Either way, this was sold as "not detected by the computer", so my bet's on the Firewire chip being toast, "as usual". Can you see a pattern emerging here?
I myself have two of these in my "recording rack", serving as 8 mic input to ADAT converters. As part of the "great bargain" aspect of these, even among the 8 mic input interfaces nowadays, having attenuator pads on all the inputs is quite a rare feature; more often you'll see that on only inputs 1/2.
Two separate headphone outputs are also a handy addition, if you're tracking, say, an acoustic guitar, or some vocals, in the control room (or if you don't have a control room at all). This would allow you to have one "full" mix for the engineer to listen to, and a separate customized one for the performer.
Another nice feature (which comes in VERY handy in a rack) is the mic inputs being all on the rear, but the (switchable) high-impedance inputs on the first two channels are on the front. They decided to keep the power supply external, which has its pluses and minuses (no electrical polarity pun intended), but that just means the unit can be quite compact (arguably).
One minor niggle might be the phantom-power being switchable only in two banks of four, but that can be easily worked around, if needed. Another one might be the decision to have a custom breakout cable for the wired digital I/O (wordclock, MIDI and S/PDIF), but that's not the end of the world either. Granted, the signal metering is perhaps a bit on the spartan side of things, but since this is an audio interface, you'll be monitoring the levels on the display anyway, so...
Right, first peek under the skirt - nothing terribly sci-fi going on here. Cute detail - they kept the "digital" board as separate from the main analog-conversion-power board. I wonder, is it a mere coincidence that the later M-Audio Fast Track 8R Ultra has a damn near identical layout, save for the digital I/O? I'll let you decide...
Two daughterboards obscure the arguably more interesting(?) bits. The one along the front edge carries the two instrument input jack sockets and nearly all of the LED indicators for the front panel. The smaller one on the back carries the DE-15 connector for the (wired) digital I/O, as well as the DC input jack for the power.
Removing the former requires taking off the front panel (along with all the knobs). On the top side (the front daughterboard being upside down) we have the high-impedance instrument inputs, handled by a TI TL072C, a pair of 74HA595 shift registers for driving the LEDs, and a pair of KSP8099 as the two switches / regulators (or rather, capacitance multipliers) for the phantom power.
The rear digital daughterboard contains a pair of 74HD04 hex inverter buffers for the S/PDIF and wordclock ins and outs, a 6N138 optocoupler for the MIDI input, and a 74AHCT1G125 single non-inverting buffer driving the MIDI output.
The main digital daughterboard is reasonably tightly packed. The main processor / controller is a TC Applied Technologies TCD2220 Dice Jr., with the associated 64Mbit (8Mbyte) K4S641632N RAM chip, and an EON EN29LV160AB 16Mbit (2Mbyte) flash storage chip for the firmware and settings. Plus the "ubiquitous" (as it were) TI TSB41AB2 Firewire transceiver.
Quite a comprehensive power supply section, but then again, when you need 5-6 different voltages out of one single DC input, things are bound to get somewhat complex. A VS-30WQ06FN-M3 3.5A Schottky diode is the first thing after the power switch, which then supplies the single (multiple-output) flyback converter. Its controller is a tiny TI LM3488, driving a Fairchild (now OnSemi) FDS8449 N-channel MOSFET. This power supply is responsible for supplying the analog side of the circuitry (+/-15V, phantom power, and with the aid of a Sipex SP7805V, +5V for the ADCs/DACs). A National Semiconductor (now TI) LM25005 buck converter is responsible for the 3.3v for powering the digital circuitry.
Also visible towards the bottom-right is a TI LC549C 8-bit ADC, for reading the position of the master volume knob (which can control several outputs, not just the first stereo pair).
Most of the capacitors are Suncap (yuck), with a light sprinkling of Lelon (ditto).
Interestingly enough, the microphone preamps are done with bespoke-designed THAT1510s (as in the Alesis io|26, just in the 8-pin package instead of the 14-pin one).
The gain adjustment potentiometers (with built-in push-pull switches), as well as the three are made by Alpha.
In fact, perhaps apart from the electrolytics, pretty much all the components used in this unit are, surprisingly enough, from pretty big brands. Amphenol XLR / 1/4" combo input sockets, Cliff UK output jacks.
Speaking of the output jacks, a nice touch - muting transistors.
All the rest of the opamps are JRC NJM4580.
The A/D and D/A conversion is handled by a quad of AKM AK4620B codec chips.
And apart from the odd glue-logic chip, that about covers the teardown. On to the repair stage...
The eBay listing for this mentioned that "unfortunately it just won’t connect to my Mac which apparently is a common problem" - it sure is. As you might've guessed, no Firewire activity.
A little diode-test probing showed the expected 500-odd mV on the four data lines of port 1. Port 0 though, slightly different story: 320mV on TPA0+, 220mV on TPA0-, and normal on TPB0+/-. Not much of a shocker. In expectation of this, just last week i stocked up, and got 10 spares from Mouser.
[~10mins later...]
Sure enough, TSB swap results in getting detected by the computer, and the mixer application starting right up. Pending double-checking of I/O functionality, i'm calling this one done.
I recently used the wrong adapter on my Fast Track Ultra 8r and now won't switch on. Had a look at the board but doesn't seem to be anything major fried. No multimeter to check just now but any pointers where to start? Relative newbie...
ReplyDeleteHow "wrong" was that adapter?
DeleteNot sure there's much you can do or find out with out any sort of test equipment, though. Unless you're a major fan of the "shotgun approach", and/or throwing money and parts at a problem...
I don't recall, nor do there seem to be any sort of fuses on the power input, but there's that DPAK-packaged diode in series (as reverse-polarity-protection), and then straight into the primary side of that flyback-converter, so... Depending on what you plugged into the input, if it was outside the controller chip's ratings, you might've toasted that.
But yeah, buy some cheapie crap $2 multimeter from your local hardware store or thrift-shop or something. Any working one will do, we're not dealing with the AC mains here.
It was the wrong polarity, same voltage and amperage but AC unfortunately. I'm glad you mentioned the lack of fuse as I've been staring at the thing for a good 20minutes wondering where it could be hiding haha. Yeah, I'll get a hold of a multimeter tomorrow so where how far, if at all, power is getting down the line!
DeleteWell, AC doesn't have a polarity, as such. Either way, with that protection diode rectifying the AC, the resulting DC would've been 1.4 times that.
