"I bought this online and it is not working or it is not compatible with my computer. My computer is not recognising this audio interface. So I am selling it as not working."Fine by me. The photo of the back showed the Firewire port as looking quite pristine, so it wasn't a case of connector-inserted-upside-down. And for the price i won the auction for, it was worth the gamble. A frontal photo did show it powered up, so at the very least, it wasn't dead.
Upon receiving and unpacking, the two preamp gain knobs felt a liiiiitle bit more wobbly than i'd like. In case they were knocked axially (which may well be the case, going by the scuffs on the tops of the knobs), it might just be that the "claws" of the metal mounting-frame of the pots got loosened off - nothing two seconds with needle-nose pliers can't fix. Master volume knob felts as firm as i'd expect, but the two headphone volume knobs were a bit loose. Well, all in due time...
On the back, the DC input jack might've seen better days - looked to me like it might've been twisted upwards a bit, and the pin in the middle looked a bit cocked-over as well.
The unit didn't come with a power supply (plug-pack, wall-wart etc.), so i plugged in a cable from my bench power supply, set to about 12V. Flicked the power switch, saw a 650mA draw (well within reason). "Power" LED came on after a second or so, then a few seconds later the "LKD" LED (clock stabilized and locked), and the two "Inst" LEDs for the front inputs. The "Dim" and "Mute" momentary buttons toggle the associated LEDs on and off, so the logic part of the circuitry should be fine indeed.
Quick input test (tapping on a dynamic mic) - input level meters react just fine, and so do the gain adjustments. Tapping on the tip of a jack plugged into line inputs 3/4 on the back just occasionaly blips the lowest LED (-48dBFS) on the level indicator. So far, so good. Time to pop the top off, then - shall we?
Aaaaand the insides greet us with... no surprises, really. On closer inspection, my initial suspicion about the input gain knobs turned out to be correct - the channel two potentiometer looks visibly "distended". The channel one is held together a bit more closely, but i can see at least one of the "claws" of the casing, on the back of the potentiometer, sitting at a 45deg angle (as opposed to flush against the plastic body).
Right then, starting with an overhead view. Power supply section on the right edge, digital side just right of the middle, and analog stuff on what's left (no pun intended).
"The brains of the operation" is a TC Applied Technologies TCD2210, aka Dice Mini, accompanied by a Samsung K4S281632K 128Mbit (16MByte) 133MHz SDRAM module. Tricky to get a picture of these two, what with being obscured by the DSP daughterboard.
A Spansion (now Cypress) S29AL016J70 16Mbit flash chip contains the firmware. The Firewire comms go through a Texas Instruments TSB41AB1.
On a slight side-note, right on the first page of the datasheet it mentions "Dual ADAT receiver/transmitter with S-MUX support". Focusrite could've easily added an ADAT output as well, but that may well have (potentially) eaten into their Saffire Pro 40 sales. But i'm only speculating...
Speaking of DSP, though - the "party piece" of this unit, offering EQ and compression for inputs 1-2, and reverb for any and all inputs - we have a Texas Instruments TMS320C6722 chip, with a second Samsung K4S281632K (mainly for the reverb, i would guess).
A Texas Instruments PCM3168A 6-in / 8-out CODEC handles all the conversions (A-D and D-A).
Most of the opamps are JRC NJM4565 apart from three exceptions - two ST TL072C handling the high-impedance instrument inputs, the two JRC NJM2122 doing the "heavy lifting" of gaining up the first two inputs, and the pair of JRC NJM4556 driving the headphone outputs.
The several NXP (now Nexperia) HEF4052 and HEF4053 visible handle the switching between the various input options (mic / line / instrument) for the first two channels, as well as switching the (line) inputs 3/4 between low and high gain.
Moving on to the power supply section. Of the two National Semiconductor LM2672 on the board, one is responsible for the 3.3V rail, while the other one does the +6.9V and phantom power; the latter being regulated to 42V by a National Semiconductor LM317LZ. A much beefier LM2670 produces the -6.9V rail, while a tiny SOT-23 packaged Microchip MIC5205 (marked "KBAA") outputs 5V.
Most of the electrolytics on the board are Jamicon SK, while the smallest ones are Jamicon SS. The rest are two 22u/35V unbranded ones, and a handful of Samxon GF 220u/50V.
Now then, with the teardown complete, time to move on to the repair stage. Three nyloc nuts hold the PCB to the front panel itself, but the knobs need to be removed in order to free up the PCB for removal.
Not a whole lot going on, on the front board - the potentiometer, the LEDs, and a couple of shift registers to drive said LEDs.
