Wednesday, July 03, 2013

Repairing an Alesis IO26

There's a really common way to kill your firewire-devices, which is to forcefully plug in the 6-pin connector, carrying both supply power and data signals, rotated by 180 degrees (which normally should be prevented by the asymmetrical shape of the connector). That way you'll send +12V, the most common supply voltage, into the chip that connects to the cable. This kills the chip pretty reliably.

This audio-interface (Alesis IO26) was damaged using this method:

From Chris’ Miscellanea

Thankfully the internal layout of the device is pretty clean. There's a stack of two PCBs (only the lower one shown) on the left side inside. The lower one has the DSP, Firewire controller and a ARM based controller. It connects with flat cables to the "analog" boards. On top, it carries a switching power supply generating the internal voltages (+3.3V, +5V, +/-15V and +48V from a single, galvanically isolated 9..30V input) which is missing in the photo.

From Chris’ Miscellanea

A good way to diagnose these kind of errors (according to some forums and FAQs) is to look at the idle-voltage on the firewire-bus lines (TPA0+...TPA0-...). They are supposed to be biased to a internally generated 1/2 Vcc (~1.8V) via 56 Ohm resistors from the twisted-pair lines to a bias pin (TBIAS0/1). It's described in the data sheet of the chip, here it's a TSB41AB2 from TexasInstruments In my case one bias-voltage-regulator spat out 3.3V, the other 0V (there's one for each of the two firewire ports). One of the ports had one receiver line stuck at 0V (chip-internal short to GND).

The only way to repair it is to desolder the chip, and replace it with a new one. Then the interface should be functioning again (I did not populate the mechanically damaged connector and associated terminating resistors, one port is enough for me).

From Chris’ Miscellanea

From Chris’ Miscellanea

The two shorted pins #1 and #2 are both GND, so I did not bother to clean up the blob.

8 comments:

El Gitano said...

Hey Chris, excellent work and great post!

The exact same thing happened to me a few days ago with my good old trusted Multimix 8 Firewire (http://www.alesis.com/multimix8firewire), don't know how i managed it, but somehow i plugged in the firewire cable 180° while the device was switched on…stupid me.

Since then, no reply from the firewire part of the device, only in Linux i got some sort of reply: when i restarted the device, dmesg told me that the firewire bus has resetted. Nothing more, no recognition or anything else.

I did some research on that matter, Alesis support told me that these mixers are not supported anymore, nor are there any replacement parts left (except from this site you can get a replacement board which costs more than i actually paid for the whole mixer.. *yuck*: http://www.instrumentalparts.com/9-79-0386.html)

So, further research led me to Your blog post here and the part which most likely must have been damaged during the action, it appears to be same PHY (TSB41AB2PAPG4) as on Your device.

http://i.ebayimg.com/t/Alesis-Multi-MixFireWire-Main-Assembly-MM-FireWire-8-/00/s/MTIwMFgxNjAw/$(KGrHqN,!gsF!pBL2GMKBQQRHU0C(Q~~60_57.JPG

Right now i am considering to give this chip a shot, i am not completely new to soldering in general, but have zero experience in the SMD field… well, lotsa old PC stuff flying around to practice on it. I surely don't want to give up this nice device, and if the Firewire PCB might get damaged i should still have a working analog mixer.

My questions to You: Did You order the chip as samples directly from the TI site (http://www.ti.com/product/tsb41ab2) and how long did it take to ship? Was the soldering process explicetly painful?

Thanks You very much!
Cheers,
Roman

Christian Vogel said...


Hi Roman,

that chip is available from a lot of vendors, I ordered it from RS components (via eibtron.de, which handles the non-company orders for RS components in Germany where I'm located). Biggest problem is the minimum order fee, the chip itself is 5€ in single quantities (maybe 7$ or so)...

With a little practice (and I had helpful advice from my colleagues who rework SMD more often than I do) it's possible to unsolder the TQFP64 using hot air, clean the pads using fresh solder, flux and copper braid, then thoroughly clean the remaining flux with alcohol and a toothbrush (which I neglegted to do, as you see on the ugly photo of the new chip), and resolder the replacement part.

It's probably a *very* good idea to first practice with stuff on an old broken PC motherboard :-). The old pentium cache-SRAMs have similar size and pin spacing.

El Gitano said...

Hi Chris,

Thanks for the reply! That's a good tip with ordering via eibtron.de, i suspect shipping directly from TI might take quite a while.

