Despite having hoped to nab an Audient interface for some time now, only quite recently did i manage to get my hands on one. Since i'm a cynical... [ahem] person, after having heard several notable(?) audio engineering channels over on Youtube ranting and raving about these, and the "Audient console preamps", and the "discrete JFET DI" and whatnot, my curiosity was indeed itching to see if and how the reality lives up to all the hype and marketing.
This one, according to the seller, basically had the whole of the analog side of things dead. The unit still connected to the computer, got recognized by its control software, and the digital ins & outs work, but no activity on the analog ins & outs. The listing also contained a couple of internal photos, and mentioned that a power chip used to have a heatsink glued on top of it, that had fallen off. Judging by the groups of components around that and the associated transformer, that looked to be in charge of precisely the analog rails. But let's not get ahead of ourselves, one thing at a time.
A reasonable complement of ins and outs, for what's marketed as a "desktop interface. Two mic / line inputs, one of which can also accept an instrument input (via the separate "DI" 1/4" jack socket; most other manufacturers' units accept instrument inputs within the same combo-XLR inputs, via FET-input opamps, but i digress) with send/return connectors for inserting effects or other processing before the analog-to-digital conversion, two pairs of analog outputs, a headphone output, and an optical input/output pair (each of which can be independently switched between ADAT and S/PDIF).
Not that i'm shallow, but... (especially) for a product that retails around $350 (if not more)... Do you really want or expect the logo to be nothing more than a crappy little sticker? Only to have it start to peel off on its own, after a couple years? I guess silkscreening just "wasn't good enough"(?).
After removing a few screws, and all the washers and nuts off the jack sockets, the faceplate lifts out with all the guts mounted to it. And we get a first spoiler, near the middle ribbon connector - "XMOS BOARD". To be fair, they've pretty much cornered the market in the low-to-midrange USB audio interfaces; MOTU and i suppose RME roll their own ARM / FPGA-based solutions (and perhaps Focusrite in the Clarett series?), but their products stretch into the 4-figure price range as well, so it's not 100% a fair comparison, perhaps. And that's without getting into the more "esoteric" of brands...
But what greets us is the analog output / power / converter board, the largest one inside the unit. The components are all on the visible side, but we'll get back to this later...
Removing said board, we end up getting mooned. Onwards and upwards, or... something-something.
The smaller of the boards is the digital / processing side. Quite cute of them to have included "diagnostic" LEDs for each of the five power rails present on the board, as well as for the PLL lock.
The DC input, apart from branching off to the converter board, is fed straight into a National Semiconductor / Texas Instruments LM2670 5V buck-regulator. The two SOT23-5 packaged parts are a pair of voltage detectors / reset generators - the "SCJRYP" is an OnSemi NCP303 3V-threshold detector, while the "SDERYK" is the same chip but with a 0.9V-threshold.
Hidden amongst much taller through-hole components is a tiny little NCP1521B buck converter, most likely in charge of the 3.3V rail.
After having spent a good few hours scouring the web, for the life of me i couldn't make out the marking of the 6-pin chip (nevermind identifying it), outside of the FairchildSemi logo. I could've sworn the "PH1AB" on the top row would be the marking code, and the "2012" below it, a week-year date code. I even ended up looking up lists of discontinued FairchildSemi chips, in the hope of getting at least a bit closer. With a bit of google-fu i managed to stumble upon this feature on the most complete(?) SMD marking database i've found so far - a pretty long list of chips, listed alphabetically, and with a short description that would ease filtering the dead-ends right off the bat. Out of some level of desperation, i ended up looking for images of a FAN2002 chip, and happened to find one on LCSC's website. The second row on that chip said "2002". Well i'll be damned - that means this little bugger is a FAN2012... This one's used to regulate a 1V rail, most likely for supplying the core of the Xmos controller.
Speaking of which, the Xmos chip that was so rudely given away right at first glance, is an XS1-L16A-128-QF124-C10 - now there's a mouthful. Took a while to decode the "SK1302L2C5" marking on top of the chip, ended up digging up Xmos's latest list of device markings. Next to it is an SMSC / Microchip USB3318 transceiver handling the USB interface (shocker, isn't it?), powered by an OnSemi NCP699 1.8V linear regulator. . The firmware is stored inside an Adesto AT25DF041A 4Mbit / 512Kbyte flash memory chip.
Off to one side we see a familiar face (as it were) - the "2PLC" marking indicating a "classic" Cirrus Logic CS2300 PLL. Between that and the 13MHz quartz crystal is a small single OnSemi NC7SZU04 inverter, employed as an oscillator.