DeleteIt's highly doubtful that 470uF capacitor failed shorted, and the rest of the power components there are rated for 40-42V, so at least in theory, they should've survived just fine, but...
Well, without some hard data, it's all just guesswork.
Hi, mine a few weeks ago worked as normal, but start taking too Long to turn On, it takes Like 5 min as it was warming up, couse the blue light indicator was blinking repeatedly until it got the blue light steady and works as normal, then was taking more time to turn on, but today did not turn On at all Just flash the blue Led one time has always and then goes off and nothing happened, i was expectin the blinking but nothing happen, do you know What could it be the problem? Thanks
ReplyDeleteAssuming the power-brick is fine, odds are pretty good that if you pop the case off, you're likely to find a few "fattened up" electrolytic capacitors in there.
DeleteI've had that happen with my TC StudioKonnekt 48 a couple times along the years, no biggie.
This comment has been removed by the author.
ReplyDeleteYippee! At last! And I hope it's not premature? You must be aware of the Profire 2626 phantom power problem? (PPPP)? Or as Biddy would say, Pish Pish Pish and Pish! It does not serve any of my condenser mics. It seems to be a fairly common problem that has never been resolved as far as I'm aware. In one forum someone suggested renewing 8 capacitors but never went into any detail. Now I find your post here and it gives me hope! Tell me how to fix this problem because I love the 2626 and want to purr through it with a big f**k off condenser mic. Will ya help? Will ya? Will ya?
ReplyDeleteOh, hey, sorry for the late reply, i must've somehow missed this one.
DeleteWell, if the boost-converter that creates the 48v from the input voltage went belly-up (or tits-up, whichever you prefer), no matter what or how many capacitors you swap, it won't magically bring it back.
So that's where i'd start investigating. In the 10th photo on this page, the one with the power supply section, near that white connector that goes to the top board (even denoted as "PHANTOM"), there's a clearly marked test point, "TP940", with "52V" under that - see if you get anywhere near that voltage there.
In case that's ok, the issue's on the top board, where there's a pair of discrete voltage regulators (controlled by the two front panel switches) feeding each set of 4 inputs.
I don't think i've snapped any photos of the bottom of the board, but i don't remember if there's any meaningful circuitry there. If C950 and C947 (almost in-line to the left of TP940) are rated at 63V, then D945 and D946 (between the two capacitors) might be the rectifiers for the phantom power rail.
No, i hadn't heard of the PPPP, but then again, after this teardown, i proceeded to re-cap the whole power section with Japanese electrolytics (and a few solid polymer ones), so... I don't predict encountering any power issues for a long, long time.
PS: No, actually that was a lie - i "only" recapped the two 2626's in my recording rack, which i got some time ago; this particular one i think i sold off together with the Presonus 1818VSL i had bought to grab the firmware from.
DeleteHey, thanks very much for taking the time to reply. I'll have to go over what you said very slowly to understand it as my brain is not tuned into electronic systems much but I'll figure it out hopefully. Thanks again. Btw They've stopped updating the drivers for this model so it seems it will not work on the latest Mac OS. I'm stuck, therefore on Yosemite for the time being. Any tips or advice? ...and, thanks again.
DeleteHi Khron, first thanks for your blog! I'm enjoying it a lot. Second, I have the device at question (Profire 2626). A few years ago I made a few recordings with it and some of them came out with odd artifacts that sounded like samples got occasionally mixed up. Since then, I have only been using it sometimes as a set of mediocre-quality preamps. Today I took it out of the closet to see if there is any way to identify the problem. I did notice a few fairly odd things. For example at 192kHz it drops samples at anything less than 512 samples buffer size (not that I need this high a rate but just for the test's sake), and at buffer size of 64-128 is completely unusable. That of course is inconclusive since it could be related to my DAW, firewire chip in the computer etc. Otherwise it seems to be working OK - I recorded a few hours of sine wave looped through its D/A and A/D into a DAW and that came out clean. I checked the wordclock signal and there are a bit more irregularities. The clock is a little off (at 96kHz my scope actually counts 96.02kHz, while for my RME interface it shows exactly 96.0000kHz), and the word clock when loaded with 75 Ohms as per spec, has a swing of just 1.6V peak-to-peak (the word clock spec defines 5V peak to peak, and I measured 4V from my RME interface when loaded). The clock is fairly solid with not a lot of jitter (given I don't have a proper technique to measure jitter but it compares fairly well with the RME clock). When slaved to RME via word clock, it sometimes won't sync and sometimes it will - I couldn't reliably reproduce the sync issue. When it does sync, it picks up the RME clock reliably and works at the correct frequency with no surplus jitter. So currently I'm not sure how to go about it and I wanted to ask you if you've seen anything similar. And also, if you think there is an easy way to fix the clock. I haven't got to disassembling it yet but I've seen on your pictures that there is a 25MHz clock that's used as a reference (and that's confirmed by the Dice spec, I've gone through the user's guide and there is a nice little schematic for suggested implementation). So I think that either the crystal itself isn't good or one of the 2 loading caps is out of spec. If I go with crystal replacement, I'm not sure how precise the new crystal should be. Looks like vendors like mouser and digikey only offer lower-precision crystals (as in 30 ppm minimum). I'm a computer tech, so I could try to extract a crystal from one of my donor boards - some of them have a higher-precision 25MHz crystals, however the package is of course different. Anyway, I'd appreciate your thoughts.
ReplyDeleteThanks for the kind words, and glad to hear this is of use to others - that's at least part of the reason this exists in the first place.
DeleteRegarding the rest though, i wouldn't blame the clocks for the not-great-stability at high sample rates. You're forgetting the DRIVERS. Which may or may not tie into the hardware used, on the logic side.
Regarding the clock accuracy aspect, i would argue it's quite irrelevant, for all intents and purposes. What matters (more) is that whatever digital equipment is connected is sync'd to each other (or to a "master clock" in one form or another).
On the wordclock front... Not quite sure what to say. The seemingly low output amplitude should only be an issue if other devices have a tough time syncing to the 2626 acting as a clock master. And concerning it acting as a slave, well... Considering it doesn't have the BNC on the unit itself, one could also suspect that dinky breakout cable. I never got one of those with either of my 2626's, nor can i say i (would)'ve had much use for those.
