Right, so, a few weeks ago (actually, well over a month ago - dang, time flies...), i stumbled upon a German eBay auction for what seemed to be an Apogee Symphony I/O, a Symphony 64 Thunderbridge, and the cable to connect the two, sold as faulty / untested. I turned out to be the only bidder (after there were none, the first time the listing was put up), and especially after seeing some "sold" prices both on eBay and Reverb.com, i decided a bit under 200e was worth taking a chance on. About a week later, i picked up the package from the post and, as it turned out, at least the main box, the Symphony I/O, was actually working perfectly fine!..
I had to cobble together a DB25 male to 8x XLR "pigtail" to be able to test the line inputs & outputs, but that was a small price to pay for confirming the functioning of this thing. Well, that, and the OLED display situated between the two knobs on the front seems to be dead (replacement's ~20 bucks) - only discovered that upon seeing a video how to switch the thing to "USB Audio" mode. But let's see what's inside this then, shall we?
Pretty fully-packed in there, one might say. Structurally, nothing particularly out-of-this-world though, at least at first glance. But on to the smallest board in there, the power supply.
OEM seems to be Astrodyne (who may or may not have made the PSU in the old silver Ensemble Firewire i fixed some years back); 24V 6.6Am so 150W in this shockingly (if not concerningly) compact brick. Pretty certain that rating requires hefty forced cooling, and even the rear panel claims "150W max", but that's quite something for what is basically, at most, a 16-channel audio interface... But maybe that's just me.
On the bright side, all Rubycon electrolytics - despite that, i've read several stories of PSU's dying in these things, so... As mentioned / implied above, these need forced cooling towards full load, but i'm still surprised at the lack of any sort of heatsinking whatsoever. Interestingly enough, the bigger transformer is actually the active PFC inductor, so there's that too. Also, plenty of mains filtering, so not much corner-cutting that i can see.
Speaking of active PFC, that's run by an ST Microelectronics L6561 controller. The main PWM looks to be an OnSemi UC3843B, but interestingly enough, there's also an International Rectifier (now Infineon) IR21531 half-bridge gate driver and a pair of OnSemi FCD5N60 MOSFETs - no doubt, for considerably greater efficiency than a mere flyback converter could allow, for the intended output power rating.
On to the main logic board then. Quite fully-packed, one might say. Decently compartmentalized, although i'm not too sure about the spread-out power supply section, but oh well - what do i know?
The first, easily recognizable chip here is a Xilinx / AMD Spartan-3A XC3S1400A (top of the range) FPGA. The two TI TPS73701 near it supply a 2.5V rail (U20) and a 1.2V rail (U46)
The next one took some digging - "thank you" Xmos, for hiding all the documentation of your even-recently-discontinued products... Good thing the format of the encoded marking (UC0944G4) looked vaguely familiar. First i had to dig up the document / page where said markings are detailed (which is no longer the PDF it used to be), then unsuccessfully search for the part number given there (XS1-G04A-FB512-C4), but on the other hand, did somehow manage to access the datasheet for the G04B version of said chip, eventually... Which also explains the mention of Xmos in the firmware update app for this.
Next to it resides a Lattice LCMXo256C-3TN100C CPLD. Interesting they'd mix manufacturers (and not just stick with Xilinx), but i guess part cost might have offset the increased development complexity? U10 near it, another TI TPS73701 provides a 1.8V rail.
Off to the side we have (some of) the digital comms interfaces - a Broadcom BCM5481A2 Gigabit Ethernet transceiver, and (apologies for the failed framing) an SMSC / Microchip USB3317 USB transceiver.
Next up there's an Analog Devices AD9547 clock generator. Quite something - just this on its own, even in bulk, is over 20eu a piece(!!!). That's being fed by a Renesas ICS512 PLL clock multiplier. The TSSOP-8 packaged IC next to that is an OnSemi NB3N2304NZ clock fanout buffer, while the 16-pin chip next to that is an Analog Devices ADT7470 10-sensor-input, 4-channel fan controller. Despite there only being a single fan in this unit. Ain't no kill like overkill, eh? Yet another TI TPS73701 provides a separate 1.8V rail to these.
Moving on to the power supply section, the largest area is taken up by what looks to be a flyback converter, based on a Texas Instruments LM3430 boost controller. Best guess, this provides the analog voltages to the headphone output circuitry, but let's not get too ahead of ourselves just yet.