The headphone volumes and the channel two gain knob firmed right up. The channel one needed to be removed, to be able to crimp down the sides of the metal mounting frame while at the same time tightening up the "ears" on the back.
Done and done. Now let's see if it does anything when hooked up to a computer. And the verdict is...
Nope. It would appear that the Firewire connection may well be busted indeed. A bit of diode-test probing (red on ground, black on test points) shows three of the four data lines as having about 520mV drop from ground, while the fourth shows a mere 120mV. That'll be it. Now, where did i put those two Focusrite LiquidMix 16 units i had?..
There, found'em quick enough. See? I'm not a hoarder, i'm just being resourceful! Sure beats waiting a week or so for the delivery of a new TSB41AB1... Anyway, swapping that means the DSP daughterboard needs to come off. Just as well, that means a better look at the digital works under there.
Post-swap, all four data lines show a much more normal 527-530mV drop from ground. Now let's plug this lil' feller in again. And there we go! "FW" LED lit up, Windows asked for drivers, and Saffire MixControl showed the mixer, along with the controls for the DSP input effects.
I'm hereby calling this a success. Now, to find a buyer for it...
Hello!
ReplyDeleteI wanna thank you for posting this. I recently bought a Saffire Pro 24 ( Non DSP ) for 35$ and it wont properly power on. I had a feeling it had issues but the buyer said it worked fine. So i searched the web hoping to find anyone repairing this type of interface to help boost my confidence and it led me to you! The interface shows 48V when pressed but no other LED light turn on/no connection to the computer.
Would you have any thoughts on why this could be?
Peace & Thanks :)
Well, the starting point would be to make sure all the power rails are on, and at (or near enough to) the values they're supposed to be.
DeleteJust because the LED for the phantom power indicator comes on, is no guarantee the phantom power itself works. Fortunately that's easily-enough determined by using a multimeter to probe between pins 1 and 2 (or 1 and 3) of the XLR connector.
That being said, you'll want to probe all the clearly-marked test-points on the mainboard. The board is identical to the DSP version, except for the DSP daughterboard, the firmware and the second headphone output, so you should be able to use the photos here as a guide.
The Dim and Mute buttons are "read" by the main processor and the associated LEDs are driven from it, so something may be unwell in that direction.
But yeah, first step, check the voltages (+/-6.9V - the "VA" LED should be lit; +3.3V; 1.8V).
If need be, i could snap some other internal photos, i still have two of these units.
Thanks for the quick reply!
DeleteI'm gonna take it over to my friend's shop during the week and perform what you suggested. ( Don't have the tools at my place at the moment ) Your post is a great guide and the Pics are really helpful.
I'll keep you posted on the process, I'm hoping to save it's life without going over the value cost.
Also would you be interested in seeing the pics of inside my Pro 24?
From what i can see there doesn't seem to be any blown caps but I wonder if the previous owner turned it ON/OFF improperly too many times causing the Fire-Wire connection to go out.
Again much appreciated, peace!
I'm not sure what i could determine from photos, really.
DeleteThe voltage measurements should be a (hopefully) enlightening first-step.
Mmm... Even if the Firewire was toast, that shouldn't prevent the unit from booting up and reacting to the Mute / Dim buttons, for example. Or locking to its internal clock (by default). That being said, if need be, IF all the power rails are ok, it might be interesting to probe the two terminals of the clock crystal (or the ceramic capacitors connected to them, if they're easier to access), to make sure the oscillator in the Firewire chip is indeed working.
Hi Khron
ReplyDeleteI'm about to head off down the same road as you with one of these. Do you have any tips on technique? The chip is in quite a tight corner.
Thanks, Morgan
Well, as pointed out near the end of the article, you'll wanna desolder the DSP daughterboard - otherwise it's damn near impossible. Other than that, you might wanna "pad" the surrounding area with some aluminium foil (i have a roll of adhesive-backed aluminium tape, which is handy for such operations), to try to keep the heat concentrated on the chip.
DeleteI like a nice chunky tip on my soldering iron, that way i can melt a big blob of solder on those pin header ends and melt each "block" at once. And i even use the same "knife"-style tip to solder the Firewire chip pins. You could probably use hot air for that too, but i kinda prefer the hands-on approach.
Hola tengo problemas con mi focusrite saffire pro 24 enciende todo bien pero el pc no la reconoce la compre con esa falla y aún no la puedo hacer que funcione.alguna ayuda de que podrá ser la falla
ReplyDeleteDid anything on this page give you the impression that i speak spanish? Hint: i don't. But fortunately, we have things like Google Translate nowadays (there's another hint). Also, you could try working on your punctuation, so your sentences can be a bit more understandable (in any language).