And on crystals - i seriously doubt that's an issue too. If it was a dud, it likely wouldn't work at all, so i wouldn't sweat it.
Do keep in mind, the wordclock male BNC on the end of your cable doesn't plug straight into the Dice...
That being said, is wordclock BNC your only option for slaving the 2626 to your RME (or whatever else you need / want to use it with)?
I appreciate your very detailed answer. I guess my main purpose here is to gain back confidence in the device so I wouldn't have to hesitate to use it again (or maybe sell it). It was working perfectly when it was new, then at some point something went wrong. Or at least that's my impression. Because the perceived problem is intermittent and it appears to be somewhere in the digital section, I've been measuring anything related that I could think of. Perhaps some of the oddities are just a red herring. Of course drivers are also a part of it but I doubt that they are to blame. I am a professional computer tech as mentioned above and my machines are well serviced. I've tested the Profire on 2 completely different computers with Windows 7. One of them has the recommended TI firewire chip and a freshly installed OS. The behavior is not significantly different between the two. Normally I record at 96kHz and I just tested this configuration - if left alone to record 8 channels, it does eventually glitch after a few hours, even with a buffer of 512 samples. That's on a computer that has
Deleteabsolutely nothing running on, no user interaction - just one instance of Reaper and a most basic recording setup with no plugins except the sine wave generator and a Voxengo SPAN (which I'm using to register spectral disturbances that show up when there is a glitch). This is the main thing that's bothering me. It should be rock solid with this setup but it isn't. I agree with you that in theory, the precise clock frequency should not matter. However I read some parts of the DICE user guide and my main takeaway from that is that the whole clocking thing is so complex there that I don't really understand it even conceptually. It's basically syncing to either internal or external clock, and then in some way it also has to sync to the firewire clock and I really don't understand how it manages both at the same time. Currently I'm thinking in the direction of a problem with the firewire chip or its clock. I don't mind investing a few bucks if that increases the chances of solving the problem so probably I'm going to make a list of anything that could be wrong and then just simply replace all of that at once. If that doesn't help then at least I've gained some experience :)
Re. the breakout cable, I don't think that there is anything wrong with it as it's been hardly used at all and spent most of its life inside a drawer. Also when it refuses to sync, fiddling with the cables doesn't help. What helps is fiddling with the settings in the software control panel of the device. So to me it looks like there might be some issue with the internal PLL or clock or something related. Or indeed with the driver. I could sync to ADAT or to SPDIF. I've been testing wordclock simply because it seems like the must some and robust way to sync with the least complexities and the most reliable connection. I'll test other methods in the following days, for the sake of completeness.
DeleteSorry, my reply registered outside the thread.
DeleteSo I finally opened the thing up. Turns out that the sysclk (generated by TSB41AB2) is 24.1614 MHz, which is a 180 ppm deviation from the nominal value of 49.152 MHz. The spec puts a limit of 100 ppm so it's x1.8 times the maximum defined by the spec. The reference clock crystal for the PHY measures 24.5779 MHz so it is within the 100 ppm spec from the nominal value of 24.576 MHz. So I think this points in the direction of the a problem with the PHY. Looking at the TI datasheet to figure out whether the frequency deviation could be attributed to some passive component. Unfortunately, I don't have another TSB41AB2 handy to just swap the thing.
DeleteEasy with the terminology there, you're getting some numbers and names mixed up :)
DeleteAs i read it, the SysClk should be 49.152MHz, which just happens to be double the frequency of the 24.576MHz crystal used with the TSB. x1.8 is nowhere near "180ppm" difference.
BUT... If SysClk would indeed be a bit under half what it's supposed to be, i'm surprised the whole thing works at all (connects to the computer, gets detected by the driver etc).
What are you taking these frequency measurements with, by the way?
Oops, I wrote the frequency of the sysclk wrong - the measured value is 49.1614 MHz.
DeleteI also made a small mistake in the calculation. The actual ppm calculation as I get it goes like this: (1 - 49.152/49.1614) * 1,000,000 = ~191 ppm
DeleteI have a Siglent SDS1202X-E scope, just using its frequency counter. It seems to be correct as for other clocks it gives values that are very close to the spec.
As a first and easiest step, i'd maybe go with swapping out the TSB's clock crystal. If i were feeling really fancy, i might even scavenge a TCXO off of one of the SB Live boards in my "parts archive".
DeleteWhat would be interesting to test is if the unit itself, not connected to the computer, still exhibited the glitches you've encountered. You mentioned you have an RME interface as well; use that as the master, and if / since the wordclock was unreliable, interconnect the two via ADAT (or S/PDIF), and do another long-term test-recording. In case that turns out fine, i'd say that rules out hardware issues inside the 2626, leaving mostly the Firewire link and/or the drivers.
Swapped the crystal (got another one from a PC motherboard that has a firewire chip). Clock frequency changed from 24.5779 MHz to 24.5774 MHz (IMO the very small change suggests that maybe the coupling capacitors are the main contributor to the deviation here). The sysclk changed from 49.1614 MHz to 49.16 MHz so still 163 ppm deviation from the expected value of 49.152 MHz and not in spec.
DeleteYeah, I thought about testing it while disconnected from a computer, so maybe it's time to do that.
DeleteThis wouldn't rule out a clock/PLL issue with the PHY inside the 2626 though. At least that's my understanding.
DeleteI suppose, but that's still just sitting and guessing. And anyway, a TSB's like $5 or so, even from a reputable source (think Farnell, Mouser, Digikey etc). Add together some other tidbits (or even a spare TSB, and/or replacement load capacitors for the crystal, and/or a brand new crystal) to pass the free-shipping limit, and call it a day :) If you're really that bothered by tolerances a smidge more loose than what "the book" says.
DeleteAnd let's not forget about the PLL and all the other circuitry inside the DICE chip...
I'll swap the TSB if I have no choice. To buy it in my country, I'll have to pay like $35 just for the shipping and wait about a week. If that's the DICE then well, no luck. In the meanwhile and before further testing, I'm trying to tune the frequency of sysclk by changing the value of coupling caps in the oscillator circuit. We don't really know how tight the practical tolerances are in this sort of communication, all we have is the official spec. The TI datasheet for TSB41AB2 explicitly warns against exceeding the tolerances. So I'm not dismissing this quite yet.