The only other chip of note is a MAX11601 4-channel analog-to-digital converter, its inputs being the two potentiometers on this board (one for the "master" volume, and one for the headphone output).
Now, it's about time to get back to the analog in/out and conversion board, and take a closer look. I've got to hand it to them, it's kinda cute how they deliniated each "stage" in its own little silkscreened perimeter, and with its own group of component designators. Easier to reference components in the design stage, i suppose, as well as prototyping and troubleshooting. One thing i'm not impressed with is the way all the through-hole capacitors look like they were soldered by a blind apprentice in a hurricane.
Even less impressive is the capacitor selection in a relatively "upmarket" unit such as this: Jamicon (bearable, i suppose), "LH.Nova" (never heard of these), Fujicon (haven't heard good things). And these are that much more baffling when contrasted with all the brand-name silicon, and Murata RFI filters etc. *sigh* Well, profits gotta come from SOMEwhere, right?
The main power controller on this board is a Linear Technology LT3439 - an interesting little beast, allegedly very low-noise, push-pull converter. Oddly enough, it seems to have been designed to run open-loop (without any feedback), the output voltages relying merely on the winding ratios of the transformer. This is in charge of creating the +/- analog voltages for all the opamps, the 5V rail for the audio converters, as well as the phantom power rail. The latter also includes a regulating Zener diode.
The two other regulators present are marked "H110" - i tried for a while, but didn't manage to figure out what part number these correlate with, so yeah... Sorry. Either way, my best guess is these handle the 5V analog voltage for the converters.
The conversion of the two analog inputs to digital is handled by a TI / Burr-Brown PCM4220, while the buffering opamps ahead of its inputs are a pair of JRC NJM8080's. As are the other nine opamps on this board, in fact.
The digital-to-analog conversion is done by a trio of TI / Burr-Brown PCM1798's.
The headphone amp is a mostly-discrete solution, employing a pair of BC317's and a pair of BC337's
And now, for the "piece du resistance" - behold, and hail to the almighty Audient console preamps, and discrete JFET DI...
As opposed to the converter board, the two opamps here are a pair of TI NE5532's. Also, i can't help but get the feeling that this board was (and needed to be) manually retouched. I can't imagine how a reflow machine could create such a glob of solder on the end of an 0603-size component (CA3, in this case, as well as CA14, DA4 etc). Or hey, maybe their subcontractor's manufacturing process is just that "finely" tuned... To once again quote AvE - "It's pretty... Pretty ugly..."
Bow down, mere mortals, or else suffer the wrath of the discrete JFET DI!... Oh wait, no, it's just a boggo J112. Wow... 10 cents/pc @ 1000pcs. How come a TL072 couldn't do the job just as well? Too undistorted, perhaps? Then again, there's also bragging rights to think of...
Now, being the curious "masochist" that i am, i took the liberty of lifting the schematic of virtually this whole board. The two preamps are identical, apart from the multiple-switch 1/4" jack on input 2, for the high-impedance buffer part. For sanity-checking, i used the component designators for channel 1, but shoehorned in the instrument buffer as it's connected on channel 2.
Not exactly what one might call "rocket science" - a variation on a classic theme, albeit with a couple twists. More like a 2-in-1 sort of deal (a compound differential pair of transistors "on top" of a differential pair), with the discrete section (at least partially) within an opamp's feedback loop.
And that about sums it up for the guts of this thing. Coming up soon, the repair. Hopefully, anyway...
Great job at reverse engineering the preamp section! It does indeed look like a similar topology to the one used in the Behringer FCA1616, and patented by Soundcraft. I wouldn't say that this is a classic circuit as a whole, though there is nothing new here at the level of the building blocks. As a disclaimer, I don't have enough perspective on preamp design and amplifiers in general, so can't make definitive claims.
ReplyDeleteAnyway, as an exercise for myself, I'm trying to analyze it. I'll add replies as I go, hope you don't mind Khron. Probably I'll make mistakes along the analysis, anybody reading this is welcome to correct me.
So Q7 and Q8 serve as a feedback current source (definition from Bob Cordell's book). This replaces a more primitive current source in the aforementioned patent which is based on a transistor and 2 diodes. The Behringer circuit downgrades from that with a barebones 1-resistor "current source". The current source dramatically improves linearity in the input stage, and for a differential amplifier also greatly improves the CMRR (reducing noise).