(Later edit) Indeed, this provides a +/-20V(!!!) supply to the opamps on the main board.
Nearby, a TI TPS60110 charge pump provides 5V to the headphone DACs, while a TI TPS40055 is in charge of the main 3.3V rail supplying most of the digital circuitry (easy to guess by the humongous 2.7uH inductor and two SMD MOSFETs).
The tiny 10-pin chip is an Analog Devices ADM1185 4-channel voltage monitor, while the QFN-packaged chip off to the side is a TI TPS54218 buck-converter (supplying 1V to at least the Xmos core).
Next up are the two DACs - a pair of Cirrus Logic CS4398, in charge of the two headphone outputs on the front. Here you can also get a better view of the USB3317 chip.
The DAC output low-pass filtering is done by a set of National Semiconductor / Texas Instruments LME49860, which also handle the driving of the headphone outputs themselves, through a bunch of BCX51/54 bipolar transistors.
Time to dive into the "Analog Optical" board then. No dearth of components here either, and the compartmentalization is also commendable.
The main slab of silicon here is a smaller-than-on-the-mainboard `Xilinx / AMD Spartan-3A XC3S700A FPGA
The 8-channel DAC is a bit of a mystery though. It had an "Apogee DAC" label on top of it, with an internal part code, and the original markings seem to have been removed (no amount of cleaning with IPA could remove what look like marker; i'll have to try some acetone as well). It must contain some PLL block as well, since the oscillator next to it is an even 40MHz, which is decidedly not a multiple of either usual sampling frequency, 44.1kHz nor 48kHz. That "should" reduce the number of potential candidates, but the search is made more difficult by the high likelihood this is an obsolete / discontinued part. Also, it's supplied with a 3.3V rail and a 1.2V rail, which should shorten the list of suspects even further.
The DAC output filtering is handled by a bunch of Texas Instruments / Burr-Brown OPA1632 differential opamps, while the outputs themselves are driven by OPA1612's.
The analog inputs, after going through the "passthrough card" (which would otherwise route the signals through the mic preamp add-in board, when installed) and the (i'm assuming) +4dBu / -10dBV switching relays, are low-passed by some more OPA1612's, then fed into OPA1632's.
After passing through another set of relays (don't ask me what for; muting perhaps?), the signal goes through a set of Analog Devices DS1882 digital potentiometers (likely for the "coarse" level adjustment), then into a pair of Cirrus Logic CS3318 analog volume controls (likely for the "fine" / calibration trim).
As for the power supply side of things, the core of this is handled by a Texas Instruments LM3430 - intended as a boost controller for LED lighting, but functioning as a flyback converter, driving an International Rectifier / Infineon IRFR4105 MOSFET, and provides a +/-15V supply to the analog circuitry. Next to the controller resides a Linear Technologies / Analog Devices LTC2927 power supply tracking controller. Also in the area is an OnSemi NCV5173 boost converter, driving a dual inductor - interestingly enough, it takes the main 24V input and puts out 7.5V. That then gets discretely linearly regulated down to the 5V the ADC's require.
Other power circuitry is on the edge of the board - mainly a Microchip MCP1826 1.2V 100mA regulator; U60 seems to be a single Schmitt trigger inverter (Texas Instruments SN74AHC1G14), and not sure what U61's supposed to be, but traces seem to run from that area to the S/PDIF in/out connectors, so it's likely another single logic gate.
Moving on to the last major assembly inside this device, the front panel.
Yet another Lattice MachXO part, this time an LCMXO2280C-3TN144C, mainly employed as an I/O expander (serial data coming in, lots of parallel data going out, driving the zillion LEDs of the level meters and so forth).
Not all that much else going on, save for powering the little OLED display, via a National Semiconductor / Texas Instruments LM3490 12V 100mA low-dropout regulator.
Speaking of said display, fortunately it doesn't seem to be anything terribly exotic or unobtainium (as opposed to some of the LCD's in MOTU mk3 units) - UG-2832ASWCG01, a 2.2" 128x32 pixel OLED display, using (possibly/hopefully) an SSD1305 controller. The closest replacement i found is one of these; we'll see in a couple of weeks whether it works or not.
But that just about wraps up this "chapter"; stand by for the teardown of the Thunderbridge next (should be a relative "quickie", especially compared to this). And after that, a teardown (and at least re-cap) of an item that compelled me to break my boycott of Presonus hardware. Bought retail too, no less!..
No comments:
Post a Comment