#6 Canon EOS 400D / Rebel XTi troubleshooting & repair

I got this old EOS 400D from a friend for about 35 bucks. She said it won't power on, even with a fully-charged battery. Might as well have a crack at it, eh?

On the outside it seems to be in pretty good shape. A bit of dust here and there, but hey, it's about a decade old, whaddya expect? Battery checks out fine, reading 8.15v, so that's definitely not it.

As chance would have it, between being offered the camera and receiving it, i did some digging around online, and i came across this little story here (scroll down to the second half of it). Could i (also) be fortunate enough to only need to swap out some fuses?

Power issues asside, i'm gonna (need to) have some real "fun" with this. Upon opening the card compartment to see if there was a CF card included (an 8GB SanDisk Ultra 50MB/s - nice!), and then removing it, a chilling view greeted me.

Oh joy - those pins sure are gonna be "fun" to straighten out again...

<sigh> Wouldn't ya know it? The furthest-bent pin was, "of course", damn near broken off. When i tried to grab it with a pair of tweezers, it felt like it rotated suspiciously easily, and sure enough, next thing i know it's well away from the connector. On the bright side though, as far as i can tell from the pinout, that was "only" pin 50, which goes to ground, so... I miiiiight just be able to get away with it (or rather, without it). 

At least the other three are still firmly-enough attached. After (reeeeeeally gingerly) inserting the card, none were stuck in it upon removal, so... I'd dare call that a moderate success so far - mechanically, at least. Not that it'll be much use if i can't revive the whole thing though.


Gotta hand it to them Japanese though - they sure do put together some really proper, thorough and well-done documentation. No actual schematics unfortunately, although the power distribution block-diagram (page 175) as well as the PCB layouts kiiinda-sorta help a bit.

Now, as far as i can tell, the power circuitry on the "mecha-block" or motor-drive (MD) PCB is all working. By near-blind probing on some of the tantalum caps visible on the "top" of the mainboard (ie. the side accessible once you remove the rear casing, with the LCD), i managed to determine that at the very least, the 4.8v and subsequent 3v linear regulators DO seem to work just fine. I'm not 100% sure if the big Canon chip around which i measured is the MPU or the LPU, but either way...

On the other hand though, despite all the fuses being intact, the DC/DC converters on the mainboard don't seem to be doing anything. During some digging around online, i came across this particular page. If you scroll down enough (or do a search in page for "XTi"), you'll get to this section:

Subject:  Canon TIi / 400D  … No Power

On Mon, Jan 30, 2012 at 4:35 PM

I have a XTi inhere that is dead.  The P fuse on the dc board is good.  It seems to have a battery drain so signs of moisture.  Any pearls?
Bart

Monday, January 30, 2012 2:28 PM

There is an "K" fuse on the Main PC Board under the right-hand metal shield. Check it. Also there is a "JH" transistor that burns out under that shield. There are pictures here...

http://camerarepairs.com/images/Position%20on%20Board.jpg   You can also see the "K" fuse on this one.

Here's a close-up...

http://camerarepairs.com/images/Burned%20Transistor.jpg

 On my camera, with the light at the right angle, i thought i saw a little bump on that particular transistor. Not exactly a crater as in the second photo, but not a perfectly flat top, as the other SOT323 components around it. After some further scouring, i managed to determine that, on that type and size of package, the JH marking corresponds to the Sanyo MCH3308 N-channel MOSFET. Thing is, it seems to measure ok - not shorted, not open-circuit, and the source-to-drain diode checks out as well.

That, in turn, is beginning to make me suspicious about the IC900, on "side B" of the mainboard (service manual, p.178). It would appear to be a QFN-packaged chip, with loads of resistors around it, and it can't be a coincidence that it's right on the other side of the board, from all the transistors, diodes, inductors and capacitors that make up the five main DC/DC converters. But to gain access to that, i'll have to pretty much take the whole camera apart, ie. remove the front and top covers respectively, to get to disconnect all the connectors on the mainboard, as well as the screws holding it to the chassis.

The PWM controller turned out to be a Fujitsu MB39A108 5-channel job. For something between "shits and giggles", thoroughness and curiosity, i googled for the name of the chip as well as Canon or EOS. The only relevant things i found were a few threads on some Russian and Ukrainian forums, about this chip being used in several PowerShot models. Apparently, at least in some cases, following sharp impacts, allegedly legs / contacts of said chip had cracked or broken off entirely. In light of this (new) information, i figured i'd reflow the bastard - what would i have to lose, y'know?

"Just in case", i decided to clean the contacts of the power switch as well, and bent the contacts a bit further, so they exert a bit more pressure on the gold-electroplated copper pads. I can't be 100% sure whether that actually did anything, but as long as i had "enough" of the camera disassembled, i figured i might as well. Not in the least so i wouldn't need to tear the whole thing apart yet again, to try this too.