Yeah... have have looked at some tuts on the net regarding SMD soldering, there seem to be quite a few approaches.
I have no hot heat gun or station at hand for desoldering the chip, so my best bet would be heating the chip from the top with the iron (chip is damaged anyways, no loss) to avoid damaging the PCB.
Option B would be cutting away the chip legs with a sharp knife carefully and then cleaning the pads.

How did You solder the new chip, pin by pin or with a single stroke and then cleaning remaining excess bridges with copper braid?

Well... time to get out and buy a new soldering tip (1mm) + flux and tin solder (0,5mm?). Always wanted to rape my old hardware waste :D

Will let You know if i actually manage to repair that device, thank You very much for the tips!

Cheers,
Roman

Christian Vogel said...

Hi Roman,

hot air is probably the best way, heating up the chip uniformly with the soldering iron is not feasible, in my opinion. Your best bet would be to cut the legs with a sharp knife, many people have reported success with this method.

Be careful not to rip off pads when putting too much force onto single pins!

I soldered the chip pin-by-pin and did no have issues with bridged pads that way, except with two pins where I was hurrying a bit. That I could fix by stroking the bridged pair with a tip from chip to the outside. Use a thin and clean tip that's only slightly "wet" with fresh solder. It will flow onto the pad and pin.

I never was successful with the "use excess solder and soak it up with copper braid" method, always turned into a mess.

Have fun, and luck, with your repair!

Chris

Jim Cottrell said...

Thank you! Thank you! Thank you!
Your post has just saved me $250.

El Gitano said...

Hey Chris!

It is been quite while... finally, i decided to give this a shot.
Spending quite some time ripping apart an old soundblaster for practice, i finally gave this a go yesterday. I gave this a 50:50 chance to work...and what can i say, turned out that have been on the lucky 50% side :D

It's been quite a pain in the ass, i did choose the cut-away-legs-and clean-mess-up method only to discover that this little bastard had been soldered to the board via power pad below as well... well, a lot of heat with the iron from the top and i was able to remove the mess:

http://www.imagebanana.com/view/zd31wj2q/P1010108.JPG

As You can see i also salvaged one of the pads (probably while cleaning them...), Number 6 from the left bottom. After studying the datasheet i found that this pin (XI) could be bridged the 24,576MhZ oscillator on the left. After studying the board with the multimeter, i found out that the correct port had to be on the top side of that oscillator. Here is how it looked, connected with a thin copper wire:

http://www.imagebanana.com/view/4q2fedry/P1010124.JPG

Well, checked everything for bridges, cleaned up the remaining flux mess, gave it a 50:50 chance, hooked everything back together....and voila:

http://www.imagebanana.com/view/zjna5xj3/P1010129.JPG

My beloved Multimix Firewire is up and running again as a soundcard :)

Thank You SO MUCH for the help and Your great blog artice, man!

Cheers!
Roman

Christian Vogel said...

Hi El Gitano,

it makes me happy to hear that you got your device working again! And a nice fix with the killed SMD-pad to the Xtal, your repair is absolutely adequate. Great work for a first try in SMD rework.

Greetings und Frohes Fest,

Chris

El Gitano said...

Thx man!

Yeah, couldn't believe it myself... freaked me out when the mixer turned up in audiosettings again :)

Btw, solder paste applied to the legs with a needle did the trick for me soldering the new chip. Together with lots of flux it was just enough to get the legs connected with a quick stroke from the tip on each pin, didn't have much problems with too much solder and bridged pins (maybe 2-3 which i had to clean up).

Yeah, have been lucky with just one pad killed, initially it has not been totally ripped of the board, just was slightly loose. But after trying to bend it back into position with a needle... i ripped it off. Luckily it was on that side where one could connect it it to another soldered point on the board... if it had been on the southside of the circuit, where all the lines go directly into the main board processor via pcb lines, i would have been out of luck, as there are no points to solder bridges.

Oh, and btw, i didn't care to solder the power pad on the chip bottom again, as it seems to be just for thermal purposes. Applied some solder paste there (probably thermal paste would have been better?) so i could have given it some heat from the pcb's backside through the via holes.. maybe i'll do that later. Didn't want to risk any damages there. At least there something between the chip on the power pad there, which might cool it a bit.

Once again, thank You very muchand happy x-mas to You as well! :)

Cheers,
Roman