#59 MOTU 828 mk2 teardown

Even though this is the 4th or 5th of these units that has gone through my hands, looks like i never documented a teardown of one - shame on me! Time to rectify that issue then, wouldn't you say? I grabbed this one off eBay for about 60 bucks or so, delivered. On one hand, because it was pretty dirt cheap at this money, and on the other hand, because it will be involved in a project idea i had a good few years ago. And yes, "as usual" (by now), this one had a busted Firewire connection, but for my own (nefarious?) purposes, i don't need that anyway.

Up front are the two microphone / instrument inputs (Neutrik combo connectors, no less), the headphone output, volume control (for the headphones / main outs), and the encoders used for controlling the built-in digital mixer, and the menu system. Plus, of course, the 16x2 character LCD, and all the metering and status LEDs.

Round the back, there's a pretty comprehensive collection of ins & outs - 8 analog line inputs and outputs, plus a pair of main outputs, as well as sends for the two front inputs (to feed into a compressor or reverb or what-have-you); MIDI in & out; footswitch input; SMPTE timecode in & out; two Firewire connectors; S/PDIF coax in & out; ADAT sync connector (which i haven't encountered anywhere else); ADAT Lightpipe in & out; and last but not least, Wordclock BNC in & out.

As i may or may not have mentioned before, i do quite like this cast aluminium enclosure design MOTU came up with - this may well have been the first generation of units to employ it, along with the 8Pre i took a look into on this blog. The same case is still being used in their current units (minor spoiler alert - some "goodies" to look forward to in the near future).

Either way, off comes the top, and we're greeted with a... "suitably" (relatively) old-school design. Not one, not two, but three(!) transformers in the power supply. These give the unit quite a bit of heft, albeit [ahem] slightly right-side-heavy. The silkscreening says "Copyright 2003-4", so you can be sure this unit has indeed "a few miles on the clock", as it were. Still works though, for the most part anyway. For the sake of clarity and a better view, i took the liberty of removing the little daughterboard holding the gain adjust potentiometers for the front inputs.

Speaking of the power supply, worth starting with that, right? The smallest transformer up top supplies the phantom-power - 4-pin DB103G 1A bridge rectifier, smoothing capacitor, and an LM317HV to regulate that to a stable and clean 48V. The right-most of the two big transformers is in charge of the analog supplies. Interestingly enough, MOTU decided to add a pair of 0.47 ohm resistors to the input of the bridge rectifier - perhaps as a sort of makeshift current-limit? Regardless, what follows is ever so slightly naughty design - the voltage is then smoothed by a pair of 4700uF / 10V electrolytic capacitors. The naughty part of that is that i'm measuring about +/-10.2V on the analog supplies, and good engineering practice is to leave at least a 20% voltage margin for a capacitor rating, versus the voltage expected across it, NOT to use them right at their rated limits, or even slightly above. Tisk-tisk-tisk...

 There's one other bridge rectifier being fed by the analog-supply transformer, but this one has an 0.47 ohm resistor hanging off its "+" output, and a 180 ohm one off its "-" output. The former goes off to the 8-pin header for the front panel. The latter though, feeds an MC7906 negative voltage regulator putting out -6V. More on this later.

Same story on the digital supply side - the left-most bigger transformer feeds a quartet of 1N5819 Schottky rectifiers, and a 15000uF / 10V capacitor. Identical transformers mean that this, as well, is being used ever so slightly above its spec, and that's on our nominal 230V AC mains. What happens when the mains may happen to go even higher than that, when the grid's under low load? Nothing good for the longevity of the caps, that's for sure. Either way, this "raw" 10.2V gets regulated down to 5V by a Burr-Brown REG103FA - arguably "fancy". This then gets further knocked down by a National Semiconductor LM1117-3.3, to feed the processors.

Speaking of electrolytics, the black ones are Meritek(?), while the blue ones are Lelon. I've yet to see Meritek used anywhere else apart from MOTU gear, which is mildly interesting (perhaps).