DeleteIf it powers on but doesn't connect to the computer, the prime suspect is the TSB41AB1 Firewire chip.
Hello Khron,
Deletecould you kindly advise where shall we order the TSB41AB1 Firewire chip from?
Bought some from china, but it seems does not work.
Thanks so much.
No big surprise there. Do you have access to ordering from one of the bigger official retailers, like Farnell / Element14, Mouser or Digikey? Getting fake components from them is far less likely than from the far east.
DeleteThanks so much. Will try to find way.
DeleteI just stumbled on this post because my Saffire Pro 14 seems to have the same issue. I'm pretty much a beginner working with electronics, but this gives me hope to at least test the chip and confirm the issue! Thanks for the detailed writeup!
ReplyDeleteNo problems, and i'm glad to hear it's "motivational" for others. Even though, at some level, this is a slightly "selfish" endeavour, fueled mainly by curiosity.
DeleteBest of luck with the repair though, i hope it won't pose too much trouble!
Dear Khron!
ReplyDeleteI'am glad to find your post. I have a Saffire PRO 26 with an issue like yours, but with a little more.Sadly it doesn't turn on at all. The On/Off switch connector on the board is kinda burnt. 'VA' led doesn't turn on. Currently i am trying to measure all the voltage rails(i am not fully done with it), but now it seems all of them are working. At first i thought its just a blown fuse or something, but i didnt find any fuse on the board. Could you give me some direction about what to do?
Thanks in advance!
Update:
DeleteI found a led on the board with a with a label(3V3),and it doesn't turn on... so i assume my 3.3V volage rail is dead. On this board there are only 3 fix value voltage regulators(7912,7812,317T), but the other regulators are variables, and i am not too confortable with them, because its legs are too close to eachother, and i am worrying about shorting them.
3.3V voltage rail fixed, now the 3V3 led turns on.
DeleteBy "26", i'm guessing you mean the "Saffire Pro 26" that's from this range, not the older, silver "26 I/O", right?
DeleteEither way, nice to hear you got the 3.3V rail back to life. For what it's worth, the "VA" LED and its associated resistor are connected between the +12V and -12V points. So between ground and either "end", you should measure about 12V.
The LM317T will be regulating the phantom-power voltage. What are the 7812 / 7912 supplied by? The Pro 26 seems to only have a single 12V DC input, so i would guess there needs to be some sort of boost converters for the positive rail (12V input -> at least 15V ahead of the 7812, and at least 52V ahead of the LM317), and an inverting boost ahead of the 7915 (12V input -> at least -15V).
An internal photo of the power supply area would help narrow things down. I've never laid hands on a Saffire 26 (albeit i owned an older 26 I/O for a brief period, many years ago).
I made a couple of photos.
ReplyDeletehttps://ibb.co/5LPm6TV
https://ibb.co/tLzsZWW
https://ibb.co/16LbVL4
https://ibb.co/cykXBF6
https://ibb.co/zsbrtbK
https://ibb.co/DCwvq6F
I can't see any signs of failure on the board, other than the burnt connector.
Tomorrow i can measure those voltages what you are talking about.
I hope these pictures gonna help for now.
Y'know, that looks strikingly much more like the older "Saffire Pro 26 I/O", rather than the newer, black "Saffire Pro 26". Which is exactly why i wrote what i did, in the previous message here.
DeleteBrowse through these photos to see the differences.
Those three LM2670's will be doing what i described in the previous post. You can find the datasheet right over here.
You'll want to check the output side of L11 and L12, and see if and what voltages you can measure there. I'm gonna guess IC62 might(?) be in charge of the +/- analog rails, in fact that's quite likely, given the transformer next to it (as opposed to a simple inductor).
You'll probably want to replace that burnt connector though. My guess is, it going high-resistance and the power circuitry drawing more and more current through it, lead to the charring.
Hi! I've had a saffire pro 14 that's been dead for almost two years now due to a wrongfully inserted Fw connector: it power ups, locks but doesn't light the FW light, so i guess it should be the same TSB41AB1 from this repair.
ReplyDeleteAt the beggining of the post you mentioned that "It wasn't a case of connector inserted upside down" so i'm wondering if that would make any difference?
Also, while looking through Mouser i find 3 different TSB41AB1, is there an specific one to buy? Thank you very much!
Well, by that i "only" meant that, in other cases, the outer shield of the Firewire ports was clearly "widened", but it didn't seem to be the case here.
