gets really interesting. In previous Dice II-based interfaces i've owned (Focusrite Saffire Pro 40) or, respectively, still own (the aforementioned TC Electronics StudioKonnekt 48), one can route almost any input signal to any output. Unfortunately, the io|26 was ever so slightly "gimped" in this area.

The main idea was to be able to use the unit stand-alone (ie. without being connected to a computer), as a crude mono "monitor mixer" of sorts. I was hoping to be able to route the signals from one ADAT input (ie. 8 channels of audio) to the 8 analog outputs, and then run wires from said analog outputs (inside the unit) to the line inputs, to allow for gain adjustment for each channel. That would all get subsequently downmixed and spat out into the headphone ouput.

But that's where i hit a bit of a snag - the "Phones 1" output is hard-wired to mirror the analog outputs 1/2, while "Phones 2" can be selected to mirror either of the four pairs of analog outs. Bugger... Ah well, no matter, i guess i'll be able to "reuse" only 6 of the 8 audio channels from the ADAT. Way better than nothing, anyway.

A couple minutes of fiddling in the HDM software to set up the signal routing (panning left/right and turning up the respective pairs of ADAT inputs' levels, for each "output mix" pertaining to analog outs 3-8, as well as turning up the levels for analog ins 3-8 for the analog out 1/2 pair, which also goes to "Phones 1"), and that was that for the software side of things. And now, for the hardware...

The intention was quite straighforward, but the... practicalities of it, shall we say, made the endeavour somewhat more annoying (or at the very least, fiddly). Granted, part of that was down to my decision to reuse a couple of pieces of shielded cable that were juuuuuust long enough for the task at hand. Hey, no point in creating waste, am i right? I'm saving the planet here!..

Right, where were we? Ah, wiring up the thing - or rather, the mods. The analog output board was easy enough to "decode". Impedance-balanced outputs (ie. the "ring" of the 1/4" TRS socket is not actively driven), and the post-DAC low-pass filter op-amps have double duty as output buffers. The board's also a single-sided load (with no components on the bottom side), so i didn't need to go around "hunting" for traces / signals everywhere.

The signal take-off points i decided would be the "inside" ends of the 220 ohm resistors in series with the outputs. As chance would have it, right next to those were some unused SOT23-footprint pads, likely for some muting transistors, as what is usually the emitter went to ground. Just as well, a handy spot to solder the shield of the signal cables.

Next step, identifying the pins associated with the 1/4" jack sections of the input "combo" connectors. Fortunately(?), Alesis went with the non-switched versions (which have 6 pins), which made things a lot more straightforward than if they were switched (12 pins). Since my signal was single-ended, and coming from within the same enclosure / device, but without any actual input jacks inserted, i decided to run the signal wire to the "tip" pin, but the shield (which was grounded) to the "ring" pin. I figured that would be the safest route; leaving the "ring" pin (aka. "cold", in a balanced line) floating would've invited the potential for noise ingress - at least in my mind. Grounding the "ring" happens by default when using an unbalanced 1/4" jack line input, so it definitely couldn't and wouldn't do any harm.

And result? A rowsing success! A -3dB FS signal on the ADAT input lights up the -24dB FS LEDs with the gain knob at minimum, so plenty of headroom left there. One nice touch is an encoder that adjusts the level between the "direct" inputs (ie. hardware inputs) and the "software" ones (from the computer. The manual says, "This knob controls the amount of direct signal (from the analog and digital inputs) that gets blended in to outputs 1/2". I found use for it basically as an overall attenuator - quite handy, if not even necessary, because at least on the first headphone output, the matching of the two sides of the volume pot is really off at the bottom of the range, and it gets pretty loud pretty fast.

A task for a later time: making a cutout in the rear of the case, and figuring out how to shoehorn in a DIY'd CAT5-to-ADAT adapter. This is part of a little brainchild / pet project of mine, that i've been brainstorming for a while now. That, and maaaaaaaaaybe adding some pad resistors where the wires connect to the line input solder pads, but i'm not even all that sure that's mandated.