Deletehttp://ww1.microchip.com/downloads/en/AppNotes/00826a.pdf
DeleteNot sure what you meant to say by that link, or is it just for information about how oscillators work?
DeleteAnyway, I swapped the original 12pF loading caps in the oscillator circuit with 20pF caps - that brought the oscillator frequency down to 24.5761 MHz (so almost spot-on) and the sysclk frequency to 49.1547 MHz (55 ppm deviation from the spec). Looks like this sort of stretched the limits of the oscillator circuit, so when trying to measure it with my probe, it won't oscillate - probably because of the additional capacitance of the probe itself. However if I lift the ground clip or use the probe in 10x mode, I can measure the oscillations and that's good enough for me. Now to test...
If you "use the probe in 10x mode"? Were you using it in 1x all this time???
Deletehttps://www.youtube.com/watch?v=OiAmER1OJh4
Hehe, I did... I'm a newb in scopes. This is my first one, still learning to use it.
DeleteWell, if that's the case, that may well have skewed your initial round of measurements. So there's a good possibility my conclusion could've been right after all - the hardware's fine, the issue is most likely in software (drivers) or at the interface between that and the hardware (somewhere along the Firewire chain).
DeleteThanks for correcting me. Actually I was surprised to see that when I was switching it to x10, I was getting an amplitude in the same order of magnitude as with x1. Dave's video explains that. Despite the facepalm, looks like the signal amplitude and the probe bandwidth still allowed me to measure these frequencies correctly.
DeleteI think the wrong probing mode could have interfered with the oscillator measurement but I doubt that it could have interfered with the sysclk measurement. Will measure again with the original components in place, just to make sure.
DeleteThought of an easier way to test that - just use the probe in both modes on the sysclk as is right now. If the probing technique would have an impact on the measured frequency before, then it should have an impact now. In fact, I've just done that and nope, no effect - 49.1546 MHz regardless of the technique. The x10 does reveal a bit more detail in the waveform though.
DeleteInterestingly, looks like the PHY oscillator tuning also had an effect on the wordclock frequency. Now it's 96.005 kHz, down from 96.02. So now it's 52 ppm from the nominal, down from 208 ppm. The improvement in wordclock precision is about the same as the improvement in the sysclock - wonder if that's a coincidence.
DeleteBTW the wordclock frequency measures the same, regardless of the probing mode :)
DeleteSorry about leaving this as a cliff-hanger. Had a few very busy days so didn't have time to do further testing. Right now it looks like the 192kHz stability improved dramatically (no immediate evidence of dropouts at 512 samples buffer). Running a long-term recording to see how it is really behaving.
DeleteDon't worry, i wasn't holding my breath, and the new day-job's keeping me pretty busy, but... Good for you, i guess?
DeleteSo it is completely stable now at 96kHz, 1024 samples buffer but with a lower buffer I'm still getting glitches. I guess it's time for TSB replacement. BTW I would also start off blaming the drivers if I'd been talking to a half-anonymous guy on the internets. In this case though it's not the drivers. I mean, if it was the drivers then there shouldn't be an improvement after the TSB oscillator tuning.
DeleteBTW if you're curious about my methodology of clock measurements, I can send you some pictures I took when doing the comparison between the RME and the M-audio clocks. Anton.
ReplyDeleteSo this is the sort of artifacts I'm getting when recording at 96kHz, 1024 samples buffer. I counted at least 7 occurences over 9 hours of recording. Looks like it just skips a few samples even without any actual dropout.
ReplyDeletehttps://ibb.co/GtsF515
Hey dude you didn't give me info on how to fix the 2626 phantom power
ReplyDeleteAre you sure?
Deletehttps://khronscave.blogspot.com/2019/04/39-m-audio-profire-2626-teardown-and.html?showComment=1570293879079#c8866982560137773532
Hi. Recently my Profire 2626 went off with a popping sound. Now a computer detects it and even sound bars works on the mixer but no sound comes from the outputs and analog inputs are also dead. Do You have any idea what can be wrong? Where i should start searching for a problem?
ReplyDeleteSounds like something related to the analog supply might've been the source of the "pop". As you can see in the teardown here, the supply rails for the analog stages are handled by the flyback converter there, so that's where i'd start looking.
DeleteThere's a chance one or more of the electrolytics might've given up the ghost, but until you pop the cover off and start with at least a visual inspection, it's just guesswork.
Thank You for the response. I've already done that and there is nothing suspicious. Nothing is blown or have any sign of physical damage. I'll take a look on the converter.
DeleteBy the way, did the popping sound come from the unit, or the speakers? Or perhaps both...
DeleteEither way, sounds like you'll need a voltmeter / multimeter to do anything more than just speculate, at this point.
I heard it coming from the speakers. I was too far away from the unit to hear if something made sound out of it.
DeleteWell, either way, without some measurements, there's no real going forward.
DeleteHello Khron! I have just had the dreaded firewire port mishap. When I am looking to get a replacement chip (TSB41AB2) there are different options, i.e. PAPR,PAP,PAPRG4... Not sure as to which replacement I should be purchasing? Any assistance in this matter would be greatly appreciated. Thanks in advance.
ReplyDeleteIf you check the datasheet, towards the end, in the "Packaging information" section (on page 48), you'll notice that the ends of those suffixes only have to do with the bulk packaging. In single-digit quantities, that's irrelevant, so for all intents and purposes, there's only one variant of the chip.
DeleteHey Khron, My Number 2 2626 is not giving me any amplitude. Audio sounds fine but its yeilds a very small amplitude on my protools waveform. I had a spare and replace the # 2 position I have 3 linked and all is good again with the new one. I was thinking about diggin into it with my Oscope and wave generator to fin out where the Gain stage goes funky. now can I get a copy of the service manual. it got issues with all of the Odd Channles 1,3,5,7 weird I know. ANy ideas would be awesome. thanks
ReplyDeleteOops...How can I get a copy...
DeleteThat's an interesting set of symptoms. If that were across all analog inputs, one might be inclined to suspect the analog power rails.
DeleteUnfortunately, i highly doubt any sort of service manual is available from anywhere (especially straight from M-Audio / Avid), but on the other hand, if you look through the datasheet of the components used (for example the THAT1510 preamp chips), i seriously doubt they would've strayed too far from the "recommended" circuitry. All dual opamps have the same pinout nowadays, which should ease troubleshooting, but your plan of signal-gen and 'scope tracing is probably the best (or at least, clearest / most certain) way to go, indeed.
