#57 Apogee Ensemble Firewire pt.2 - the repair

I was considering making the previous post about the repair as well, but once the teardown part of it was done, i decided it had gotten long enough already, so here's the testing and (hopefully) repair of it, separately.

As pointed out in the teardown, i bought this as faulty, with inputs 1/2 and outputs 1/2 allegedly dead. Seeing as the ins & outs are handled by the four CS4272's, i applied the so-called "Occam's razor" principle, which entails that the simplest explanation is, more often than not, the correct (or true) one. I figured the odds of the first pair of inputs and outputs being handled by one of the four codec chips (and it being dead) far outweighed the chances of the analog input pair of one chip being faulty AND the analog output of another one, at the same time. That, and the low likelihood of the channel order being jumbled up quite so counter-intuitively, by design.

But that's enough yapping on and supposition and whatnot - onwards to hands-on testing!

First impressions - the front panel LEDs are effin' BRIGHT. As in, you-wouldn't-want-to-look-at-them-straight-on, bright - not for long, anyway. Once it's done with the boot-up LED "animation" and things settle down, the input channel 2 level meter keeps sporadically flashing, as if some noise is passing the minimum-level-for-indication now and then. Keeping in mind the level meters are driven from the digital sides, ie. from whatever the signals are post-analog-to-digital conversion... If the  codec that handles inputs 1/2 is faulty, no wonder it may well be putting out garbage.

That being said, most of the power supplies have been identified as well. In my description in the teardown, i had forgotten that the PGA2500 preamp chips need their own +/-5V supplies. They were sneaky, though - i think they're deriving the -5V through some sort of "analog" voltage-inverter type deal (think along the lines of voltage multipliers).

Probably not the greatest annotations, but there's only so much i'm able (and/or willing) to do in Paint, so there. During the rest of the poking around, i've also identified the mystery circuit group over near the edge of the board - a pair of discrete voltage regulators, for the great majority of the opamps (the group of eight that are "south" of the preamps in the overall view, plus the headphone amps), but not quite all of them (the send/returns for inputs 1/2, and the four handling the line outputs).

Haaaaaaaaang on a second - i think i just found one potential culprit. A tiny little resistor that looks a bit darker than its corresponding pair. The two 10 ohm resistors are in series with the +/-14.6V rails going to the opamp (U6), to provide some extra filtering in conjunction with the capacitors "downstream" of them, and the one on the negative rail is a bit toasted - getting a mere 0.12V "after" it, with -14.6V at the other end of it. That'll surely mess up things, wouldn't you say?

Now i'm wondering if that's all that's wrong... Nah, that'd be way too easy. Nothing's ever that easy. Well, except the cases where you buy something as faulty and it turns out to be fine, but that's a "fringe case"...

That RC filter seems to supply all three quad opamps in the area of the inputs 1/2. Supplies look fine on all the output opamps, so the codec being dud can't be ruled out entirely just yet. Either way, that needs to be replaced - the pair, in fact, just for good measure.

Some more probing, first. Sadly, looks like there's something else shorting the negative rail to ground, downstream of the toasted resistor. Best case, it's one of those [insert bad word here] little yellow tantalum caps, or even one of the ceramics in parallel with them; worst case, it's one (or more) of the opamps. Easiest thing first though: might as well start with the tantalums. And easiest thing first on the easiest thing first: instead of removing the caps altogether, lifting one end is enough to remove them from the circuit for further probing, and the folded metal-strip terminals of these tantalum caps allow enough flex to raise one end clear of the board with no lasting trauma.

Diode-test mode on the multimeter, black probe on the chassis (ground / earth), red one for probing. First off, the "output" pad of the removed-toasted resistor showed a 1.2V drop - a massive improvement compared to the 0.022V previously indicated (ie. a short between the negative rail and ground). That's a huge step forward right there, and a glaring sign that my "paranoia" was justified and correct, at least so far, and at least in this case. Move probe to the lifted end of the tantalum right next to that, open-circuit. Next, the tantalum near the channel 1 opamp, open-circuit as well. Last but not least, the one near the channel 2 opamp - 0.022V.

BINGO!!! Well, that settles it, we're halfway home on this one. And dodged a major bullet, too - i've got TL074's, but not in this wide-body SOIC package, so i would've had to order some, and wait for delivery etc. Lots of extra hassle, so i'm real glad that's not necessary.

These tantalums are marked "106D", which i'd guess is "106" for the capacitance value (ie. "10" and six zeroes, measured in picofarads, so 10000000pF / 10000nF /10uF), and "D" might be a code for the voltage rating. Standard values in that area are 16V (a bit too close for comfort, for a 14.6V rail) and 25V. Not sure i have any exact replacements, both as value and footprint, but i might have some suitable ceramics. And yes, i'm replacing all six of them, because reasons.

There we go, perfect - looks like i had scavenged a handful of 1206-size 10uF 25V SMD ceramic caps some time ago. Similar length with these tantalums, roughly half the width. Yes, they probably have higher ESR than the tantalums, but at audio frequencies (as opposed to MHz-range), that's not really an issue.

Post-cap replacement, but before the new 10 ohm resistors, a 1.2V drop between negative rail and ground, same as between ground and the positive rail - i'm happy with that, and it's a promising step forward. After installing the resistors as well, a quick power-up test and... Whaddya know, channel 2 level indicator quit flickering, surprise-surprise...

For testing the outputs though, i'll have to get my test-machine up and running first...