Speaking of processors, quite a bit of "fat" silicon going on here. Running the show (controls, LCD, metering LEDs) is an SST89V564RD 8051-based microcontroller. Next to it resides an Altera ACEX EP1K30TC144-2N programmable logic device (PLD; think "FPGA lite") - most likely for marshalling all the audio data, as well as whatever format conversions may or may not be necessary, to get the data properly fed into the two other big chips in the vicinity.

One of those is a Texas Instruments TMS320VC5402 digital signal processor - this most likely handles all of the mixing duties (channel level adjustments, panning, muting, mixing), since there are no DSP effects (or what we might think those are, nowadays) available. The other large pin-count chip on the board is a Philips PDI1394L40 AV link-layer controller. Between this and the Firewire connectors we have the now-classic Texas Instruments TSB41AB2 Firewire PHY chip.

Speaking of which - Houston, i think we have a boo-boo. I wonder if that's how and why the Firewire connection is dead... More often than not, a crater in the chip and a pin (if not even two) vaporized right off it, tend to hint at the fact that things are not quite all well. On the bright side though, the problem seems to be isolated to the "outside" in/out sector; the oscillator and "internal" data-comms seem to still be in one piece, which is fine by me.

Staying on the digital side of things, the audio data I/O side employs a few jelly-bean logic gate chips, so nothing too space-age going on here. The ADAT Sync seems consist of three optocoupler-isolated inputs, using a Fairchild Semiconductor HCPL2530 dual unit, and a FOD817C single one.

Hidden in-between the power supply and the ribbon cables going off to the front panel is a Texas Instruments TLC2933 PLL chip.

Time to take a look at the analog and conversion side of things. The opamp complement is quite varied, to say the least. Even the preamp design is... perhaps a bit more complicated than it needs to be, but what do i know?

Regardless, the two microphone inputs are handled by a pair of NJM2068's, obscured by the two DB103G bridge rectifiers used as input protection clamps. The instrument inputs are fed into a pair of Texas Instruments / Burr-Brown OPA2134 JFET-input opamps. Both preamps are basically instrumentation amplifiers with variable gain.

Said gain is adjusted for both (microphone and instrument) at the same time, using a pair of pretty much custom-made potentiometers. They're both "stereo" and anti-log taper ("C" taper), but one "side" has a 2Kohm resistance (for the mic preamp), while the other is 20Kohm (for the instrument input) - go figure. The kicker is, these potentiometers are notorious for going bad and causing noises, pops and clicks, so... Yeah.

The outputs of the preamps go into the trio of LM833's, which split the signals in two, one feed going into the first of five AKM AK4528VF codecs, while the other goes via the header and ribbon cable to the output daughterboard, and to the "Send" outputs on the back. The smaller multi-legged chip is a lone AKM AK4382A DAC for the "Main Out" pair of outputs. And speaking of the first of five similar codecs, the DAC side of that seems to feed the headphone output, driven by a pair of NJM4580's.

 The other four codecs are in charge of the eight line inputs and outputs. Pretty much all of the "output" opamps are NJM4580's, handling DAC output filter duties as well as driving the actual outputs. On the input side though, things get a bit more interesting. On one hand, mostly hidden under the output daughterboard (so you'll have to take my word for it), each pair of line inputs goes into a Philips 74HCT4066D quad SPST analog switch, and then three NJM2115 opamps. Since the output feature list in the manual quotes selectable +4/-10dB selectable input sensitivity, i'm pretty sure that's the explanation. Also, the -6V rail seems to be used for powering these input NJM2115's, interestingly enough.

And a quick look at the front panel - not a whole lot going on here. On the front there are the encoders, the LCD, all the LEDs, a little LM78L05 regulator to power the logic chips, three SN75HC595 shift registers and a handful of transistors for driving the LED matrices.

Even less going on around the back - a lone Elna(!!!) capacitor on the input of the 5V regulator, and three SN74HC165N's for reading the encoders and associated buttons.

And that about wraps it up. Arguably "old" tech, but still very much usable. Stand by for whenever i manage to get around to putting some of my plans including this, into practice (but don't hold your breath)...