Done and dusted - i fluxed all four sides of the chip, heated the area until all the passives around the chip that i dared poke, were "floating" on solder, even if the chip itself wasn't. I let the board cool off, stuck it back into the chassis, plugged in all of the flexes into their respective connectors, and installed a couple of the mounting screws, to have some positive ground connections. The card compartment door switch is still jumpered shorted; in addition to this, i also manually held down the battery door switch, flicked the power to "On" and... Whaddya know, we have a power LED!!!

After turning it off, i plugged in the LCD flex as well, flicked the power switch again, and i got a prompt to input the time and date. We have a live one!..

... But it would seem i'm not quite out of the woods just yet. Autofocus doesn't seem to want to work with either of the two lenses i have (an old Canon 18-55mm that came, i suspect, in kit with the camera, and a Sigma 18-125mm HSM OS). The 18-55mm works fine with my other camera (EOS 550D), so that's not it. That being said, if i switch the lens to manual focus, keep the shutter button half-pressed, the camera DOES beep when focus is achieved, so... As far as i can figure, the AF sensor is ok, but communication with the lens isn't happening (properly).

Also, the display in the viewfinder (ie. the digits and graphics along the bottom edge) only blinks on for a fraction of a second, every second or so, accompanied by a "tick" sound from somewhere inside the camera. The oscillator driving the red-eye-reduction lamp (between the lens and the grip) is pretty noisy as well hissing & whining when it's on.

Right, took (almost) the whole thing apart again, but this time i'm focusing (no pun intended) on the "motor driver" board, ie. the one under the grip area (service manual, p.179). On the lower-right side of the board (top view, left side of page), there are the three pads named G, B and R, with wires going to the first three pins of the EF lens mount. G is the "lens installed" signal, normally pulled up to 2.8v or so, and goes to the second pin (which, on the lens, is the wider contact that grounds it). B goes to pin 3 of the EF mount, and is ground, while R is the Vbat that goes to pin 1. The latter powers the focus motor in the lens.

Now, between what i can see on the board (traces etc), and the component designations and whatnot, what should be Vbat coming straight from the 4.75v buck converter, is switched through the SOT23-6 packaged MOSFET (i suspect), with the "ZDMIF" diode underneith it being either a Zener clamp, and/or a reverse-polarity protection. When probing the red wire without a lens installed, i couldn't measure anything, but with a lens installed, the voltage is all over the place, bouncing anywhere between 2.4v and close to 0v.

Now that i look a bit more carefully, it looks / feels more like a sawtooth wave, with a period of a second and a bit, or so. The signal's the same before the above-mentioned MOSFET switch, so it's definitely coming from somewhere upstream. The power diagram from p.175 shows CVBAT coming from the 4.75v buck converter on the mainboard, through a switch. The same voltage rail supplies the shutter (which, as far as i can tell, works fine), the red-eye reduction lamp (which is oddly noisy), as well as something called "FLED" and "SI-LED". I wonder if one of these last two has anything to do with the viewfinder display. The rate of this voltage going up and down seems to fit with the ticking noise i mentioned a bit earlier. That buck converter's not doing too great, it would appear.

When poking around the motor-driver board, i also measured the reverse-diode-drop in various points (ie. multimeter on diode test, red probe on ground, black probe in various points). I noticed a suspiciously low 65mv reading on the CVBAT line, before the MOSFET that passes that along to the EF mount, but a more normal 750mv after said transistor. The 65mv reading went away upon disconnecting the 3-wire cable going to the mainboard. Then again, that might be plausible due to there being a Sanyo SBS010M Schottky diode employed in that buck converter, which does indeed have uber low forward-voltage. The switching element is a Sanyo CPH3338 P-channel MOSFET.

But still, something's grossly overloading that voltage rail, that's for damn sure. The SI-LED, i believe, comes through the "SIL" connector on the mainboard. Disconnecting that didn't change anything. Neither did disconnecting any of the others, one by one. Then again, the ones i couldn't do without was the "TSW" one, going to the power switch, and the "DRSW" one, which contains the battery door and CF slot door switches. Without either of these, due to the interlocks and whatnot, the camera doesn't power up at all.

I guess the next step is to remove the above-mentioned MOSFET and hooking up that particular rail to my bench supply, and see how much current the camera would want to draw...

Actually, i never got to that point - quite fortunately, in fact. The PWM chip only has drivers for P-channel MOSFETS on two channels of the five (ch2 and ch3). When looking a bit more closely at the photo of the PWM above, i noticed a / the potential issue. Let me just crop out the suspect area. See anything out of the ordinary?

As the old saying goes - the road to hell is paved with good intentions. After reflowing the PWM, i tried to "make sure" and add some leaded solder onto the IC contacts. Well, in the process, it would seem i managed to bridge together two capacitors, namely the ones connected to the FB3 (ch3 feedback) and CS3 (soft-start) signals. Once that was cleared, the camera fired up 100% normally - no more ticking, no more flashing viewfinder overlay! SCOOOOORE!!!...

Well, there ya go - a reasonable little 10Mpx backup camera, for merely a few smackers, plus "a few" hours of elbow-grease, as it were. Me likey...