Delete(of course the single-resistor "current source" in the Behringer does a much worse job than the proper current source in the id22)
DeleteThe 1st input stage here is a cascode configuration, similar to the one employed in the Behringer. There are a couple of differences. First, in the id22 the cascode resistor is replaced with a complementary feedback pair (Q3-Q4 and Q5-Q6). Second, in the Behringer, there is a network of passive components across the base-emitter junction of the amplifying transistors (not sure about their function), and in id22 there is only a diode across that junction. The diode seems unusual and I can only make sense of it as an overvoltage protection feature.
DeleteFor reference, the Behringer schematic is here:
https://drive.google.com/file/d/17W-MV_yg0dkG0bn6FMaWxGqrGunVHH1H/view
And the patent is here:
https://patentimages.storage.googleapis.com/52/76/46/8792fb86ea4215/GB2242089A.pdf
* meant to write "in the id22 the cascode *transistor* is replaced with a CFP"
DeleteWell, that's much more in depth than i could've (or would've been willing to) analyzed this circuit. But as i mentioned, no "alien technology" to be found here. Great job on "dissecting" the preamp(s), though!
DeleteI am still confused as to why is there no feedback path from opamp output to left side (Q3), only to right side (Q5). Wont this make the left side "static" ?
DeleteThanks for the excellent writeup! I'm working on a defunct ID22 now, seems like there is an issue in the power-supply section on the "analog" board. If you have any notes/schematics etc. on that section I would really appreciate it. Trying to source the part id on the power transformer, but no luck so far. Mine is also labelled Best B5680 on the metal housing. Seems like I get no voltage on the secondary side, so that's my focus for now.
ReplyDeleteLow-power transformers very rarely fail. I'm not saying it's impossible, just that it's very uncommon. Silicon devices, on the other hand, fail very frequently. For this reason, you would generally want to rule out every silicon-related failure mode before concluding that the transformer is to blame. A simple test you could make would be measuring continuity of the primary and the secondary of the transformer out of circuit. Anyway, before tinkering with the transformer, I would test all related circuits, focusing on LT3439 on the primary side, and all silicon on the secondary side, including diodes, regulators and what have you. BTW I believe there is some analysis of this area in the above writeup.
DeleteThanks for the reply! I was thinking the same thing about the unlikeliness of the transformer crapping out, but I don't know the circuit well enough to make sense of it. The LT3439 is a beast I haven't encountered previously, so I need get up to speed on that one. What appears to be multi layered PCB doesn't help when it comes to mapping out the circuit. Thanks again for the feedback!
DeleteGenerally first things to check are shorts on power rails. In circuits around switching regulators, there is a coil and it's usually quick and easy to check for shorts to ground on these, then for voltages if no short found. For an LDO, do the same on the output pin. If no fault found this way, it makes sense to trace the circuit back to a point where there is an input voltage but no output, and then check that subcircuit in detail, including enable signals if relevant, feedback paths, shorted silicon devices, continuity etc. Another shortcut would be to use a thermal camera in order to visually check the board for hot spots (which would indicate a shorted component). If you don't have a thermal camera, in case there is a major short, you might be able to sense it with your finger.
DeleteAnd i'm back, after being away for the weekend.
DeleteAnton pretty much mentioned what i was going to say - silicon is far, FAR more likely to fail than a (relatively chunky) piece of wire. Especially in such low-voltage and low-power situations as this.
Not to mention, if the LT3439 in your unit is as "well" soldered as in this one (as in, it wasn't), then it's no surprise it crapped out. That metal belly-pad is SUPPOSED TO be soldered to a nice fat ground-plane, precisely for heatsinking purposes. I'm still embarrassed FOR Audient's engineers, trying to place that piddly heatsink band-aid on top of a shoddily-installed power supply chip.
As i believe i mentioned here, the faulty output transistor (of the two inside the LT3439) was actually NOT short-circuited. So without power applied, there was no short to ground or anything. But while the chip was working, it seemed that the faulty-side transistor was kept in an "on" state, conducting. Very odd failure mode, but there you go.
I've since sold off this unit, and i've yet to hear of any complaints from the buyer, so i'm calling it good. Nor do i expect it to fail anytime soon, not after soldering the new chip on as it was suppsed to be in the first place, AND adding that chunk of heatsink on top.
Thanks a lot for answering in great detail, both of you. More than I could've asked for! :)
DeleteWill do some short testing and more voltage testing for sure.