Pls I have a profire 2626 but the firewire controller is fried...Pls how do I order the controller online...or beta still how do I get it work. ...I'm in Nigeria
ReplyDeleteIt seems the closest to Nigeria that the big reputable component distributors (Farnell, Digikey, Mouser) can get is South Africa, so you're left with taking your chances on eBay or other such online places.
DeleteRegarding the "how do i get it work" part, you'll need a hot-air station at least for removing the old chip. I've read about using a sharp blade (box-cutter or scalpel) to cut the pins off, remove the body and then desolder the pins, but that runs the risk of cutting through circuit traces under the pins.
I've reverse engineered the input stage of channel 1 (including the INST input). Here is the link:
ReplyDeletehttps://drive.google.com/file/d/1dGPiuR8Nxl3WggaFIEf31sXjP5qdBD2v/view?usp=sharing
If anybody has a substantiated idea for an upgrade of this stage, I would love to hear. So far it all looks pretty good to me on paper. May eventually get to taking some noise and distortion measurements. One questionable thing is the ceramic coupling cap on the INST input.
I've been discussing this circuit with a friend who is an electronic engineer and she pointed out that the TVS diodes configuration is not following the recommended practice. The drawbacks of using TVS diodes are half-assed protection, increased distortion and potentially added noise. The THAT 1510 datasheet offers a different configuration (with general rectifier diodes) that basically sinks excess current to the power rails. A side effect of that is that values of R110 and R111 (resistors in the signal path) can be reduced to 10Ohms, thus reducing source impedance. So I'm considering a mod to implement this. If anybody has any thoughts about this, I would be interested to hear.
DeleteAn additional mod I'm considering would be implementing an input DC servo, as described in the "48 Volt Phantom Menace" document.
Really nice work with the reverse-engineering there.
DeleteWhat would you hope to achieve with whatever "upgrade" you'd make?
Re: noise measurements, it's pretty standard to terminate the mic input with a 150 ohm resistor, if memory serves (between pins 2 and 3).
I'm half-kidding here, but wouldn't "increased distortion" count as "mojo", especially since we're talking about preamps here? :P
And regarding the 48V thing... How detrimental to anything would adding, say, a couple resistors to ground, right on the XLR pins 2/3? Something like 100k or so, just to discharge the DC-blocking capacitors.
Maybe i'm weird, but i kinda prefer the simpler solutions. What would the DC servo do, when you do need to turn on the phantom power for a mic? Or where would you even connect it, anyway?
That convention paper kinda got me skeptical from the first sentence in the abstract, with "anecdotal evidence". If it were a widespread issue, i would imagine it would've been adressed more commonly..?
Hopefully I don't get mixed up with the audiophile crowd here. I don't consider distortion on a general-purpose mic preamp a good thing. Implementing a DC servo is at this point just an idea. I didn't go into all details there and I don't have anything to contribute here before I do. I just saw the implementation shown on that paper that you didn't like and it looked like a fairly easy upgrade, the same way as displayed in their example. As to the paper itself, I might be naive but I got an impression of a serious work there. The anecdotal evidence was just a starting point for their research. They offer experimental evidence and measurements, all of which sounded plausible to me.
DeleteAs to discharging resistors:based on the THAT datasheet recommendation and on the situation described in the whitepaper, they won't help here. Because the issue to begin with is a too-rapid discharge of these same capacitors anyway.
I don't know how widespread (or not) the issue is. Considering that even m-audio implemented a (cheaper version of) protection mechanism in their product, it looks like it might be, in fact, a recognized problem. Could be interesting to investigate other manufacturer's recommendations.
DeleteLooking at the recommended application in the INA163 datasheet, you find an exact same basic configuration of 4 diodes sinking excessive current to the power rails. So if 2 different major manufacturers provide the exact same recommendation, I think it is reasonable to assume this is, in fact, a recognized problem, and there is a generally recommended way to implement a solution. Which is not at all like the Profire implementation.
Delete
ReplyDeletefriend i speak of brazil, and i needed help with my fast track ultra 8r i plugged in a very old mac book pro that a friend took in an exchange and he had a problem, he was shocking the gnd and didn't know but the usb was giving shock too and when I took it to the church and went to record it was all shocking and I saw that someone had removed the grounding from the table and when I put it in place it gave an explosion and the plate does not turn on but the source is good but the interface looks short. if you can help me i will be very grateful ... a hug !!!
Hi there! Do you happen to know the values of the pot for the master controll? Mine is working baddly in lower volume settings (tough it works normally after '11 o clock'). Do you think this would be an easy repair/easy to find part?
ReplyDeleteThere's a good chance it's nothing more "fancy" than a mono 10k or something. Feel free to pop the cover off and take a peek yourself (my two units are in my recording rack).
DeleteIn the photos, i can see the traces going straight into that 8-pin ADC, so there's no analog audio going through that pot.
There is a pretty common problem with these devices- issues where the headphone outs aren't playing back full stereo signal (and no, its not an internal routing issue)
ReplyDeletein my personal experience, the left or right headphone channel will cut or dim if the channel 1& 2 main outs on the back are in use, or the 2nd headphones if 3 & 4 are in use, but I am currently using 7&8 out on the back (all stereo out pairs being fed with the mix out, only 5&6 and 7&8 being controlled by the master fader knob) and the left channel on my headphone 1 out sounds quiet and distorted (like theres dirt and/or disconnection in the signal path). i know that often times playing signal through these problematic channels will, given time, pop back up to full volume, but that isnt helping this time. I did just switch over to a new computer, but everything else is exactly the same as it was before, and before everything functioned perfectly (other than the channel outs being used at the same time as the headphones).
My gut tells me that there is some issue with power delivery or something with firewire, but my gut is neither an electrical engineer nor a computer scientist, and no power cycling or firewire switcherooing has changed the issue in its current form.
any advice or tips or ideas on where this problem may be coming from would be greatly appreciated
Hello there! I have a stupid question, but it would help me a lot! I have a Profire 610, a smaller version of this one. The potenciometer of the input 2 is making a terrible noise, even at minimum. I am working at a plataform that receives all the inputs from the interface, so even if I am using input 1, I need input 2 to be silent. How do I kill input 2? What should I take from the board to silence input 2 without damaging the interface? I know nothing about eletronics, so I apologize again for the question. Thanks in advance!
