24 I/O works after "cold" start, but then sample rate starts flashing endlessly

[motucitizen]

I have a 24 I/O rack unit hooked to a PCIX-424 card.

When I first turn the unit on, the sample rate indicators do their normal "scan" before eventually settling on a single rate, which remains solidly lit (for example, "44.1", or whatever was last configured). I can then also configure the unit to switch to any of the four available rates, as normal.

However, after the unit has been on a little while (just minutes) the sample rate indicator will suddenly start flashing by itself and then never stops, no matter what I do. (The unit is set to use the internal clock of the PCI-424, so this is not due to a missing clock signal.)

Once in this "endless flashing" state, the unit is still detected by the computer (PCI Audio Setup can see it, with no error messages) and the unit still responds to sample rate configuration changes (in a way) but the sample rate indicator never stops flashing. Curiously, attempting to set the rate to either 44.1 or 48 always makes the unit switch to flashing only 44.1. Likewise, attempting to set the rate to 88.2 or 96 always makes the unit switch to flashing only 88.2.

Now, I have three other 24 I/O rack units which are working fine with the same PCI-424 card. I have tried every unit on all four AudioWire ports on the card, as well as swapping all the cables around between them. The issue does not appear to be the card or the cables... the three working units always behave normally (no matter the port or cable) while the one "endless flashing" unit always exhibits the problematic behavior described.

The flashing behavior, since it occurs after the rack unit has been on for a while, seems possibly to be heat related? Once the unit starts flashing, power cycling it does not change anything (already too warm?) -- the indicator starts out flashing and never stops (never goes solid). The only thing that seems to restore normal behavior (briefly) is to leave the unit powered off for many hours (cooled off?), at which point, the sample rate will light solid as normal, albeit only for a short time (until too warm again?).

I have opened up the unit and tried changing the large 15000uf capacitor on the main board, as it seemed possibly domed (though not clearly) as someone else had posted that a similar behavior with their 24 I/O was magically fixed by replacing this capacitor. But, this made no difference in my case.

Anyway, has anybody experienced this behavior with the 24 I/O, and has any advice?

[motucitizen]

Some things I have tried/noticed:

- One of my other (working) units has firmware 1.1, while the "flashy" unit has 1.0. I moved the 1.1 chip into the flashy unit, but this did not help (was a long shot).

- I measured voltage output of both regulators (that I could find) in the flashy unit and they seemed fine…. 3.3v and 5v (presumably feeding different parts of the digital circuits) and identical to those measured in a working unit.

- While the two units were open, I noticed that the flashy unit has a "REV 1" main board, while the other has "REV 2". Interestingly, the REV 2 unit has additional heat mitigation measures (note the beefy heat sink):



Seems that MOTU maybe realized that heat was an issue and revised later hardware? Fwiw, the smaller voltage regulator (see photos) is actually the component that seems to be the warmest on the board (on my flashy REV 1), even though the heat sink is near the larger one (on the newer REV 2 board).

If my voltage regulators are fine, perhaps it is one of the other nearby components (for example, the smaller surface-mounted capacitors) that has been compromised by heat exposure?

[motucitizen]

Interesting too is that the REV 2 main board has a large 12000uf 16V capacitor, whereas the REV 1 has 15000uf 10V. Another design change.

[mikehalloran]

These were discontinued eleven years ago. That you have both a gen 1 and 2 should help a tech isolate the failing component(s).

I remember buying cold spray from Radio Shack to diagnose such problems by freezing them until the problem was isolated.

[motucitizen]

Thanks Mike. That's a great idea to use a chilling spray.

[motucitizen]

Update:

I haven't done any chilling spray diagnostics (as Mike suggested), but I did try slapping a heat sink on the larger voltage regular (since that is the one "with" a heat sink in the REV 2 board):



...and, lo and behold, the flashy unit maintains clock sync for much, much longer... indefinitely as far as I can see. It lasted for 3 hours or so, before I finally turned it off before bed (just didn't feel like leaving a potentially overheating device on unattended).

I will still perform some chilling spray diagnostics, since the heat sink that I added could be affecting multiple components at once, or affecting the "true" problem component merely as a by-product of keeping the regulator cooler, while the spray could potentially narrow things down in a more targeted way.

The heat sink is a just a cheap aluminum block of fins (20x20x16mm), attached with thermal tape, purchased from Amazon (20 for $14, here: https://www.amazon.com/dp/B07ZC65NJL):



If my unit stays working indefinitely because of the heat sink (I haven't actually put everything back into the enclosure; its just boards spread out on the bench) I might go that route if/until the unit stops working.

The only other thing that concerns/bothers me is the buzzy transformers (which is an issue with all my units).

[mikehalloran]

There are a ton of posts about voltage regulators and noisy transformers on old MOTU units here and o Gearspace.