The heatsink on the LT3439 is darn hot to the touch. It could very well be somewhat the same fault, with being in an "on" state. I don't have an SMD soldering station though, so I've been a bit wary to do anything to it (and the other smd-components). But I see that you managed without one :)
Since the heatsink is so hot, I'm not sure if it's the best thing to pry it off to test the voltages in/out since I'm a bit worried that it might burn out. Either way if it comes to it, I might have to replace it altogether. Fingers crossed. Maybe the chip just falls out? Wouldn't be surprised seeing the quality of some of the soldering elsewhere on the board. I managed to crap out one of the 33uH filters while desoldering (heat transfer, too much heat etc...) The plastic casing on filter was partially melted so will have to replace that one. Some of the capacitors looks quite dented on the outside (no bulging though), so I'm suspecting another blind trainee in a blizzard wielding a hot soldering iron around on the board...
Somehow, i'm not surprised it's hot. I'm willing to bet that chip hasn't been properly soldered in the factory either, otherwise it would be sinking the bulk of the heat into the groundplane on the top & bottom of the board, underneith it. That's what it's for; that heatsink must've been a last ditch attempt to not need to manually rework all those board, nor have them scrapped and written off, and respin a new batch.
DeleteAs you can see, mine came with the heatsink already "baked" off so... Not much to lose.
Yes, i managed without using hot-air, but that's only because i was stubborn, as well as having invested a sum in a not-quite-bottom-of-the-barrel soldering iron, a few years ago.
Just because the plastic shell of an inductor might've melted, doesn't mean it's broken or dead. That's just there for protection / isolation. If you can still measure continuity between the two pins, i'd call it good.
I'd suggest to check for shorts on the cathode of each diode rectifying the current on the secondary side of the transformer. If no shorts detected then, considering the fact that the chip is getting very hot for no apparent reason, I would assume that the chip is faulty.
DeleteA couple of tips, though. I would not recommend removing the chip the way Khron did, unless you have tons of experience with this sort of stuff. This way it is very easy to lift some pads, in which case you would be looking at a more complex repair which could be quite difficult to do without professional soldering iron (and some experience). Also re. your soldering iron, if plastic casing of that coil melted without direct contact with the iron, that means that the temperature of the iron tip is way too high for this sort of delicate job. If it's impossible to regulate the temperature of the iron, I'd recommend to not use it for this purpose. Sorry if this is a bit discouraging, my intent is to give you a warning that improper equipment may easily turn an easy and quick repair into a complex and expensive one. If your goal here is to gain some DIY electronics repair skills then maybe it's a good time to consider getting better equipment (including a hot air station). If the goal is fixing this device then maybe borrowing proper equipment or simply hiring someone who has it might be a good idea. This sort of request comes up quite frequently in my local electronics-related Facebook group and there always are a few people willing to do the job.
The filter casing was already partially melted when I received the interface. That's why I tried to desolder it in the first place. But I made a dumb mistake with the wrong temperature selected on my soldering station. My bad.
ReplyDeleteI've been building quite a bit and doing some electronic repair, so I'm fairly confident in most of my soldering. But we'll see when it comes to the LT3439. It might be just out of my league :)
Thanks again for all the helpful tips!
If that turns out to be the case, depending on where you live, i might be tempted to make you an offer on it... :D
DeleteI'm in Norway :)
DeleteI suppose that's workable, but first let's see if you manage to revive it yourself :)
DeleteMaybe a typo, but I wanted to ask about the phantom power portion of the preamp schematic -- Is R7 meant to connect to +48V (vs. +24V listed)?
ReplyDeleteI'm trying to track down the source of some dodgy phantom power.
Eagle didn't have a "+48V" supply symbol, so...
DeleteYou'll note the analog supplies are "V+" and "V-". No point in being too-too nitpicky with what a supply net is labeled as :)
"Dodgy" how?
Sir i have id22 its age is almost 3 years old nowadays the volume pots are loosing their toughness i mean to say its not feels like healthy and one of them is sounds crackling while moving the gain up or down . Can you please tell me where i can buy new potentiometer for this id 22 i searched alot on (RK09 B) this but part number is not on online as well as ofline . Where i can find oem parts of audient . Please provide a link of these pots.
ReplyDeleteThank you .
I doubt Audient will sell you any spare parts. In a pinch, you could try some contact-cleaner spray, and work the pots from end to end a few times.
DeleteBut which exact pots are giving you trouble?
Hi at the beginning you said the analog i/o was completely down, i assume you mean the headphones side too ? if so thats exactly the same as mine, it connects to the mac and the optical I/O works, just everything else up shit creek.
ReplyDeleteThe headphone output is indeed an analog one, and is as such included in the "analog i/o".
DeleteOdds are good the LT3439 has given up the ghost in yours too. Not a major shocker, given the things i've seen...