ReplyDeleteI assume you've already tested that switching the pad on or off has no effect on the noise?
DeleteBut if you just want to silence input 2 "permanently", check the THAT1510 datasheet and follow the trace going from the output of the one on input 2, and remove the first DC-blocking cap that's in series with that. There's bound to be at least one between that, and the input of the associated ADC channel.
Thank you very much!
DeleteNo problem, good luck with the operation!
DeleteI am sorry for bothering you but you seem to have expertise and knowlede on the subject. I have a 2nd hand M-audio FTU 8R that no longer works. If it is powered on, it doesn't show all the green leds, and then all the red ones (some self test fails I assume). Sometimes all the leds are lit. Sometimes even only the green led for a few inputs light up.
ReplyDeleteThe unit is not recognised over USB. No visibly blown components on the board. Ever had of heard of this issue before?
I wonder if you perhaps have any suggestions what the problem could be?
Thanks in advance for any response.
I can't say i've heard of, or encountered anything like that. Might be worth checking that the supply rails are all what they're supposed to be, for starters. If the analog rails are wonky, the level-detectors might see that as "false" signal levels.
DeleteThe "not recognised over USB" part is a bit more concerning, though. Could be a sign of corrupted firmware, but as mentioned above, checking the power would be the first step. No (correct) power -> no operation.
Thanks for your reply! I've checked the power rails which seem to be supplying the right voltage. This particular unit has a Dream SAM3308B processor which appears to have a serial firmware debugging option. It shall probably be some proprietary binary protocol but I'll try to see if that shows any sign of life. I will also try to figure out what chip contains the firmware and see if I can find a way to verify the data on there. (This will probably prove to be tricky, and I suspect that it will probably be a write-once/read-many chip.) I do have a unit that /does/ work so I do have some reference material.
DeleteYou seem to not be impressed by the brand of capacitors that are used? This one also uses Suncap, which I never encountered before. Would it make sense to de-solder the ones near the power distribution to verify these are still within tolerances?
SAM3308... Isn't that the same processor that's in the M-Audio C600? After a quick check of that teardown's photos, yes it is. At least in that unit, it had an external flash memory chip containing the firmware (or at least the bulk of it, i'd imagine).
DeleteI'm very much unimpressed, indeed. The cap brands i trust are a handful of Japanese ones (Panasonic, Elna, Nichicon, Nippon / United Chemi-Con and Rubycon... plus Sanyo, whatever "NOS" ones might still be found, or scavenged).
Neverminding the power filtering uses (where they're bound to fail, sooner rather than later), i've "even" seen CapXon caps fail in signal coupling roles (on the outputs of a dbx digital crossover).
I did discover a chip that may store the firmware in 256Kbit (32Kbyte) of space. It is marked ATMLH122, can't find any datasheet for it, but it appears the same as an AT24C256. I haven't been successful yet to read the contents from it in situ: my EEPROM reader returns all 0xff. So either (a) my reader doesn't read it properly, (b) it's broken, or (c) I can't read it properly because it is in circuit. So, I am afraid that I need to de-solder that chip to know for sure. And even if I do get what seems real data from it, I'm a bit hesitant to do the same to my still working unit to actually compare. :)
DeleteI understand that the SAM3308B also has on-board firmware which also could have gotten corrupted somehow.
I tried connecting to the serial test pins but I don't speak the proprietary protocol that it uses, so no interesting developments there.
The previous owner mentioned the unit worked fine before it was put into storage 5 years ago. So it may have gotten a bit of heat, hot or cold but I can't imagine that would or could cause firmware chip to get corrupted.. The way it behaves I'm tempted to say it got an EMP from solar flares :-D
The quest continues. Sadly it has proven impossible to find a service manual or even schematics for the unit. I see many markings on the PCB like TP1 et cetera which I suppose are test points.
Not too sure trying to use a programmer in situ is a great idea anyway. Depending on how the memory chip is powered, the programmer might need to supply more current than it can (since everything else powered from the same voltage rail as the memory chip, will start working to some extent, and thus need "juice").
ReplyDeleteWell, whether you end up being brave enough or not to dig into your working unit, the first step you'll want to do is read out the contents of the known-good-unit's chip, so you have a known-good dump of that. I've taken that route with most of the interfaces i've torn down here, whenever practical (ie. "only" a SOIC-8 flash chip).
Hello!
ReplyDeleteAfter a storm here where i live my profire starts to make a strange white noise when using condenser mics. Its usable but it used to be so clear. What could possible be?
Thanks
That's not really a whole lot to go on, if i'm honest. Is it only condenser mics? And the same with all the condenser mics you have?
DeleteHello Khron, congratulations on your blog, very good information. I have a Profire 2626 with the problem that it is not recognized by the PC. I replaced the TSB41AB2 IC and the failure continues, power supply voltages ok, I have oscillation in the xtal, I suspect firmware failure, will you have it available? Thanks in advance
ReplyDeleteI guess that's a possibility, assuming everything else is ok. I have a 2626 that's waiting for a new power supply controller chip, but i could read out the firmware in the flash memory in the mean time. Do you have (or can get access to) a programmer that supports the flash memory chip, along with the TSOP48 adapter?
DeleteThanks for the reply, I have a TL866II programmer and a TSOP48 adapter. I think it could be the firmware because looking at the signals in the memory it seems like they have cycles of 3 or 4 seconds and then they reboot, as if a watchdog timer was activated. If you could read the firmware I would appreciate it
DeleteHi Khron, it seems I have the same problem. I don't want to be misplaced, but did you get a chance to read the firmware? Thanks
DeleteIt took a while, but i got my hands on another of these, and i did dump the known-good firmware. It can be found over here:
Deletehttps://drive.google.com/file/d/1vihy4Nr66TUDhSMtJc3qaLaN9vRQZXim/view?usp=sharing
Thank you very much Khron for the file, I just saw it, I will reflash the firmware and tell you the results
DeleteNo problem, and good luck! I myself am waiting for a new TCD2220 from AliExpress, to try to revive another 2626 (where replacing the flash and the RAM didn't help).