DeleteRegarding the chips from Mouser, if you'll look in the datasheet, towards the end there's a page with "Packaging information". There are three choices there too, but one is for a 64-pin package (not suitable here), and two with 48 pins. Not even sure what the "G4" ending signifies, but either one of the 48-pin options should work.
Thank you a lot! I thought it might've had another implication, like something else would've burnt. Will buy the chip ASAP, it sucks to be in southamerica lol
DeleteNo problem, glad to be able to help :)
DeleteHi,
ReplyDeleteI've just stumbled upon your blog when trying to repair a Saffire 26 I/O.
While +/- 12V and 3V3 rails are okay I have no 5V and two of the three LM2670s doesn't seem to output anything.
The input of the LM2940-5.0 is about 0.6V and nothing gets hot so I don't think there is a short.
I desoldered the 5V regulator and put 5V in the pins where the LM2940 went but the current draw is minimal (about 50mA) and I don't see any change.
The card was given to me as it was not being recognised by the previous owner's computer.
I could order a couple of LM2670s and replace the ones that are not outputting anything but I don't think these fail so easily.
Any ideas on where I should look or what the lm2670s should output in a working unit?
Thanks:)
The 26 I/O is the older, silver unit, right? Did they not have nicely-labeled test points in that generation, yet?
DeleteRegardless, i would imagine it would (or should) be relatively easy to sort out what's powered from what. I briefly had one of these back in the day, but i don't recall what the power brick is supposed to supply the unit with (a single voltage, two voltages?).
Can you trace out / figure out what each chip is supposed to be handling? It sounds odd that three voltage rails are fine, but only two chips seem to be working?
A well-lit (no flash glare) and well-focused photo of the area would be helpful.
So, after a bit of delay I got back to this today.
DeleteWhat I did first what to check if the lm2670-adj buck converters are faulty.
Two of the LM2670s in this board are not outputting anything.
The third one that's outputting the 3v3 rail is fine.
So I swapped the 3.3V one with another and I got 3v3 alright while I'm still getting nothing on the other.
The one I swapped is probably responsible for the +6.8V rail because its output goes straight to a lm2940-5.0 (which is definitely faulty so I took it out of the circuit).
What I also noticed is that the output of -6.8V lm2670 has a dead short to ground.
I lifted pin 1 of the converter and the short remains but nothing is getting hot throughout the whole unit.
What I also noticed is that the 3v3 regulator has its enable pin floating == always on (that pin sits at about 6V) while the other two have their enable pins going somewhere else (there's a via there and I haven't yet flipped the board to see where this is going).
So I guess I need to find what the -6.8V rail is powering and start from there..
Pretty sure that's only powering the opamps; none of the digital stuff needs negative voltages nowadays.
DeleteAlthough, since that's a positive buck regulator that's being used to create that negative voltage, i'm thinking its output is supposed to be at ground, and its "ground" will actually be at -6.8v.
Measure the voltages on pins 4 and 8 of the opamps to see if you're getting the correct power there. Or actually, aren't there test-pads for the positive and negative analog voltages, right next to the LED indicator? I can see them in one of the photos here.
That's correct -6.8V goes to some opamps near the converter cirrus chips.
ReplyDeleteIt doesn't get there directly from the buck converter though there's some kind of LC filter in series.
I also checked the diodes that every lm2670 has right beside it and they all tested good.
No -6.8V or +6.8V rail unfortunately.
I would also suspect the capacitors but there's no signs of bulging, leaking, discoloration or anything.
I'm still reluctant to unscrew everything to be able to remove the board (I need to unscrew everything in order to pull it off).
Only the 3v3 lights on.
There's another LED that probably refer to the -6.8V rail, I don't see any LEDs for the +6.8V or 5V rail.
My guess is that either high ESR on a cap or two or maybe a control signal is preventing the two LM2670s from powering up.
They're probably both good so I'm not going to bother to swap the -6.8V one with the 3V3 rail one again to confirm.
I already did that with the +6.8V lm2670.
The only thing that was indeed faulty so far was the lm2940-5 which was responsible for the 5V rail.
I'm not sure if this regulator managed to burn something really important down the line with higher voltage as most of the digital stuff are powered by 3.3V or 1.5-1.8v.
So to recap, the 20 or so initial DC from the rectified transformer input reaches all lm2670s and also the 7812/7912 pair.
I have 3.3V and I also have the phantom power (48V) as well as 12/-12V.
It's only the 6.8V and -6.8V rails that are missing and this seems to hold the whole interface down.
I only get the power on led light up and the interface seems otherwise dead.
No signal input led or anything.