DeleteThanks to your help the interface came back to life !! After programming the flash memory now the computer detects the interface and works perfectly. If you have a similar case, check the signals in the flash memory, if they seem to restart every 2 or 3 seconds it is surely a firmware corruption and it restarts due to the watchdog timer. I hope I can return the favor one day
DeleteHey, that's great news! I'm glad to hear that, and nice to hear i was able to help.
DeleteThe one i'm having trouble with doesn't really do anything at all. The Firewire / "main" / RAM clocks are active, but nothing else is (I2S clocks are quiet etc).
Unfortunately I don't have the schematic diagram, according to the TCD2220 datasheet it requires two voltages 1.8V for the core and 3.3V for I / O, I suppose both are correct, and the xtal also works correctly. If you have replaced the RAM and Flash chips is probably the TCD damaged, good luck
DeleteI didn't expect you to have it, but that's as far as i got myself. Right now i've got 3 TCD2220's on the way from AliExpress (yeah, i know, i know...), so i should see what happens "soon enough" - hopefully this month.
DeleteHere are a couple of documents that may be useful:
Deletehttps://docplayer.net/58442740-Dice-hardware-evm004-fw843-rev-1-0-users-guide.html
ftp://beta.zephyr.com/~tlscorp/pub/Omnia/Service/Omnia%209/1960-00290-102_Main%20IO%20Board_CN9000.pdf
Dear Khron,
ReplyDeletethank you very much for all this extremly useful info about the profire 2626 unit. I would like to ask you about my unit's problem. Almost each output has a different volume. I meassured it (with exactly the same signal in) and difference is up to 10db. Please do you have an idea, what part/parts are responsible for output level so I can attepmt to fix that. Thank you very much, Jiri
"Exactly the same signal in" = the same signal coming out of your playback software? And what did you use to measure / monitor each output?
DeleteThis comment has been removed by the author.
DeleteHello,
Deletethanks for the reply. Yes, tested with the same signal out of software. And measured again by the same app. Basically, I just used external FX on that channel, so the signal go out / in. I tested all outs with the same IN. Then I watch peak level.
It might not be totally impossible for some of the coupling capacitors to be aging. Might be worth doing some frequency sweeps (with something like Room EQ Wizard) to see if there's any anomalies there, or just level differences.
DeleteHi Khron,
ReplyDeleteWhat a wonderful resource page to stumble upon.
I have three profire 2626’s and one of them will no longer connect to the computer—i suspect the firewire has died on it. This isn’t the end of the world, since I use two of the three via ADAT only in standalone mode. The broken unit seems to still pass audio via ADAT, despite the FireWire problems, but I don’t think it’s syncing its clock from the adat port. Since I can’t connect it to a computer to change its settings, is there any way you can think of physically forcing it to clock from the adat? Or otherwise changing its standalone settings? I wish it just had some switches externally, like many other similar units.
I don’t have the chops to desolder and resolder a replacement firewire chip, I’m afraid.
Hoping you have an idea,
Hal
To my knowledge, there's no way other than a computer, to change the clock / sync source for these units.
DeleteGotcha, thank you for the reply. I think to ebay this malfunctioning unit will go. It may be time to replace all three altogether, they’re all finicky under Mojave and if I’ll be running them all as adats eventually anyway with a Raydat, I don’t want to have to rely on their unreliable firewire connections in order to change settings. Bummer! Thanks again.
DeleteNo worries.
DeleteNot sure it's necessarily worth selling off all three - i have two in my recording rack (slaved via ADAT to a MOTU 8M). The mic inputs all on the back are nice for keeping things tidy, while still having the switchable DI's in the front, and the pads on all eight inputs are really welcome too. But yeah, there's the catch of the "set and forget" clock source...
Hello Khron,
ReplyDeleteNice write-up! And great to see that this thread is still alive and doing well. :)
Besides the 2626 I also have the Profire 610, which is giving me trouble. I once inserted a damaged XLR cable too forcefully into the preamp and since then the preamp has had a lot of noise. It doesn't matter if anything's connected to the port or not, there's always noise in a wavy-kind of way (sometimes less, sometimes more). Today I finally got to open the device and look inside, but can't seem to find anything visually wrong. I was hoping to find a bad solder or something.
I wonder if you (or someone else) could guide me to the right tracks in finding out where the fault lies. Where should I start measuring continuity?
Here are a couple of images. The faulty preamp is input nr 1, so in the first image it's in the left lower corner and in the second image in the left upper corner:
Image 1, view from top: https://www.dropbox.com/s/lbaeirkczalre6w/Foto%2015.4.2021%2017%2000%2048.jpg?dl=0
Image 2, view from bottom:
https://www.dropbox.com/s/yqg9vt558gnv9wq/Foto%2015.4.2021%2017%2009%2036.jpg?dl=0
Thank you!
- K.
How exactly do you mean, "forcefully"? And how exactly was it "damaged"?
DeleteI suppose there's a chance you might've bent something inside the connector itself? There are four switched contacts inside those combo jacks (tip/ring/sleeve for the 1/4" jack section, and the shell for the XLR), which may or may not now be more or less out of place.
Hi,
DeleteThank you for your reply.
I didn't notice that the cylindrical metal enclosure of an XLR cable's male connector that I inserted was slightly bent so it didn't fit the female connector of my ProFire. I tried to force it in and that's when the damage happened. The first preamp is practically unusable as a result. I have to mute it from the software mixer so it doesn't make noise.
I can't see any visible damage to the device so I think I'll just keep on using it without the first input. Or I might try to solder a new connector to the PCB.
That's unfortunate... Well, that just goes to prove yet again - violence is not the answer.
DeleteYou'll probably want a capable hot-air station / gun for that, and try to heat-shield the other nearby connectors. I'd also add some lead-based solder to the joints, just to help heat transfer and (hopefully) slightly lower the melting point of the pre-existing lead-free solder.
You could start off simply trying to re-touch all solder joints connecting the TRS/XLR combo jack to the board. If that doesn't help, you could put both preamps this device has into the same state (zero gain, de-select INST, de-select phantom power) and measure resistance from Sleeve to all other points where each jack connects to the board. Then compare what's different. I reverse-engineered the preamp section, you can view the schematic here:
Deletehttps://drive.google.com/file/d/1dGPiuR8Nxl3WggaFIEf31sXjP5qdBD2v/view
The schematic includes the pin-out of this connector (which I copied over from the datasheet). It displays the pinout as viewed from top, so you'll need to mirror flip it in order to have the pins correspond to what you see on the bottom of the board.
Of course, measure resistance only after you removed all power from the device (and preferably gave it ten minutes to discharge the capacitors).
Good morning everybody. Could some of your friends make the service manual for profire2626 available for download? If possible, could you send me by e-mail?
ReplyDeletejuniordosteclados@hotmail.com
Thank you very much in advance.
I'm afraid there is no such thing in existence, as far as i know. Anton kindly reverse-engineered the analog section of input 1, but that's about it.
DeleteIs there any particular problem you're faced with?
PS: It's usually a bad idea to post an email address so plainly, anywhere online. It's just inviting spam...
Hi, Khron! I'm having some weird problems with my Profire 2626. The interesting part is that it's limited to the first 2 channels. The 1st channel just clips and pops. Even in the lowest gain and with the padding it still sounds distorted and ear bleeding loud (Both instrument and XLR inputs). The 2nd channel just doesn't work with instruments, but works fine with a mic through XLR. Other channels working as expected. Already opened it and found no blown or stuffed caps. Any ideas?
ReplyDeleteRe. 1st channel, you didn't write anything about the balanced line input. Did you test it? If you didn't then I'd suggest testing it. The reason is that XLR and INST inputs go through the THAT1510 chip for amplification, while the line input bypasses it. So testing line-in could help isolate the issue.
DeleteYup, line inputs are working. So that means that the 1510's are dead? Channels 1, 2 and 5 are not working with XLR but fine with line inputs. Each 1510 has 2 channels, am I right? If that's the case, is replacing them worth the headache (I live in Brazil so I'll have to import the parts)? or just getting an external preamp and using through the line inputs (I already have an ADA8000 hooked through ADAT, I could get another one...)?
DeleteEach 1510 serves 1 channel. The repair might be not as simple as replacing the 1510's. What we can tell for sure is that something in the signal path where 1510 is involved is not working properly. I can not say for sure that the fault is withing 1510 itself, or that the fault is limited to the 1510. If you have 3 channels with faults, I would think that there might be a precursor issue, for instance with the supply (internal or external). IMO proper diagnostics is required in this case. Blindly replacing parts may or may not help. I'd suggest you to take the device to a local electronics repair shop that has a good reputation and preferably some experience with fixing audio equipment.
DeleteIn case you have experience fixing audio equipment and want to diagnose and fix it yourself, here is a link to the schematic of the input section which I reverse engineered some time ago:
Deletehttps://drive.google.com/file/d/1dGPiuR8Nxl3WggaFIEf31sXjP5qdBD2v/view
I'm not fully experienced with AUDIO equipment, but I have some experience with power inverters and valve amps. So I'll be trying to fix it. Thanks for the schematics! I'll keep you guys posted on my findings!
DeleteLovely weekend to you. My computer no longer detects my maudio profire 2626, I found out that the problem came from the audio interface and not from the cable or the computer. Can this problem be solved cheaply please?
ReplyDeleteSounds like there's a good chance the TSB41AB2 might need replacing. The chip itself is around $5-6 from official suppliers (Mouser, Digikey etc), but the shipping costs for just one chip will be a bit more than that.
DeleteAnd then there's the issue of finding someone with the equipment and skill to perform the operation (in case you can't / aren't willing to).
It's always the TSB. Always. :)
ReplyDeleteThis is 2022 and this thread is fresh. I'm glad I found this. Thanks guys
ReplyDeleteI like to say a very BIG THANK YOU to you Khron for taking the pain to share this profound truth. I bought a used Profire 2626 from ebay. The seller was kind enough to declare that the firewire port wasnt connecting to the laptop. After i read through this post, i was convinced and encourage to take the plunge by purchasing the device regardless. As soon as i got the delivery of the device, i didnt waste anytime, i order for the chip at aliexpress, and contracted someone to remove the spoilt one fix thi one. to my greatest joy, the Profire 2626 came up live. Connected to my PC. I am the happiest man NOW. I just want to say thanks for sharing this post. it has been a blessing to me. Thanks again
ReplyDeleteGlad to hear my "adventures" have helped others as well! I've still got two of these in my recording rack, and recently sold another one to my fiance's band, along with a laptop, for playing backing tracks at rehearsals and gigs. It's done at least two gigs and close to a dozen rehearsals, and i've yet to hear any complaints.
DeleteA really good blog. Thank you. I have a problem with Profire 2626. Phones 1 feels only left, phones 2 feels very low and distorted. I don't know how it happened because I realised it after months I was using the card but not the headphones. Advice on how to intervene? Do we need to open it? I'm not practical but do you think it's an internal contact problem? Thank a lot. Lenzo
ReplyDeleteYes, opening it up is quite unavoidable. Could be some capacitors, could be the connectors themselves...
DeleteHi Khron...thanks for your work! in my 2626 the analog section has stopped working, if I connect an external adat converter it works fine and communicates with the pc,I think can be the power (analog part)supply...which component you think can I check?
ReplyDeleteWell, the first commandment of troubleshooting is "Thou shalt measure voltages". Fortunately in this case, the various supply rails have conveniently clearly marked test points, so you could start there. Who knows, there might be even visibly damaged components that could offer a more definite indication of where the trouble lies...
DeleteHello Mr Khron. On the Internet, I only found your blog on the repair of this device. could you tell me, recently my interface started to turn on for a very long time. visual inspection I did nothing suspicious. I checked r033 at the power input, it works, but if you unsolder it, it turns on immediately, but then the analog inputs and outputs do not work. if you wait 20 minutes, it turns on and everything works like a clock. but if she stands for a minute, she turns on for a long time. Please tell me which way to look and what should be replaced first. Thank you.
ReplyDeleteSounds like it's time to replace the electrolytics in the power supply section. Just be warned - the holes for the capacitor legs are TIGHT, and getting the ground-connected ones cleaned out completely can be very frustrating and time-consuming...
DeletePerhaps I'm repeating myself here as it's been a while since my last enquiry...Has anyone overcome the out-of-date-drivers for this interface? Mine has been sitting idle this past 4 years as it seems no longer compatible with the latest OS on my MacBook Pro.
ReplyDeleteSwitch to PC? :) Jokes aside, worst case use it as extra I/O via ADAT, with a supported interface? But you'll still need to configure it to get its clocking from ADAT while stand-alone...
Delete