I obviously haven't tried to connect it to a computer but I guess now it's not the right time anyways.
Even if the firewire IC is bad I would still get the +/- 6.8V.
Is it possible that it needs some kind of user interaction to initiate these rails?
Should I connect it to a computer perhaps?
I don't really think so as I think it also works as a standalone preamp unit..
You still haven't mentioned if your unit is the older silver Saffire 26 I/O, or the newer black Saffire Pro 26. Also, a well-lit (no flash glare) and well-focused hi-res photo of the power supply area would help a lot.
DeleteIt's the older silver model.
ReplyDeleteI'll try to upload some photos later and a proper explanation of what I have done.Thanks a lot for your replies.
Hey, I just plugged the firewire cable upside down in a broken port!
ReplyDeleteNow it doesn't even work as Preamp, still showing it's linked even if it's not plugged.
Do you know how To DEBUG??? Can you please help me to use the DBG port?
Step one would be to replace (or fix) the connector itself, then likely the TSB41AB1.
Deletethe port in th interface is not broken. when I turn it on (without connecting it ), everything seems normal but it show shows the LED LKD on like if it's connected... I've tried to remove the part where there is the DSP ship but seems impossible with the tools I have. I've also played with the DBG port and the RST pins... but nothing changes... if you have any Idea how to reset everything... or how to use the DBG port plz
ReplyDeleteThe "LKD" LED only means that it has locked to its (by default) internal audio clock. But you haven't really described the actual problem you're encountering..?
DeleteReally neat stuff! I've had to get back into this board repair stuff, but I need a quick course on test gear for chips and diodes etc since I started out on vacuum tubes. My Pro 24 1. has started emitting a low level click in the audio output and 2. the sound has momentary dropouts. I've tried the 800 FireWire on different Mac computers so the problem is in the Pro 24. They could be two different issues, one a leaky cap and the second a glitchy FW circuit. I'd appreciate any guidance and Thank you for your posts!
ReplyDeleteNeither is impossible. Is that click audible on any and all audio outputs? Analog and/or digital?
DeleteIf you really wanted to, i guess you could replace the FW chip (or have it replaced); other than that, i'd consider looking at the audio buffer settings for the dropout issue.
Khron, I have a Saffire Pro 24 with one side of the headphone output dead. Upon inspection I see one of the caps on the headphone jack board cracked and kinda blown apart. I need to know if you have an idea of the value of the cap and type. My cap tester shows the good cap to be 100uf. Can you help. Thank you.
ReplyDeleteAre there no markings whatsoever on the still-good capacitor?
DeleteI also have the same problem as you above, power via firewire is fine but no FW light and didn't appear in windows, the LKD light is on. I tried to put music through the first line input and put the monitor knob all the way up and I can hear the music, which means that the Analog side is fine
ReplyDeleteI open the unit and everything looks super clean, no burn IC smell ect
Sounds like the Firewire interface chip might be shot, then.
DeleteSo I did manage to fix it last month but because of me being clumsy, I dropped the measuring probe down to the TDice cpu pins and killed it (the LD4 led stopped blinking and the LKD led never lit back).
DeleteBut the problem was not the FW chip, it was this array cap thingy
https://imgur.com/a/amR10GD
I measured and did have the same problem as you, 3/4 data lines show 525mV drop from ground, but the 4th one shows nothing.
Switching the multimeter to continuity test, I found the 4 data lines pin on the FW interface chip, and when measuring directly from the chip, it was fine, showing all four data lines have 525mV drops like normal.
But oh well, I learned my lesson. The interface was given to me free so I wasn't that sad.
I'm kinda interested in MOTU 828 mk3, found one seller selling that baby for $140, might pick one up in the future so I can replace my current TC konnekt 6 interface lol
rip 24dsp cus the TCD 2210 cpu is gone
DeleteThat's really unfortunate... And those are actually inductors / common-mode chokes, in series with the data lines. But best of luck on the future purchases!
DeleteIt was actually working and detected in computer + saffire control after what I did like in this picture
Deletehttps://imgur.com/a/i0RdK7A
Basically bypass the componence
The only thing is dead the TCD2210, everything else, even the TSB41AB1 chip is fine. So I'll prob use this inteface as spare parts, I guess.
That makes sense - those ferrite beads are virtually wires (they just "look like" higher resistances at higher frequencies).
DeleteHaven't measure anything yet but the VA, 3.3v LED lights up correctly while LD4 blinks (is that normal?)
ReplyDeleteQuite sure LD4 blinking is normal, i think i recall seeing that in Saffire 24's and 40's.
Delete